Journal of Membrane Science 323 (2008) 28–36

Contents lists available at ScienceDirect

Journal of Membrane Science

journa l homepage: www.e lsev ier .com/ locate /memsci

Copper recovery by polymer enhanced ultrafiltration (PEUF)and electrochemical regeneration

Javier Llanos ∗, Angel Perez, Pablo Canizares

Chemical Engineering Department, University of Castilla-La Mancha, Faculty of Chemical Sciences,

al ofPEUFnece

inuoun thepH thtestsstepn onsureotenvoltaH is l

Avda. Camilo Jose Cela 12, 13005 Ciudad Real, Spain

a r t i c l e i n f o

Article history:Received 19 December 2007Received in revised form 9 April 2008Accepted 28 May 2008Available online 12 June 2008

Keywords:Copper recoveryPartially ethoxylated polyethyleniminePolymer enhanced ultrafiltrationElectrochemical regenerationCyclic voltammetry

a b s t r a c t

In this work, the removenhanced ultrafiltration (polymer. Overall, the tworecirculation and discontconfronted individually. Ometal–polymer ratio andobtained by ultrafiltrationthe polymer regenerationthe metal electrodepositiohave been carried out to astests, a cathodic working pAg/AgCl). Working at thisdifferent pH values. This p(pH 2).

1. Introduction

Metal ions recovery is a key challenge from both, environmentaland economical points of view. Copper is one of the most impor-tant metal ions for the global economy as it is a key componentin building and electrical industries, with a market share of 48and 17%, respectively of its world consumption. Moreover, its pricehas increased by almost a factor of three within the last 3 years,according to London metal exchange data.

Several methods have been applied to face the treatment ofdiluted aqueous effluents with heavy metals. Amongst them, poly-mer enhanced ultrafiltration (PEUF) technique has been extensivelystudied and used in the separation and concentration of metalloaded water effluents with different aims, either analytical ortechnological. On the one hand, this technique has been appliedas a method for the concentration of heavy metals as a previ-ous step for the application of analytical techniques [1,2]. On theother hand, this separation method can be used in the treatment ofwater effluents with heavy metals [3], either sewage water, under-

∗ Corresponding author. Tel.: +34 926 29 53 00x3511; fax: +34 926 29 52 56.E-mail address: javier.llanos@uclm.es (J. Llanos).

0376-7388/$ – see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.memsci.2008.05.073

Cu2+ from a synthetic effluent has been tested by means of polymer), using partially ethoxylated polyethylenimine (PEPEI) as water-solublessary steps of a hypothetical continuous process, metal retention (in totals mode) and polymer regeneration (in discontinuous mode), have beenone hand, the values of temperature (T), transmembrane pressure (�P),at maximize both, permeate fluxes and rejection coefficients, have been, reaching Cu2+ retention coefficients higher than 97%. On the other hand,has been carried out by the electrochemical technique, which consists inthe cathode of an electrochemical cell. In a first step, cyclic voltammetriesthe polymer does not suffer any oxidation or reduction process. From thesetial has been selected to minimize hydrogen evolution reaction (−0.7 V vs.ge in deposition tests, a pH of 3.3 has been selected from experiments atess extreme than the pH necessary if this step was carried out chemically

© 2008 Elsevier B.V. All rights reserved.

ground water or with radionucleides [4]. Recently, several workshave studied the optimization of design parameters for both, themetal retention and the polymer regeneration steps [5,6]. More-

over, the heavy metal retention by means of polymer enhancedultrafiltration has been successfully modelled using equilibriummodels [7], as well as carried out and modelled in semi-continuousmode [8] and simulated in continuous mode [9].

A wide variety of water-soluble polymers have been utilizedin PEUF processes for the recovery of heavy metals. Amongstthem, polyethylenimine has been one of the most extensively used[10–13]. This polymer has been applied in such a separation pro-cess mainly because of its high water solubility, its high capacityto bind metal ions and its physical and chemical stability. One ofthe main disadvantages of working with this polymer in PEUF pro-cesses is that it is necessary to reach very low pH values to get thecomplex breakage, due to the high stability of the macromolecu-lar complex formed. This fact can entail complexity and high costof the regeneration stage. In order to minimize this drawback andimprove polymer selectivity, in the present study partially ethoxy-lated polyethylenimine (PEPEI) has been used as water-solublepolymer. In this polymer, 80% of primary amines (the most activefunctional groups) have been substituted by hydroxyl groups. Thispolymer has not been deeply studied as water-soluble polymer inPEUF processes.

mbran

J. Llanos et al. / Journal of Me

In respect of economy and technical viability of the process,polymer regeneration stage is a key step, as it represents one of itsmain costs [13,14]. In PEUF processes, this stage has been carriedout by two main different techniques. On the one hand, a chemicalprocess involving a decrease on pH and a subsequent ultrafiltrationstage can be used to face this aim. In this stage, permeate streamcontains the free target metal ion meanwhile retentate streamretains the water-soluble polymer that can be recycled [5,8,9,13,15].Polymer regeneration by means of a chemical procedure has someadvantages such as operative simplicity or low energetic cost. Onthe contrary, the metallic content of the final permeate stream, lowrecovery percentage of the treated solution and high water con-sumption are the main disadvantages of this kind of regenerationtechnique [8,15].

