Selectivity of Ion Exchange Membranes: A Review Tao Luoa, Said Abdua, Matthias Wesslinga,b* a
Chemical Process Engineering AVT.CVT, RWTH Aachen University, Forckenbeckstraße 51, 52074 Aachen, Germany b
DWI – Leibniz-Institute for Interactive Materials, Forckenbeckstraße 50, 52074 Aachen, Germany
*Phone: +1-49-241-8095488, Fax: +1-49-241-8092252,
[email protected], Forckenbeckstraße 51, 52074 Aachen
1
Abstract Ion exchange membranes have been established as a key component in industrial water desalination and electrolysis processes, and nowadays are being studied and developed for new energy conversion and storage, efficient desalination and wastewater treatment processes. These processes include redox flow batteries, reverse electrodialysis, membrane capacitive deionization, microbial fuel cells and ion exchange membrane bioreactors. Ion permselectivity, as the most essential property of ion exchange membranes, makes these processes possible and/or efficient. Now, not only the permselectivity between counter- and co-ions but also the ion selectivity between different counter-ions is required for the efficiency of these novel processes. This review aims to provide a comprehensive overview of the ion exchange membrane permselectivity by summarizing the developments in this field over the past decade. Membrane microstructure and possible mechanisms for ion transport in the membrane phase are discussed with respect to permselectivity, along with the influence of current density and related membrane - solution boundary conditions. A selectivity order for the transport of common anions through conventional anion exchange membranes is generalized, and the same is done for common cations. Two types of experimental methods for the determination of ion permselectivity ̶ electrodialysis and the membrane potential method ̶ are summarized. Relevant membrane preparation methods and the surface modification of ion exchange membranes reported in the past decade are classified and discussed. In summary, we conclude that synthetic methods have evolved rapidly, however, fundamental knowledge development on the many complex phenomena occurring in the adaptive membrane bulk environment as well as the solution-interface environment is missing and requires significant efforts combining experimental, theoretical and simulation efforts. Keywords: ion exchange membranes, permselectivity, ion selectivity order, ion transport, boundary layers, electrodialysis, membrane potential, membrane surface modification
2
1.
Introduction
1.1. Established and emerging applications of ion exchange membranes As an important member among all types of membranes, ion exchange membranes (IEMs) refer to a class of dense polymeric membranes bearing fixed charges in the polymer matrix. IEMs can selectively allow the passage of oppositely charged ions (counter-ions), while obstructing similarly charged ions (coions). This permselectivity of IEMs for counter ions was first elucidated by Donnan [1], and the mechanism is called Donnan effect or Donnan exclusion (towards co-ions) thereafter. Due to this ion permselectivity, industrial processes based on IEMs like electrodialysis (ED), diffusion dialysis (DD) and electrolysis have been established. Apart from ED and DD, IEMs have also been widely studied as a key component in a variety of systems: flow batteries, reverse osmosis (RO), and also some emerging new applications like membrane capacitive deionization (MCDI), reverse electrodialysis (RED), microbial fuel cells (MFC) and ion exchange membrane bioreactors (IEMB). As summarized in Table 1, these different processes employ different driving forces and they frequently aim to help address the increasing global concern of energy shortage [2, 3], environment issues and drinking water sources. Table 1. Different processes employing IEMs Process Driving Force for Ion Selectivity Application References Ion Transport ED U* co- and counter-ions water desalination, [4-6] NaCl production Flow Batteries µ*(discharge) Na+/H+ and other energy storage and [7] U (charge) ions conversion DD c* H+ and OH- leakage recovery of acids [8] MCDI U co- and counter-ions water desalination [9, 10] RED c co- and counter-ions energy conversion [11-13] RO c, P* retention of salt water desalination [14, 15] MFC µ H+ and Na+ waste biomass treatment, energy [16, 17] conversion IEMB µ co- and counter- water treatment [18] ions, different counter-ions U, electrical potential; µ, chemical potential; c, concentration difference; P, pressure difference. 1.2. Requirements of better ion selectivity With the growing number of application of IEMs, the permselectivity not only between counter- and coions but also between counter-ions of different valence (monovalent and multivalent) or even equal valence, such as nitrate and chloride, are desired. As shown in Table 1, processes like flow batteries, DD, MFC and IEMB have such a higher requirement of membrane permselectivity between counter-ions of different valence. In flow batteries, the essential function of an IEM is to isolate ions of redox couples that are involved in electrode reactions to prevent self-discharge, while allowing the transfer of specific ions (charge carriers) across the membrane at a high rate to complete the electric circuit. The 3
permselectivity of IEMs between redox-active ions and charge carriers determines the coulombic efficiency, ultimately the energy efficiency of a flow battery. Table 2 lists the redox-active ions (anolyte and catholyte), charge carriers and IEM types used in several types of flow batteries. As can be seen, the permselectivity requirement can be solely the separation between counter- and co-ions as in the case for polysulfide bromine battery with a cation exchange membrane (CEM). But more often, it is desirable to have the permselectivity between metal cations and H+. The special “H+ leakage” through anion exchange membranes (AEMs) bearing fixed positive charges is utilized in all vanadium [19] and chromium bromine batteries leading to very high coulombic efficiency. Table 2. Ion selectivity requirements in redox flow batteries [7] Battery Type
Anolyte Polysulfide Bromine Br3 /3Br- All Vanadium VO2+/VO2+ Vanadium Bromine ClBr2-/ Cl-,2Br- Chromium Bromine Fe2+/Fe3+ Zinc Cerium (non-aqueous)
Ce3+/Ce4+
Catholyte 2S4 /S22- V2+/V3+ V2+/V3+
Charge Carrier
Membrane Permselectivity
Na+ H+ H+
CEM CEM, AEM CEM
Na+ and anions H+ and V(II~V) H+ and V(II, III)
Cr2+/Cr3+
H+
CEM, AEM
Zn/Zn2+
H+
CEM
H+ and metal cations + H and metal cations
For the application of IEMs in IEMB and MFC, a high requirement of permselectivity between counter- ions is crucial for the system efficiency. In the case of IEMB, the selective transport of toxic oxyanions (ClO4-, NO3-, BrO3-) over other multivalent anions (SO42-, HPO42-) under Donnan dialysis conditions is desired to treat drinking water polluted with these monovalent anions in an efficient way [20]. Recent studies with the commercial monovalent anion selective membrane show the best performance in terms of selective removal of monovalent anions to the recommended safety levels [18]. Even while normal AEMs actually have larger fluxes of toxic oxyanions (ClO4-, NO3-, BrO3-), the monovalent anion selective membrane still displays better efficiency for having SO42-, HPO42- fluxes almost two orders of magnitude lower compared with those in normal AEMs [18]. However, the high price of the monovalent anion selective membrane is a hurdle for the application of this technology. In microbial electrochemical systems like MFC, the requirement of selective H+ transport in the presence of Na+ with a substantially higher concentration is even a greater challenge [17]. As implied by the “H+ leakage” of AEMs for the flow battery application, the selective transport of H+ and OH- ions through IEMs is very important for several technical applications. In diffusion dialysis (DD) for the recovery of acids from metal finishing solutions, AEMs with high H+/metal cations selectivity is desired [8]. While in electrochemical water splitting with bipolar membranes, AEMs with high H+ retention and CEMs with high OH- retention are superior for this application [21, 22]. 1.3. Past research motivations In the past, the practical motivation for the study of tuned IEM permselectivity was the seemingly impossible task to retain more mobile H+ ions while transporting Na+ ions in the electrolysis process of NaCl solution to produce Cl2, NaOH and H2 - the so-called chlor-alkali process [23]. This process is actually until today still the most relevant industrial application for IEMs. The challenge was tackled with 4
a perfluorosulfonic acid type CEM bearing a surface layer of perflurocarboxylic acid groups to retain H+ ions, after the finding of faster Na+ transport compared with H+ in perflurocarboxylic acid type CEMs [24]. Another motivation was originally to produce table salt (NaCl) in Japan by electrodialysis of sea water that has multivalent Mg2+ and Ca2+ ions [5, 6]. The CEMs used here should have higher transport affinity towards Na+ than Mg2+ and Ca2+, which is however not an intrinsic property of traditional CEMs. Meanwhile, the AEMs should also ideally be more permeable for Cl- than for other anions (SO42-). Sata and co-workers had performed systematic investigation on the modification of IEMs to achieve such monovalent ion selectivity and studied the electrodialysis performance. These experimental works are summarized in two reviews for AEMs [6] and CEMs [5], respectively. A recent review summarizes the patents about preparation methods of CEMs with monovalent cation selectivity [25]. This topic of ion permselectivity is also briefly touched in a very recent review about IEMs [26]. The work of Sata and his co-workers shows that counter-ions of different valence can be separated in a kinetic manner in electrodialysis by a membrane surface layer bearing the same charge as the counterions. The commercialized monovalent ion selective IEMs could also find some further niche applications like demineralization of whey [27] and electro-acidification of milk to produce high-purity bovine milk casein isolates by bipolar membrane electrodialysis [28]. However, the high fabrication cost of such IEMs impedes their utilization in wider applications like flow batteries and IEMB mentioned above. Even the NaCl production by sea water electrodialysis in Japan is heavily subsidized by the government [29]. Therefore, there is a need to advance the development of IEMs with tuned ion selectivity, including further understanding of the ion transport in the IEM system and the pursuit of fabrication methods that can lower the membrane cost. 1.4. Scope and the aim of this review The purpose of this review is to give a comprehensive overview of the ion selectivity of IEMs, aiming to stimulate the research work in this field for the development of IEMs with tailor-made ion selectivity. The first part of this review briefly introduces IEM types and the Donnan effect, followed by a contemporary model of membrane microstructure, as well as the ion transport mechanisms occuring in the bulk membrane and at the membrane - solution interface. We suggest a selectivity order of cations and anions through CEMs and AEMs, respectively. The second part provides a review of two groups of experimental methods to determine the counter- and co-ion selectivity (permselectivity), and also the selectivity between different counter-ions. These two groups of methods include the electrodialysis method and the membrane potential method. The set-ups used and the relationship of ion selectivity obtained by these two methods are discussed. Then the last part summarizes the advancement of membrane modification and fabrication methods in the past decade to achieve better ion selectivity.
