Facilitated ion transport through polyelectrolyte multilayer films containing metal-binding ligands

Journal of Membrane Science 459 (2014) 169–176

Contents lists available at ScienceDirect

Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci

Facilitated ion transport through polyelectrolyte multilayer films containing metal-binding ligands Chunjuan Sheng, Salinda Wijeratne, Chao Cheng, Gregory L. Baker 1, Merlin L. Bruening n Department of Chemistry, Michigan State University, East Lansing, MI 48824, USA

art ic l e i nf o

a b s t r a c t

Article history: Received 21 October 2013 Received in revised form 22 January 2014 Accepted 24 January 2014 Available online 5 February 2014

Alternating layer-by-layer adsorption of poly[(N,N0 -dicarboxymethyl) allylamine] (PDCMAA) and protonated poly(allylamine) (PAH) at low pH yields thin films with abundant Cu2 þ -binding sites. When deposited on porous alumina substrates, (PDCMAA/PAH)3PDCMAA or (PDCMAA/PAH)4 films show average Cu2 þ /Mg2 þ diffusion dialysis selectivities around 50. PDCMAA/PAH membranes also exhibit Cu2 þ /Ni2 þ and Cu2 þ /Ca2 þ selectivities 410. The high Cu2 þ /Mg2 þ selectivity despite similar aqueous diffusion coefficients and equal charges for the two ions suggests a facilitated transport mechanism. For comparison, poly(styrene sulfonate)/PAH films with 7 bilayers show Cu2 þ /Mg2 þ selectivities around 10, perhaps also due to facilitated transport via PAH. With PDCMAA/PAH membranes, Cu2 þ flux increases nonlinearly with increasing CuCl2 concentrations in the feed. Sorption isotherms show that PDCMAA/ PAH films contain both strong and weak binding sites, and the nonlinear increase in flux with increasing CuCl2 feed concentration likely represents hopping between weak binding sites, probably the amine groups of PAH. Strong binding of Cu2 þ to PDCMAA may displace ionic cross-links in the film and create free amine groups for facilitated transport. Additionally, Cu2 þ binding to the film suppresses Mg2 þ transport, either through electrostatic exclusion or occupation of hopping sites. & 2014 Elsevier B.V. All rights reserved.

Keywords: Facilitated ion transport Polyelectrolytes Layer-by-layer adsorption Diffusion dialysis

1. Introduction Layer-by-layer adsorption of oppositely charged polyelectrolytes [1,2] offers a simple and versatile way to form functional thin films on porous membrane supports. Moreover, variation of the polyelectrolyte composition [3], deposition conditions [4] and post-deposition treatments [5] affords control over the thickness, permeability and charge density of the polyelectrolyte multilayers (PEMs). Several research groups investigated the permeability of PEM-coated membranes in pervaporation [6,7], gas separations [8,9] and various separations with dissolved ions [10–17]. In particular, in nanofiltration [18–20] PEMs allow selective passage of monovalent ions over multivalent ions, which is important in water softening, and the minimal PEM thickness affords high water permeabilities [20,21]. The monovalent/divalent ion selectivity may stem from differences in either ion hydration or electrostatic exclusion for the two types of ions [22]. However, for ions with the same charge and similar hydrated radii, e.g. Cu2 þ and Mg2 þ , PEMs will likely show minimal

n

Corresponding author. Tel.: +1 517 355 9715x237; fax: +1 517 353 1793. E-mail address: [email protected] (M.L. Bruening). 1 Professor Baker passed away unexpected on October 17, 2012. We dedicate this work as a memorial to him. http://dx.doi.org/10.1016/j.memsci.2014.01.051 0376-7388 & 2014 Elsevier B.V. All rights reserved.

selectivity. Thus, typical PEM membranes would likely not prove useful in applications such as selective metal recycling and removal of specific transition metal ions from waste streams. Facilitated transport through membranes can address the challenge of obtaining selectivity when separating ions with the same charge and comparable hydrated radii. This type of transport relies on ion complexation in the membrane along with either ligand diffusion across the membrane or ion hopping between immobile binding sites [23–26]. The development of facilitated transport has progressed from liquid membranes [27] to supported liquid membranes [28,29] to polymer inclusion membranes [30] and molecularly imprinted membranes [31]. Nevertheless, most facilitated transport systems still suffer from limited stability, selectivity or permeability. PEM membranes may overcome some of the stability issues that plague liquid membranes, and the low thickness of these films will enhance permeance relative to thicker solid membranes. Development of facilitated transport through PEMs requires polyelectrolytes that bind metal ions and allow hopping between binding sites. Additionally, the rate of diffusion of uncomplexed ions through these films must be much lower than the rate of facilitated transport. Schlenoff et al. suggest that ions pass through PEMs via hopping between ion-exchange sites created by dissociation of polymeric ion pairs at high ionic strength [32]. Unfortunately, such a mechanism will not likely provide high selectivities among ions with the same charge unless complexation occurs.

170

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

Tieke et al. described the formation of membranes through alternating adsorption of charged macrocyclic compounds, i.e. calixarenes, azacrowns and cyclodextrins, and oppositely charged polyelectrolytes [3,11,33–35]. However ions that specifically interact with the macrocycles, especially calixarenes, pass through these membranes more slowly than through the macrocycle-free PEM membranes. Additionally, the selectivities among ions with the same charge, either monovalent or divalent, are o5. This paper reports the formation of membranes through layerby-layer adsorption of poly[(N,N0 -dicarboxymethyl) allylamine] (PDCMAA) and protonated poly(allylamine) (PAH). Fig. 1 shows the structures of these polymers. PDCMAA contains iminodiacetic acid (IDA) groups that strongly bind Cu2 þ , and at high concentrations Cu2 þ may also interact with the amine groups of PAH. In prior work we found that PDCMAA/PAH films bind Cu2 þ at levels as high as 2.5 mmol per cm3 of film [36]. Remarkably, this study shows that PDCMAA/PAH membranes exhibit Cu2 þ /Mg2 þ diffusion dialysis selectivities around 50. For comparison, poly(styrene sulfonate) (PSS)/PAH multilayer films show typical selectivities around 10. Sorption isotherms for Cu2 þ and measurements of transport rates as a function of feed concentration suggest that facilitated transport may proceed via the amine sites in PDCMAA/ PAH films.