On the other hand, electrochemical regeneration seems to be aclear alternative to traditional chemical regeneration [16–18]. Inthis method, a pH decrease is produced initially to weaken themacromolecular complex. Next, the target metal ion electrode-position is carried out in an electrochemical cell. This alternativeoffers three clear advantages: (1) the metal ion is recovered inits most valuable form, that is, as metal; (2) does not exist anyfinal effluent with metal ions content, achieving high water recov-ery percentages; (3) the main reagent, the electron, is a “cleanreagent”, making this technique fully compatible with environmen-tal regulations [19]. Although this method offers clear advantages

from the environmental and operational points of view, very fewworks have been focused on coupling electrochemistry and reactivemembrane separation techniques. Amongst them, some attemptshave been done with micellar enhanced ultrafiltration [20,21] orpolymer enhanced ultrafiltration [22,23]. One additional electro-chemical regeneration technique consists in using electrodialysiswith bipolar membranes [24]. The main disadvantage of this lastmethod is the presence of a final mud with the target metal ionhydroxide.

The main aim of the present work is the study of the copperrecovery from a synthetic effluent by means of a PEUF process, usingPEPEI as water-soluble polymer. In the present work, a rising watereffluent from an electrolytic copper plating process has been sim-ulated, consisting of acidic baths with CuSO4 and Na2SO4. In thiscontext, the two key steps of the whole process, metal retentionand polymer regeneration, have been studied individually. First, theeffect of the most important operational variables (temperature,ionic strength, transmembrane pressure, pH and loading ratio) ondesign parameters (permeate flux and rejection coefficients) hasbeen studied and optimized. Furthermore, the main phenomenathat affect permeate flux (fouling and concentration polarization)

Fig. 1. Electrochemical regenerat

e Science 323 (2008) 28–36 29

have been studied in discontinuous mode. Secondly, in respect ofpolymer regeneration step, the viability of an electrochemical tech-nique has been researched. This study comprises two stages. First ofall, several cyclic voltammetries have been carried out with differ-ent solutions. With these tests it can be checked if PEPEI sufferseither oxidation or reduction processes and the optimal opera-tion voltage can be selected. This first stage is a clear advance incomparison with previous studies dealing with electrochemicalregeneration of macromolecular ligands in PEUF processes. Finally,the influence of pH on both, deposition rate and current efficiency,has been studied with constant voltage experiments.

2. Experimental

2.1. Materials and apparatus

Ultrafiltration experiments, both in total recirculation and indiscontinuous mode, were carried out in a laboratory scale ultrafil-tration installation which details are gathered in previous articles[13,25]. A MicroCarbosep 20 UF module with an inner ceramicmembrane (MWCO = 10 kDa, A = 0.004 m2, i.d. = 6 mm, ZrO2–TiO2active layer) was used. A stainless steal rod (outer diameter = 5 mm)was placed inside the membrane with the aim of improving thesystem hydrodynamic behaviour.

The electrochemical cell is schematised in Fig. 1. It consists of adiscontinuous stirred reactor, with a thermostatic jacket, in whichthe working electrode (cathode) was a ceramic ultrafiltration mem-brane, discarded after its use in ultrafiltration experiments due toits low permeate flux. The electrodeposition is carried out upon theouter surface of its graphite support, which acts as cathode. The areain contact with the solution is 31.42 cm2. An inert platinum elec-trode was used as counter electrode. The potential was measuredand controlled in respect of the working electrode using a Ag/AgClreference electrode. In order to control and visualize both, oper-ating voltage and intensity, a potentiostat/galvanostat VOLTALABPGP-201 was used.

The polymer used was 80% ethoxylated polyethylenimine (Mw

50,000) in aqueous solution (37%, w/w) supplied by Aldrich. Its con-centration was measured by a total organic carbon (TOC) analyzerShimadzu 5050A. The metallic salt used was copper sulphate 5-hydrate of analytical grade from Panreac. Ionic strength was fixedby adding sodium sulphate 10-hydrate of analytical grade fromPanreac. Copper concentration was measured by atomic absorp-tion spectrophotometry (Varian, SpectrAA 220). All solutions wereprepared using ultrapure water.

ion schematic installation.

30 J. Llanos et al. / Journal of Membran

used in the present work [13]. Here, we suggest a similar foulingmechanism as for PEI. At very acidic pH (pH 2) a cake layer foul-

Fig. 2. Effect of transmembrane pressure and pH on permeate fluxes and polymerrejection coefficients. T = 50 ◦C, [PEPEI] = 0.06% (w/w), v = 3.2 m s−1.

2.2. Procedure

Polymer concentration in all ultrafiltration experiments was0.06% (w/w). A higher polymer concentration (0.1%, w/w) wasused in polymer regeneration experiments in order to simulate theconcentrate stream from a retention stage. Feed flow rate in ultrafil-tration experiments was 100 l h−1 (tangential velocity of 3.2 m s−1).The initial solution of macromolecular ligand is circulated throughthe system without pressure until working temperature is reachedand keeps constant. Working pH was adjusted by adding the nec-essary amount of sulphuric acid.

In electrochemical regeneration experiments, 0.15 M Na2SO4was added to assure good electrolyte conductivity. Cyclic voltam-metries were carried out between voltage values from −1.5 to 2 V.The rest of the electrochemical tests were setup at a constant cath-ode voltage of −0.7 V in respect of the Ag/AgCl reference electrodeand at constant volume. In order to keep a constant volume withinthe experiment, 5 ml of fresh solution, without metal and withthe same polymer concentration and ionic strength, was added toreplace each sample taken.

