Highly selective extraction of metal ions from dilute solutions by hybrid electrodialysis technology

Accepted Manuscript Highly selective extraction of metal ions from dilute solutions by hybrid electrodialysis technology Salah Frioui, Rabah Oumeddour, Stella Lacour PII: DOI: Reference:

S1383-5866(16)30373-2 http://dx.doi.org/10.1016/j.seppur.2016.10.028 SEPPUR 13298

To appear in:

Separation and Purification Technology

Received Date: Revised Date: Accepted Date:

13 May 2016 19 October 2016 19 October 2016

Please cite this article as: S. Frioui, R. Oumeddour, S. Lacour, Highly selective extraction of metal ions from dilute solutions by hybrid electrodialysis technology, Separation and Purification Technology (2016), doi: http:// dx.doi.org/10.1016/j.seppur.2016.10.028

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Highly selective extraction of metal ions from dilute solutions by hybrid electrodialysis technology Salah Frioui1, Rabah Oumeddour 1, Stella Lacour2* (1) Laboratoire d’Analyses Industrielles et Génie des Matériaux (LAIGM), Université 8 Mai 1945 de Guelma, B.P.401, 24000, Algérie. (2) Institut Européen des Membranes (IEM, UMR 5635 CNRS-ENSCM-UM), Université Montpellier CC047, Place Eugène Bataillon, 34095 Montpellier Cedex 5, France. *Corresponding author: [email protected] (tel: +33467149116; Fax: +33467149119)

Abstract: This study examines the feasibility of coupling electrodialysis (ED) to an in-situ complexation reaction (CR) step (named as ED-CR process), to selectively recover and concentrate various metallic cations with different or similar valences hardly separable by conventional ED. In a first stage, the complexing behavior of different chemical agents towards four metal ions (Ag, Zn, Cu and Cd) have been studied and modeled from ionic forms distribution calculations. EDTA was found to be the best agent which combines advantageous characteristics: the formation of thermodynamically stable negative charged complexes in a large range of metal ion concentration, a discriminant complexation ability, and the possibility of ligand displacement for the ultimate recovery of the released metal ion (and the complexing agent as well). In a second stage, electro-extraction performances have been investigated under various operating conditions: pH of the feed to be treated, solution flow rates and concentrations, electric voltage applied, concentration of electrolyte solutions used, continuous or batch mode applied. The results showed high electro-separation performances both for the Ag/Zn and Cu/Cd systems. Some limitations of the process have been identified and well-delimited operating conditions could be proposed to achieve both optimal extraction and selectivity. This original hybrid technology would extend the potentialities of ED for the treatment of added-value metal ion containing wastewaters. Keywords: hybrid electrodialysis; complexation; selective separation; metal ions recovery; water reuse. Abbreviations / Acronyms: AAS AEM A400 AFNOR CEM CR DC EC ED EE EDBM EDTA HEDP IEM NF UF

atomic absorption spectroscopy anion exchange membrane special grade of anion exchange membrane for large organic ion transport association française de normalisation cation exchange membrane complexation reaction direct current energy consumption electrodialysis electro-extraction efficiency electrodialysis with bipolar membranes ethylene diamine tetra acetic acid 1-hydroxyethane-1,1-diphosphonic acid ion exchange membrane nanofiltration ultrafiltration

1

1. Introduction During the last decades, many industries of metal and surface coating (galvanizing, electroplating, anodizing, painting) and other industries such as photography, water softening, textile and paper manufacturing, industrial cleaning, generate large and dangerous amounts of effluents. These effluents are liquid, sometimes a little dense and complex medium, which contain a multitude of toxic metals such as silver, cadmium, copper, zinc, platinum, …. These inorganic pollutants are of considerable concern because they are nonbiodegradable, accumulated in living tissues, causing various diseases and disorders [1]. Conventional methods for heavy metals removal from effluent generally involve chemical and physical methods, such as precipitation, adsorption, coagulation-flocculation, solvent extraction… [2-4]. These technologies have relatively low operating costs. They proved their ability to achieve regulatory effluent limits for several metals. The main drawbacks are the generation of solid wastes or big sludge volumes that are usually hazardous or unstable in terms of potentially toxic metals re-mobilization and require either safe land filling or disposal, while added-value metals are not re-used. As the environmental standards on releases are increasingly being stringent, actual and future heavy metal treatment technologies have to fulfill several criteria: minimize waste volume, maximize recycling (water and matter) and thereby optimize concentration and purification steps. Membrane and electro-membrane processes are known to be highly attractive technologies in terms of recovery and reuse of waste process streams [3-8] but, still today, they present only a partial response when dealing with the treatment of mixtures, particularly due to the non-ideal selectivity of the membranes [9, 10]. Important research activities have been devoted to solve this problematic. The majority concerns the chemical modification of membrane surfaces or the synthesis of new materials in order to increase the selectivity of membranes [7, 11-16]. Other propose multistage process or pre, post, or coupled treatment operations [12, 1726], and/or more simply act on specific operating conditions [3, 5, 12, 27]. Very few papers deal with the use of electro-membrane techniques for separating ions from multi-metallic mixtures [3, 5, 15, 28-31]. Notwithstanding, electrodialysis (ED) has significant potentialities such as limited consumption of chemicals, low waste rejection, operation at room temperature. Moreover, thanks to its high modulability, ED can easily integrate in-situ treatment steps in order to facilitate molecules transport and enhance selectivity: UF or NF steps [20], complexation step [9, 10, 28-30, 32]. For the latter, the role of the complexing agent is to modify the ionic transport of a given metallic cation within the electrodialysis process, through charge and/or size metal ion modifications. Thus, the “modified ion” will have its own affinity with one electrode and the separation will be done. Various complexing agents (with strong, medium or weak complexing behaviour towards metal ions) can be used depending on the nature of the ion to be separated. The use of ethylene diamine tetra acetic acid (EDTA), in the conventional ED, is fairly widespread for the selective complexation of various metal ions (Sr2+, Ni2+, Zn2+, Cu2+, Co2+…): Already in 1966, a first attempt in the separation of strontium and cesium using EDTA was successful [33]. Other authors evidenced later the formation of a chelate with EDTA only with strontium [9], zinc [29] or cobalt [30] from a given metallic salt mixture. Various studies stated the relevance of using other complexing agents such as oxalic acid and citric acid (for Ni2+-Cu2+ and Cu2+-Fe3+ separation, respectively) [32], hydroxyethane diphosphonic acid (HEDP) (for Cu2+ recovery from cyanide solutions) [34]. Also, several polymeric agents bonded with metal ions have been tested using electrodialysis with bipolar membrane (EDBM) technique, with the view to regenerate ligand and re-use metals [6]. If the ion transfer by the way of complexation can easily be done in the electrodialyser, the prevention of fouling phenomena and/or the transport of a voluminate complexed ion through ion exchange membranes (IEM) can be considered as a daring challenge. Previous work showed severe limitations of the ED process using standard ion exchange membranes (IEM) [6, 29, 30, 35, 36]. Emerging coupled membrane techniques (ED-UF or ED-NF) [18-20, 37] or the synthesis of ion-exchange membranes specially adapted to large organic ions transfer (PCA GmbH Company [38]) may take on the challenge for specific applications. Also, the selected ligand must have discriminant complexing behaviour towards a given metal ion but also it is important that the complexation reaction can be easily displaced in order to recover and, in-situ recycle, the free ligand. In this context, the present study examines the potentialities of a hybrid electrodialysis technology for the selective treatment of a metal ions mixture solution with different or identical valences (Ag + and Zn2+; Cu2+ and Cd2+). This hybrid technique named as ED-CR process, couples ion-exchange and complexation reaction (CR) step under electrical field applied. For the complexation step investigation, different chemical agents were considered and modeled from ionic forms distribution calculations and the best discriminant ligand was selected for electro-extraction experiments. For the latter, the influence of various parameters on the electro-separation performances was especially investigated: pH, flow rate and concentration of the feed stream, electric voltage applied, concentration of the electrolyte solution used, continuous or batch mode applied. Monitoring pH, conductivity and metal ion concentration changes in all hydraulic circuits helped the understanding of ion transport and of physico-chemical interactions that took place in the bulk solution as well as at membrane/solution interfaces, during the course of electrodialysis process.