2. Experimental 2.1. Materials Poly(allylamine hydrochloride) (Mw ¼120,000–200,000 Da) was purchased from Polysciences, and PDCMAA was synthesized as described previously [36]. Poly(sodium 4-styrene sulfonate) (Mw ¼ 70,000 Da), CuCl2  2H2O, CaCl2  2H2O, NiCl2 and 3-mercaptopropionic acid (MPA) were obtained from Sigma-Aldrich or Spectrum, and all reagents were used without further purification. Ethylenediaminetetraacetic acid disodium salt (EDTA-Na2) was purchased from Jade Scientific, and aqueous solutions were prepared with deionized water (18 MΩ cm, Milli-Q). Alumina membrane supports (Anodisc, pore diameter¼ 0.02 μm) were purchased from Whatman, and colloidal silica nanoparticles (70–100 nm, SNOWTEX-ZL) were obtained from Nissan Chemical Industries. 2.2. Film preparation and characterization PEMs were first deposited on Au-coated wafers (200 nm of Au sputtered on 20 nm of Cr on Si(100)) for determination of ellipsometric film thicknesses. The Au-coated wafers were cleaned with UV/ozone for 15 min before a 30 min immersion in 5 mM MPA in ethanol, followed by rinsing with ethanol and then water for 1 min each, and drying under a N2 stream. The MPA-modified wafer was immersed in a PAH solution for 5 min, rinsed with water from a squirt bottle for 1 min, immersed in a polyanion solution (PSS or PDCMAA) for 5 min, and rinsed again. The dip-and-rinse process was continued to deposit the desired number of polyelectrolyte bilayers. Adsorption of PEMs on porous alumina membranes followed essentially the same procedure starting with polyanion adsorption. The alumina membrane was placed in a holder that exposed only the top of the membrane to the solutions. All polyelectrolyte solutions contained 0.01 M polymer repeating units and 0.5 M NaCl. The pH of these solutions was adjusted to 3 with 0.1 M HCl or 0.1 M NaOH. The addition of NaCl to deposition solutions results in differences in IR spectra and ellipsometric thicknesses compared to our previous work that employed adsorption from solutions containing no added salt [36]. The 0.5 M NaCl in deposition solutions may also lead to more IDA sites that do not form electrostatic complexes with PAH [37]. In one

Fig. 1. Structures of the polymers employed to prepare PEMs.

case, we used the same membrane in more than 8 consecutive dialysis experiments without losing high selectivity. Previous work showed that in cross-flow nanofiltration PEMs are stable for several days [13]. In modification of silica nanoparticles, 50 mL of a 0.01 M PAH, 0.5 M NaCl solution (pH ¼3) was added to 0.1 g of silica colloids in a centrifuge tube. This mixture was sonicated for 5 min, and the adsorption solution was left to stand for 10 min with continuous stirring. The solution was then centrifuged for 10 min at 3000 rpm. After the supernatant was removed, 20 mL of water was added to the sample, and the solution was sonicated for 5 min and again centrifuged prior to removal of the supernatant. Subsequently, 50 mL of a 0.01 M PDCMAA, 0.5 M NaCl solution (pH ¼3) was added to the remaining colloidal solution, and adsorption and rinsing proceeded as with PAH deposition. Similar adsorption and washing steps were performed to prepare PAH/PDCMAA coatings with 10 bilayers. The procedure was performed with 12 centrifuge tubes to obtain the required quantity of modified particles. Thicknesses of films deposited on Au-coated wafers were determined using a rotating analyzer ellipsometer (J.A. Woollam model M-44), assuming a refractive index of 1.5 for the dry films. Reflectance Fourier transform infrared spectra of these films were obtained with a Thermo Scientific Nicolet 6700 FT-IR spectrometer (801 incident angle, Pike grazing angle apparatus) with a MCT detector. A UV/ozone-cleaned Au-coated wafer served as a background. UV–vis spectra of Cu2 þ -containing solutions were acquired with a PerkinElmer UV/VIS spectrophotometer (Lambda 25).

2.3. Diffusion dialysis Diffusion dialysis studies were carried out in a home-made apparatus that consists of feed and permeate chambers separated by a membrane with the PEM facing the feed solution [38]. Initially, 90 mL of salt solution and 90 mL of deionized water were added to the feed and permeate chambers, respectively, and the two solutions were stirred vigorously throughout the experiment. The specific compositions and pH values of the feed solutions are provided for each experiment in the results and discussion section. An aqueous 0.1 M MgCl2 solution has a pH of 6.4, whereas 0.1 M CuCl2 has a pH of 3.6 due to water hydrolysis by Cu2 þ . A solution containing both 0.1 M MgCl2 and 0.1 M CuCl2 also has a pH of 3.6. For analysis, 1-mL aliquots were removed from the permeate chamber approximately every 5 min for 40–60 min. To balance the water level in the two chambers, 1 mL of solution was simultaneously removed from the feed. The permeate aliquots were diluted 10-fold with 2% nitric acid and analyzed by Inductively Coupled Plasma Optical Emission Spectroscopy (Axial ICPOES, Varian 710-ES). Over the course of these experiments, the salt concentration in the permeate was always small compared to that in the feed so the concentration gradient and the flux across the membrane were essentially constant. Thus when normalized to the exposed membrane area (2.1 cm2), the slopes in plots of moles of ion passed through the membrane versus time gave values for ion fluxes. Between diffusion dialysis with different solutions, the membrane and cells were rinsed with deionized water. Unless

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

stated otherwise, uncertainties in flux values represent standard deviations of experiments with 3 different membranes. 2.4. Sorption studies Sorption studies were carried out with (PDCMAA/PAH)10-modified nanoparticles. The high surface area of the nanoparticles and the large number of bilayers give the high number of binding sites needed for these studies. The modified nanoparticles were dried and ground into a fine powder with a mortar and pestle (dried, modified nanoparticles aggregate). Weighed amounts (  0.1 g) of these nanoparticles were incubated for 15 h at room temperature in stirred, 2-mL solutions containing varied initial CuCl2 concentrations. Diffusion into the film likely controls binding rates [36], and studies of binding kinetics show that the copper sorption approaches saturation after 8 h (see Fig. S1). The pH of the CuCl2 solutions was adjusted to 3.6 with dilute HCl or NaOH prior to mixing with the nanoparticles. The mass of the nanoparticles and the solution volume were chosen to achieve sorption of at least 20% of the Cu2 þ in the loading solution. The total sorption was determined from atomic absorption spectroscopy (Varian AA240) analysis of the Cu2 þ concentration in the source solution before and after sorption. After sorption, the residual solution was decanted, and the nanoparticles were rinsed with 1.5 mL of deionized water three times (with centrifugation and decanting) to remove the remaining solution and the weakly adsorbed ions. These particles were dried under vacuum and subsequently mixed with 1.5 mL of 0.1 M EDTA (pH ¼6.4) and incubated overnight (without stirring). The resulting eluate was diluted and analyzed by atomic absorption spectroscopy using standards prepared in 0.1 M EDTA. The Cu2 þ sorption calculated from either the loading or eluate solutions was normalized by the initial mass of nanoparticles and then plotted against the equilibrium (residual loading solution) concentration to give the sorption isotherm.