Fig. 3. Hypothetical model of the interaction between PEPEI

e Science 323 (2008) 28–36

3. Results and discussion

3.1. Ultrafiltration experiments

3.1.1. Hydrodynamic membrane conditions optimization withpolymer solutions

Firstly, operation variables were optimized in order to maximizeboth, permeate fluxes and rejection coefficients, with ultrafiltrationexperiments carried out in total recirculation mode. Based on previ-ous works [8,15], the feed flow rate was fixed at 100 l h−1 (tangentialvelocity of 3.2 m s−1). In respect of the rest of working variables,pH and transmembrane pressure were first optimized with exper-iments at 50 ◦C. At this temperature, four experiments were setupat different pH values, from 2 to 9.4 (original polymer solution pH),and within the range of transmembrane pressures from 0 to 4 bar.Fig. 2 gathers the influence of pH and transmembrane pressure onpermeate fluxes and polymer rejection coefficients. As it can beobserved, a maximum flux value is reached at pH 6. The perme-ate fluxes obtained at different pH values are lower, in the sameway as it has been observed and described for non-ethoxylatedpolyethylenimine, using the same ceramic membranes than those

ing phenomenon occurs due to the coordination of polymer withacid molecules [26]. At moderately acid pH values, PEPEI is stillslightly protonated but it is not coordinated with acid molecules.Polymer charge favours intramolecular and intermolecular repul-sion forces and the maximum value of flux is reached. At higherpH values, polymer is not protonated and its electrodonor char-acter originates that PEPEI behaves as a Lewis base, promoting itscoordination with ZrO2 acid sites on the membrane active layer.Moreover, another kind of interactions can occur as Bronsted orhydrogen bonds. The hypothetical interaction polymer–membraneactive layer is proposed in Fig. 3. This behaviour leads to a strongadsorption of polymer molecules over membrane surface that pro-duces the appearance of a fouling phenomenon.

In the present study, flux decline is not as marked as it has beenreported in previous studies for PEI. This result can be explainedmainly for two reasons: (1) the substitution of primary aminegroups by hydroxyl groups. Primary amines are the most activegroups as it can be observed from the value of the protonation con-stants of the different types of amines [27]. This can lead to the

and ZrO2 acidic centres on the membrane active layer.

J. Llanos et al. / Journal of Membrane Science 323 (2008) 28–36 31

125 ppm, is a new evidence that no copper hydroxide precipitationis taking place. If it would exist, the ultrafiltration membrane wouldreject it easily and the rejection decline would not be observed insuch an extent.

Considering the above-mentioned criterion, the optimal loadingratio is 208 mg Cu/g PEPEI, equivalent to a copper concentra-tion of 125 ppm. This loading ratio is similar to the experimentalcoordination capacity determined in bibliography for PEI–Cu sys-tem (246 mg Cu/g PEI) [29]. Taking into account this result, onecan establish that the partial substitution of primary amines byhydroxyl groups does not produce a clear decrease in polymercapacity to retain Cu2+ ions.

3.1.2.2. Ultrafiltration experiments with the optimal loading ratio.Once optimal loading ratio has been determined in total recircu-lation mode, different experiments were carried out at differentpH values (from 2 to 6). In these tests (gathered in Fig. 6) the influ-ence of pH and transmembrane pressure on permeate fluxes andmetal rejection coefficients were studied. As it can be observed,metal rejection coefficient diminishes when pH decreases, from0.97 at pH 6 to 0.2 at pH 2. This result can be explained

Fig. 4. Influence of transmembrane pressure and temperature on permeate fluxesand polymer rejection coefficients. pH 6, [PEPEI] = 0.06% (w/w), v = 3.2 m s−1.

strongest interaction with membrane functional acid groups and,consequently, to the highest fouling phenomenon; (2) the workingconditions used here (higher temperature and tangential velocityand lower polymer concentration) diminish flux decline because ofeither concentration polarization or fouling problems.

Regarding polymer rejection coefficients, values higher than 0.9have been reached within the pH interval used in the present work.Although the influence of pH is not as significant as on permeatefluxes, the maximum polymer rejection (R = 0.968) is obtained at pH6. As operation conditions must be those that maximize permeateflux without affecting rejection coefficients [14], a pH value of 6 isselected as the optimal working pH.

Next, a new experiment was setup at 25 ◦C and at the optimal pHvalue set previously with the aim of check the effect of temperatureon design parameters (Fig. 4). As it can be deduced, although thereexists a clear increase on permeate fluxes because of the viscos-ity drop, rejection coefficients keep constant. Therefore, selectedworking temperature for the rest of the experiments carried out inthis work is 50 ◦C.