2

2. Materials and Methods 2.1. Solutions All solutions used were analytical grade reagents and prepared with ultra-pure water (Millipore Milli-Q Plus System).The mixture solutions (feed) were prepared from reagent grade nitrate salts for metal ion, and sodium salt for the selected ligand. In the case of EDTA, the readily soluble disodium salt was used. a) A volume v of ligand Lx- at a concentration of 3x10-3 eq.L-1 (or 3x10-4 eq.L-1) was introduced into a beaker with stirring. It must be introduced in an amount necessary to complex the target ion M1Z1+, but should not be excessive so as not to interact with the other metal ion M2Z2+. b) An identical volume v of M1 nitrate salt at 3x10-3 eq.L-1 (or 3x10-4 eq.L-1) was slowly poured to form the complex M1Ln- at a pertinent pH (depending on the thermodynamic previsions). Adjustment of pH value was made by micro additions of NaOH (0.1 or 1.0 M). c) Finally, an identical volume v of M2 nitrate salt at 3x10-3 eq.L-1 (or 3x10-4 eq.L-1) was added, then pH was adjusted again if needed. The solution thus prepared had a final concentration of every ion M1Z1+, M2Z2+ and Lx- equal to 10-3 eq.L-1 (or 10-4 eq.L-1). 2.2. Membranes Three types of commercial ion exchange membranes from PCA GmbH manufacturer [38] were used and their characteristics are given in Table 1.  Three AEM “PC 400 D” named A400 dedicated to the transport of large organic anions with a mass of about 400D;  Three standard MEA “PC SA” for the transport of inorganic anions;  Six standard MEC “PC SK”, for the transport of inorganic cations. Prior to ED experiment, physico-chemical properties of ion exchange membrane were stabilized according to AFNOR specifications (NF X 45-200) but replacing chloride by nitrate ions (because of silver ions use in the process). The experimental protocol is described below:  AEM membrane: Each membrane was immersed in 0.1 M HNO3 for one hour, then rinsed with ultra-pure water, dried slightly with filter paper and finally flushed with 0.1 M NaNO 3.  CEM membrane: Each membrane was immersed in 0.1 M HNO3 for one hour, then rinsed with ultra-pure water, and dried on filter paper. It was then packed in 0.1 M NaOH for one hour and then rinsed with 0.1 M NaNO3. At the end of this protocol, AEM and CEM were equilibrated in NaNO3 electrolyte solution (10-2 eq.L-1 or 10-3 eq.L-1) before entering the ED stack. Table 1 Characteristics of the ion exchange membranes used in the ED-CR process [38]. Series

Ion-exchange capacity CE (meq.g-1)

Water content (%)

Thickness (µm)

Functional group

Mobile counter-ion

Chemical stability (pH)

Operating temperature (°C)

Organic anion

1.0

~48

160-200

Strongly alkaline (ammonium)

NO3-

0-10

Max 40

AEM standard “PC-SA”

Standard

1.5

~14

180-220

Strongly alcaline (ammonium)

NO3-

0-9

Max 60

CEM standard “PC -SK”

Standard

0.9

~9

160-200

Strongly acid (sulfonic)

Na+

0-11

Max 50

Membrane

A400 “PC 400D”

2.3. Equipments The electrodialysis unit used is a PCCell ED 64 0 04 manufactured by Polymerchemie Altmeier (PCA) GmbH [38]. It is composed of two electrode blocks (platinum plated titanium for anode and stainless steel for cathode) and a membrane stack between them. The stack consists of three cell units; each unitary cell (Fig.1) is composed of one diluate compartment (metal ion mixture as the feed solution) and two different concentrate compartments, with an alternate location of anion exchange (AEM standard or A400) and cation exchange (CEM) membranes. Electrode compartments are independent circuits as well as “buffer” compartment, the latter allowing each cell unit to be separated. A power supply device TTi EX 752 M was used in potentiostatic mode. Two multimeters METRA Hit 29s were used for external control of voltage and current. A peristaltic pump (Watson-Marlow 302) allowed to deliver constant flow of the feed in the range 3.0-9.0 L.h-1 and four centrifugal pumps SIEBEC M7 (10 W, Qmax=700 L.h-1) were used for the other tanks. Monitoring 3

conductivity was made respectively by means of five portable conductivity sensors (WTWLF 320); pH and temperature measurements were made by portable pH-meters (WTW pH 320) equipped with combined pHelectrode with integrated temperature sensor SenTix 41. The calibration of pH and conductivity sensors was done against standardized pH buffer and conductivity solutions. The schematic layout of the full-equipped electrodialysis setup is shown in Fig.2. Table 2 summarizes all the operating conditions applied.

Fig.1. Schematic layout of a unitary cell in the ED-CR process (initial feed solution is composed of M1Z1+ , M2Z2+, Lx- with an in-situ Complexation Reaction: M1Z1+ is complexed selectively as M1Ln- form).

Fig.2. Scheme of the full-equipped ED pilot. pHI: pH indicator; FI: flow indicator; DC: direct current supplier; CI: conductivity indicator; V: voltameter ; (+): anode; (-): cathode. For reproducible ED experiments, an “initialization-running-rinsing” protocol was applied: The “electrolyte” solution (NaNO3, 10-2 or 10-3 eq.L-1) flowing in concentrate circuits for the initialization and running steps, was also used for the rinsing ED procedure (three cycles of 5-15 minutes) in all hydraulic circuits. HNO3 solution (10-2 or 10-3 eq.L-1) was used in electrode and buffer compartments. For the running step, the application of the electric field corresponded to the time t0 of every ED experiment. Data monitoring (pH, flow rate, conductivity, temperature) and sampling (5 or 7mL) were done in all the reservoirs as a function of time, every 15 minutes. Voltage input and electric current evolution involved were also monitored. Analyses of metal ion concentration evolution in all compartments were carried out by AAS, with a Spectraa 220FS (Varian). Table 2 shows the operating conditions investigated in this work.