171

UV–vis spectra (Fig. 2) demonstrate that PSS, PAH and PDCMAA have very different affinities for Cu2 þ in solutions at pH 3.6 (the pH of a 0.1 M CuCl2 solution). Although for the spectra in Fig. 2 the polymer repeat unit is in a 10-fold excess with respect to the 1 mM CuCl2, the presence of PAH or PSS does not significantly alter the UV–vis spectrum of the solution. Thus, complexes between PAH or PSS and Cu2 þ do not form at this pH and Cu2 þ concentration. In contrast, the formation of a PDCMAA–Cu2 þ complex gives rise to a dramatic change in the Cu2 þ UV–vis spectrum. Reflectance FT-IR spectra (Fig. 3) provide evidence for Cu2 þ sorption in PAH/PDCMAA films on gold. After Cu2 þ sorption, the – COO  stretches observed at  1590 cm  1 (shoulder) and 1645 cm  1 merge into a single peak at 1637 cm  1. The small shoulder due to the –COOH carbonyl stretch ( 1720 cm  1) also decreases in intensity after Cu2 þ sorption. Elution of Cu2 þ from the film using an EDTA solution, and subsequent equilibration in pH 3 water yield a spectrum comparable to that of the pristine film. 3.2. Selective Cu2 þ transport through (PDCMAA/PAH)n films adsorbed on porous alumina As Table 1 shows, the equilibrium constant for formation of Cu2 þ –iminodiacetate complexes is 47 orders of magnitude higher than the corresponding constant for Mg2 þ –iminodiacetate. (Iminodiacetate is the binding functionality in PDCMAA.) Additionally, the formation constant for Cu2 þ –ammonia complexes is 4 orders of magnitude higher than that for Mg2 þ –ammonia complexes, and these trends should be similar to those for metal-ion complexation by PAH. Thus if ion transport through

3. Results and discussion 3.1. Preparation and characterization of Cu2 þ -binding PDCMAA/PAH films Fig. 1 shows the structures of the polymers we employed to create thin PEMs on porous alumina supports. Partial deprotonation of PDCMAA allows its adsorption as a polyanion [36]. Moreover, film thickness varies with the adsorption pH. Protonation of the two –COOH groups of PDCMAA occurs primarily below pH 4, so we chose a deposition pH of 3 to obtain a relatively low charge density on this polymer. This low charge density leads to thicker films than adsorption at higher pH values [36]. Under the selected deposition conditions, (PAH/PDCMAA)4 films adsorbed on MPAmodified gold-coated wafers have dry ellipsometric thicknesses of 42 nm. Assuming that the film contains 50% PDCMAA and has a density of 1 g/cm3, a (PAH/PDCMAA)4 coating contains  12 nmol of IDA groups per cm2. The concentration of IDA groups in the dry film is around 2.9 M but should decrease upon swelling in water. We chose to focus on films containing 3.5 or 4 bilayers because when deposited on porous supports, such coatings allow high ion fluxes while still fully covering the support [39]. Our previous studies showed that ellipsometric thicknesses of polyelectrolyte films on porous alumina supports and metal-coated Si wafers are similar [40], and the supporting information shows cross-sectional SEM images of film-coated membranes (Fig. S2). Unfortunately, the film did not fracture cleanly, so we could not use these images to determine thicknesses.

Fig. 2. UV–vis spectra of 1 mM CuCl2 in water or aqueous solutions containing the various polyelectrolytes indicated in the legend. The concentrations of the polyelectrolyte repeat units were 10 mM, and the solution pH was adjusted to 3.6.

Fig. 3. Reflectance FT-IR spectra of a (PAH/PDCMAA)3PAH film before and after immersion in 0.1 M CuCl2 (pH ¼ 3.6), and after subsequent immersion in 0.1 M EDTA (pH ¼6.4) and equilibration at pH 3. The film was rinsed with deionized water and dried with N2 prior to obtaining each spectrum.

172

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

Table 1 Ion diffusion coefficientsa [41] and equilibrium constants for formationb of ligandmetal ion complexes [42,43]. Ions

Cu2 þ

Ni2 þ

Diffusion coefficient (10  5 cm2/s) Formation constant (iminodiacetate–M2 þ , log K) Formation constant (NH3–M2 þ , log K)

0.714 10.63

0.661 8.19

4.24

2.81

Ca2 þ 0.792 2.59  0.2

Mg2 þ 0.706 2.94 0.23

a

Value at infinite dilution and 25 1C. Temperature and ionic strength are 20 1C and 0.1 M for iminodiacetate–M2 þ , and 25 1C and 0.5 M for NH3–M2 þ . b

Fig. 5. Evolution of permeate concentrations with time during diffusion dialysis through (PDCMAA/PAH)4-modified porous alumina membranes. The mixed-salt feed solution contained 0.1 M CuCl2 and 0.1 M MCl2 (MQCa or Ni, pH¼ 3.5), and the permeate was initially deionized water.

higher affinities of IDA and amine groups for Ni2 þ than for Mg2 þ (Table 1). This does not, however, explain why Ca2 þ shows higher fluxes than Mg2 þ , but a related ligand, EDTA, shows a Ca2 þ complex formation constant that is 2 orders of magnitude higher than that for Mg2 þ [42]. 3.3. Comparison of fluxes and selectivities in mixed- and single-salt diffusion through different polyelectrolyte multilayer films

Fig. 4. Evolution of permeate concentrations with time during diffusion dialysis through (PDCMAA/PAH)n-modified porous alumina membranes. The mixed-salt feed solution contained 0.1 M CuCl2 and 0.1 M MgCl2 at pH 3.6, and the permeate was initially deionized water. Filled and open symbols represent dialysis through (PDCMAA/PAH)4- and (PDCMAA/PAH)3PDCMAA-modified membranes, respectively.