3.1.2. Metal rejection coefficients optimization3.1.2.1. Loading capacity influence. Firstly, working at the optimalconditions previously set (pH 6, T = 50 ◦C, �P = 4 bar), keeping

polymer concentration constant and working in total recirculationmode, metal concentration was increased from 25 to 250 ppm. Inthese conditions, one can assure that does not exist any copperhydroxide precipitate. As an example, with a copper concentra-tion of 2 mM and 25 ◦C, it is necessary a pH near 6 to precipitatecopper hydroxide (pKs = 19.32). Furthermore, taking into accounttwo important aspects, the formation of insoluble hydroxides canbe considered negligible: (1) on the one hand, working at 50 ◦Cincreases hydroxide solubility what make necessary a higher pHvalue to produce hydroxide precipitation; (2) on the other hand, thepresence of chemical compounds with the ability of forming solublecomplexes with the target metal ion, as is the case for the polymer,clearly reduces the amount of free Cu2+ ions and, consequently,hinders the formation of hydroxide precipitates [28].

The optimal loading ratio (mg metal/mg polymer) will not leadto copper rejection below 0.97 at 4 bar of transmembrane pressure.All results are gathered in Fig. 5. As it can be deduced, on the onehand, rejection coefficients descend when transmembrane pres-sure increases, due to the increase of convective transport throughthe membrane. On the other hand, as expected, rejection coef-ficients reach lower values at higher metal concentrations. The

Fig. 5. Effect of transmembrane pressure and metal concentration on metal rejec-tion coefficients. T = 50 ◦C, pH 6, [PEPEI] = 0.06% (w/w), v = 3.2 m s−1.

noticeable descent that takes place over copper concentrations of

Fig. 6. Influence of transmembrane pressure and pH on permeate fluxes and metalrejection coefficients working at the optimal loading ratio. T = 50 ◦C, [PEPEI] = 0.06%(w/w), v = 3.2 m s−1.

32 J. Llanos et al. / Journal of Membran

�M1/3w

The values calculated for K and Cm using the stagnant layermodel as well as the residues (

∑(Jt

P − JP)2), where JtP is the per-

meate flux calculated using Eq. (3), are gathered in Table 1. Resultsobtained applying dimensional analysis are presented in Table 2.

Although Eq. (3) accurately adjusts experimental data, themass transfer coefficient calculated by dimensional analysis(7.47 × 10−6 m s−1) is considerably lower than that obtained using

Fig. 7. Evolution of permeate fluxes and polymer rejection coefficients during a110-min discontinuous experiment. Exp. 1: T = 50 ◦C, pH 6, [PEPEI] = 0.06% (w/w),v = 3.2 m s−1. Exp. 2: Exp. 1 + [Cu] = 125 ppm.

considering the competitive reaction between copper ions andprotons to bind polymer repetitive units. These reactions areschematised in Eqs. (1) and (2), in which polymer functional unitsare represented as L. From results plotted in Fig. 6, one can deducedthat it would be necessary to diminish solution pH till a value of 2 inorder to carry out the chemical polymer regeneration (pH decreaseand subsequent ultrafiltration step).

Cu2+ + nL ↔ [LnCu]2+ (1)

H+ + L ↔ [LH]+ (2)

3.1.3. Study of permeate flux affecting phenomenaWith the aim of studying the phenomena that cause a drop

on permeate flux (concentration polarization and fouling), twoexperiments were done in discontinuous mode and working at theoptimal conditions determined in previous steps (T = 50 ◦C, pH = 6,Qa = 100 l h−1, and �P = 4 bar). One test was carried out with a poly-mer concentration of 0.06% (w/w) and the other one with the samepolymer concentration and a metal concentration of 125 ppm (opti-mal loading ratio). All results are plotted in Fig. 7.

3.1.3.1. Concentration polarization. Concentration polarization isone of the phenomena that more clearly affects membrane pro-

cesses performance as it produces a deep drop in permeate flux. Todescribe this phenomenon, the stagnant layer model can be applied[30]:

Jp = k lnCm − Cp

Cb − Cp(3)

where Jp is permeate flux (m s−1), k is mass transfer coefficient(m s−1) and Cm, Cb and Cp are polymer concentration (% w/w) atthe membrane active layer (concentration of polarization), in thefeed stream and in the permeate, respectively. By adjusting perme-ate flux experimental data vs. polymer concentration in feed stream(in this work using Marquardt algorithm), the value of mass transfercoefficient and concentration of polarization can be calculated.

Furthermore, mass transfer coefficient can be calculated bymeans of dimensional analysis using empirical correlations basedon analogies between mass, heat and momentum transport equa-tions [30]. This method has been successfully applied in ourresearch group [5]. The dimensionless equation used is

Sh = ˇReaScb(

dh

L

)c

(4)

e Science 323 (2008) 28–36

Table 1Concentration–polarization parameters obtained for experiments in discontinuousmode (Eq. (2))

Without metalCm (% w/w) 0.61k (×106 m s−1) 26.53Average error (%) 1.7

With metalCm (% w/w) 3.81k (×106 m s−1) 9.78Average error (%) 2.0

[PEPEI] = 0.06% (w/w), pH 6, v = 3.2 m s−1. With metal: [Cu2+] = 125 ppm.

where dimensionless numbers needed are the Sherwood number(Sh), the Reynolds number (Re) and the Schmidt number (Sc), dh isthe membrane hydraulic diameter (m) and L (m) is the membranelength. Parameters ˇ, a, b and c depend on each system flux con-ditions and geometry. For developing laminar flow these constantsare equal to 0.664, 0.5, 0.33 and 0.5, respectively. In our system,hydraulic diameter is equal to the difference of the membrane innerdiameter and the steel rod diameter. Sh, Re and Sc can be calculatedby the following equation:

Sh = kdh

D(5)

Re = �vdh

�(6)

