Proceedings of the 13 th International Conference on Environmental Science and Technology Athens, Greece, 5-7 September 2013 A NEW ENVIRONMENTALLY FRIENDLY PROCESS FOR REMOVING HEAVY METALS FROM WASTEWATER C. FERREIRA ESMI *, L. SCHRIVE *, Y. BARRE *, J. PALMERI **, A. DERATANI *** * CEA, DEN, DTCD/SPDE/LPSD - Marcoule, F-30207 Bagnols-sur-Cèze, France. caue.ferreira-esmi@cea.fr; ** Laboratoire Charles Coulomb, Université Montpellier 2, 34095 Montpellier cedex 05, France. *** Institut Européen des Membranes, Université Montpellier 2, 34095 Montpellier cedex 05, France. EXTENDED ABSTRACT Developing new sustainable process is a key issue in wastewater treatment. We present an innovative process constituted by the association of nanofiltration (NF) and a new electrochemical controlled ion complexation (ECIC) technique. NF is a cross-flow membrane separation technique known for high rejection of divalent ions (>96%). ECIC would use chelating polymers to capture low concentration metal cations coming from the NF filtrate. Elution of these cations is made by an imposed electric current. Concentrated eluted solution could then be recycled in NF stage. No secondary effluents would be generated as final waste is only constituted by nanofiltration retentate solution. In this work we investigate the application of the above described process in the treatment of a solution representing wastewater from a nuclear waste treatment facility. Such solution is composed by Co 2+ (380 mg.l -1 ) and Ni 2+ (2 mg.l -1 ) cations in a complex concentrated multi-element matrix Nanofiltration step was successfully studied by the combination of very few experiments associated to a simulation software containing commercial membranes database (Nanoflux). The complexity of the waste solution required complementary studies in order to improve software predictions which were then used to determine the best operational conditions for NF. Ionic rejections could be predicted as a function of permeate flux, transmembrane pressure and volume concentration factor (n, ratio of retentate volume by waste treated solution). 98% of divalent rejection was achieved for n = 5 with Dow s NF-90 membrane. For the ECIC technique, the pertinence of applying a high surface carbon felt grafted with polyacrylic acid (PAA) as a treatment matrix was studied. Complexation by grafted PAA was observed to be selective towards divalent cations in solutions of composition similar to NF permeate. A complexing capacity (Q eq) of 10 mg of divalent cations per gram of carbon felt was obtained. High partition coefficients (K d > 250) were achieved in NF permeate solution. Regeneration of carbon felts was carried out by in situ protons production coming from solution electrolysis. Up to 65% of the complexed ions were eluted when a current of 1 ma was applied to the system. Operational conditions are being optimized in order to attain higher rates. No losses in Q eq or elution rates were noticed after several complexation and elution cycles. The association of NF and the new ECIC technique in a coupled process can be an efficient way to treat wastewater contaminated with Co 2+ and Ni 2+ cations. KEYWORDS: sustainable process, heavy metals removal, Ni 2+, Co 2+, wastewater, nuclear effluent, nanofiltration, water electrolysis
1. INTRODUCTION In recent years, stricter regulation motivated by healthy concerns has increased the research in water treatment field [1]. Different techniques such as chemical precipitation and ion exchange might be used to treat wastewater containing heavy metals. Each of them presents their own advantages and limitations. Chemical precipitation, for example, is extensively used because of its simple and inexpensive operation. However, the large amounts of sludge generated may be a problem. Ion exchange, on the other hand, does not present any sludge disposal problems but it can only be applied to less concentrated solutions [2]. A well-designed association of different treatment techniques would make it possible to create processes requiring fewer chemical and operational resources. In this paper we present a new environmentally friendly process composed by the combination of nanofiltration (NF) and the electrochemical controlled ion complexation (ECIC) technique. NF