(PDCMAA/PAH)n films involves hopping between ion-binding sites, Cu2 þ should move through the film to the exclusion of Mg2 þ . Fig. 4 presents permeate ion concentrations as a function of time during diffusion dialysis through porous alumina membranes coated with (PDCMAA/PAH)4 and (PDCMAA/PAH)3PDCMAA films. The feed solution contained 0.1 M CuCl2 and 0.1 M MgCl2, and the receiving phase was initially deionized water. The data in Fig. 4, indicate that Cu2 þ diffuses through these membranes 43-fold and 64-fold faster than Mg2 þ for (PDCMAA/PAH)4 and (PDCMAA/ PAH)3PDCMAA, respectively. Average Cu2 þ /Mg2 þ selectivities for three replicate membranes were 53 738 and 53714, for (PDCMAA/PAH)4 and (PDCMAA/PAH)3PDCMAA films, respectively. The aqueous diffusion coefficients of Cu2 þ and Mg2 þ differ by only 1% (Table 1), so this high Cu2 þ /Mg2 þ selectivity most likely reflects selective Cu2 þ binding to functional groups in the film and not size-based selectivity. Regardless of whether films terminate with PDCMAA or PAH, they show high Cu2 þ /Mg2 þ selectivities, so if facilitated transport is responsible for the high selectivity, it does not require a large excess of PDCMAA at the membrane surface. In addition to Cu2 þ /Mg2 þ selectivity, the especially high formation constants for iminodiacetate– and NH3–Cu2 þ complexes (Table 1) suggest that PDCMAA/PAH films should also exhibit Cu2 þ /Ni2 þ and Cu2 þ /Ca2 þ selectivities in diffusion dialysis. Based on the data in Fig. 5 and replicate measurements with two other membranes, (PDCMAA/PAH)4-coated membranes with a Cu2 þ /Mg2 þ selectivity of 617 20 show Cu2 þ /Ca2 þ and Cu2 þ /Ni2 þ selectivities of 13 72 and 18 74, respectively. (Table S2 in the Supplementary material gives results for the individual membranes). Higher fluxes for Ni2 þ than for Mg2 þ may stem from the

If Cu2 þ binding to coordination sites limits the transport of other ions through polyelectrolyte films, fluxes of these other ions in diffusion dialysis with single- and mixed-salt solutions should differ greatly. Table 2 compares single- and mixed-salt ion fluxes in diffusion dialysis through bare alumina membranes and alumina coated with (PDCMAA/PAH)4, (PDCMAA/PAH)3PDCMAA (termed (PDCMAA/PAH)3.5 in the tables), and (PSS/PAH)4 films. In the control experiment with bare porous alumina, the Cu2 þ and Mg2 þ fluxes are essentially the same within experimental uncertainty, regardless of whether the feed solutions contain single or mixed salts. Thus the transport selectivity through the PEMmodified membranes results exclusively from the PEMs. Comparison of ion fluxes in single- and mixed-salt solutions provides further evidence for Cu2 þ complexation in (PDCMAA/ PAH)4 and (PDCMAA/PAH)3PDCMAA films. For membranes coated with these films, Cu2 þ fluxes increase by a factor of  1.8 in mixed-salt solutions, perhaps because the higher ionic strength due to MgCl2 decreases electrostatic exclusion of Cu2 þ from the film. Streaming potentials resulting from the high permeability of Cl  compared to Mg2 þ may also enhance Cu2 þ transport [44]. In contrast, the Mg2 þ fluxes decrease two orders of magnitude on going from a single-salt to a mixed-salt solution. If the partition coefficient for Mg2 þ in (PDCMAA/PAH)4 were 1 and there were no significant concentration polarization in the solution, the Mg2 þ flux from the mixed-salt solution would correspond to a diffusion coefficient of only 7  10  13 cm2/s in the film (assuming a film thickness of 42 nm). Binding of Cu2 þ may remove Mg2 þ hopping transport pathways in the film and could also introduce positive charge that contributes to electrostatic exclusion of divalent cations from the membrane, which would lead to a lower partition coefficient. In contrast to the PDCMAA/PAH systems, the difference in single- and mixed-salt ion transport through (PSS/PAH)4-modified membranes was less dramatic. Although there is significant variation in Mg2 þ fluxes among different membranes, the Mg2 þ flux varied less than 2-fold between single- and mixed-salt solutions for each of the individual membranes (see Table S3). Presumably this reflects minimal Cu2 þ complexation by PSS. Even for Cu2 þ , flux decreases from  30 to 0.6 nmol/(cm2 s) (mixed-salt solutions) after coating the membrane with (PDCMAA/

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

173

Table 2 Cu2 þ and Mg2 þ fluxes (10  10 mol cm  2 s  1) and Cu2 þ /Mg2 þ selectivities in diffusion dialysis through bare and PEM-modified alumina membranes. Dialysis employed either single- or mixed-salt solutionsa in the feed. Film composition

Mg2 þ (single)

Cu2 þ (single)

Mg2 þ (mixed)

Cu2 þ (mixed)

Cu2 þ /Mg2 þ selectivityb

Bare substrate (PDCMAA/PAH)4 (PDCMAA/PAH)3.5 (PSS/PAH)4c

213 787 8.8 73.0 20.6 74.9 11.0 79.1

303 746 3.6 70.9 3.1 71.1 20.4 76.6

2577 9 0.167 0.10 0.137 0.08 7.3 7 4.8

2757 22 6.17 0.7 6.0 7 2.1 26.8 7 6.4

1.077 0.07 537 38 537 14 4.3 7 1.4

a The pH values of the feed solutions were 6.4, 3.6, and 3.6 for 0.1 M MgCl2, 0.1 M CuCl2, and a mixture containing 0.1 M MgCl2 and 0.1 M CuCl2, respectively. These are the natural pH values of these solutions. Table 3 presents the flux with 0.1 M MgCl2 solutions adjusted to pH 3.6. b Selectivity was calculated based on mixed-salt fluxes, and the uncertainty is the standard deviation of selectivities for 3 different membranes. c Table S3 gives the fluxes for the individual membranes.