5
2.
Structure of ion exchange membranes and ion transport mechanisms
2.1. Ion exchange membrane types and the Donnan effect Based on the charge sign and the distribution of fixed ionic groups, IEMs are generally classified into five groups: cation exchange membranes, anion exchange membranes, amphoteric ion exchange membranes, bipolar membranes and mosaic ion exchange membranes [30, 31]. An illustration of these membranes is given in Fig. 1. The cation- and anion exchange membranes are also called monopolar ion exchange membranes, to be distinguished from the bipolar membrane (Fig. 1 d). Since an IEM contains hydrophilic/hydrated ionic groups and relatively more hydrophobic polymer chains, the membrane is essentially heterogeneous at a length scale equivalent to the polymer chain segments, and micro-phase separation between ionic groups and the hydrophobic domain is assumed to occur when the membrane is brought into contact with water. The several structure models of perfluorosulfonic acid Nafion membrane at hydrated state is a typical example [32].
Figure 1. Schematic illustration of different IEM types: (a) anion exchange membranes, (b) cation exchange membranes, (c) amphoteric ion exchange membranes, (d) bipolar ion exchange membranes and (e) mosaic ion exchange membranes. The circles depict the counter-ions in the membrane matrix, and the background patterns differentiate different fixed ionic groups. Since most research work about ion selectivity has been focused on monopolar IEMs, this review will also focus on these two basic types of IEMs, CEMs and AEMs. By treating the IEM matrix as a solution with homogeneously-distributed fixed charges, Donnan derived the thermodynamic membrane equilibrium equations [33]. An electric potential at the membrane - solution interface was found to be responsible for the exclusion of co-ions from the membrane matrix. This potential is now called Donnan potential, and the exclusion of co-ions by IEMs is termed Donnan effect. As shown in Fig. 2, the 6
distribution of the co- and counter-ions between the membrane matrix and bulk solutions, and the resulting Donnan potentials at a cation and an anion exchange membrane are shown.
Figure 2. Schematics of the distribution of the co- and counter-ions between the membranes and the bulk solution (a, b) and the resulting Donnan potentials at a cation- and an anion-exchange membranes (c). Only mobile ions in the membranes are depicted in (a). The Donnan potential is expressed in such a mathematical form [33]: 𝜑"#$ = 𝜑 & − 𝜑 ( =
*+ ,- .
ln
1-2
1-3
(1)
Where 𝜑"#$ is the Donnan potential, which is the potential difference between the membrane 𝜑 & and the solution 𝜑 ( ; R is the universal gas constant; T is the absolute temperature; F is the Faraday constant; 𝑧5 , 𝑎5( and 𝑎5& refer to the valence, and the activity in the solution and in the membrane of the ion i, respectively. The general conclusions of Donnan´s derivation are: the larger the fixed charge density is, the smaller is the co-ion sorption in the membrane phase; co-ion amount in the membrane phase increases with the external electrolyte concentration [33]. Until today the Donnan theory serves as the basis for the explanation of ion sorption phenomena in IEMs, despite some discrepancies between the experimentally-determined co-ion sorption and the values calculated according to the Donnan theory. This discrepancy has been thought to be due to the heterogeneous microstructure of IEMs [34, 35]. Recently, Kamcev et. al [36-38] suggest a model that combines the Donnan theory and ion activity coefficients in both the solution and the membrane phase. The Manning's counter-ion condensation theory [39] was used to calculate the ion activity coefficients in the membrane phase. The remarkable agreement between the predicted and experimentally determined co-ion sorption data demonstrates that the discrepancy mentioned above might be due to the simplification of ion activity coefficients in Donnan´s derivation [36]. 2.2. Microstructure of ion exchange membranes and ion transport mechanisms 7
The transport of ions through IEMs under the driving forces of concentration and/or electrical potential gradient can be phenomenologically treated with the solution - diffusion model, disregarding the complex transport mechanisms of ions through ion exchange membranes [6]. In this regard, the ion selectivity between two counter-ions could be understood in terms of the ion solubility (ion exchange) and the ion mobility in the membrane phase. The definition of the ion selectivity between two counter- ions A and B, 𝑃89 , will be introduced in Section 3.2. Membrane microstructure and ion transport in membranes For the quantitative treatment of IEM properties and transport processes, three types of models regarding the membrane microstructure have been suggested [40, 41]. The first type of models considers a membrane as a homogeneous solution, and it is based on such models that the classical Donnan effect was first elucidated [33]. The second type of models takes into account the structure inhomogeneity on the submicroscopic scale, the classical cluster - channel network model for Nafion membrane is among them [41]. The third is on the micro-phase scale. As shown in Fig. 3(a), a two - phase model of the third type divides the membrane structure into the gel phase (the grey part) and the interstitial phase (the blue part). The gel phase is composed of hydrophilic ion exchange groups, and the polymer chains the ion exchange groups are bound to; the interstitial phase is the void between elements of the gel phase. The source of the void could be interstices, pores and structure defects. The interstitial phase is also assumed to be the space filled with electro-neutral solution when the membranes are hydrated with salt solutions, and is also responsible for the transport of co-ions through IEMs. Zabolotsky et al. estimate that if the void radius exceeds 1.5 - 2.0 nm, the electro-neutral solution would appear in the inner part of the void [41]. It is necessary to stress that the void size is not a fixed membrane parameter, but rather dependent on the hydration level of the membrane as discussed by Kreuer [42]. Besides the gel and the interstitial phase, a third inert phase may also result from the hydrophobic parts of the polymer matrix, and this is noticed in perfluorosulfonic acid membranes especially at low hydration degrees [41]. The importance of the membrane water content, in terms of the volume fractions of water in the membrane, Φ; , in influencing the co-ion (mobile salt) sorption has also been pointed out by Freeman and his coworkers [14, 43]. A phenomenological model that considers the IEM as a sum of ideal Donnan portion and non-charged portion was used to fit the experimental coion (Cl-) sorption data of sulfonated polymer membranes [43]. The increase of Φ; in the same series of polymers leads to the decrease of the ideal Donnan portion´s influence on the co-ion sorption, even though the averaged volumetric fixed charge concentration actually increases with the Φ; in the sulfonated polymers [43, 44]. The results suggest that the increase of Φ; contributes to the nonuniform distribution nature of the fixed charges in IEMs [43].