Sc = �

�D(7)

Finally, polymer diffusion coefficient, D (m2 s−1), can be calcu-lated from Eq. (8) [31], using solution temperature, T (K), solutionviscosity, � (Pa s), and polymer molecular mass, Mw. This equationhas been slightly modified to use international system units:

D = 8.34 × 10−15 T(8)

the stagnant layer model (26.53 × 10−6 m s−1). This result, joined tothe low value obtained for Cm, means that concentration polariza-tion is not the controlling phenomenon that causes the flux decline.On the contrary, mass transfer coefficient obtained in the presenceof copper (9.78 × 10−6 m s−1) is in the range of that calculated bydimensional analysis. Furthermore, the value of Cm (3.81%, w/w) issimilar to those obtained from the literature for non-ethoxylatedpolyethyleneimine [5]. This means that concentration polarizationis playing an important role as the controlling mechanism whencopper is present in the feed stream. As it can be observed fromFig. 7, the initial flux decline with respect to the pure solventflux is much higher when PEPEI–Cu solutions are filtered. As the

Table 2Results derived from the dimensional analysis

Re 5808Ds (×10−10 m2 s−1) 1.34Sc 4141.8Sh 362.676k (×106 m s−1) 7.47

T = 50 ◦C, [PEPEI] = 0.06% (w/w), pH 6, v = 3.2 m s−1.

mbran

(

(

calculated using Eq. (11), based on the Faraday law:

mt = MImFn

t (11)

where Im (a) is the average intensity in the considered time interval,t (s). As intensity is not a constant value, Im can be calculated usingthe following equation:

Im =∫ t

0I dt

t(12)

On the other hand, the average deposition rate, Vd (g h−1), wouldexpress the amount of copper deposited per hour during eachexperiment.

3.2.1. Cyclic voltammetriesFirst of all, several cyclic voltammetries were performed with

two objectives: (a) determine if the polymer suffered either oxi-dation or reduction processes; (b) calculate the optimal workingvoltage at which competitive cathodic reactions were minimized.With this aim, voltammetric studies were carried out with increas-ing complexity solutions, gathered in Table 4. The amount of

J. Llanos et al. / Journal of Me

Table 3Fouling parameters obtained for experiments in discontinuous mode (Eq. (8))

With metal Without metal

Standard pore blockingKs (l) 0.195 0.264Average error (%) 1.87 1.71

Intermediate pore blockingKi (l) 0.205 0.286Average error (%) 1.71 1.55

Cake formationKc (l2 h−1) 0.363 0.368Average error (%) 1.41 1.21

[PEPEI] = 0.06% (w/w), pH 6, v = 3.2 m s−1. With metal: [Cu2+] = 125 ppm.

initial flux decline is mainly due to concentration polarization,these results confirm the conclusions previously explained.

3.1.3.2. Fouling. Fouling is such an important problem as concen-tration polarization in respect of the performance of ultrafiltrationprocesses. Fouling causes an irreversible descent on permeate fluxdue to the precipitation or accumulation of solutes on the mem-brane surface and/or on the pores caused by physical or chemicalinteractions between the solute and the membrane active layer[8,32].

From the results plotted in Fig. 7, one can deduce some con-clusions. On the one hand, the flux decline cannot be explainedby the increase in the polymer concentration and solution viscos-ity. On the other hand, the permeate flux does not suffer a deepdecrease within time and this decrease is considerable lower thanthe initial flux decline. These results mean that, although does exista fouling phenomenon, this is not as important as concentrationpolarization.

To describe mathematically the fouling phenomenon, three dif-ferent models have been applied corresponding to different typesof fouling:

a) standard pore blocking

Jp = Jv0

(1 + 1

2KsAJv0t

)−2(9a)

b) intermediate pore blocking

Jp = Jv0(1 + KiAJv0t)−1 (9b)

(c) cake layer formation

Jp = Jv0(1 + 2Ks(AJv0)2t)−1/2

(9c)

where Jv0 is the initial permeate flux (l h−1 m−2), Ks the standardpore locking constant, Ki the intermediate pore blocking constant,Kc the cake layer formation constant and A is the membrane area.

The results obtained in experiments with and without metal aredisplayed in Table 3. As it can be seen, the differences obtained forthe three models are negligible. This is a new clue that fouling is notnoticeable under the experimental conditions of the present work.

3.2. Electrochemical regeneration

In this stage of the work, copper concentration was increasedto 200 ppm mainly for two reasons: (a) cover a higher metalconcentration range during the discontinuous electrodepositionexperiments; (b) simulate the retentate stream obtained in a realultrafiltration stage (feed and bleed or discontinuous) where theconcentration of the retentate is higher than the concentration in

e Science 323 (2008) 28–36 33

Table 4Concentration of the different solutions tested in the cyclic voltammetry analyses

H2SO4 Na2SO4 (M) CuSO4 (ppm) PEPEI (%)

1 pH 2 – – –2 pH 2 – – 0.13 pH 2 0.15 – –4 pH 2 0.15 – 0.15 pH 2 0.15 200 –6 pH 2 0.15 200 0.1

the feed stream. To keep the polymer/metal optimal ratio constant,the polymer concentration was increased to 0.1%.