PAH)3PDCMAA or (PDCMAA/PAH)4 films. Despite the small thickness of (PDCMAA/PAH)4 compared to the alumina substrate ( 40 nm and 60 μm, respectively), the film provides the vast majority of the resistance to ion mass transport. Thus, these membranes allow fluxes that are 1–2 orders of magnitude lower than those in supported liquid membranes [45,46]. However, as mentioned in the introduction, supported liquid membranes suffer from instability. Fluxes through (PDCMAA/PAH)4-coated membranes are similar to those with imprinted polymer membranes [47–49]. Mixed-salt selectivities are an order of magnitude higher with the (PDCMAA/PAH)4 and (PDCMAA/PAH)3PDCMAA membranes than with (PSS/PAH)4. However, the high thicknesses of (PDCMAA/ PAH)4 and (PDCMAA/PAH)3PDCMAA (20–40 nm) relative to (PSS/ PAH)4 (11 nm, see Table S1) may lead to fewer film defects and higher selectivity for PDCMAA/PAH films. Three (PSS/PAH)7-coated membranes showed Cu2 þ /Mg2 þ selectivities of 9.870.2, and one (PSS/PAH)10 film exhibited a selectivity of 14. Thus, the extra bilayers enhance selectivity. Nevertheless, 10-bilayer PSS/PAH films on Au-coated wafers have thicknesses ( 33 nm) between those of (PDCMAA/PAH)3PDCMAA and (PDCMAA/PAH)4 coatings, but in diffusion dialysis (PSS/PAH)10 shows  4-fold lower selectivity than (PDCMAA/PAH)3PDCMAA. If selectivity stems from Cu2 þ binding to PAH, selectivity in both PDCMAA/PAH and PSS/PAH systems may result from facilitated transport, but this effect is more prominent with PDCMAA/PAH (see below). In addition to the formation of Cu2 þ complexes, decreases in the feed solution pH from 6.4 to 3.6 upon addition of CuCl2 to MgCl2 solutions might alter Mg2 þ fluxes. To test this possibility, we adjusted the pH of 0.1 M MgCl2 feed solutions with HCl and performed diffusion dialysis. As Table 3 shows, for (PDCMAA/PAH)4 and (PDCMAA/PAH)3PDCMAA films, decreasing the feed pH from 6.4 to 3.6 does not significantly change the Mg2 þ flux. This insensitivity of Mg2 þ permeability to pH may reflect the pKa values of PDCMAA. Presuming that in PDCMAA/PAH films the –COOH group pKa values are similar to those of IDA, 2.6 and 1.8, they lie out of the range of the feed solution pH change (from 6.4 to 3.6). Titration of IDA also suggests that protonation of the –COOH groups primarily occurs below pH 3.6 [36]. Thus for these films the main effect of CuCl2 on the Mg2 þ flux should stem from complexation. (As Table 3 shows, there is a  2-fold increase in Mg2 þ flux on going from pH 3.6 to 6.4 for (PDCMAA/PAH)3PDCMAA films only. The flux increase may reflect a change in conformation for the acid-terminated film [50], but the ratio of 1.8 for these Mg2 þ fluxes at pH 6.4 and 3.6 is much smaller than the ratio of 160 for Mg2 þ fluxes in the absence (pH 6.4) and presence of 0.1 M CuCl2 (pH 3.6), see Table 2.) The large uncertainty in fluxes through PSS/PAH membranes at different pH values seems to preclude a firm conclusion about the effect of pH on transport (Table 3), but the Mg2 þ fluxes for each individual membrane do not vary significantly with the feed pH (Fig. S3). The degree of protonation of either PAH (pKa in the range of 8–9 [50,51]) or PSS (pKa of protonated PSS around 1.0 [52]) will not change greatly on going from pH 6.4 to pH 3.6.

Table 3 Mg2 þ fluxes (10  10 mol cm  2 s  1) as a function of feeda pH during diffusion dialysis through PEM-modified porous alumina membranes. Film composition

(PDCMAA/PAH)4 (PDCMAA/PAH)3.5 (PSS/PAH)4c a b c

Feed pH 6.4

3.6

6.4b

8.8 73.0 20.6 74.9 11.0 79.1

12.0 7 4.2 24.47 4.2 8.1 7 7.2

7.2 7 2.5 43.3 7 9.0 8.8 7 8.0

The feed solution contained 0.1 M MgCl2. After exposure of the film to pH 3.6 feed solution. Fig. S3 shows the data for the individual membranes.

The effect of CuCl2 on the Mg2 þ fluxes through (PDCMAA/ PAH)4 and (PDCMAA/PAH)3PDCMAA films is reversible, but only fully reversible after eluting Cu2 þ from the film. As Table 4 shows, after experiments with feed solutions containing 0.1 M CuCl2 and 0.1 M MgCl2, diffusion dialysis of just 0.1 M MgCl2 yields Mg2 þ fluxes lower than in the same experiment with a fresh membrane. Subsequent exposure of membranes to EDTA (pH ¼ 6.4) restores Mg2 þ fluxes in diffusion dialysis, further suggesting that Cu2 þ sorption inhibits Mg2 þ flux. This kind of gate effect accompanying facilitated transport also occurs in molecularly imprinted facilitated transport membranes [31]. 3.4. Flux as a function of feed concentration For ions undergoing facilitated transport, as their feed concentration increases eventually their flux should plateau due to saturation of binding sites in the membrane [23,24]. However, Fig. 6 shows that the Cu2 þ flux through (PDCMAA/PAH)4 films continuously increases with increasing CuCl2 feed concentration. Based on fluxes at even higher feed concentrations (Fig. S4) and the trends in fluxes through individual membranes (Fig. S5), the increase in flux with CuCl2 feed concentration is superlinear, which suggests an increased hopping rate or increased number of transport sites at high Cu2 þ concentrations [32]. Isotherms of Cu2 þ sorption in similar films on silica nanoparticles confirm the presence of weak binding sites that fill only at high Cu2 þ concentrations (see below for further discussion). 3.5. Isotherms for sorption of Cu2 þ in (PDCMAA/PAH)10-modified nanoparticles Sorption isotherms may help explain trends in diffusion dialysis fluxes as a function of feed concentration. We chose to examine sorption on PEM-modified nanoparticles because their large surface area enables quantitation of binding even with high (0.2 M) Cu2 þ concentrations in solution. The use of 10 rather than 4 polyelectrolyte bilayers also increases the total sorption. In these experiments, the decrease in the Cu2 þ concentration in a loading

174

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

Table 4 Mg2 þ fluxes (10  10 mol cm  2 s  1) in sequential diffusion dialysis with single- and mixed-salt feed solutions. Film composition

Before Cu2 þ exposurea

Mixed with Cu2 þ b

After Cu2 þ exposurec

After EDTA elutiond

(PDCMAA/PAH)4 (PDCMAA/PAH)3.5

8.8 7 3.0 20.6 7 4.9

0.167 0.10 0.137 0.08

1.17 0.76 2.4 7 1.6

12.8 7 1.9 14.0 7 9.0

a

Diffusion dialysis of 0.1 M MgCl2 at pH 6.4 with a freshly prepared membrane. Subsequent diffusion dialysis of 0.1 M MgCl2, 0.1 M CuCl2 at pH 3.6. c Diffusion dialysis of 0.1 M MgCl2 at pH 6.4 after dialysis of the mixed salt solution. d Diffusion dialysis of 0.1 M MgCl2 at pH 6.4 after immersion of the same membrane in 0.1 M EDTA solution (pH adjusted to 6.4) for 30 min and rinsing with deionized water. b