4

Table 2. Operating conditions applied to the electrodialysis unit. Diluate (feed)

Solution

; -1

Concentration (eq.L )

10 or 10

Volume v (L) Global dead volume (L) Flow

rate

Q

-1

(L.h )

(peristaltic pump (PP) centrifugal pump (CP))

;

-4

-3

Concentrate M1 or M2

Electrode

Buffer

NaNO3

HNO3

HNO3

-3

10 or 10

-2

-3

10 or 10

-2

10-3 or 10-2

1.80

1.00

1.00

1.00

0.10

0.30

0.25

0.18

3.0-6.0-9.0

10.0

(PP)

20.0

(CP)

10.0

(CP)

(CP)

or

Active membrane area (cm2) per cell unit

64

64

64

64

Membrane spacing (cm)

0.05

0.05

0.1

0.05

Processing length (cm)

8

8

8

8

Fluid flow section (cm2/ compartment)

0.4

0.4

0.8

0.4

Circulation rate (cm.h-1/ compartment)

47-94-140

156

312

156

Batch or Single pass

Batch

Batch

Batch

Circulation mode Voltage U applied (V)

0-5-10-20

The experiences of electro-separation were started at room temperature (24±2 °C). During the process, an increase in the temperature of the solutions was observed due to joule effects. In the case of excessive variation, this could significantly affect ionic conductivity and thereby ion transport. Nevertheless, temperature never exceeded 30 °C for a 6 hour-experiment. 3. Data computation and analysis From AAS ion analysis, data collected in all the compartments were converted into number of moles by Eq.1 (concentrate compartments) and Eq.2 (diluate compartments): (1)

and

(2)

Where : mol of ions at time t; : mass concentration of ions (g.L-1) at time t from AAS analysis; : molar -1 mass of the ion (g.mol ); : mass concentration of ions (g.L-1) at previous time (t-1); Vt: volume (L) at time t in the tank; Vp: volume (L) of the sampling = 0.007 L. Faradic efficiency Rf represents the fraction of current carried by an ion during its transfer from the diluate (feed) to the concentrate. Generally, Rf is expressed as a percentage, calculated from either flux of extracted ions from the diluate (Eq.3) or flux of transfered ions in the concentrate compartment (Eq.4), under potentiostatic regime. (3)

and

(4)

With F: Faraday constant (96485 C.mol-1); z: charge of the ion; N: number of ED cell units (N=3); moles in the diluate (feed); : moles in the diluate at time t; : initial moles in the concentrate; the concentrate at time t; i: electric current (A) measured at time t.

: initial : moles in

On the other hand, the overall mass balance for metal ions is written according to Eq.5: (5) : initial moles in the diluate (feed); : moles in the diluate at time t; : moles in the concentrate at time t; : moles in electrode at time t; : moles in buffer at time t; : moles inside membranes at time t. The metal ion electro-extraction efficiency ( to Eq.6:

) from the diluate compartment (feed) was determined according

(6) 5

Where

: initial moles in the diluate (feed) and

: moles in the diluate (feed) at time t.

In all studied cases, experimental time never exceeded 240 minutes. Energy consumption (kWh.m-3) involved for the treatment of one meter cube of feed solution (diluate) was calculated according to Eq.7: (7) U: constant electric voltage applied (V); i: electric current (A) measured at time t (sec-1); Vfeed: volume (L) in the feed tank (diluate compartment). Statistical t-test for paired comparisons was used to check for experimental measurement repeatability (Eq.8). where

(8)

With di: difference between two measured values (=one pair); p: number of pairs; : mean difference observed for the p pairs of di values; Sd: standard deviation; t at (p-1) degrees of freedom. A first order kinetic for the electro-extraction process was proposed (Eq.9) and confirmed against the calculated determination coefficient R2. (9) Where : mass concentration of ions (g.L-1) at time t (sec) from AAS analysis; : mass concentration of -1 -3 2 ions (g.L ) at previous time (t-1); S: total active membrane surface (1.92x10 m ) involved for every metal ion transfer (three MEC or three A400); K: first order kinetic constant per membrane surface unit (sec-1.m-2). 4. Complexation Reaction (CR) step: speciation modeling The performance of electro-extraction process firstly depends on the quality of the complexation step implemented and then on the operating conditions applied in the ED-CR process. For the complexation step, the complexing behaviour of various chemical agents towards binary metallic solutions (Ag and Zn, Cu and Cd) was studied and ionic forms distribution calculations were performed to identify the most appropriate ligand to use in the hybrid ED-CR process as well as the optimal operating pH-range. The role of the complexing agent must combine the formation of negatively charged complexes (valence opposed to that of the other free metal ion, or at least, neutral one), a discriminant complexation ability (M1complex more stable than M2-complex), and a complexation reaction that can be later displaced in order to recover and concentrate the free M1 metal ion, while the free ligand being in-situ recycled. The choice of the ligand depends on the stability of the formed complexes, the degree of the complexation selectivity obtained, and the operating pH-range. Chemical speciation modeling allow us to inform those criteria by screening the effect of physico-chemical conditions (metal ion concentration, pH of the solution, metal ion competition for complexation) on the complexing ability of the ligand. The behaviour of different ligands was investigated with the copper-cadmium binary system, the most difficult system to separate due to identical metal ion valence. Once selected, the best agent was experimentally tested on both systems: copper-cadmium and silver-zinc ones. Ionic form distribution calculations were performed for citrate (H3C), glycine (H2G+), phosphate (H3P) and EDTA (H4L) ligands symbolized under their most acidic form (Table 3), at various individual initial ion and ligand concentrations (5.0x10-3, 5.0x10-4, 5.0x10-5 mol.L-1 corresponding to 10-2, 10-3, 10-4 eq.L-1 respectively for divalent species). Table 3. Presentation of the different ligands under study, written in their most acidic form. Ligand name

Chemical formula (acidic form)

Citrate Glycine

H3C H2G+

HOOC-C(OH)(CH2-COOH)2 H3N+-CH2-COOH

Phosphate

H3P

O=P(OH)3

EDTA

H4L

(HOOC-H2C)2N-C2H5-N(CH2-COOH)2

Equilibrium was assumed to be reached for all the species (both soluble and not soluble). The reference temperature was 25°C and ionic strength was selected at 0.1 mol.L-1 (pH range investigated from 2 to 12). Table 4 (in Supplementary Information Section) refers to all the thermodynamic constant values taken from the literature (ligand acidity, hydroxide metal ion and complex formation species) for the two binary systems studied 6

(Cu/Cd system, Ag/Zn system). Computation used thermodynamic data, the related mass action equations, mass and charge balances. Equations solve was made either with MapleTM mathematical software [39] or HySS program [40] (identical results from both were obtained).

5. Results and discussion 5.1. Copper-Cadmium-ligand system modeling All ligands (glycine H2G+, citrate H3C0, EDTA H4L0) except phosphate H3P0 exhibit a preferential complexation pattern towards copper compared to cadmium, and this is irrespective of the equivalent concentration applied. Octahedral copper (II) complexes are usually subject to the Jahn-Teller effect, affording a complex extra stability [43]. Regarding all the simulations and their evolution as a function of metal ion concentration and pH range modulations (see Fig. 3-5 in Supplementary Information Section), EDTA appears to have the best discriminant action for the Cu/Cd electro-separation, in terms of both metal ion charge modification and complex stability (Fig.6): A dominant anionic complex CuL2- is formed in the pH range (5.0-6.4), while Cd2+ remains mainly as free ion form. These results are identical for all metal ion and ligand concentrations investigated (from 5.0x10 -4 to 5.0x10-2 mol.L-1). It will therefore be selected for the complexation step. However, because of close values of stability constants for CuL2- and CdL2-, the Cu/Cd separation looks not complete: a 90/10 ratio mixture of CuL2/CdL2- and Cd2+/Cu2+ is expected in anodic and cathodic compartments respectively, suggesting some ion competition for selective transport in the electrodialysis process.

Fig.6. Cu/Cd/EDTA speciation. Individual metal ion concentration: 5.0x10-5 mol.L-1. Molar ratio (metal ionligand) 1:1. (Identical results are obtained at 5.0x10-4 and 5.0x10-3 mol.L-1). 5.2. Silver-Zinc-EDTA system modeling Compared to the Cu/Cd/EDTA system, an ideal selective complexation is obtained as shown in Fig.7: Zn 2+ ion is totally complexed by EDTA as a unique divalent anionic form (ZnL2-) while Ag+ ion remains totally as a free ion form, in a large pH range (5 < pH <10).