is a cross-flow membrane separation technique known for higher rejections of divalent ions compared monovalent ones. ECIC uses Polyacrylic acid (PAA) grafted into high surface carbon felts electrodes to capture heavy metal cations present in wastewater at low concentration. An electrical dissociation of water molecules at the interface polymer/solution creates a localised acid medium which is then used to regenerate the system [3]. In this process (presented at figure 1), ECIC is used as a post treatment technique to NF filtrate. Concentrated eluted solution would be recycled in NF stage and no secondary effluents would be generated. Final waste is only constituted by nanofiltration retentate solution. The application of the process NF + ECIC in the treatment of Co 2+ (380 mg.l -1 ) and Ni 2+ (2 mg.l -1 ) cations present in a multi-element matrix representing wastewater from a nuclear waste treatment facility will be described in the next sections. Special attention will be given to computer calculated NF rejection predictions. Figure 1. Schematic illustration of the NF + ECIC process. 2. THEORETICAL BACKGROUND 2.1. Nanofiltration The extended hindered electro-transport (EHET) theory is used to calculate the rejection of the wastewater ionic components. In this theory, ionic separation at the interface membrane/solution is a combination of steric blocking, Donnan (electric) forces and dielectric exclusion forces. A detailed description of the model can be found in reference [4]. Nanofiltration simulation software Nanoflux is used to solve EHET theory equations. These calculations require the use of the following five experimental parameters: initial ionic feed concentration (C if, mol.l -1 ), effective membrane thickness (L eff, nm), pure water hydraulic permeability (Lp 0, L.h-1.m - ².bar -1 ), membrane pore radius (r p, nm) and membrane charge density (X i), equivalent to a membrane charge (X m, mol.l -1 ) normalized by the solution ionic force (mol.l -1 ) [4], [5]. 2.2. Electrochemical controlled ion complexation Polyacrylic acid (PAA) is known for the formation of metal complexes with divalent metals [6]. The K d approach [7] will be used to characterize the application of PAA-grafted
carbon felts in a solution containing Co 2+ and Ni 2+. The partition coefficient K d represents the distribution of one ion between an aqueous and a solid phase. It is calculated by: K d Q f eq i eq (ml.g -1 ) C eq ( C C m. C eq ). V (1) Were Q eq (mg.g -1 ) is the amount of metal present on the solid at equilibrium, C eq (mg.l -1 ) is the amount of metal remaining in solution at equilibrium, V is the treated volume solution (L) and m is the mass of the grafted carbon felt (g). Metal (M 2+ ) uptake by PAA (RCOOH) is described by the following equation: 2 RCOOH M ( RCOO) 2 M 2 H (2) Once reaction 2 is in equilibrium, the polymer can be regenerated by protolysis. In the ECIC, the required proton arises from water electrolysis [3]: 0 2 H 2 O( l) O2( g) 4H( aq) 4e E red 1. 23V (3) 3. MATERIALS AND METHODS 3.1. Nanofiltration step The two main objectives of this experimental step are: Reduce the volume of a solution representing wastewater from a nuclear waste treatment facility by using nanofiltration in a batch operation mode. Such solution is low concentrated in Co 2+ (380 mg.l -1 ) and Ni 2+ (2 mg.l -1 ), but has a high concentration in Na + (19.6 g.l -1 - ), NO 3 (47.7 g.l -1 2- ) and SO 4 (2.3 g.l -1 ). Simulate the above described operation by using experimental data from a NaNO 3 solution filtration. Using simulation models significantly facilitates the study of ion rejections in nanofiltration. However, almost every model presents a certain number of experimental adjusted parameters. Obtaining such parameters in a multi-element solution like the target wastewater would require multiple experiments. A procedure explained in reference [5] has shown that it is possible to predict ionic rejections of a multi-element complex solution in a constant volume nanofiltration experiment by using only experimental data from a NaNO 3 filtration. This method will be extended to the batch wastewater nanofiltration experiment whose set-up is schematized in figure 2-a. Figure 2. Experimental set up: a) batch nanofiltration; b) ECIC.