8
Figure 3. Two - phase structure model and possible transport mechanisms in IEMs. (a) Two - phase structure model composed of the interstitial and the gel phase (modified from Ref. [45] showing a Nafion membrane of intermediate water content). (b) Possible mechanisms for the ion transport in IEMs. Diffusion, migration and convection occur predominantly in the interstitial phase of free water (Grotthus mechanism for H+ transport comes later in the next graph); surface site hopping of counter-ions occur along the fixed ionic groups at the interface (adapted from Ref.[46]). The exact transport mechanisms of ions in the membrane are complicated, which is exemplified in the transport of H+ in proton exchange membranes [47, 48]. For common ions, several possible ion transport mechanisms are illustrated in Fig. 3(b). Diffusion, migration, convection, and surface site ‘‘hopping’’ are the four possible mechanisms for the ion transport in the membrane phase. In the quantitative theoretical consideration of ion transport in IEMs, the extended Nernst - Planck equation describes the ionic flux 𝐽5 as the sum of three terms [31]: @A , .A " @C 𝐽5 = 𝑣𝐶5 − 𝐷5 - − - - - (2) @B *+ @B Where 𝑣 is the convective velocity of solvent (water); 𝐶5 , 𝐷5 and 𝑧5 are the concentration, the diffusion coefficient and the valence of the ion i, 𝑥 is the distance coordinate across the membrane. The first term represents convective transport of ions imposed by the electro-osmotic solvent (water) transfer, the second term ion diffusion due to concentration gradient, the third term ion migration as a result of electric potential gradient [30]a. Even though diffusion and convection can contribute to the ion transport in the membrane, their effect is believed to be small compared with migration when electric potential is involved as the driving force, since IEMs are dense and the electrolyte diffusion coefficient in the membrane is 1 to 3 orders of magnitude lower than that in the bulk solution [49, 50]. Despite certain limitations of the traditional Nernst - Plank equation (without the convective term) in describing ion transport through inhomogeneous media, the wide applicability of the equation has been proved [51]. Today, the extended Nernst - Plank equation serves as the basis of quantitative treatment. In the extended Nernst - Planck equation, the diffusion term is described by the Fick´s law. It comes to know that the Fick´s law is limited in describing the diffusion of multicomponent systems, especially in 9
describing the diffusion of individual ionic species [52]. In this case the Maxwell - Stefan approach is more general and suitable [52, 53]. Heintz and co-workers have derived the equations to determine the diffusion coefficients of counter-ions in IEMs, based on the Maxwell - Stefan formalism [54-56]. It is found that the counter-ion diffusion coefficients determined by this method are larger than the ones calculated with the Fick´s law, which is due to the fact that in the Maxwell - Stefan approach the friction between different counter-ions are taken into account [56]. In addition to the extended Nernst - Plank equation mentioned above, there are other two theoretical approaches that deal with the transport of ions in IEM systems [57]. One is based on the principles of thermodynamics of irreversible processes, the other is the theory of rate processes which is not often utilized now. While the approach based on the Nernst - Plank equation is restricted to isothermal systems, the irreversible thermodynamic approach can deal with the non-isothermal problems, and it is more rigorous and realistic [57]. However, one major limitation is that the phenomenological coefficients, 𝐿, in the models depend on the chosen frame of reference [57]. To solve this, the treatment utilizes the relative velocity difference of species to develop the frictional forces between species is suggested [57]. This idea is quite similar to the Maxwell - Stefan approach to describe diffusion. For the detailed discussion of these two approaches, the readers are referred to the excellent review of Lakshminarayanaiah [57]. Ions carry the current through IEMs, and it follows from Faraday’s law that 𝑖=
G 9
=𝐹
$ 5 𝑧5 𝐽5
(3)
Where 𝑖 is the current density; I is the current and A is the effective membrane surface area for the ion transport. Part of the current that is carried by a certain ion is expressed by the term transport number (𝑡5 ), which is given by 𝑡5 =
,- JK - ,- J-
(4)
The permselectivity of commercial IEMs for counter-ions is normally high when the external electrolyte concentration is smaller than the fixed charge concentration in the membrane, as indicated by Donnan’s derivation [33, 58]. This means the transport number of the counter-ions is close to 1 while the transport number of co-ions is approximately 0. Surface site ‘‘hopping’’ of counter-ions from one fixed ionic site to a neighboring one is also mentioned in literature as one possible mechanism for the ion transport, as illustrated in Fig. 3(b). The hopping of counter-ions is thought to be of secondary process on the basis of length scales [46]. Fixed ionic groups confined at the interface are ionized in the presence of water. It is observed that the effective dielectric constant (ԑ) in ion exchange resins is lower than that in the bulk solution, so the coulombic interactions are stronger in the resin phase [59]. The ionized fixed groups are also bound to water (forming a hydration shell), as are for mobile ions in the solution. The strong coordination of water molecules around the fixed ionic groups renders a reduction of the possibility of coulombic interaction between mobile ions and fixed groups. Calculations according to quantum-mechanics by Shaposhnik et al. predict 10
the substantial weakening of coulombic interaction between fixed ionic groups and mobile ions when water molecules are present between them [60]. Therefore, as a whole, the interaction between counter-ions and fixed ionic groups is influenced by both the coulombic force and the hydration effect of ions. In the investigations of Sata [6], the effect of alkyl substitutes of quaternary ammonium type AEMs on the ion selectivity between different anions is understood based on the correlation between the hydrophobicity of fixed ionic groups and the hydration energy of anions. Compared with monovalent ions, the possibility of simultaneous abstraction for multivalent ions from several fixed ionic groups to accomplish the hopping is very low [60], which is the reason for the high ion solubility but low mobility of multivalent ions in traditional IEMs. The distance between the neighbouring fixed groups, and ion exchange capacity are parameters that would influence the transport rate of counter-ions via the surface site hopping mechanism. The importance of distance between fixed groups for the monovalent and divalent ion selectivity has been pointed out in the ion exchange study of resins [61]. Data about the ion exchange kinetics in [61] would have provided more information about the transport processes within the resins. As might be noticed in the extended Nernst - Plank equation (Eq. 2), there is no term calculating the contribution of surface site hopping to the transport rate of ions. The wide validity of the extended Nernst - Plank equation in calculating ion transport in IEMs might suggest that the contribution from surface site hopping is negligible compared to the contribution from diffusion and migration of ions, even though there is evidence in the study of polyelectrolyte that counter-ions localized at the fixed charge groups are still mobile [62]. To which degree this surface site hopping contributes to the transport of ions in the membrane phase is still not clear yet. Very recently, Freeman and his co-workers have introduced the Manning`s counter-ion condensation theory to account for the non-ideal thermodynamic effect (ion activity coefficient) in the membrane [36, 37, 63], and predicted successfully the salt permeability in highly charged and swollen IEMs via the classical solution – diffusion model [64, 65]. In their studies, the salt (NaCl, MgCl2) partition coefficients could be predicted with Manning theory to calculate the ion activity coefficient in the membrane, together with the Donnan theory for ion partitioning between the membrane and the contiguous solution [36, 37]. For the diffusion of salt in membranes, they considered the effect of tortuosity of ion diffusion paths in membranes and the effect of electrostatic interactions between the mobile ions (both counter- and co-ions) and the fixed charge groups of the membrane [64]. The tortuosity effect is described by the Mackie and Meares model, with water volume fraction in the membrane, Φ; , as the key parameter to calculate the ion diffusion coefficient in the membrane [66]. At the same time, the ion diffusion equations derived by Manning are used to account for the effect of electrostatic interactions [39]. The Manning’s counter-ion condensation model proposes one dimensionless linear charge density (of the polyelectrolyte polymer chain), 𝜉, defined as [39]: 𝜉=
MN
OPQR QS+T
=
UV T
(5)
Where 𝑒 is the electron charge, 𝜀# the vacuum permittivity, 𝜀 the dielectric constant, 𝑘 the Boltzmann’s constant, 𝑇 the absolute temperature, 𝑏 the average distance between the fixed charge groups on the polymer chain, and 𝜆8 the Bjerrum length. In the theoretical framework proposed by Freeman and his co-workers, only three important parameters describing the IEM property are present in the simplified analytical expression for salt permeability coefficients [64, 65]. These three membrane property parameters are the dimensionless linear charge density ( 𝜉), the concentration of fixed charge groups ( 𝐶9&,^ ) and water volume fraction in the membrane ( Φ; ) [64]. The importance of membrane fixed charge group concentration, water contents and the distance between neighboring fixed charge groups is described quantitatively with these three parameters. Because in concentration-gradient driven transport process, the diffusion of salt (counter-ion and co-ion pairs) in IEMs is almost controlled by the 11
co-ion diffusion [67]. So it is not unexpected that electrostatic interactions, compared to the tortuosity effect, account for a relatively small portion (5 – 15%) of diffusion coefficient decrease relative to values in aqueous solution [64]. While the approach is very rigorous and gives systematic insight into the concentration of ions inside the bulk membranes as well as their molecular mobility, the approach for the moment cannot account for effects of ion selectivity of different materials, nor the substantial effect of membrane surface properties as elaborated below in great detail. For H+ and OH-, besides the transport mechanisms illustrated in Fig. 3(b), an extraordinary transport mechanism exists [68]. In a solution, the H+ is transported mostly from one hydronium ion to the next water molecule via re-construction of the H-bond network as indicated in Fig. 4(a), and only to a small extent are H+ ions transported as individual ions surrounded by a hydration shell [31]. In the confined aqueous environment of the interstitial phase in IEMs (Fig. 3(a)), the proportion of the total transport that structural diffusion or simple diffusion has is dependent on the hydration level of the membranes: structural diffusion contributes most at high hydration degree, and the contribution from simple diffusion increases as the water content decreases [45]. For OH- transport in a solution, a similar but slightly different mechanism holds. As shown in Fig. 4(b), OH- is also transported through local recombination of the H-bond network. This process is accompanied by the involvement of one water molecule and the leave of another water molecule from the H-bond network, and the hydration situation for OH- ion is also different from that of H+ ion. These mechanisms make the transport of H+ and OH- special in IEMs. It is also the structure diffusion mechanism and the small size of the ions that makes the ‘H+ leakage’ through AEMs and ‘OH- leakage’ through CEMs possible.