7

Fig.7. Ag/Zn/EDTA speciation. Individual metal ion concentration: 1.0x10-4 eq.L-1. Equivalent ratio (metal ionligand) 1:1 ([Zn2+]=[H2L2-]=5.0x10-5mol.L-1; [Ag+]=1.0x10-4 mol.L-1). (Identical results are obtained at 1.0x10-3 and 1.0x10-2 eq.L-1). Thermodynamic predictions at higher concentrations (1.0x10-3 and 1.0x10-2 eq.L-1) of metal ions and ligand gave similar results. For electro-extraction experiments, the selected working pH range is large but will be taken close to the neutrality (pH=7) to limit chemical consumption. The selectivity of the electro-extraction process is predicted to reach 100%. Then, for ED-CR experiments, only the best complexing agent (EDTA) was tested for the metal ion separation, under various operating conditions investigated. Results were evaluated in terms of selectivity, electro-extraction rate, and extraction kinetic characteristics.

5.3. Screening the operating conditions for ED-CR process 5.3.1. Effect of feed pH on the selective electro-extraction efficiency of Ag and Zn For the Ag/Zn/EDTA system under fixed ED-CR operating conditions, electro-extraction of Ag+ as free ion and Zn2+ as anionic complex species was studied with or without controlling the pH of the feed solution. As shown in Fig.8, when the pH is adjusted at a value close to the neutrality, really efficient extraction of both Ag + and ZnL2is obtained from the feed (%EE >95% from 60 minutes for Ag+ and 80 minutes for ZnL2- ; 100% from 120-150 minutes for each of them). When the pH is not maintained to 7.0, the pH of the diluate significantly decreases even before applying any voltage and within the first twenty minutes of the experiment (from pH 7.0 to 5.0), then keeps decreasing more slowly until pH 4.0 through the course of the experiment. The pH drop of the feed (diluate compartment) is caused by proton leakage from the acidified zinc concentrate to the feed (diluate) through the anionic A400 membrane. This would also be observed with a more dense membrane than A400. Also, at the cathodic side, a back-diffusion of protons through AEM (lack of permselectivity) occurs as mentioned in Fig.1. This feed pH change induces a slower extraction rate of both Ag+ and ZnL2- from the feed and a global lower efficiency only for Ag+ removal, even at the end of treatment (88% at free pHfeed compared to 100% at pHfeed adjusted to 7), due to competition for current transport between H+ and Ag+. It should be noted that the feed pH drop observed does not induce any change in the quantity of zinc complexed by EDTA as absolutely no zinc transfer occurs at the cathodic side. These results agree with thermodynamic predictions which showed the dominance of free Zn 2+ only at pH2 and a possible change in the complexation state of zinc at pH slightly below 4 (formation of protonated but still anionic complex ZnHL-) (Fig.7). A total selectivity of the process is obtained as neither silver nor zinc is found in anodic and cathodic concentrates, respectively.

8

Fig.8. Percentage of Ag+ and ZnL2- electro-extraction from the diluate as a function of time related to the change in pH of the feed (Experiment P1: Initial feed solution [Ag+]=[ZnL2-]=10-3 eq.L-1;Qfeed=6 L.h-1; U=10 V; recirculation mode; free evolution of pHfeed or continuously adjusted pHfeed to 7.0). As far as current efficiencies are concerned, low values for Ag+ and ZnL2- were obtained and did not exceed 20%. Preferential transport of Na+/H+ and NO3- across CEM and AEM, respectively, occurs and limits that of larger ions (free Ag+ hydrated metal ion and anionic ZnL2- metal complex). When pH of the feed is adjusted, micro-additions of NaOH did not change significantly ion transport efficiency. In the next ED-CR experiments for the Ag/Zn electro-extraction, it was decided to maintain the pH of the feed close to 7. 5.3.2. Relation between mass transfer, pH and conductivity In the case of Ag/Zn/EDTA system and under controlled feed pH of 7, the change in pH (Fig.9) and conductivity (Fig.10) in all hydraulic compartments was monitored and related to the mass transfer of Ag and Zn. In Zn concentrate compartment, a significant pH drop of initial NaNO3 solution (pHinitial=5.3) is observed at t0 and this is due to the evident proton diffusion and migration through CEM (cation exchange membrane) from electrode and/or buffer compartments (Fig.1). Accordingly, the change in conductivity in Zn concentrate has an identical, but opposite, evolution as it corresponds to both ZnL2- ions and proton inputs. For Ag concentrate, pH drops more slowly at t0 but keeps decreasing with time (related to a continuous increase in conductivity, and this is caused by the back diffusion of proton through anion exchange membrane (AEM) (Fig.1). The lack of AEM permselectivity to proton is a well-known membrane limitation. pH in electrode and buffer compartments logically increases slightly. Checking for proton mass balance was laborious as protons that transfer in the feed (diluate compartment) are continuously neutralized by microadditions of NaOH to maintain the feed pH to 7. From mass transfer efficiency point of view, we observe that fast extraction for both Ag+ and ZnL2- can be achieved, together with a total selectivity of the concentration step (absolutely no silver species observed in anodic Zn-concentrate and no zinc species observed in cathodic Ag-concentrate). It should be noted that ED experiments started with 1 L NaNO3 electrolyte flowing in concentrate compartments. This protocol allows us to track finely individual transfer of ZnL2- and Ag+ ions that occur from the feed to their receiver compartments, and also to evidence the saturation process of ion exchange membrane sites during their transport. As expected, the decrease of metal ions quantities in the feed compartment is correlated to their increase in their receiver compartment. Moreover, the increase in metal ions in the concentrate is always lower than the decrease of metal ions in the feed. This discrepancy is obviously attributed to the progressive saturation of ion exchange membrane sites (initially conditioned with Na + (CEM) or NO3- (AEM) counter-ions), during the transport of ZnL2- and Ag+. Finally, the end of this first run did therefore not lead to concentrated Zn and Ag compartments. Instead, their concentration in their respective receiver compartments became constant as the depletion of the ions from the feed was completed. By renewing the feed solution at every batch run, more concentrated streams would be obtained.

9

Fig.9. Milliequivalent of Ag (left) and Zn (right) as a function of time related to the change in pH in diluate, concentrate, electrode and buffer compartments (Experiment P1: Initial feed solution [Ag+]=[Zn2+]=[H2L2-]=10-3 eq.L-1; pH feed=7; Qfeed= 6.0 L.h-1; U=10 V; recirculation mode).

Fig.10. Milliequivalent of Ag (left) and Zn (right) as a function of time related to the change in conductivity in diluate, concentrate, electrode and buffer compartments (Experiment P1: initial feed solution [Ag+]=[Zn2+]=[H2L2-]=10-3 eq.L-1; pHfeed =7; Qfeed = 6.0 L.h-1; U=10 V ; recirculation mode).

5.3.3. Effect of voltage, feed flow rate and feed concentration on electro-extraction efficiency Fig.11 compiles the evolution of of Ag+ ion obtained according to various operating conditions applied. 2Results obtained for ZnL ion followed the same trend (results not shown here). Voltage and feed flow rate have an interactive effect with the concentration of the feed solution to be treated and significantly affect both electroextraction kinetics and electro-extraction efficiency ( ).