The batch experiment described above was simulated by the method described as it follows. First of all, Nanoflux is used to calculate approximate ionic rejections (R, %) of the solution as a function of its ph. After choosing a convenient ph value, constant volume nanofiltration of a simple NaNO 3 solution with a similar concentration to the target solution is made. Values of ionic rejections as a function of permeate flux (J v) are used to adjust X i and L eff so the difference between experimental and simulation points is minimized. Since Nanoflux cannot be used to directly calculate a batch operation, simulation of the wastewater nanofiltration experiment has to be divided in a series of step-like constant volume experiments. For each of them, the same NaNO 3 adjusted X i and L eff are used to calculate membrane solution permeability (L p, related to J v by transmembrane pressure, P) and then ionic rejections. Both R and J v are considered to be constant during each step and are used in mass/volume balance equations [8] to calculate feed volume, ionic feed concentration, permeate tank concentration and permeate tank volume of the next step. Results are expressed as a function of volume concentration factor (n, ratio between initial feed volume and retentate volume) and compared to experimental results. A total of 36 steps were used in the simulation for a n equal to 5. All nanofiltration experiments were carried out on a SEPA CF II (GE Water & Process) filtration cell, equipped with one shim piece of 0.64 mm and one feed spacer of 0.79 mm. A NF-90 (Dow filmtec) membrane was used. Active area is 138 cm². Pure water (15 M) hydraulic permeability (L p0 ) is 6.3 L.h -1.m - ².bar -1. For all nanofiltration experiments, feed solution was pumped at a constant rate of 250 L.h -1. P is adjusted when performing NaNO 3 filtration experiments and solution s ph is 5.8. For batch experiments, 10.5 L of a synthetic solution simulating the wastewater was prepared. P is fixed at 27.5 bar and ph is 6.2. The concentration of the feed solution and the produced permeate solution are measured from 50 ml of samples collected at concentration factors of 1; 1.8; 3.4; 4.4 and 5. Volume of treated solution is followed by measuring permeate tank mass. Temperature of the feed solution was regulated at 25.0±0.5 C. Concentrations of Na +, NO 3-, SO 4 2- and NO 2 - were measured by ionic chromatography. Ni 2+ and Co 2+ concentrations were determined by ICP-MS. 3.2. Electrochemical controlled ion complexation step The objective of this step is to evaluate the grafted carbon felts performance in the uptake of Co 2+ and Ni 2+ from the solution stored at the NF permeate tank after the batch experiment (n is equal to 5). K d is calculated from equation 1. Two cycles of complexation/acidic elution were carried out. The study of the electrochemical regeneration promoted by water electrolysis was performed on a Co 2+ saturated carbon felt. Two cycles of complexation and electrical elution were effectuated. The constant current chronopotentiometry was employed for such reaction. A solution composed by 5.10-3 M Co(NO 3) 2 and 10-3 M Na 2SO 4 was used for both complexation and regeneration. Elution is calculated from the ratio between the mass of Co 2+ added to the solution after the electrolysis and Q eq. Both of complexation and electrochemical regeneration studies were effectuated in the electrochemical pilot unit presented in figure 2-b. Such device can be seen as an electrolytic cell in which the anode is composed by the grafted carbon felt and the cathode is composed by a platinised titanium grid. Because reference electrode is electrically connected to the counter electrode, measured electric potential (ΔE, V) values cannot be directly compared to standard reduction potentials (Ered 0 ). ACTITEX FC 1201 PAA-grafted carbon felts (6.9 g from Pegastech ) were used. Measured BET-surface area is 590 m².g -1. Total volume of electrolytic cell is 460 ml. In the NF permeate treatment experiments, 7.9 L of solution was treated in a recycling mode with a 5 L.h -1
flow for four hours. Carbon felts were regenerated during 3 hours with 2 L of a HNO3 acid solution (ph 3) in a 3 L.h -1 recycling mode and during one hour with other 2 L of ph 3 HNO3 in a column mode flow. Acid regenerated felts were then rinsed with water until a ph value of 6 was obtained. Saturation of grafted carbon felts was made by filtering during 30 minutes a solution containing 5.10-3 M of Co(NO3)2 and 10-3 M of Na2SO4 at a 25 L.h -1 recirculation flow. Electric regeneration is made at the same solution by the application of a constant 1 ma electric current during 20 min using an EC-Lab (Bio-logic) potentiostat. 4. RESULTS 4.1. Nanofiltration Step Ionic rejections in NF are ph dependent. This parameter is important because it is related to membrane charge formation and its electrical interaction with ions in solution. For a given ph value, known as isoelectric point (IEP), the membrane presents no charge. A reduction in ionic rejections is then observed because steric blocking will be the main mechanism of the ionic separation. Results of simulated ionic rejections as a function of wastewater ph prior to any nanofiltration experiment are presented in figure 3: low rejections performances are clearly represented around IEP for ph 6.6. Ions are highly retained at other points. Figure 3. Simulated rejection as a function of wastewater ph. Membrane Dow NF-90. Nanoflux database : r p = 0.42 nm and L p 0 = 5.3 L.h-1.m - ².bar -1. These high ionic rejections presented by the membrane NF-90 might be disadvantageous for a solution concentrated in monovalent ions such as the target wastewater. In nanofiltration, retained ions create an important ionic concentration polarization layer on the membrane surface. This layer acts as a mass transfer resistance and is responsible for an important drop in J v. The wastewater natural ph value of 6.2 was kept for the batch experiments. At this ph membrane still presents an electrical charge and divalent ionic rejections are near its maximum while monovalent rejections are considerably reduced. Table 1 presents the results of the nanofiltration batch experiment performed at the wastewater natural ph of 6.2 and n = 5. The higher rejections (> 98%) for the divalent ions compared to the monovalent ones (40 %) is a typical nanofiltration characteristic. Simulation of ionic rejections as a function of n was performed with X i and L eff values of 0.443 and 0.374 nm. This set of parameters was obtained from initial fit of experimental ionic rejections as a function of J v to a 0,6M NaNO 3 solution. Figure 4 compares the simulation and experimentally obtained values.