Fig. 6. Cu2 þ and Mg2 þ diffusion dialysis fluxes through (PDCMAA/PAH)4 films deposited on porous alumina. The feed solution contained both CuCl2 and MgCl2 at the indicated concentrations with solution pH values ranging from 3.7 at the highest concentration to 4.6 at the lowest concentration. The error bars show the standard deviation of flux values for three different membranes. A fourth membrane showed 85% lower fluxes for both Cu2 þ and Mg2 þ , but the evolution of flux as a function of feed concentration remained the same (Fig. S5 in the Supplementary material shows fluxes through the individual membranes).

solution after equilibration with (PDCMAA/PAH)10-modified nanoparticles allows an estimate of the amount of Cu2 þ sorption per mass of particles. A control experiment with bare nanoparticles showed that less than 20% of the Cu2 þ sorption is due to the Cu2 þ adsorbing to the underlying nanoparticle. We also rinsed the loaded particles with deionized water and eluted the Cu2 þ with EDTA. The amount of eluted Cu2 þ may be less than the bound Cu2 þ if rinsing removes some of the Cu2 þ from the particles. Fig. 7 shows the Cu2 þ sorption isotherms as calculated from both the decrease in Cu2 þ concentration in the loading solution and the amount of Cu2 þ eluted after rinsing. The isotherm based on eluted Cu2 þ approaches saturation at a  25 mM equilibrium concentration in the loading solution, and the maximum sorption is 800 μmol/g. However, for total sorption (determined from the loading-solution analysis), Cu2 þ binding nearly triples on increasing the equilibrium concentration from 37 mM to 193 mM. These data and the shape of the total sorption isotherm suggest two types of sorption sites with different affinities for Cu2 þ . We speculate that the strong binding sites, which saturate at Cu2 þ concentrations around 25 mM, are the IDA functionalities of PDCMAA and that the weak binding sites are amine groups of PAH. At sufficiently high concentrations, the Cu2 þ ions effectively compete with protons and bind to the amines of PAH. These sites are not saturated because of the competition with protons, and increasing ionic strength may also lead to fewer electrostatic cross-links in the film and the formation of more binding sites [53]. If facilitated transport involves hopping between weak binding sites (perhaps amines), increased binding to these sites at high Cu2 þ concentrations should enhance flux. Moreover the

Fig. 7. Sorption isotherm for Cu2 þ binding to (PDCMAA/PAH)10-modified nanoparticles. Sorption incubation time was 15 h at each concentration. Each point represents a fresh set of modified nanoparticles immersed in a new solution adjusted to pH 3.6, and the Cu2 þ concentration is what remains after sorption. Displacement of protons by Cu2 þ decreases pH values during sorption, so typical final pH values are o2. Diamonds represent the total sorption determined from analysis of the solution before and after mixing with nanoparticles for 15 h. Triangles show sorption determined from the amount of Cu2 þ eluted from the particles after rinsing with water.

flux should increase nonlinearly with the number of binding sites as more percolation pathways become available [32]. We note that the isotherm in Fig. 7 shows saturation at higher Cu2 þ concentrations than previous isotherms determined with (PAH/PDCMAA)10 films on Au-coated substrates [36]. This difference likely stems in part from a lower adsorption pH in the current work. Because of the high number of binding sites in the nanoparticles, displacement of protons by Cu2 þ should decrease pH values to 2 or less when the Cu2 þ saturates IDA binding sites. In contrast, studies with thin films examined only low Cu2 þ concentrations (o 1.2 mM) and employed fewer binding sites per volume as well as a weak buffer (pH 4.0). Thermogravimetric analysis of the SiO2 nanoparticles coated with (PAH/PDCMAA)10 films shows that these materials contain  60% polymer. Assuming that the mass of PDCMAA accounts for 50% of the polymer film, the maximum sorption of 800 μmol/g (eluted Cu2 þ after rinsing) corresponds to about 40% of the IDA groups. We suspect that IDA that interacts electrostatically with protonated amines may not participate in Cu2 þ binding, or that some IDA groups are not accessible. The total Cu2 þ sorption of 2300 μmol/g (based on removal of Cu2 þ from the sorption solution) would exceed the number of IDA groups, assuming that a film contains 50% PDCMAA and 50% PAH. However, should binding occur to amines, the film would contain more than enough functional groups to capture 2300 μmol of Cu2 þ per gram of (PAH/PDCMAA)10-coated SiO2. If transport occurs via the amine sites, why do (PSS/PAH)4 films on alumina show lower Cu2 þ /Mg2 þ selectivities than (PDCMAA/ PAH)4 films on alumina? Table 2 shows that the Cu2 þ fluxes through (PSS/PAH)4 membranes are even higher than those through (PDCMAA/PAH)4 films. However when normalized to ellipsometric film thicknesses, the Cu2 þ permeabilities through (PSS/PAH)4 films are similar to those through (PDCMAA/PAH)4 films. The biggest difference between transport data for (PDCMAA/ PAH)4 and (PSS/PAH)4 films is the low Mg2 þ flux through (PDCMAA/PAH)4 in the presence of 0.1 M CuCl2. Because Cu2 þ binds weakly to amines and negligibly to sulfonates, the presence of 0.1 M CuCl2 does not greatly affect Mg2 þ or Cu2 þ transport through PSS/PAH films, and ion permeabilities are relatively independent of the feed concentration (flux increases approximately linearly with concentration, see Fig. S6). With (PDCMAA/ PAH)4, Cu2 þ binding to PDCMAA should break ionic cross-links and create free ammonium groups that may bind Cu2 þ at high

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

CuCl2 concentrations in the feed [53]. Such binding should both increase Cu2 þ flux and limit Mg2 þ flux, either by electrostatic exclusion of Mg2 þ or occupation of hopping sites.