Fig.11. of Ag+ according to the operating conditions applied at initial feed concentrations of 10 -3 eq.L-1 -4 (left) and 10 eq.L-1 (right) (similar trend were obtained for Zn species). 10

Also, the influence of the experiments chronology has been examined together with punctual membrane integrity controls, as it can be seen in Table 5. Table 5. ED-CR Pilot experiments (P): Operating conditions, chronology and punctual membrane integrity controls. Ag/Zn (10-3 eq.L-1) 6L.h-1

10V 20V P 1

P 2

9L.h-1

5V

5V

P 3

P 4

3L.h-1

10V 20V P 5

P 6

5V P 71

10V 20V P 8

P 9

6L.h-1

3L.h-1

Cu/Cd (10-4)

Ag/Zn (10-4)

Cu/Cd (10-3)

6L.h-1

3L.h-1

6L.h-1

6L.h-1

10V

5V

5V

P 15

P 161

P 162

5V

10V 20V

0V

5V

10V

10V 10V 10V

P 72

P 101

P 12

P 13

P 141

P 142

P 11

Feed flow rate falls suddenly during the rinsing step after P7 1. Damaged AEM replaced.

Stack control after P11: Damaged AEM and CEM replaced.

P 14sp

P 102

Stack control after P141: Membranes Ok.

10V 10V 10V P 17

P 18

P 17sp

Stack control after P15: Membranes Ok.

For the highest initial feed concentration studied (Fig.11, left), increasing voltage and/or feed flow rate promote ion transfer and thereby increases metal ion extraction kinetics (comparisons of P7 2 and P8 or P72 and P3; P1 and P2; P2 and P9). The use of suitable flow rates allows the reduction of the concentration polarization phenomenon (lower thickness of the boundary layers at the membrane surfaces) and thus, ion transfer through membranes is favored [44]. However, at the highest flow rate applied (P4, P5 and P6), electro-extraction kinetic seems to be altered: Because of too short residence time, the ion-exchange reaction might become the limiting step. Attention must be paid on these results because poor performances were obtained after every 20 V experiment performed (i.e. after P2, P6, P9, P11) and irreversible damage of membranes occurred (Table 5). Actually, the highest voltage used at P2 (20 V corresponding to applying 1.6 V per membrane) could be responsible of lower extraction performances observed from P4 to P6. To refine those observations, a t-test for paired comparisons (Table 6, Supplementary Information Section) was used to test the significance of the observed differences between duplicated experiments, one performed after a 20 V experiment and the other after checking for membrane integrity and replacing any damaged membrane if needed. Results for P7 and P10 show significant differences at both 95 and 99% of confidence, between the two identical treatments Px1 and Px2 and highlight the negative effect of using extreme voltage (20V). On the contrary, replicated P16 performed without any extreme voltage applied before them, lead to identical results (at 95% and 99% confidence). At lower initial feed concentration studied (Fig.11, right), feed flow rate seems to have no significant effect on electro-extraction efficiency and kinetics, unlike voltage which still enhances ion transfer kinetics (the higher the voltage, the higher kinetics). However and as previously observed, using too high value of voltage (20V) altered membrane integrity (P11, Table 5). Finally, note that experiment performed with no voltage applied (P12) shows that Ag+ extraction occurs with simple diffusion mechanism involved (dialysis mode), but kinetics is lower than in ED mode. For P12, an increase in the feed conductivity (factor 2 within 2 hours of treatment) is observed and corresponds to Ag+ permutation with ions of higher conductivity (Na +, H+, NO3-). Dealing with Zn species, ZnL2- permutation appeared even less effective in terms of achieved and kinetics. Moreover, without any electric field applied, when ion concentrate compartment exceeds that of dilute compartment, back diffusion would severely limit both process efficiency and electro-extraction kinetics.

5.3.4. Influence of the electrolyte concentration in electro-extraction performance Fig.12 (left) and (right) shows the evolution of metal ion species in the feed and adjacent concentrate compartments, during ED-CR experiment (recirculation mode applied), at two different electrolyte concentrations used in the concentrate, buffer and electrode compartments. The overall electrical resistance of the system is the sum of membrane, solution and electrode material resistances. Results from Fig.12 evidence that the lower the electrolyte concentration (or conductivity), the higher the resistance of the system, the slower Cu/Cd electro-extraction from the feed results. Increasing electrolytes concentrations to 10-2 eq.L-1 allow more efficient Cu/Cd electro-extraction in terms of kinetics (comparison of P17 and P18 in Table 7). As already observed for Ag/Zn system, a total selectivity for the concentration step of Cu and Cd is obtained: absolutely no transfer of Cu species was observed in cathodic Cd-concentrate (100% of selectivity, Fig.12 left) whereas only trace of Cd species appeared in anodic Cu-concentrate (99% of selectivity within the time range 11

20-90 minutes, Fig.12 right). These selectivity results are higher than those expected from the thermodynamic previsions (Fig.6) in the corresponding experimental pH range investigated (pHfeed=5.4±0.4) which demonstrate that Cu-EDTA complex is kinetically favored in dynamic regime compared to Cd-EDTA complex.

Fig.12. Milliequivalent of copper (left) and cadmium (right) as a function of time, in the feed and in Cu- and Cdconcentrate compartments, at two electrolyte concentrations used: 10 -3 eq.L-1 (P18) or 10-2 eq.L-1 (P17) in concentrate, buffer and electrode compartments. (Experiments P17 and P18: initial feed solution [Cd2+]=[Cu2+]=[H2L2-]=10-3 eq.L-1; pHfeed =5.4; Qfeed = 6 L.h-1; U=10 V ; recirculation mode).

5.3.5. Electro-extraction kinetics parameters of Ag/Zn/EDTA and Cu/Cd/EDTA systems The decay of metal ion amount in the feed with time suggests that the overall extraction process (through migration and diffusion transport) is governed by a first order kinetic (Eq.9). Determination coefficients R2 for the first order equations and electro-extraction kinetic constants K could then be deduced and are presented in Table 7 for the most relevant experiments: Whatever the concentration of metal ion and electrolyte investigated, the observed K values look higher for free metal ion than for complexed metal ion and this is consistent with evident higher mobility values of free ions compared to complexed one. However, with regard to experimental error values (estimated to 15% from time and concentration uncertainties), no significant difference could be highlighted. Also, electro-extraction kinetics of Cu/Cd and Ag/Zn systems look very similar. As far as ion fluxes are concerned, differences observed between diluate and concentrate are attributed to metal ion sorption (as free or complexed form) in ion exchange membranes during their transport. Flux values obtained depend on metal ion concentration in the feed and their diffusivity in the ion-exchange membranes, and consequently, on their relative affinity of the exchanger [31]. Note that the flux of free Cd2+ transferred in its concentrate compartment is very low compared to the free Ag+ flux and could be explained by its much higher interaction with ionexchanging sites of CEM than Ag+. This well agrees with the general affinity series observed for sulfonate strong acidic cation exchange materials (Cd2+ > Cu2+ > Zn2+ >>Ag+) [44]. Table 7. Rate and kinetics parameter results for the electro-extraction of Ag/Zn/EDTA and Cu/Cd/EDTA systems, as a function of electrolyte concentrations used in the ED-CR process. Concentrations of metal ions and electrolytes

P

Ion

Time at 95% EE

K

(min)

(sec-1.m-2)

Ion flux extracted from the feed

Ion flux transferred in concentrate

Energy consumption at 95% EE

(mol.sec-1.m-2)

(mol.sec-1.m-2)

(kWh.m-3 of feed)