Figure 4. Mechanisms of H+ and OH- transport in the solution. (a) H+ transport through the H-bond network of water [31], which is called Grotthuss mechanism or structural diffusion. (b) OH- transport in aqueous solution by fluctuations in the local H-bond network [69]. Influence of the boundary layers The counter-ion transport number is not only a material property of an IEM, but depends also on the operational and process conditions. The operational conditions will have influence on the concentration boundary layers adjacent to a membrane, namely the influence of concentration polarization [70]. As predicted by the Donnan effect, the co-ion leakage increases with the electrolyte concentration in the 12
bulk solution [33], and becomes profound when the electrolyte concentration in the solution is close to the fixed charge concentration in the membrane. In technical processes with IEMs, such as electrodialysis, the local solution concentration at the membrane - solution interface is strongly influenced by the current density and fluid dynamics [49]. With the passage of a current through the system, a concentration boundary layer featuring low electrolyte concentration builds up at the depleted side of the membrane, as illustrated in Fig. 5 (a). Further increase of the current density decreases the boundary layer concentration Cs furthermore until a certain point at which the electrolyte concentration at the membrane - solution interface becomes zero, as depicted in Fig. 5 (b). The corresponding current density is called the limiting current density of the system, 𝑖_5& .
Figure 5. Schematics of concentration boundary layers adjacent to IEMs. (a) Salt concentration profile in the two diffusion boundary layers (DBL) adjacent to a CEM and cation concentration in the membrane. The concentration denoted in the membrane phase is the cation concentration. (b) An enlarged view at the depleted side of (a) showing cation and anion concentration profiles of a CEM at limiting current density of the system. The DBL can be divided into a space charge region deviated from local electroneutrality, and an electroneutral region with linear concentration profiles of both ions (simplified from [49, 71]). The influence of these boundary layers on the membrane permselectivity (for counter-ion and co-ion) is not thoroughly studied yet, possibly due to two reasons: (a) the applied current density in the measurement of the counter ion transport number is normally well below the limiting current density of the system; (b) complex phenomena accompany the emergence of limiting current density such as water splitting, membrane discharge, and electro-convection. Abu-Rjal et al. [72] analyzed numerically the effect of concentration polarization on permselectivity with a 1-dimensional three-layer membrane system model at steady-state, similar to Fig. 5 (a). Their simulation results concerning the effect of boundary layer thicknesses, fixed membrane charge density and current density reveal the following 13
observations [72]: with the increase of current density the counter-ion transport number gains at the depleted side of the membrane however loses at the enriched side of the membrane, and the overall change of transport number is dependent on the thickness ratio between the enriched and depleted boundary layers, 𝐿. When 𝐿 = 0.1 (thin enriched boundary layer), the overall counter-ion transport number increases with the current density, and this increase is more significant for membranes with smaller fixed charge density [72]. Under the opposite condition when the enriched boundary layer is thicker (𝐿 = 10), the transport number decreases with the current density [72]. For symmetric boundary layer thicknesses on both sides of the membrane, the transport number is nearly independent of the current density [72]. This effect of boundary layer thickness also implies the influence of fluid dynamics, as the boundary layer thickness is determined by both the current density and fluid dynamics. The authors also investigated the additional influence of membrane heterogeneity on the permselectivity, and found similar results [72]. For systems with binary counter ions, the influence of boundary layers on the ion selectivity has been studied theoretically and experimentally [73-76]. Zabolotsky et al. [74] investigated the competitive transport of Ca2+ and Na+ in the overlimiting current density regime, by extending a mathematical model taking into account the effect of the space charge region. As shown in Fig. 5 (b), the space charge region is one part of the depleted diffusion boundary layer which has deviation from the local electro-neutrality near the depleted solution - membrane interface ̶ it emerges only when the current density is equal to or above the limiting current density [74]. The other parts of the depleted diffusion boundary layer are near the solution bulk and have linear concentration profiles for both counter- and co-ions. As shown in Fig. 6 (a), the calculated transport number of Ca2+ decreases with the current density as a result of the boundary layer development, while the transport number of Na+ has an opposite trend [74]. When the current density is close to or larger than the limiting one, the transport number ratio (counter-ion selectivity) approaches a stable value, which is independent of the membrane properties and also the space charge region, and is determined solely by the transport numbers in the region of depleted diffusion boundary layer that has linear concentration profile [74]. The transport numbers in this region are controlled by the diffusivity of the Ca2+ and Na+ ions in the solution (Table 4). Normally, a stationary boundary layer of a defined thickness is considered in these theoretical calculations. In a practical application with IEMs, such as electrodialysis, the boundary layer conditions of a membrane are determined by the local hydrodynamics induced by spacers under a specific fluid velocity. Further, Kim et al. [75] did similar calculations, also considering the alteration of boundary layer thickness with a statistical standard deviation. Their simulations indicate that large standard deviations in boundary layer thickness substantially decrease the transport number of monovalent K+ ions in a binary counter ion system with K+ and Ca2+ ions [75]. Meanwhile, both their simulations and electrodialysis experiments confirm the transport number increase of K+ ions in the binary counter ion system with the current density (cell pair potential drop) as shown in Fig. 6 (b, c), and that the ion selectivity in the overlimiting current density regime is determined by the diffusivity of the two ions in the depleted solution boundary layer near the membrane [75].