In order to compare the process behaviour at different exper-imental conditions, two parameters have been calculated. On theone hand, current efficiency, �c (%), Eq. (10), expresses the ratiobetween the deposited copper mass, me (g), and the theoreticalmass that would be eliminated if all the electrons fed to the systemwere used in the target reaction of electrodeposition, mt (g):

�c = me

mt× 100 (10)

Deposited copper mass can be calculated from the evolution ofcopper concentration with time. Theoretical deposited mass can be

Fig. 8. Cyclic voltammetries for H2SO4 solutions (pH 2). Solid line: without polymer.Dash line: [PEPEI] = 0.1% (w/w). Voltage is expressed vs. Ag/AgCl.

34 J. Llanos et al. / Journal of Membrane Science 323 (2008) 28–36

at the working electrode surface. In all cases, an intensity increase

Fig. 9. Cyclic voltammetries for H2SO4 (pH 2) and Na2SO4 (0.15 M) solutions. Solidline: without polymer. Dash line: [PEPEI] = 0.1% (w/w). Voltage is expressed vs.Ag/AgCl.

sulphuric acid added was that necessary to lower pH until a valueof 2. At this pH, the polymeric complex PEPEI–Cu must be weak-

ened, as it has been demonstrated in ultrafiltration experiments.This condition has been observed to be decisive in previous works[23]. All voltammetric analyses were carried out from −2 V to 1.5 Vrespect to Ag/AgCl reference electrode. No higher oxidation poten-tials were applied in order to avoid the oxidation of the graphiteelectrode [33]. All results are presented in Figs. 8–10. Fig. 8 repre-sents the results obtained for solutions 1 and 2 from Table 4, Fig. 9gathers results reached for solutions 3 and 4 from Table 4 and Fig. 10shows experiments for solutions 5 and 6.

First of all, as it can be deduced for all the tests, at values near−1.5 V, a big peak is observed corresponding to hydrogen evolution,this is, the reduction of protons to hydrogen (Eq. (13)). In the sameway, at a voltage of 1.5 V, the beginning of a new peak is observedcorresponding to oxygen evolution (Eq. (14)).

2H+ + 2e− ↔ H2 (13)

OH− ↔ 12 O2 + 2e− + H+ (14)

Secondly, an important result is that all the shapes for I–V curvesare similar for tests with and without polymer. This result showsthat polymer does suffer neither oxidation nor reduction processes

Fig. 10. Cyclic voltammetries for H2SO4 (pH 2), Na2SO4 (0.15 M) and CuSO4 (200 ppmCu2+) solutions. Solid line: without polymer. Dash line: [PEPEI] = 0.1% (w/w). Voltageis expressed vs. Ag/AgCl.

Fig. 11. Intensity vs. time for experiments carried out at a constant voltage of−0.7 V vs. Ag/AgCl at different pH values. [Cu] = 200 ppm, [PEPEI] = 0.1% (w/w),[Na2SO4] = 0.15 M.

for a given voltage is observed when the polymer is present, whichcan be explained by higher conductivity due to the presence of apolyelectrolyte in the target solution.

Considering the curves in the presence of CuSO4 (Fig. 10), thereexists a little reduction peak at a voltage of, approximately, −0.5 Vwhich can be assigned to Cu2+ ions reduction. Likewise, a new peakis observed at a voltage of 0.5 without PEPEI and at 0.7 V with PEPEI,corresponding to the oxidation of Cu to Cu2+.

Taking into account all the previous results, a working potentialof −0.7 V was selected with the aim of allowing copper reductionmeanwhile hydrogen evolution is minimized.

3.2.2. Deposition tests at constant voltageCopper deposition was studied at a constant voltage of −0.7 V

with solutions of CuSO4 (200 ppm), Na2SO4 (0.15 M) and PEPEI(0.1%, w/w), varying pH from 1.8 to 4.8. Figs. 11 and 12 repre-sent, respectively, the intensity values (average value calculatedeach 10 min) and the evolution of copper concentration with time.Fig. 13 gathers the influence of pH upon deposition rate and currentefficiency.

Fig. 12. Removal of dissolved copper for experiments carried out at a constant volt-age of −0.7 V vs. Ag/AgCl at different pH values. [Cu] = 200 ppm, [PEPEI] = 0.1% (w/w),[Na2SO4] = 0.15 M.

J. Llanos et al. / Journal of Membran

Fig. 13. Influence of pH on average deposition rate and current efficiency.[Cu] = 200 ppm, [PEPEI] = 0.1% (w/w), [Na2SO4] = 0.15 M.

As it can be deduced from Fig. 11, intensity clearly diminisheswith time on stream. During the first 10 min a deep descent isobserved coinciding with the formation of the first copper layerdeposited upon the electrode. This result allows us to affirm thatmetallic copper surface presents worse performance than graphiteto act as cathode under the conditions tested in the present work.Furthermore, there exists a decrease in the available electrode areabecause copper recovery can cover porous structure of graphite.During the rest of the experiment, a continuous but very much lowdescent is observed, which can be explained by the gradual cooperconcentration descent [34].

Moreover, as it can be observed in Fig. 12, the concentration dropis almost constant with pH until a value of 3.3. For higher pH val-ues, a deep fall in deposition rate is obtained. This experimentalresult has been previously described in bibliography [23,35] andcan be explained by the fact that the strength of PEPEI–Cu complexincrease with pH, what hinders the Cu2+ ion transference to theelectrode surface. Therefore, a clear descent on deposition rate isproduced when pH is increased. This result is also observed in termsof current efficiency results, gathered in Fig. 13. Current efficiencyis lowered because competitive reactions (hydrogen evolution)become preferential when copper is complexed by the polymer.Base on these results, the optimal working pH is 3.3, higher than

the pH necessary to carry out the chemical polymer regeneration.