Cd

105

0.024 ± 0.004

0.998

5.5x10-6

<10-9

CuL2-

150

0.014 ± 0.002

0.988

5.9x10-6

1.9x10-6

Cd

60

0.046 ± 0.007

0.985

4.6x10-6

7.8x10-7

CuL2-

70

0.042 ± 0.006

0.996

6.7x10-6

4.9x10-6

Ag

60

0.042 ± 0.006

0.999

1.8x10-5

1.2x10-5

ZnL2-

80

0.032 ± 0.005

0.999

1.0x10-5

6.8x10-6

Cd

100

0.032 ± 0.005

0.998

6.8x10-7

8.2x10-9

CuL2-

100

0.030 ± 0.005

0.994

7.2x10-7

6.2x10-7

Ag

90

0.039 ± 0.006

0.988

1.4x10-6

5.2x10-7

0.982

-7

4.2x10-7

(eq.L-1)

Initial metal ion (10-3) -3

Electrolyte (10 )

Initial metal ion (10-3)

P18

P17

Electrolyte (10-2) P1

Initial metal ion (10-4)

P15

Electrolyte (10-2) P141

R2

2-

ZnL

100

0.034 ± 0.005

8.1x10

0.28

0.54

0.55

0.30

0.42

Optimal operating conditions could be found for both Cu/Cd and Ag/Zn systems that led to a metal ion electroextraction efficiency %EE ≥ 99%, with fast kinetics and highly selective recovery ≥ 99% in their respective concentrate compartments (i.e. P1 or P17, P14 or P15). 12

From results in Table 7, the energy consumption (EC) depends on the initial feed concentration and also on electrolyte concentration in electrode and buffer compartments. The energy needed to treat 1 m3 of feed solution increases with increasing metal ion feed concentration (as an expected consequence of higher current density involved) or decreasing electrolyte concentration (due to a higher resistance of the solutions) [13, 45, 46]. Under similar feed concentration and voltage applied, our EC values are in the range of that obtained in previous work for the electro-extraction of Ag [46] (EC < 1 kWh.m-3) but look higher than that obtained for Pb removal [45] (EC < 0.1 kWh.m-3). Note that the stack design and both membrane area and type were different.

5.3.6. Influence of hydraulic feed mode (batch or single pass) applied Fig.13 shows the effect of the hydraulic feed mode applied (single pass or batch) on the electro-extraction process, experimented on the Ag/Zn/EDTA (P14sp) and Cu/Cd/EDTA (P17sp) systems. It can be observed that the electro-extraction efficiency of the ED-CR process strongly depends on the hydraulic feed mode applied: The results reveal that the batch operating mode allows increasing the residence time of metal ions in the feed which in turn promotes their selective transfer in the corresponding concentrate compartments [5, 9, 45]. Therefore, batch mode gives better results in terms of electro-extraction efficiency and concentration step.

Fig.13. Effect of the hydraulic feed mode applied on the % of electro-extraction of metal ions from the feed solution: (left) Ag/Zn/EDTA system (P14sp); (right) Cd/Cu/EDTA system (P17sp). P14sp and P17sp operating conditions identical to P141 and P17 respectively, except hydraulic feed mode. In order to enhance performances of single pass mode, the insertion of bipolar ion exchange materials (ionic promoters) in the feed compartment would both promote ionic transport (by decreasing the feed compartment resistivity during demineralization process) and would reduce metal ions leakage in the outlet, thus promoting higher electro-extraction efficiency [8, 24, 29, 31]. Under these conditions, the single pass mode would become a very attractive process, compared to batch mode, as it would provide constant quality of the demineralized outlet while promoting metal ion transfer in the concentrate compartments.

6. Conclusion In this study, a hybrid electrodialysis technique named as ED-CR process, which couples ion-exchange and insitu complexation reaction (CR) step, was investigated with the view to selectively recover metal ions from a mixture solution. For the complexation step investigation, the complexing behaviour of various chemical ligands (citrate (H3C), glycine (H2G+), phosphate (H3P) and EDTA (H4L), towards binary metallic solutions (Ag/Zn and Cu/Cd) was investigated through thermodynamic speciation modeling. The best complexing agent was found to be EDTA for both metal ion systems, in terms of discriminant metal ion charge modification and complex stability. This ligand was therefore selected for electro-extraction experiments. Moreover, to promote the transport of large size complexed species while avoiding membrane fouling, a special anion exchange membrane (PC 400D from PCA GmbH Company) was used. The results obtained showed that the process efficiency (i.e. electro-extraction rate (%EE) and kinetics), was significantly affected by the operating conditions applied (initial feed concentration and flow rate, feed pH, electric voltage applied, electrolyte concentration in electrode and buffer compartments). Under given operating conditions, excellent electro-extraction (≥99%) together with a total selective metal ion recovery could be achieved for both Ag/Zn and Cu/Cd systems: If low feed flow rate is selected, one should increase the applied voltage to get similar results as those obtained at higher flow rate and lower voltage applied; sufficient conductivity of the electrolyte concentration should be selected in electrode and buffer compartments; maximal voltage of 1V per membrane should be applied to the membrane stack in order to prevent membrane damage. 13