14
Figure 6. (a) Calculated transport numbers of Ca2+ and Na+ in a CEM system as a function of the current density ratio 𝐼 (reproduced from [74]). (b) Current density vs voltage curves for electrodialysis experiments with a binary K+ and Ca2+ system for varying electrolyte flow rates and (c) the K+ transport number increase as a function of the applied voltage (reproduced from [75]). The surface hydrophilicity of IEMs is sometimes considered for the explanation of ion selectivity alteration. However, the actual reason would be the density of ion exchange groups on the membrane surface (reflected by the overall hydrophilicity). Since normal IEMs surface have both hydrophilic and hydrophobic parts at microscopic scale, this is especially true for old-type heterogeneous IEMs. The hydrophobicity increase of the membrane surface (actually the hydrophobic parts) could possibly induce flow-focusing and early emergence of electro-convective flow in the overlimiting current density regime [49]. This would have an influence on the ion selectivity, however the exact trend is still unclear. There is A_ g another interesting idea: the ion selectivity 𝑃de Ng increase observed by the hydrophobization of ion f exchange groups in AEMs was interpreted by the increased energy barrier of partial dehydration for ions when they transport from the solution into membranes possessing more hydrophobic anion exchange groups [77]. These interpretations all need further investigations to develop a comprehensive picture of ion selectivity. 2.3. Ion selectivity order The ion transport rate through IEMs is determined by both the ion exchange (solubility) and the mobility of ions in the membranes. The ion exchange is related to the valence and size of ions: ions with higher valence or ions of larger size when the valence is the same are preferentially ion exchanged into IEMs [78]. The mobility of ions in the membrane matrix depends on the size of the ions and also their interaction with the fixed ionic groups. Tables 3 and 4 list characteristic properties of common anions and cations in aqueous solution, respectively. Even the reported hydrated radius of ions varies in literature as summarized in Ref [79], so the choice of values of the effective hydrated radius of ions is very important, and the Stokes radius of ions suggested in Ref. [80] is widely adopted. As can be seen from Table 3, the electrochemical mobility of monovalent anions at infinite dilution (µ) is inversely proportional to the Stokes radius, just as defined [81]; divalent anions also have high mobility due to stronger coulombic force even though they possess larger Stokes radius. A permselectivity order for the anion transport through normal AEMs is hereby generalized as follows: I- > (NO3- ~ Br-) > NO2- > Cl- > OH- > SO42- > F- 15
This order is concluded from systematic electrodialysis experiments with two kinds of AEMs by Sata et al. [6], and bi-ionic membrane potential measurements from AEMs based on brominated poly(2,6dimethyl-1,4-phenylene oxide) (PPO) and pyridine [82]. It is obvious that this order follows neither the trend of Stokes radius, nor the ion valence (ion exchange), let alone the self-diffusion coefficients. However, it seems that this anion permselectivity order correlates well with the Gibbs hydration energy of anions [6] and also follows the Hofmeister anion series [83], except for the divalent sulfate anion. The Hofmeister ion series, generally known for representing ions’ ability to change the precipitation behavior of different materials in a colloidal system and revealed first by F. Hofmeister, actually stands for the specific ion effect prevalent in vastly diverse systems [83-85]. The definition of the ion selectivity between two counter-ions and the experimental methods to measure it will be introduced in Section 3.2. It is necessary to note that the transport preference for NO3- and Br- is very close and the actual order varies even among the three types of membranes mentioned above. Table 3. Characteristics of anions in aqueous solution anions 𝑟( [80] −∆jkl 𝐺 # [6] −∆no@ 𝐺 # [86] µ[87] 𝐷[88, 89] -8 2 -1 -1 -1 -1 Å (10 m s V ) (10-9 m2 s-1) (kJ mol ) (kJ mol ) OH- (0.46) 430 20.64 5.27 Br 1.18 303 315 8.09 2.01 I- 1.19 257 275 7.96 2.00 Cl 1.21 317 340 7.91 2.03 NO2 330 1.91 NO3- 1.29 270 300 7.40 1.90 SCN 280 1.76 - F 1.66 434 465 5.70 1.46 S2O32- 1.13 CH3COO 365 4.24 1.09 2SO4 2.30 1000 1080 8.29 1.07 C2O42- 0.99 2CO3 2.66 1315 7.46 0.96 # 𝑟( , Stokes radius of ions; ∆jkl 𝐺 , standard Gibbs hydration energy of ions; µ and 𝐷, electrochemical mobility and self-diffusion coefficients of ions at infinite dilution in the solution. The temperature is 25 o C. Li et al. also studied the permselectivity of different monovalent anions (F-, Cl-, Br-, I-, OH-, NO2-, NO3-) through AEMs based on bromomethylated PPO (BPPO) polymers [90]. Three series of AEMs were prepared by amination of BPPO with pyridine (A series), cross-linking with ammonia and amination with pyridine (B series), further amination of the B series with trimethylamine (C series). The permselectivity evaluated by the membrane bi-ionic potential of different anion pairs (cf. 3.2.2) confirmed part of this anion selectivity order. The study also confirmed the vital role of membrane hydrophilicity and hydration energy of anions over the effect of Stokes radii of anions in determining the ion selectivity [90]. For common cations, a general transport order through normal CEMs with fixed sulfonic acid groups is hereby generalized, based on the flux comparison of ions in electrodialysis with mixed binary salts: 16
Ba2+ > Sr2+ > Ca2+ > Mg2+ > H+ > (Cu2+ ~ Zn2+ ~ Ni2+) > K+ > Na+ > Li+ > Fe3+
The detailed transport number ratios, membrane types and experimental conditions are provided in the Supporting Information. The three cations in the bracket have quite close fluxes through the CEMs ̶ the actual selectivity order varies among CEMs of different types. For ions of the same group in the periodic table, smaller ions are preferentially transported across the CEMs (Ba2+ > Sr2+ > Ca2+ > Mg2+, K+ > Na+ > Li+). Due to the very low mobility in the membrane matrix, Fe3+ lies behind monovalent cations even though Fe3+ has very high solubility in the membrane matrix. From this order, it can be seen that an ion with higher valence and smaller Stokes radius has higher permselectivity over an ion with lower valence and larger Stokes radius. Table 4. Characteristics of cations in aqueous solution at 25 oC −∆no@ 𝐺 # [86] cations 𝑟( [80] 𝜇 [87] 𝐷 [88] -8 2 -1 -1 -1 Å (10 m s V ) (10-9 m2 s-1) (kJ mol ) H+ (0.28) 1050 36.23 9.31 + NH4 1.25 285 7.63 1.98 + K 1.25 295 7.62 1.96 Ag+ 1.48 430 6.24 1.66 + Na 1.84 365 5.19 1.33 + Li 2.38 475 4.01 1.03 Pb2+ 2.83 1425 0.945 Ba2+ 2.90 1250 0.848 2+ Ni 2.92 1980 0.679 Sr2+ 3.10 1380 0.794 2+ Ca 3.10 1505 6.17 0.793 2+ Cu 3.25 2010 5.56 0.733 Fe2+ 3.44 1840 0.719 2+ Mg 3.47 1830 0.705 2+ Zn 3.49 1955 5.47 0.715 3+ Fe 4.06 4265 0.607 Al3+ 4.39 4525 0.559 This selectivity order of cations through CEMs with fixed sulfonic groups is quite similar to the ion exchange order of these cations in sulfonic CEMs as concluded by Strathmann [31]a: Ba2+ > Pb2+ > Sr2+ > Ca2+ > Mg2+ > Ag+ > K+ > NH4+ > Na+ > Li+ The two ions whose permselectivity has not been investigated by means of electrodialysis (Pb2+ and NH4+) may also possess the same positions as they are in the ion exchange sequence. There might be one exception that Ag+ would fall behind Li+ in the cation selectivity order, because Ag+ was found to diffuse very slowly in resins of fixed sulfonic groups [58]. The similarity between the selectivity order and ion exchange order originates from the determining role of ion exchange in ion transport through IEMs under mild conditions, which means the current density is well below the limiting one in electrodialysis. 17
There has been continuous effort to model the ion exchange (ion partition) of single and multicomponent cation mixtures into CEMs like Nafion 117 by Pintauro and his co-workers [91-93]. In their partition coefficient model, all the mutual interactions, the cation – membrane (electrostatic), membrane – water (change of the water dielectric constant) and cation – solvent (ion hydration effects) are included [92]. Despite that the ion activity coefficients in the membrane are simplified to unity, this model predicts nicely the experimental results of preferential partition of alkali metal cations [92]. One important point about this model is the approach to calculate the Gibbs free energy change associated with the transfer of ions from the bulk solution to the membrane pore fluid where the dielectric constant is less than in the bulk. The authors find the linear correlation between this free energy change q q and the difference of the reciprocal dielectric constant of the solvent ( − , 𝜀T is dielectric constant in Q
QT
the bulk) [92, 93]. Actually the effect of this difference in dielectric constant between the bulk solution and the pore liquid is closely related to the so-called dielectric exclusion [94], which has been incorporated into the mass transport models for nanofiltration membranes [95]. This free energy change from the bulk solution to the membrane pore fluid is more realistic than the Gibbs hydration energy of ions (from vacuum to bulk solution, shown in Tables 3 and 4) in correlating ion transport selectivity. As a result, this approach could be applied to anions [96] to further test the significance of anion hydration in controlling its transport selectivity. For anion transport through AEMs with quaternary ammonium groups, the anion permselectivity order (shown above) is exactly the same as the ion exchange sequence [31], and that order is consistent with the Hofmeister series [97], except for the SO42- anion. Though ClO4- transport through normal AEMs has not been studied by electrodialysis yet, it can be expected that ClO4- is even more selectively transported than I- as implied from the results of membranes with mobile quaternary ammonium carriers for anions [97]. The study of ClO4- permselectivity by electrodialysis would be interesting for the ion exchange membrane bioreactor (IEMB) application, together with the transport selectivity of BrO3-, HPO42- ions that have not been studied thoroughly. It is worthwhile stressing that the cation and anion selectivity orders are generalized based on electrodialysis experiments with equivalent amount of two counter-ions in the underlimiting