4. Conclusions

The viability of a PEUF process to recover copper using PEPEIas water-soluble polymer has been demonstrated in the presentwork. In metal retention stage, rejection coefficients over 97% hasbeen obtained working with a loading ratio of 208 mg Cu/g PEPEI, at50 ◦C, �P = 4 bar and pH 6. From this result it can be concluded thatthe substitution of –NH2 groups by –OH in polyethylenimine doesnot have a marked negative effect on its capacity to retain copperions.

The polymer electrochemical regeneration via has been testedchecking that PEPEI does suffer neither oxidation nor reduc-tion processes when the target solution is regenerated under ourworking conditions. Moreover, the optimal working pH has beendetermined with studies at constant voltage. It has been checkedthat the optimal working pH is higher than that necessary if theregeneration were carried out chemically. This represents a clearadvantage as lower reagents consumption and less extreme work-ing conditions are required. Furthermore, a used ultrafiltration

e Science 323 (2008) 28–36 35

membrane has been used as cathodic material. Although this optionoffers the possibility of re-use this material, low deposition veloc-ities have been obtained which diminish with time on stream. Forthis reason, in future works a new electrochemical cell that maxi-mizes mass transfer will be designed in order to get higher valuesof both, current efficiency and deposition rate.

Nomenclature

c fouling phenomenological model constant(l h−1 m−2)

Cb feed polymer concentration, % (w/w)CCu copper concentration in electrodeposition experi-

ments, % (w/w)Cm polymer concentration of polarization, % (w/w)Cp permeate polymer concentration, % (w/w)d fouling phenomenological model constant (m−1)dh membrane hydraulic diameterD polymer diffusion coefficient (m2 s−1)F Faraday constant, 96,485 C mol−1

Im average intensity in the considered time interval (t),A

j current density (mA cm−2)Jp permeate flux (l h−1 m−2)J∞ permeate flux at steady state (l h−1 m−2)k mass transfer coefficient (m s−1)L membrane lengthme electrodeposited copper mass (g)mt theoretical copper electrodeposited mass consider-

ing 100% current efficiency (g)M copper atomic mass, 63.54Mn polymer molecular massn number of electrons involved in the considered elec-

trochemical reaction�PTM transmembrane pressure (bar)Re Reynolds dimensionless numberSc Schmidt dimensionless numberSh Sherwood dimensionless numberv tangential velocity (m s−1)Vd average deposition rate (g h−1)

Greek letters�c current efficiency (%)

� solution viscosity (kg m−1 s−1)� solution density (kg m−3)

References

[1] K.E. Geckeler, E. Bayer, B.Y. Spivakov, V.M. Shkinev, G.A. Vorob’Eva, Liquid-phasepolymer-based retention, a new method for separation and preconcentrationof elements, Anal. Chim. Acta 189 (1986) 285.

[2] J. Buffle, C. Staub, Measurement of complexation properties of metal ions innatural conditions by ultrafiltration: measurement of equilibrium constants forcomplexation of zinc by synthetic and natural ligands, Anal. Chem. 56 (1984)2837.

[3] M. Rumeau, F. Persin, V. Sciers, M. Persin, J. Sarrazin, Separation by couplingultrafiltration and complexation of metallic species with industrial water sol-uble polymers. Application for removal or concentration of metallic cations, J.Membr. Sci. 73 (1992) 313.

[4] K. Volchek, E. Krentsel, Y. Zhilin, G. Shtereva, Y. Dytnersky, Polymer bind-ing/ultrafiltration as a method for concentration and separation of metals, J.Membr. Sci. 79 (1993) 253.

[5] P. Canizares, A. Perez, R. Camarillo, Recovery of heavy metals by means ofultrafiltration with water-soluble polymers: calculation of design parameters,Desalination 144 (2002) 279.

mbran

[

[

[

[

[

[

[

[

[

[

[

[

Sci. 232 (2004) 99.

36 J. Llanos et al. / Journal of Me

[6] J. Barron-Zambrano, S. Laboire, Ph. Viers, K. Rakib, G. Durand, Mercury removalfrom aqueous solutions by complexation–ultrafiltration, Desalination 144(2002) 201.

[7] J. Llorens, M. Pujola, J. Sabate, Separation of cadmium from aqueous streams bypolymer enhanced ultrafiltration: a two-phase model for complexation bind-ing, J. Membr. Sci. 239 (2004) 173.

[8] P. Canizares, A. Perez, R. Camarillo, J.J. Linares, A semi-continuous laboratory-scale polymer enhanced ultrafiltration process for the recovery of cadmiumand lead from aqueous effluents, J. Membr. Sci. 240 (2004) 197.

[9] J. Sabate, M. Pujola, J. Llorens, Simulation of a continuous metal separationprocess by polymer enhanced ultrafiltration, J. Membr. Sci. 268 (2006) 37.

10] J. Muslehiddinoglu, J. Uludag, H. Onder Ozbelge, L. Yilmaz, Effect of operating

parameters on selective separation of heavy metals from binary mixtures viapolymer enhanced ultrafiltration, J. Membr. Sci. 140 (1998) 251.