This original hybrid technology would make possible to obtain highly desalinated solutions with final metal ion concentration below than 0.3 mg.L-1 for the four ions studied at 90 minutes of treatment, and no detectable anymore after 120-150 minutes. The quality of the treated water obtained could make possible its recycle and reuse in some industrial processes. Moreover, during the process, the simultaneous acidification of the complexed metal ion concentrate (anodic compartment) allows to dissociate drastically the metal ion-EDTA complex (pHconcentrate < 2.3 at 60 minutes for Zn or Cu), with regard to thermodynamic speciation prevision. After ED-CR treatment, the complete ligand recovery (and thereby the second metal ion) would be easily achieved using bipolar electrodialysis (EDBM), as previously investigated [6] on a copper-complex system. Therefore, ligand and metals could be completely recycled within the treatment system and the production process respectively. Future studies will be devoted to concentration step investigation by applying multiple ED-CR treatments with regularly renewed feed solution, as well as ligand recovery efficiency by EDBM. References [1] C.J. Williams, D. Aderhold, G.J. Edyvean, Comparison between biosorbents for the removal of metal ions from aqueous solutions. Water Research 32 (1998) 216–224. [2] J. Pavlovié, S. Stopié, B. Friedrich, Z. Kamberovié, Selective removal of heavy metals from metal-bearing wastewater in a cascade line reactor. Environment Science Pollution Research 14 (7) (2007) 518-522. [3] T. Mohammadi, A. Razmi, M. Sadrzadeh, Effect of operating parameters on Pb2+ separation from wastewater using electrodialysis. Desalination 167 (2004) 379-385. [4] J-F. Blais, S. Dufresne, G. Mercier, State of the art of technologies for metal removal from industrial effluents. Revue des Sciences de l’Eau 12(4) (1999) 687-711. [5] M. Sadrzadeh, A. Razmi, T. Mohammadi, Separation of different ions from wastewater at various operating conditions using electrodialysis. Separation and Purification Technology 54 (2007) 147-156. [6] B. Schlichter, V. Mavrov, T. Erwe, H. Chemiel, Regeneration of bonding agents loaded with heavy metals by electrodialysis with bipolar membranes. Journal of Membrane Science 232 (2004) 99-105. [7] R. K. Nagarale, G. S. Gohil, V. K. Shahi, Recent developments on ion-exchange membranes and electromembranes processes. Advances in colloid and interface science 119 (2006) 97-130. [8] S. Ezzahar, A. T. Cherif, J. Sandeux, R. Sandeux, C. Gavach, Continuous electropermutation with ionexchange textiles. Desalination 104 (1996) 227-233. [9] A. J. Chaudhary, J. D. Donaldson, S. M. Grimes, N. G. Yasri, Separation of nichel from cobalt using electrodialysis in the presence of EDTA. Journal of applied Electrochemestry 30 (2000) 439-445. [10] V. J. Violleau, Determination by electrodialysis in the presence of a complexing agent: Application to lactoserum. Doctoral thesis, N°1627, INP de Toulouse, France, 1999. [11] C.Vallois, P. Sistat, S. Roualdès, G. Pourcelly, Separation of H+/Cu2+ cations by electrodialysis using modified proton conducting membranes. Journal of Membrane science 216 (2003) 13-25. [12] Y. Zhang, B. Van der Bruggen, L. Pinoy, B. Meesschaert, Separation of nutrient ions and organic compounds from salts in RO concentrates by standard and monovalent selective ion-exchange membranes used in electrodialysis. Journal of Membrane Science 332 (2009) 104-112. [13] H. Strathmann, Electrodialysis, a mature technology with a multitude of new applications. Desalination 264 (2010) 268-288. [14]S. Mulyati, R. Takagi, A. Fujii, Y. Ohmukai, T. Maruyama, H. Matsuyama, Improvement of the antifouling potential of an anion exchange membrane by surface modification with a polyelectrolyte for an electrodialysis process. Journal of Membrane Science 417-418 (2012) 137-143. [15] T. Chakrabarty, B. Shah, N. Srivastava, V. K. Shahi, U. Chudasama, Zirconium tri-ethylene tetra-amine ligand-chelator complex based cross-linked membrane for selective recovery of Cu2+ by electrodialysis. Journal of Membrane Science 428 (2013) 462-469. [16] M. Vaselbehagh, H. Karkhanechi, R. Takagi, H. Matsuyama, Surface modification of an anion exchange membrane to improve the selectivity for monovalent anions in electrodialysis - experimental verification of theoretical predictions. Journal of Membrane Science 490 (2015) 301-310. [17] V. A. Shaposhnik, N. N. Zubets, I. P. Strygina, B. E. Mill, High demineralization of drinking water by electrodialysis without scaling on the membranes. Desalination 145 (2002) 329-332. [18] J. F. Poulin, J. Amiot, L. Bazinet, Improved peptide fractionation by electrodialysis with ultrafiltration membrane: Influence of ultrafiltration membrane stacking and electrical field strength. Journal of Membrane Science 299 (2007) 83-90. [19] L. Firdaous, P. Dhulster, J. Amiot, A. Gaudreau, D. Lecouturier, R. Kapel, F. Lutin, L.P. Vezina, L. Bazinet, Concentration and selective separation of bioactive peptides from an alfalfa white protein hydrolysate by electrodialysis with ultrafiltration membranes. Journal of Membrane Science 329 (2009) 60-67. [20] L. Bazinet, M. Moalic, Coupling of porous filtration and ion-exchange membranes in an electrodialysis stack and impact on cation selectivity: A novel approach for sea water demineralization and the production of physiological water. Desalination 277 (2011) 356-363. [21] L. Ge, B. Wu, Q. Li, Y. Wang, D. Yu, L. Wu, J. Pan, J. Miao, T. Xu, Electrodialysis with nanofiltration membrane (EDNF) for high-efficiency cations fractionation. Chemical Engineering Journal 498 (2016) 192-200. 14

[22] V. Mavrov, H. Chmiel, B. Heitele, F. Rögener, Desalination of surface water to industrial water with lower impact on the environment, Part 2: Improved feed water pretreatment. Desalination 110 (1997) 65-73. [23] V. Mavrov, H. Chmiel, B. Heitele, F. Rogener, Desalination of surface water to industrial water with lower impact on the environment, Part 4: Treatment of effluents from water desalination stages for reuse and balance of the new technological concept for water desalination. Desalination 124 (1999) 205-216. [24] V. D. Grebenyuk, R. D. Chebotareva, N. A. Linkov, V. M. Linkov, Electromembrane extraction of Zn from Na-containing solutions using hybrid electrodialysis-ion exchange method. Desalination 115 (1998) 255-263. [25] A. Bouchoux, H. Roux-de Balmann, F. Lutin, Investigation of nanofiltration as a purification step for lactic acid production processes based on conventional and bipolar electrodialysis operations. Separation and Purification Technology 52 (2006) 266-273. [26] O. Souilah, D. E. Akretche, M. Amara, Water reuse of an industrial effluent by means of electrodeionisation. Desalination 167 (2004) 49-54. [27] A. H, Galama, G. Daubaras, O. S. Burheim, H. H. M. Rijnaarts, J. W. Post, Seawater electrodialysis with preferential removal of divalent ions. Journal of Membrane Science 452 (2014) 219-228. [28] A.T. Cherif, A. EL-Midaoui, C. Gavach, Separation of Ag+, Zn2+ and Cu2+ ions by electrodialysis with monovalent cation specific membrane and EDTA. Journal of Membrane Science 76 (1993) 39-49. [29] S. Lacour, J. Sandeaux, Enhancing mass transfer in multi-electroextraction technique using ion exchange grafted textiles. Desalination 199 (2006) 57-58. [30] A. Iizuka, Y. Yamashita, H. Nagasawa, A. Yamasaki, Y. Yanagisawa, Separation of lithium and cobalt from waste lithium-ion batteries via bipolar membrane electrodialysis coupled with chelation. Separation and Purification Technology 113 (2013) 33-41. [31] S. Abdelaziz , R. Delimi, E. Chainet, J. Sandeaux, Removal of heavy metals from diluted mixtures by a hybrid ion-exchange/electrodialysis process. Separation and Purification Technology 57 (2007) 103–110. [32] T. C. Huang, J.K. Wang, Selective transport of metal ions through cation exchange membrane in the presence of a complexing agent. Industry and Engineering Chemistry Research 32 (1993) 133-139. [33] H. C. Hershey, R. D Mitchell, W. H. Webb, Separation of cesium and strontium by electrodialysis. Journal of Inorganic Nucleid Chemistry 28 (1966) 645-649. [34] T. Scarazzato, D. C. Buzzi, A. M. Bernardes, D. C. R. Espinosa, Treatment of wastewaters from cyanidefree plating process by electrodialysis. Journal of Cleaner Production 91(2015) 241-250. [35] F. Aouad, A. Lindheimer, M. Chaouki, C. Gavach, Loss of permselectivity of anion exchange membranes in contact with zinc chloride complexes. Desalination 121(1999) 13-22. [36] M. A. S. Rodrigues, C. Korzenovski, E. Gondran, A. M. Bernardes, J. Z. Ferreira, Evaluation of changes on ion-selective membranes in contact with zinc-cyanide complexes. Journal of Membrane Science 279 (2006) 140147. [37] E. Husson, M. Araya-Farias, A. Gagné, L. Bazinet, Selective anthocyanins enrichment of cranberry juice by electrodialysis with filtration membrane: Influence of membranes characteristics. Journal of Membrane Science 448 (2013) 114-124. [38] http://www.pca-gmbh.com/membrane/membrane.htm (last online access: 09/14/2016). [39] MapleTM mathematical software: A division of Waterloo Maple Inc (2001-2008) Maple Getting Started Guide. Copyright © Maplesoft. [40] HySS version 4.0.31 protonic software : http://www.hyperquad.co.uk/ (last online access: 04/14/2016). [41] L. D. Pettit, K. J. Powell, Stability Constants Database, Academic Software: Yorks, United Kingdom, 2000. http://www.acadsoft.co.uk/ (last online access: 04/14/2016). [42] S. Lacour, V. Deluchat, B. Serpaud, J. C. Bollinger, Influence of carbonate and calcium ions on the phosphonate complexation with Cu, Zn, Cd and Ni in fresh waters: An evaluation of thermodynamic constants and a chemical model. Environmental Technology 20 (1999) 249-257. [43] M. Gruden-Pavlovic, M. Zlatar, C. W. Schläpfer, C. Daul, DFT study of the Jahn-Teller effect in Cu(II) chelate complexes. Journal of Molecular Structure: Theochem 954 (2010) 80-85. [44] Z. Hubicki, D. Kołodyńska, Selective removal of heavy metal ions from waters and wastewaters using ion exchange methods. In Ion Exchange Technologies’, Chapter 8, A. Kilislio lu Ed., ISBN 978-953-51-0836-8, November 7, 2012. [45] C.V. Gherasim, J. Krivcík., P. Mikulasek, Investigation of batch electrodialysis process for removal of lead ions from aqueous solutions. Chemical Engineering Journal 256 (2014) 324-334. [46] A. Güvenç, B. Karabacako lu, Use of electrodialysis to remove silver ions from model solutions and wastewater. Desalination 172 (2005) 7-17.