current density regime. It does not necessarily mean the relative amount of counter-ions transported in a specific system, which is also influenced by the current density as discussed before (boundary conditions, cf 2.2.2), and the relative concentrations of ions in the system. Although the Hofmeister effect is prevalent in a variety of different electrolyte-based systems [83] and is generally more pronounced for anions than for cations [84], it is interesting that the generalized anion transport selectivity order through IEMs almost follows the Hofmeister series, while the cations not. The original thought was that the Hofmeister effect might be caused by the change of bulk water structure induced by the electrolyte (cations and anions) [98]. However, modern experiments and simulations have made it clear that the Hofmeister effect is more likely a result of the direct ion – macromolecule interactions and interactions with water molecules in the first hydration shell of the macromolecule [98]. Nowadays, the underlying physicochemical origin of the Hofmeister effect is still a topic under research, as discussed in a recent Perspective [85]. With newly developed experimental methods to probe the anion and cation concentration difference at the air - ion solution interface, and with simple models to decouple the ion - ion interaction (electrostatic) and ion - water interaction, Song et al. could experimentally quantify the energy difference of an ion in the solution bulk and at the air - ion solution interface [85]. By comparing the experimental energy difference and theoretical ones based on different models, it is speculated that anions consisting of a single atom (Cl-, Br-, I-) have different hydration 18
structure from that of NO3- anion [85]. It is also experimentally demonstrated that the ion - water interaction is asymmetric, dependent on the sign of ions, which in turn means that the interaction is at least partially electrostatic [85]. Although several models have been proposed to describe the ion - water interaction, including image charge potential, dispersion energy, hydrogen-bond breaking energy, the nature of the interaction is more complicated beyond the scope of any of these models [85]. Even though the nature of the ion – water interaction is not fully clear [85], a simple empirical rule that matches water affinity of different ions has been proposed by Collins as an effective way to explain the specific ion effect [99]. As shown in Fig. 7, ions to the left of Cl- are called cosmotropes, and those to the right are referred to as chaotropes [84]. This differentiation is related to the hydration of ions [84] (Table 3). The cosmotropes are strongly hydrated in water bulk, and they were believed to be ‘water structure makers’ (but has been disproved) [84]. While the chaotropes, ‘water structure breakers’, are less hydrated [84]. The charge density of ions is central to its hydration, and an ion with large charge density and resulting strong hydration can be considered as a cosmotrope, as compared to the reference ion (Cl- for anions) [99]. The Collins rule suggests that the interaction between ion pairs of similar hydration behavior is strong, for example chaotropes with chaotropes [99, 100]. But the chaotropes do not come into close contact with cosmotropes [99, 100]. The underlying idea is that the strength of ion – counter ion interaction is correlated to the strength of ion – water interaction [100]. The Collins rule is successful in explaining the specific ion effect in various systems [98, 101]. This empirical rule was extended by Kunz et al. to ionic groups with large alkyl chains, as shown in Fig. 7 [98]. The trimethylammonium headgroup is classified as a chaotrope in cation series, and the large arrow in Fig. 7 indicates the direction in which the interaction strength between the anions and the trimethylammonium headgroup increases [98]. Because this quaternized headgroup is chemically similar to the fixed positive charge groups in AEM, so we assume the Collins rule is also applicable to IEM systems. The difference of the anion and cation transport selectivity through IEMs is therefore, in our opinion, partially due to the hydration difference of fixed anion exchange group (ammonium) and cation exchange group (sulfonic). As inferred from the smaller hydration free energy of NH4+ cation (Table 4) compared with half the hydration free energy of SO42- (Table 3), the hydration free energy of substituted ammonium groups in AEMs might be very likely smaller as comparison to the sulfonic groups in CEMs. This difference makes the counter ion hydration contribute to the ion transport, to different degrees among other factors in CEMs and AEMs.
Figure 7. The Hofmeister series (for anions) with the classification of cosmotropes and chaotropes (reproduced from [84]). The ordering of anions (shown in italic) with respect to their affinity for the 19
trimethlyammonium headgroup in surfactants (modified from [98]), which is similar to the fixed positive charge groups in AEMs. The ordering is based on Collins empirical rule that ions of similar water affinity have strong interaction [99, 100] (indicated by the vertical arrow). 3.
Experimental methods to determine the permselectivity of ions
3.1. The permselectivity of counter-ions in ion exchange membranes The permselectivity of a counter-ion through an IEM is defined as [102, 103]: 𝑃 =
r-3 srqsr-
(6)
where 𝑡5& , 𝑡5 are the transport numbers of the counter-ion in the membrane phase and in the bulk solution, respectively. There are two general types of analytical methods to determine the 𝑡5& : calculations from the membrane concentration potential (potentiometric method) or electrodialysis [30]b. Besides, it is thought that chronopotentiometry can also be applied to evaluate the 𝑡5& [104]. The potentiometric method is based on the static concentration potential across an IEM when the two surfaces of the membrane are in equilibria with two solutions of the same electrolyte however possessing different concentrations. The static concentration potential is generated due to the interdiffusion of counter-ions across the IEM. In the study of Gohil et al., a membrane concentration potential was formed when 50 mM electrolyte on one side and 5 mM electrolyte on the other side of the membrane were placed [103]. The 𝑡5& value can be calculated from the membrane concentration potential ∆𝐸 by the following relation [103]: *+ 1 ∆𝐸 = (2𝑡5& − 1) ln x (7) ,.
1y
where 𝑎z and 𝑎@ are the activities of the concentrated and the diluted solution, respectively; z is the valence of the counter-ion i. Eq. 7 is valid when a 1:1 type electrolyte (for example NaCl) is used, for a 2:1 (for example MgCl2) type electrolyte the membrane concentration potential is as follows [30]b: { *+ 1 ∆𝐸 = ( 𝑡5& − 1) ln x (8) |
,.
1y
From the membrane potential data, the permselectivity 𝑃 is obtained according to Eq. 6 provided the corresponding 𝑡5 value is known. The transport number of ions in the bulk solution 𝑡5 can be calculated from the ion concentrations and respective self-diffusion coefficients in the bulk solution, as discussed in [57, 105]. Another way to estimate the permselectivity by the membrane potential measurement was suggested [106]. Two cells are separated by the membrane under investigation, through one cell flows a 0.1 M NaCl solution and through the other cell a 0.5 M NaCl solution. The membrane is pre-conditioned in a 0.1 M NaCl solution for 24 h outside the membrane cell. The experimental concentration potential across the membrane at steady state ∆𝐸MB} is recorded. In practice, this ∆𝐸MB} should be offset by the 20
potential difference between the two reference electrodes, which can be determined when both reference electrodes are placed in the same 0.5 M NaCl solution [44]. Then the permselectivity of the counter-ion (Na+ for CEM, Cl- for AEM) can be estimated as [106]: 𝑃 =
∆~•€• ∆~‚ƒ•
(9)
The theoretical concentration potential ∆𝐸rnM is calculated from the Nernst equation for the 1:1 type NaCl: *+ 1 ∆𝐸rnM = ln x (10) .
1y
The experimental factors in this kind of permselectivity measurement, including variations in temperature, inaccurate solution concentrations and fluctuations in membrane potential measurements, were investigated by Ji et al. to study their influence on the measured permselectivity.[107] Temperature had a small influence, by approximately 2%, on the measured permselectivity of two commercial CEMs [107]. In generally, the influence of these experimental factors on the measured permselectivity is small, and comparable to the magnitude of variability between different replicate measurements [107]. The second approach to measure the permselectivity is the electrodialysis method, of which the traditional one is called the Hittorf method. The calculation is based on the amount of a counter-ion transported through the membrane under investigation and the amount of electricity passed through the electrochemical cell. The following relation correlates the 𝑡5& with the experimental variables: ,.∆$ 𝑡5& = (11) 59r where ∆𝑛 denotes the amount of counter-ions transported through the membrane, 𝑖 the current density, 𝐴 the effective membrane surface area and 𝑡 the duration of the electrodialysis experiment. The major objective of the experiments is determining the ∆𝑛. A setup for the Hittorf method is a two-compartment cell with reversible electrodes like the Ag-AgCl electrodes (Fig. 8 a). This set-up can only be used for limited types of salts because of the electrode reaction. The electrodialysis cell made up of four or six compartments can be applied for all types of salts. The membrane under investigation is placed in the middle of the six-compartment cell as illustrated in Fig. 8 b [108]. The cell has one pair of auxiliary compartments (1 and 6) specifically for the electrode reaction, and another pair of compartments near the auxiliary compartments (2 and 5) to eliminate the influence of H+ and OH- on the pH of the two central compartments whose ion concentrations are to be analyzed. A cation exchange membrane (CEM) near the anode can prevent the entrance of Cl- into the anodic compartment and the consequent evolution of chlorine. It is preferable to use one solution of the same concentration in the two compartments adjacent to the central membrane (3 and 4), to diminish the effect of electrolyte diffusion through the membrane on the ion transport number during the electrodialysis experiments.