[11] R. Molinari, S. Gallo, P. Augurio, Metal ions removal from wastewater or washingwater from contaminated soil by ultrafiltration–complexation, Water Res. 38(2004) 593.

12] B.L. Rivas, S.A. Pooley, E.D. Pereira, R. Cid, M. Luna, M.A. Jara, K.E. Geckeler,Water-soluble amine and imine polymers with the ability to bind metal ions inconjunction with membrane filtration, J. Appl. Polym. Sci. 96 (2005) 222.

13] P. Canizares, A. de Lucas, A. Perez, R. Camarillo, Effect of polymer nature andhydrodynamic conditions on a process of polymer enhanced ultrafiltration, J.Membr. Sci. 253 (2005) 149.

14] K.E. Geckeler, K. Volchek, Removal of hazardous substances from water usingultrafiltration in conjunction with soluble polymers, Environ. Sci. Technol. 30(3) (1996) 725.

15] P. Canizares, A. Perez, R. Camarillo, J. Llanos, M.L. Lopez, Selective separationof Pb from hard water by a semi-continuous polymer-enhanced ultrafiltrationprocess (PEUF), Desalination 206 (2007) 602.

16] E. Bayer, Recovering precious metals from aqueous solutions with solublepolyamine derivatives, Ger. Offen. DE 3,002,883 (July 30, 1981).

[17] M. Rumeau, J.P. Mangeolle, Method and apparatus for continuous recovery ofmetal cations from dilute solution, FR 2,567,914 (January 24, 1986).

[18] M. Rumeau, Complexation, ultrafiltration, electrolysis. A process for the recov-ery of metals from impure and dilute solutions, Inf. Chim. 272 (1986) 143.

19] K. Juttner, U. Galla, H. Schmieder, Electrochemical approaches to environmentalproblems in the process industry, Electrochim. Acta 45 (2000) 2575.

20] C.K. Liu, C.W. Li, Simultaneous recovery of copper and surfactant by an elec-trolytic process from synthetic solution prepared to simulate a concentrate

[

[

[

[

[

[

[

[

[

[

[

e Science 323 (2008) 28–36

waste stream of a micellar-enhanced ultrafiltration process, Desalination 169(2004) 185.

21] C.K. Liu, C.W. Li, Combined electrolysis and micellar enhanced ultrafiltration(MEUF) process for metal removal, Sep. Purif. Technol. 43 (2005) 25.

22] R.S. Juang, C.H. Chiou, Ultrafiltration rejection of dissolved ions using variousweakly basic water-soluble polymers, J. Membr. Sci. 177 (2000) 207.

23] J. Barron-Zambrano, S. Laboire, P. Viers, M. Rakib, G. Durand, Mercury removaland recovery from aqueous solutions by coupled complexation–ultrafiltrationand electrolysis, J. Membr. Sci. 229 (2004) 179.

24] B. Schlichter, V. Mavrov, T. Erwe, H. Chmiel, Regeneration of bonding agentsloaded with heavy metals by electrodialysis with bipolar membranes, J. Membr.

25] P. Canizares, A. Perez, R. Camarillo, M.T. Villajos, Modelling and processimprovement of a batch polyelectrolyte enhanced ultrafiltration process forthe recovery of copper, Desalination 184 (2005) 357.

26] I.H. Park, E.J. Choi, Characterization of branched polyethyleneimine by laserlight scattering and viscometry, Polymer 37 (2) (1996) 313.

27] N.V. Jarvis, M.N. Chen, Measurement of binding constants of PEI with metal ionsand metal chelates in aqueous media by ultrafiltration, Ind. Eng. Chem. Res. 35(1996) 1935.

28] S. Kotrly, L. Sucha, Handbook of Chemical Equilibria in Analytical Chemistry,Ellis Horwood Ltd., Chichester, 1985, p. 339.

29] V.N. Kislenko, L.P. Oliynyk, Complex formation of polyethylenimine with cop-per(II), nickel(II) and cobalt(II) ions, J. Polym. Sci. Polym. Chem. 40 (2002) 914.

30] W.S. Winston Ho, K.H. Sirkar, Membrane Handbook, Chapman & Hall, New York,1992, pp. 398–407.

31] M.E. Young, P.A. Carroad, R.L. Bell, Estimation of diffusion coefficients of pro-teins, Biotechnol. Bioeng. 22 (1980) 947.

32] J. Kim, F.A. DiGiano, A two-fiber, bench-scale test of ultrafiltration (UF) forinvestigation of fouling rate and characteristics, J. Membr. Sci. 271 (2006) 196.

33] E.A. Ticianelly, C.R. Derouin, A. Redondo, S. Srinivasan, Methods to advancetechnology of proton exchange membrane fuel cells, J. Electrochem. Soc. 135(9) (1998) 2209.

34] F.C. Walsh, A First Course in Electrochemical Engineering, Electrochemical Con-sultancy, United Kingdom, 1993, (Chapter 4).

35] P. Baticle, C. Kiefer, N. Lakhchaf, O. Leclerc, M. Persin, J. Sarrazin, Treatment ofnickel containing industrial effluents with a hybrid process comprising of poly-mer complexation–ultrafiltration–electrolysis, Sep. Purif. Technol. 18 (2000)195.