15

Supplementary Information Section

16

Table 4. Overall thermodynamic constant values for ligand acidity, hydroxide-metal ion and metal-ligand complex formation.

OH- species (symbolized as H-1-) Species

Log βa

EDTA species (L4-)

Citrate species (C3-)

Species

Log β

Species

Log β

Glycine species (G-) Species

Log β

Phosphate species (P3-) Log β

Species

H-1-

-13.78*

HL3-

10.22

HC2-

5.69

HG0

9.60c

HP2-

11.68

CuH-1+

-7.68

H2L2-

16.38

H2C-

10

H2 G+

12.0 c

H2P-

18.43

CuH-20

-12.98

H3L-

19.08

H3C0

12.92

CuG+

8.20 c

H3P0

-26.84

0

21.08

-

0

15.1

c

11.0

c

CdG+

4.31 c

0

7.86

c

10.7

c

CuH-3 CuH-4

-

-2

CuH-2(s) CdH-1

+

CdH-2

0

CdH-3 Cd2H-1

-

H4 L

2-

4.53 0

-39.52

CuL

19.19

CuHC

9.9

CuHL-

22.47

CuH2C+

0

-10.38

CuH2L

-20.56

2-

CdL

-

-32.54

3+

CuC

CdHL

0

+

24.46

Cu2C

16.54

Cu2C22-

19.47

9.31 11.58b 8.1

2-

CuH-1C -

14.72 1.61

-9.18

CdH2L

21.1

CdC

3.65

Cd4H-44+

-32.32

AgL3-

7.3

CdHC0

7.8

AgH-10

-12.0

AgHL2-

13.7

CdH2C+

11.22

AgH-2-

-23.78

AgH2L-

18.2

CdC24-

5.3

ZnH-1+

-9.18

Ag2L

7.6

ZnH-20

-17.96

ZnL2-

16.52

ZnH-3-

-28.04

ZnHL-

19.22

2-

-40.02

ZnH2L0

20.6

ZnH-4

Zn2H-13+

2-

2-

CdH-1C

CuCdH-2C24-

-3.81 0.33d

CuG2 CuHG

CdG2 CdHG

2+

2+

20.23 0

12.01

+

CuH2P

19.63

CdHP0

14.47

CuHP

+

CdH2P

20.67

All the data except for hydroxide (H-1) speciesa were selected from SC-database 5.4 [41], at t=25 °C and I=0.1 mol.L -1 unless otherwise specifiedbcd. *. Value of ionic product of water (Log Ke). a. hydroxide formation constants [42]. b. value determined or estimated at 20 °C. c. values determined or estimated at 0.09 mol.L-1 and 25 °C. d. values determined at 0.2 mol.L-1 and 25 °C.

-8.28

Table 6. t-test results for paired comparisons to test the significance of the observed differences between replicated experiments P7, P10 and P16 (see Eq.(8)). Metal ion P71 & P72 P101 & P102 P161 & P162

Sd

n

tcalc at (n-1) d.f

Ag+

11.44

3.02

7

10.04

ZnL2-

14,62

3.60

7

10.73

Ag+

5.95

4.14

8

4.06

ZnL2-

11.15

6.02

8

5.24

Ag+

2.44

2.40

5

2.12

ZnL2-

2.44

2.56

5

2.13

t95

t99

Any difference?

YES at 95 and 99% 1.94

3.14

1.89

3.00

2.13

3.75

YES at 95 and 99% YES at 95 and 99% YES at 95 and 99% NO at 95 and 99% NO at 95 and 99%

17

Fig.3. Cu/Cd/Glycine speciation. Individual metal ion concentration: 5.0x10-5 mol.L-1. Molar ratio (metal ionligand) 1:1. (Similar results were obtained at 5.0x10-4 and 5.0x10-3 mol.L-1, with a precipitation of Cu(OH)2 species taking place at C0(Cu)> 0.00336 mol.L-1=0.00168 eq.L-1).

Fig.4. Cu/Cd/Citrate speciation. Individual metal ion concentration: 5.0x10-5 mol.L-1. Molar ratio (metal ionligand) 1:1.

Glycine ligand (H2G+) Glycine is the smallest amino-acid with wellknown amphoteric properties (the HG0 form). As shown in Fig.3, copper complexation with glycine does not lead to the formation of any anionic complex in the whole pH-range, but rather positive CuG+ or neutral CuG20 ones. The formation of the dominant neutral complex CuG2°, while Cd2+ remains in solution as free ion form, would make possible its use for the complexation step in the ED-CR treatment. However, the corresponding pH range (7.4-8.6) is located at slightly basic pH, for which the formation of poorly soluble hydroxide species occurs (Cu(OH)20) with other interfering reactions which would complexify the electroseparation process (i.e. formation of competitive cationic species CuG+ and CdG+ for current transport, with regard to Cd2+).

Citrate ligand (H3C0) Citric acid is considered as a triprotic species (three carboxylic groups) despite the possible displacement of the fourth proton from the hydroxyl group that would occur in highly basic solutions. From Fig.4, at low metal ion concentration (5.0x10-5 mol.L-1), predictions show a discriminant complexing behaviour towards copper compared to cadmium, with the formation of a dominant anionic complex CuH2in the pH range (5.4-7.6), while cadmium 1C remains mainly as free metal ion form Cd2+. Citrate ligand would be an interesting candidate to change copper charge compared to cadmium. However, attention must be paid when dealing with the treatment of more concentrated solutions (5.0x10-4 or 5.0x10-3 mol.L-1 investigated) because of the formation of polynuclear molecules (i.e CuCdH-2C24-, Cu2H-2C24-, Cd2H42C2 ) with large size and low electrophoretic mobilities which may alter the electro-extraction efficiency of the technique (membrane fouling, poor faradic efficiency Rf (Eqs.3 or 4)).

Phosphate ligand (H3P0) The phosphate ligand has almost no complexing ability to any affect the speciation of copper and cadmium species (Fig.5). At higher metal ion concentration (5.0x10-4 or 5.0x10-3 mol.L-1), cadmium complexes are preferentially formed compared to copper ones but this leads, in the whole pH range, to a mixture solution of free Cu2+ and Cd2+ ions, protonated phosphate species, with the formation of the sparingly soluble hydroxide copper at pH >5. Fig.5. Cu/Cd/Phosphate speciation. Individual metal ion concentration: 5.0x10-5 mol.L-1. Molar ratio (metal ionligand) 1:1. 18

HIGHLIGHTS listing

H1: Original coupling of electrodialysis and complexation reaction for large ion transfer H2: Highly selective extraction of metal ions from a binary mixture solution H3: Relationship between performance, operating condition and experiment chronology

19