21
Figure 8. (a) Illustration of a two-compartment cell with reversible Ag-AgCl electrodes used in the Hittorf method (Reproduced with permission from [30]). (b) A six-compartment cell with four auxiliary membranes for the investigated CEM placed in the middle (Reproduced with permission from [108]). The ion transport number measured by the membrane potential method is smaller than the one measured by the electrodialysis method, which is due to the presence of water transport through a membrane during the dynamic electrodialysis measurement [30]b. When the water transport is corrected for the permselectivity calculated from the measured membrane potential, the transport numbers obtained by both methods are almost identical [30]b. Actually, there could be an eigenvalue of the transport number for a certain counter ion - membrane system, when the transport number in the membrane phase is defined as [109]: ,.∆$ [𝑡5& ]∗ = [ ] (12) 59r ‰z,‰}Š‹ This is the transport number provided there are no concentration and pressure gradients in the membrane, and it is called electro-migration transport number [109]. This transport number can be determined from the two experimental methods when certain corrections are made for each experimental value. However, this generally is not considered in experiments. Normally, the potentiometric method is less influenced by the water transport and ion diffusion during the measurement. When the solution concentrations on both sides of the membrane are small, there are considerable random errors in the membrane potential measurements [109]. The electrolyte frequently used in the potentiometric method is NaCl, and the determined permselectivity is for Na+ in the case of CEMs and for Cl- when AEMs are studied. If a different electrolyte is used, the measured permselectivity alters to different extents that depend on both the nature of counter- and co-ions [105]. So, ideally, the permselectivity should be measured with the electrolyte dominant in the application that the IEMs are going to be applied. On the other hand, in electrodialysis one should be careful in choosing the current density to ensure that the measurement is performed under the overlimiting current density of the system (cf. Section 2.2.2). For both methods, the solutions in both cells adjacent to the membrane under investigation should be vigorously agitated during measurement to eliminate the effect of the diffusion boundary layers on the measured transport numbers [30]b. A less widely adopted method for the transport number determination is the chronopotentiometric study of IEMs. In a chronopotentiometric study, the membrane potential is followed at the instant of a 22
certain current density 𝑖 is applied, and the transition time 𝜏 at which the electrolyte concentration at the membrane surface declines to zero, can be obtained from the potential - time curve. The counter- ion transport number in the membrane phase 𝑡5& is calculated by the following equation [70, 110]: P" A ,. q 𝜏 = ( 3R ) N (13) O
r- sr- 5
where 𝐶# is the initial electrolyte concentration, D the diffusion coefficient of the counter-ion, 𝑖 the current density. Besides the experimentally determined transition time, D is also needed for the calculation of 𝑡5& . Similar to the dynamic electrodialysis measurement, 𝑡5& determined by chronopotentiometry is also larger than those determined by the static membrane potential measurement. This difference is attributed to either the water transport across the membrane or the membrane surface heterogeneity [104]. 3.2. The ion selectivity between different counter-ions As for the measurement of ion selectivity between different counter-ions, the basic principle is the same as for a single counter-ion. When the co-ion is the same, the measured transport number of individual counter-ion through the same IEM can be indicative of the relative ion selectivity between these different counter-ions. However, the virtual ion selectivity between different counter-ions should be measured with mixed salt solution of these two ions, because the interaction between one counter-ion and the IEM will be influenced when another counter-ion is simultaneously present in the system [111]. The ion selectivity between two counter-ions A and B, 𝑃89 , is defined as [5, 6]:
𝑃89 =
r• rV A• AV
(14)
where 𝑡5 represents the transport number of ion 𝑖 in the membrane phase, 𝐶5 the concentration of ion 𝑖 at the membrane surface of the desalting side. To facilitate the comparison of the ion selectivity among different ions, Na+ and Cl- ions are generally selected as the reference ion for cations and anions, respectively [5, 6]. Below, the methods for the static and dynamic transport number measurements are reviewed. Electrodialysis with a mixed salt solution Electrodialysis directly with the two counter-ions of interest and the same co-ion in a four- or sixcompartment cell is used to evaluate the ion selectivity between these two counter-ions. A fourcompartment cell illustrated in Fig. 9 [112] was employed for the measurement of the anion transport number. The two electrodes 1, 4 are Ag-AgCl electrodes to supply a current. Electrodialysis is performed at a constant current density for a known time period at constant temperature. The concentration change of counter ions in the two middle compartments is analyzed and the ion selectivity is calculated according to Eq. 14. If inert electrodes are preferred, the six-compartment cell as shown in Fig. 8 b should be used instead [108].
23
Figure 9. Illustration of a four-compartment cell for the measurement of the ion selectivity between Cl- and a second anion X- (SO42-, NO3-, F-, Br-). 1, 4 are reversible Ag-AgCl electrodes; 2, 3 are Ag-AgCl wire probe electrodes (reproduced with permission from [30]c). The experimental conditions that need to be considered in the electrodialysis to evaluate the ion selectivity include the salt concentration, current density, temperature and time period. In the systematic investigations based on the interpolymer membrane, Sata et al. performed the electrodialysis under such experimental conditions: for anions, with 1:1 mixed salt solution (concentration 0.01 N or 0.04 N) at a current density of 1 mA cm-2 for 1 h at 25 oC under vigorous agitation [6]; for cations, also 1:1 mixed salt solution of concentration 0.250 N at a higher current density of 10 mA cm-2 and otherwise identical conditions as for anions [5]. The simultaneous consideration of the salt concentration and the current density should guarantee the experimental current density is below the limiting current density of the system. The salt concentration should not be too low in case that selective sorption of ions into the IEM dominates the ion selectivity, therefore the mobility characteristics of ions through the membrane phase is concealed. During the electrodialysis, the salt concentration at the desalting side of the membrane is actually decreasing with time, and 𝐶5 in Eq. 14 is the averaged salt concentration. So, the electrodialysis period is suggested to be just enough to observe considerable concentration change of ions in the solution, but not so long that quite differing salt concentrations at the desalting side exist during the investigated period. The bi-ionic membrane potential Similar to the emergence of the concentration potential across IEMs, a bi-ionic membrane potential is generated when both surfaces of an IEM are in contact with solutions containing different counter-ions [113-115]. This bi-ionic potential is due to the difference in the inter-diffusion rates of the two counterions across the IEM, and the generated bi-ionic membrane potential is rather a stationary value other than a thermodynamically stable one [114]. Even though a relation between the bi-ionic potential ∆𝐸+ and the transport numbers of the two counter-ions in the membrane phase was derived [115], the concentration gradient in the membrane and Donnan potentials at the two membrane - solution interfaces were not included, which made it difficult to relate directly the ion selectivity between two counter-ions in the electrodialysis to the ion selectivity obtained by the bi-ionic potential [30]d. Nonetheless, with suitable assumptions and carefully selected experimental conditions, the ion selectivity estimated from the bi-ionic potential can also provide valuable information [114]. 24
Xu et al. developed a simple method based on the bi-ionic membrane potential and applied it for the ion selectivity measurement for anions [82]. By assuming a linear concentration distribution in the membrane phase, the bi-ionic potential across the membrane ∆𝐸+ is calculated as the sum of two interfacial Donnan potentials and the diffusion potential within the membrane. The experimental cell is composed of two half-cells separated by an AEM in the middle (cf Fig. 8, a), the two half-cells are filled with electrolyte YA and YB of the same cation Y. For the two anions, assuming the charge 𝑍9 ≤ 𝑍8 , a correlation was found [82]: exp
“• . s∆~” *+
− 1 = 𝑇89
“• A•,•
“V AV,•
(15)
where 𝐹, 𝑅 and 𝑇 have the usual meaning, 𝐶5,— means the concentration of counter-ion 𝑖 on the left side; 𝑇89 is the ion selectivity between counter-ions A and B, and is defined as: ˜ ™ ™ 𝑇89 = • • V (16) ˜V ™V ™•
where 𝜇5 denotes the mobility of counter-ion 𝑖 in the membrane phase, 𝑞5 and 𝑞5 are the activities of counter-ion 𝑖 at one surface of the membrane and in the corresponding solution, respectively. Compared with the ion selectivity definition by Sata [6], 𝑇89 is actually 𝑃89 .