4. Conclusions PDCMAA/PAH films adsorbed on porous alumina allow selective diffusive transport of Cu2 þ over Mg2 þ , Ni2 þ , or Ca2 þ in mixed-salt solutions, and Cu2 þ /Mg2 þ selectivities are around 50. These high Cu2 þ /Mg2 þ selectivities do not occur with PSS/PAH films or in single-salt experiments with PDCMAA/PAH films. Selectivity appears because binding of Cu2 þ to PDCMAA/PAH films greatly decreases the Mg2 þ flux, presumably by saturating hopping sites or inducing electrostatic exclusion. The high Cu2 þ /Mg2 þ selectivity of these films in mixed salt solutions suggests facilitated transport, as the aqueous diffusion coefficients of these ions differ by only 1%. Unlike most facilitated transport membranes, however, the Cu2 þ flux through PDCMAA/PAH films increases with the concentration of Cu2 þ in the feed solution, and sorption isotherms suggest that facilitated transport occurs through binding to weak sorption sites, perhaps amine groups. Binding of Cu2 þ to PDCMAA could disrupt ionic crosslinks and create free amine groups that serve as hopping sites for Cu2 þ transport. These Cu2 þ -selective membranes may serve as a model for future membrane-based separations in areas such as metal recycling.

[11]

[12]

[13]

[14]

[15]

[16]

[17]

[18]

[19]

[20]

[21]

Acknowledgments [22]

We gratefully acknowledge funding from the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences of the U.S. Department of Energy through Grant DE-FG0298ER14907.

[23] [24] [25] [26]

Appendix A. Supplementary material [27]

Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.memsci.2014.01.051. References [1] G. Decher, Fuzzy nanoassemblies: toward layered polymeric multicomposites, Science 277 (1997) 1232–1237. [2] G. Decher, J.D. Hong, J. Schmitt, Buildup of ultrathin multilayer films by a selfassembly process: III. Consecutively alternating adsorption of anionic and cationic polyelectrolytes on charged surfaces, Thin Solid Films 210–211 (2) (1992) 831–835. [3] A. Toutianoush, J. Schnepf, A. El Hashani, B. Tieke, Selective ion transport and complexation in layer-by-layer assemblies of p-sulfonato-calix[N]arenes and cationic polyelectrolytes, Adv. Funct. Mater. 15 (2005) 700–708. [4] S. Bhattacharjee, J. Dong, Y. Ma, S. Hovde, J.H. Geiger, G.L. Baker, M.L. Bruening, Formation of high-capacity protein-adsorbing membranes through simple adsorption of poly(acrylic acid)-containing films at low pH, Langmuir 28 (2012) 6885–6892. [5] A.M. Balachandra, J.H. Dai, M.L. Bruening, Enhancing the anion-transport selectivity of multilayer polyelectrolyte membranes by templating with Cu2 þ , Macromolecules 35 (2002) 3171–3178. [6] L. Krasemann, A. Toutianoush, B. Tieke, Self-assembled polyelectrolyte multilayer membranes with highly improved pervaporation separation of ethanol/ water mixtures, J. Membr. Sci. 181 (2001) 221–228. [7] J. Li, G.J. Zhang, S.L. Ji, N.X. Wang, W. An, Layer-by-layer assembled nanohybrid multilayer membranes for pervaporation dehydration of acetone–water mixtures, J. Membr. Sci. 415 (2012) 745–757. [8] Q. Ji, S.B. Yoon, J.P. Hill, A. Vinu, J.-S. Yu, K. Ariga, Layer-by-layer films of dualpore carbon capsules with designable selectivity of gas adsorption, J. Am. Chem. Soc. 131 (2009) 4220–4221. [9] D.M. Sullivan, M.L. Bruening, Ultrathin, gas-selective polyimide membranes prepared from multilayer polyelectrolyte films, Chem. Mater. 15 (2002) 281–287. [10] A. El-Hashani, A. Toutianoush, B. Tieke, Use of layer-by-layer assembled ultrathin membranes of dicopper-[18]azacrown-N-6 complex and

[28]

[29]

[30]

[31] [32] [33]

[34]

[35]

[36] [37]

[38]

[39] [40]