Table 1. Comparison between initial feed, final mean permeate and retentate concentrations. NF batch mode, n = 5. Ion Initial feed Final mean permeate Retentate Total rejection 384.3 mg.l -1 7 mg.l -1 2055 mg.l -1 > 98% 1.7 mg.l -1 < 0.1 mg.l -1 8.1 mg.l -1 > 99% 19.6 g.l -1 11.6 g.l -1 43.6 g.l -1 40% 47.7 g.l -1 26.6 g.l -1 84.9 g.l -1 44% Co Ni Na 2 2 NO 3 NO 2 2 SO 4 1.1 g.l -1 0.7 g.l -1 2.7 g.l -1 36% 2290 mg.l -1 < 40 mg.l -1 11580 mg.l -1 > 99% Figure 4. Ionic rejections as a function of concentration volume factor (n). ph = 6.25 ; X i = 0.443; L eff = 0.374 nm; r p 0.42 nm, L p 0 6.3 L.h -1.m - ².bar -1 ; Simulation/experimental results: / Co 2+, / Ni 2+, / SO 4 2-, / NO 3-, / NO 2 - et / Na +. Two distinct behaviors are observed for ionic rejections. Divalent ions are almost completely retained (99%) during the entire duration of the experiment. Monovalent ions, however, show a significant rejection decrease as n increases. Apart from SO 4 2- ions, simulation and experimental results are in good accordance. Both trends are correctly simulated by Nanoflux. Detailed Co 2+ and Ni 2+ concentrations of both retentate and instantaneous permeate as a function of n are presented in figure 5. Figure 5. Ionic concentration as a function of volume concentration factor (n). Simulation/experimental results: / Co 2+, / Ni 2+. a) Retentate ; b) Permeate.
A good agreement between experimental and simulated values is obtained. Such result is especially remarkable when one considers that all batch simulations were done without any experimental information other than membrane charge density and effective membrane thickness from the NaNO 3 filtration experiment. At the permeate, Ni 2+ concentration is under the quantification limit of the analytical technique (0.1 mg.l -1 ) for n equivalent to 1; 1.8 and 3.5. A slight overestimation of divalent cations rejections is the origin of the underestimated permeate concentration. It does not affect, however, the retentate concentration prediction. 4.2. Electrochemical controlled ion complexation In order to evaluate the performance of the ECIC step in the proposed process, the NF final mean permeate solution stocked at the permeate tank when n = 5 was treated in two complexation experiments. An HNO 3 rinse (4 L, ph = 3) between each experiment permit to recover 95% of the ions. Results obtained are presented in table 2. Table 2. Partition coefficients (K d) obtained for the treatment of NF permeate tank solution at n = 5. Solution volume treated = 7.9 L; carbon felt mass = 6.9 g. Ion Feed concentration 1 st experiment Kd 2 nd experiment Kd 2 Co 6.7 mg.l -1 240 260 0.1 mg.l -1 303 354 10.7 g.l -1 38 47 Ni Na 2 Expressing the ionic uptake in terms of K d is specially useful to demonstrate the selectivity of PAA towards divalent cations. Nuclear wastewater is usually highly concentrated in Na + cations. The elevated partition coefficient (K d > 240) obtained for the low concentrations of Ni 2+ and Co 2+ in comparison to Na + (K d < 50) in the NF permeate reveals the grafted carbon felt as a convenient material to treat such solutions. Moreover, the acid elution does not alter the polymer as evidenced by the relatively similar K d values from the two cycles. The study of the electrochemical regeneration step was performed by two complexation/elution experiments. The treated solution presented a high (300 mg.l -1 ) concentration in Co 2+ so the maximum uptake of this cation on a ph value similar to the NF permeate could be obtained. Na 2SO 4 assured the electrical conductivity of the solution as the supporting electrolyte. Obtained results are presented in table 3. Table 3 Complexation/electrochemical regeneration experiments; uptake capacity is expressed in mg of complexed Co 2+ per gram of PAA grafted carbon felt. Experiment Co 2+ uptake capacity Elution 1 10.0 mg.g -1 67% 2 10.5 mg.g -1 62% For both experiments, constant uptake capacities of 10 mg of Co 2+ per gram of carbon felt and approximate 65% rate of electrochemical elution were obtained. This indicates that, other than the acidic, the grafted PAA is also not affected by the electrochemical regeneration. The water electrolysis reaction can be followed by the electrical potential curve presented in figure 6. The rapid increase followed by a stabilization in the measured potential is coherent to other water electrolysis reaction presented in [3]. Even though incomplete, the ensemble of results confirms the electrochemical regeneration concept as the water electrolysis was the only source of the H + used in the carbon felt recovery. Besides, compared to the 4 L of acid eluted solution, only 0.5 L of electrically eluted solution was produced. The interest in this type of regeneration is readily noticed.
Higher elution rates are expected to be obtained by establishing optimal operational conditions such as imposed current and elution time. Figure 6. Chronopotentiometric curve from the electrochemical regeneration experiment. 5. CONCLUSIONS A new wastewater treatment process was studied. The batch nanofiltration pre-treatment step was successfully studied by the combination of very few experiments associated to numerical simulation. Ionic rejections were predicted as a function of volume concentration factor. A good estimation of retentate concentration was obtained. ECIC step confirmed the high selectivity of PAA grafted carbon felts towards divalent cations. High partition coefficients (K d > 250) were achieved in NF permeate solution. Regeneration of carbon felts was carried out by in situ protons production coming from solution electrolysis. Up to 65% of the complexed ions were eluted when a current of 1 ma was applied to the system. The results obtained confirmed the interest in the association of these two techniques in a single wastewater treatment process. Acknowledgements Authors would like to thank Dr. Pascal Viel from CEA/DSM/SPCI for allowing the use of the electrochemical semi pilot unit. REFERENCES 1. Shannon, M.A., et al., Science and technology for water purification in the coming decades. Nature, 2008. 452(7185): p. 301-310. 2. Lutzenkirchen, J., et al., Comparison of Various Models to Describe the Charge-pH Dependence of Poly(acrylic acid). Journal of Chemical and Engineering Data, 2011. 56(4): p. 1602-1612. 3. Le, X.T., et al., Electrochemical-switchable polymer film: An emerging technique for treatment of metallic ion aqueous waste. Separation and Purification Technology, 2009. 69(2): p. 135-140. 4. Palmeri, J., et al., Modeling of multi-electrolyte transport in charged ceramic and organic nanofilters using the computer simulation program NanoFlux. Desalination, 2002. 147(1-3): p. 231-236. 5. Ferreira Esmi, C., et al., Using nanofiltration in a zero-rejection process: the removal of Ni2+ and Co2+ from salty wastewater. Desalination and Water Treatment, 2012. 51(1-3): p. 476-484. 6. Roma-Luciow, R., L. Sarraf, and M. Morcellet, Complexes of poly(acrylic acid) with some divalent, trivalent and tetravalent metal ions. European Polymer Journal, 2001. 37(9): p. 1741-1745. 7. Zagorodni, A.G., Ion Exchange Materials: Properties and Applications. 2007, Stockholm: ELSEVIER. 496. 8. Kovacs, Z., M. Discacciati, and W. Samhaber, Numerical simulation and optimization of multistep batch membrane processes. Journal of Membrane Science, 2008. 324(1-2): p. 50-58.