175

polyvinylsulfate for water desalination under nanofiltration conditions, J. Membr. Sci. 318 (2008) 65–70. B. Tieke, A. Toutianoush, W.Q. Jin, Selective transport of ions and molecules across layer-by-layer assembled membranes of polyelectrolytes, p-sulfonatocalix[N]arenes and prussian blue-type complex salts, Adv. Colloid Interface Sci. 116 (2005) 121–131. B.W. Stanton, J.J. Harris, M.D. Miller, M.L. Bruening, Ultrathin, multilayered polyelectrolyte films as nanofiltration membranes, Langmuir 19 (2003) 7038–7042. R. Malaisamy, M.L. Bruening, High-flux nanofiltration membranes prepared by adsorption of multilayer polyelectrolyte membranes on polymeric supports, Langmuir 21 (2005) 10587–10592. B. Tieke, L. Krasemann, A. Toutianoush, Tailor-made membranes for alcohol/ water pervaporation and ion separation prepared upon layer-by-layer adsorption of polyelectrolytes, Macromol. Symp. 163 (2001) 97–111. S. Qi, W. Li, Y. Zhao, N. Ma, J. Wei, T.W. Chin, C.Y. Tang, Influence of the properties of layer-by-layer active layers on forward osmosis performance, J. Membr. Sci. 423 (2012) 536–542. Q. Saren, C.Q. Qiu, C.Y. Tang, Synthesis and characterization of novel forward osmosis membranes based on layer-by-layer assembly, Environ. Sci. Technol. 45 (2011) 5201–5208. L. Setiawan, R. Wang, S. Tan, L. Shi, A.G. Fane, Fabrication of poly(amideimide)-polyethersulfone dual layer hollow fiber membranes applied in forward osmosis by combined polyelectrolyte cross-linking and depositions, Desalination 312 (2013) 99–106. K. Hoffmann, T. Friedrich, B. Tieke, Layer-by-layer assembled polyelectrolyte blend membranes and their use for ion separation and rejection, Polym. Eng. Sci. 51 (2011) 1497–1506. K. Hoffmann, B. Tieke, Layer-by-layer assembled membranes containing hexacyclen-hexaacetic acid and polyethyleneimine N-acetic acid and their ion selective permeation behaviour, J. Membr. Sci. 341 (2009) 261–267. L. Ouyang, R. Malaisamy, M.L. Bruening, Multilayer polyelectrolyte films as nanofiltration membranes for separating monovalent and divalent cations, J. Membr. Sci. 310 (2008) 76–84. M.L. Bruening, D.M. Dotzauer, P. Jain, L. Ouyang, G.L. Baker, Creation of functional membranes using polyelectrolyte multilayers and polymer brushes, Langmuir 24 (2008) 7663–7673. M.L. Bruening, D.M. Sullivan, Enhancing the ion-transport selectivity of multilayer polyelectrolyte membranes, Chem. Eur. J. 8 (2002) 3832–3837. E.L. Cussler, R. Aris, A. Bhown, On the limits of facilitated diffusion, J. Membr. Sci. 43 (1989) 149–164. R.D. Noble, Analysis of ion-transport with fixed site carrier membranes, J. Membr. Sci. 56 (1991) 229–234. R.D. Noble, Facilitated transport mechanism in fixed site carrier membranes, J. Membr. Sci. 60 (1991) 297–306. R.D. Noble, Generalized microscopic mechanism of facilitated transport in fixed site carrier membranes, J. Membr. Sci. 75 (1992) 121–129. H.C. Visser, D.N. Reinhoudt, F. de Jong, Carrier-mediated transport through liquid membranes, Chem. Soc. Rev. 23 (1994) 75–81. A.J.B. Kemperman, D. Bargeman, T. van den Boomgaard, H. Strathmann, Stability of supported liquid membranes: state of the art, Sep. Sci. Technol. 31 (1996) 2733–2762. D. de Agreda, I. Garcia-Diaz, F.A. Lopez, F.J. Alguacil, Supported liquid membranes technologies in metals removal from liquid effluents, Rev. Metal. Madrid 47 (2011) 146–168. L.D. Nghiem, P. Mornane, I.D. Potter, J.M. Perera, R.W. Cattrall, S.D. Kolev, Extraction and transport of metal ions and small organic compounds using polymer inclusion membranes (PIMS), J. Membr. Sci. 281 (2006) 7–41. M. Ulbricht, Membrane separations using molecularly imprinted polymers, J. Chromatogr. B 804 (2004) 113–125. T.R. Farhat, J.B. Schlenoff, Ion transport and equilibria in polyelectrolyte multilayers, Langmuir 17 (2001) 1184–1192. A. El-Hashani, B. Tieke, Electrostatic layer-by-layer assembly of ultrathin films containing hexacyclen and p-sulfonatocalix[N]arene macrocycles, J. Nanosci. Nanotechnol. 6 (2006) 1710–1717. B. Tieke, A. El-Hashani, A. Toutianoush, A. Fendt, Multilayered films based on macrocyclic polyamines, calixarenes and cyclodextrins and transport properties of the corresponding membranes, Thin Solid Films 516 (2008) 8814–8820. A. Toutianoush, A. El-Hashani, J. Schnepf, B. Tieke, Multilayer membranes of psulfonato-calix 8 arene and polyvinylamine and their use for selective enrichment of rare earth metal ions, Appl. Surf. Sci. 246 (2005) 430–436. S. Wijeratne, M.L. Bruening, G.L. Baker, Layer-by-layer assembly of thick, Cu2 þ -chelating films, Langmuir 29 (2013) 12720–12729. Y. Ma, J. Dong, S. Bhattacharjee, S. Wijeratne, M.L. Bruening, G.L. Baker, Increased protein sorption in poly(acrylic acid)-containing films through incorporation of comb-like polymers and film adsorption at low pH and high ionic strength, Langmuir 29 (2013) 2946–2954. J.H. Dai, A.M. Balachandra, J.I. Lee, M.L. Bruening, Controlling ion transport through multilayer polyelectrolyte membranes by derivatization with photolabile functional groups, Macromolecules 35 (2002) 3164–3170. J.J. Harris, J.L. Stair, M.L. Bruening, Layered polyelectrolyte films as selective, ultrathin barriers for anion transport, Chem. Mater. 12 (2000) 1941–1946. D.M. Sullivan, M.L. Bruening, Ultrathin, cross-linked polyimide pervaporation membranes prepared from polyelectrolyte multilayers, J. Membr. Sci. 248 (2005) 161–170.

176

C. Sheng et al. / Journal of Membrane Science 459 (2014) 169–176

[41] P. Vanýsek, Ionic conductivity and diffusion at infinite dilution, in: David R. Lide (Ed.), CRC Handbook of Chemistry and Physics, 94th edition, Internet Version, CRC Press, Boca Raton, Florida, 2013. [42] A.E. Martell, R.M. Smith, Critical Stability Constants, Amino Acids, vol. 1, Plenum Press, New York, 1974. [43] A.E. Martell, R.M. Smith, Critical Stability Constants, Inorganic Complexes, vol. 4, Plenum Press, New York, 1976. [44] C. Cheng, A. Yaroshchuk, M.L. Bruening, Fundamentals of selective ion transport through multilayer polyelectrolyte membranes, Langmuir 29 (2013) 1885–1892. [45] A.P. de los Ríos, F.J. Hernández-Fernández, L.J. Lozano, S. Sánchez-Segado, A. Ginestá-Anzola, C. Godínez, F. Tomás-Alonso, J. Quesada-Medina, On the selective separation of metal ions from hydrochloride aqueous solution by pertraction through supported ionic liquid membranes, J. Membr. Sci. 444 (2013) 469–481. [46] P.A. O'Hara, M.P. Bohrer, Supported liquid membranes for copper transport, J. Membr. Sci. 44 (1989) 273–287. [47] V. Vatanpour, S.S. Madaeni, S. Zinadini, H.R. Rajabi, Development of ion imprinted technique for designing nickel ion selective membrane, J. Membr. Sci. 373 (2011) 36–42.

[48] Y. Zhai, Y. Liu, X. Chang, X. Ruan, J. Liu, Metal ion-small molecule complex imprinted polymer membranes: preparation and separation characteristics, React. Funct. Polym. 68 (2008) 284–291. [49] K. Araki, T. Maruyama, N. Kamiya, M. Goto, Metal ion-selective membrane prepared by surface molecular imprinting, J. Chromatogr. B 818 (2005) 141–145. [50] J. Choi, M.F. Rubner, Influence of the degree of ionization on weak polyelectrolyte multilayer assembly, Macromolecules 38 (2005) 116–124. [51] K. Itano, J.Y. Choi, M.F. Rubner, Mechanism of the pH-induced discontinuous swelling/deswelling transitions of poly(allylamine hydrochloride)-containing polyelectrolyte multilayer films, Macromolecules 38 (2005) 3450–3460. [52] S.R. Lewis, S. Datta, M. Gui, E.L. Coker, F.E. Huggins, S. Daunert, L. Bachas, D. Bhattacharyya, Reactive nanostructured membranes for water purification, Proc. Natl. Acad. Sci. 108 (2011) 8577–8582. [53] T.R. Farhat, J.B. Schlenoff, Doping-controlled ion diffusion in polyelectrolyte multilayers: mass transport in reluctant exchangers, J. Am. Chem. Soc. 125 (2003) 4627–4636.