Before the bi-ionic potential measurement, the membranes are conditioned in 1 M YB solution for 2 days. In the experiment, both half-cells are filled with 100 ml YB solution of 0.01 M concentration in the beginning. Both half-cells are free of YA in the beginning. Then the concentration of YA in the left halfcell is increased gradually by adding solutions of YA while at the same time adding pure water in the right half-cell to keep the solution volume on both sides equal, after which a series of bi-ionic membrane A potentials ∆𝐸+ and the corresponding concentration ratio •,• are acquired. Finally, the ion selectivity AV,• “• . s∆~” 9 𝑇8 can be gained from the linear regression of [exp *+
− 1] vs
A•,• AV,•
as shown in Fig. 10 [82].
25
Figure 10. The linear relationship between the bi-ionic membrane potentials and concentration ratios for the 1:1 type counter-ion pairs. The reference anion B- is Cl-, and the co-ion is Na+. The ion selectivity between different anions can be calculated from Eq. 15 (reproduced with permission from [82]). For 𝑍9 = 𝑍8 , it does not matter which salt solution is used for the membrane conditioning; however for the counter-ion pairs 𝑍9 < 𝑍8 , the membrane should be conditioned with YB solution. Because during the derivation of Eq. 15, an assumption is made that the activities for the counter-ion of higher valence (𝑍8 ) are equal on both sides of the membrane. Second, because the activity of the electrolyte is replaced by the concentration during the derivation of Eq. 15 and the absence of co-ions in the membrane is also assumed, so the concentration of YB should not be high. The validity of this method was also verified by common 2:2 and 2:1 type counter-ion pairs. This method is very simple and easily accessible. It predicts such an ion selectivity order for anions: S2O32- > C2O42-> SO42- > SCN- > CO32- > I- > OH- > Br- > NO3- > Cl- > F- > CH3COO- , which is similar to the results obtained from electrodialysis [82] except for SO42- and OH-. The anion selectivity order obtained here does not follow any trend in the change of ion properties (Stokes radii, electrochemical mobilities and self-diffusion coefficients at infinite dilution) as listed in Table 3. The general validity of this convenient method for cations in CEMs, and also other AEMs mentioned above for which the selectivity order were given (cf. 2.3) remains to be tested. Another potentiometric method was suggested for the cation transport number determination in CEMs with binary cations [116]. Okada et al. used this method to measure the transference numbers of H+ and alkaline metal cations in the Nafion membranes [111, 117]. In these studies, the cation transference number is actually identical to the transport number. As shown in Fig. 11, two Nafion membranes that each had been equilibrated with a binary electrolyte solution respectively were overlapped with half of their area as a membrane stack, leaving the remaining two membrane surfaces in contact with corresponding equilibrating solutions [111]. The contacting solution on one side of the membrane stack is set as the reference solution with equivalent molar concentration of both salts (ACl and BCl). The contacting solution on the other side is the test solution that has the same ionic strength as the reference solution but varying molar ratios of the two salts. An electromotive force (emf) was formed due to the difference in chemical potentials of the two salt species across the membrane stack. Ignoring the chemical potential difference of water at two sides of the membrane stack, the transport number of cation A in the membrane phase, 𝑡9& , can be derived as [111]: @(~.) 𝑡9& = 𝑥9A_ − (1 − 𝑥9A_ ) (17) @∆˜•œ•
where 𝑥9A_ is the molar ratio of ACl in the test solution, E the formed emf across the membranes, F the Faraday constant and ∆𝜇9A_ the chemical potential difference of ACl in the two contacting solutions. The activity coefficients necessary for the calculation of ∆𝜇9A_ are obtained by the Debye-Hückel equation. When a series of test solutions with different 𝑥9A_ are employed to record respective emf values, the @(~.) ratio of the two gradients, , in Eq. 17 can be obtained. Then 𝑡9& corresponding to a specific ACl @∆˜•œ•
concentration (𝐶𝑥9A_ ) in the test solution is calculated with the determined The transport number of the other cation, 𝑡8& , is then obtained as:
𝑡8& = 1 − 𝑡9&
26
@(~.)
@∆˜•œ•
(18)
value at this 𝑥9A_ .
The obtained transport numbers of the two cations in the membrane phase, 𝑡9& and 𝑡8& , can be normalized by their concentrations in the test solution. An estimate of the ion selectivity between these two cations in the tested membrane can then be obtained by Eq. 14. To keep Eq. 17 valid, low electrolyte concentration in both solutions is preferred to avoid the presence of the co-ion Cl- in CEMs [116]. The solutions have a total salt concentration of 0.03 M in these studies [116, 117]. This method is more convenient compared with the Hittorf method for membrane stacks because of the short measuring time and also high precision, and it can also avoid problems of concentration polarization, diffusion and water transport [116]. However, this method has only been applied for CEMs, possibly due to the limited selection of reversible electrodes with suitable co-ions.
Figure 11. A schematic representation of the setup and the membrane arrangement for the measurements of the cation transport number in Nafion membranes (reproduced with permission from [117]) .
4.
Methods to improve the ion selectivity of membranes
4.1. Surface modification of ion exchange membranes Ion selectivity is a membrane property that can come into play in particular at the membrane solution interfaces [115]. Hence the surface modification of an IEM is the most studied approach to tune the membrane ion selectivity, and it is also proved to be the most effective approach till now. Generally, four types of membrane surface modification can be distinguished, which are illustrated in Fig. 12. In Fig. 12 only one surface modification is depicted, and whether both surfaces of an IEM should be modified or not depends on the specific application. For example, the IEMs used in the chlor-alkali industry make use of this two-layer structure with a perfluorocarboxylic acid cation exchange layer on one surface of a perfluorosulfonic acid CEM [30]e.
27
Figure 12. Illustration of the surface modification types of IEMs. (a) A highly cross-linked surface layer with the same ion exchange groups as the membrane bulk, (b) a surface layer with fixed ion exchange groups of the sign opposite to those of the membrane bulk, (c) an LbL film and (d) a surface layer formed by dense and (mostly) neutral polymers. Dimensions are not to scale. Surface layer with a high degree of cross-linking The direct way to improve ion selectivity by highly cross-linked membrane surface layer (Fig. 12 a) is based on the idea of steric sieving of ions that possess different hydrated radii. Among the first trials has been to improve Na+/Ca2+ selectivity for the NaCl production from sea water by electrodialysis. Here a surface layer of high cross-linking degree in old fashion CEMs polycondensed from phenol, mphenolsulfonate and formaldehyde was proved to be effective to sieve smaller Na+ ions; while for A1 NŸ another type of CEMs based on copolymers from vinyl monomers (styrene and DVB), the 𝑃ž1 Ÿ decreased only slightly with increasing the cross-linker content [5]. This difference originates from the fact that in condensation-type CEMs both the mobility ratio and the ion exchange equilibrium constant between Ca2+ and Na+ ions decrease with the increase of cross-linker content, however in the other type of CEM the decrease of the mobility ratio is offset by the increase of ion exchange equilibrium constant between Ca2+ and Na+ ions [5]. Since AEMs prepared by polycondensation of monomers are rare, it is not clear whether different trends of anion selectivity with surface cross-linking degree exist between the two corresponding types of AEMs. Oppositely-charged surface layer Among the various surface modification methods, creating a thin oppositely-charged surface layer on an IEM (Fig. 12 b) can improve the monovalent ion selectivity without significantly increasing the membrane electrical resistance, which has thus been considered as the most effective method. Monovalent ion selective CEM having a thin positively-charged surface layer is the key technical 28
component that makes possible the NaCl production from electrodialytic concentration of sea water [5]. Table 5 lists the ion selectivity values of commercial monovalent ion selective CEMs and AEMs. For example, the Na+-selective CMS membrane is composed of negatively charged polystyrene-divinyl benzene (PS-DVB) matrix and a positively-charged polyethyleneimine (PEI) surface layer [118]. Table 5. Reported selectivity of commercial monovalent ion selective IEMs CMX* CMS CSO ž1 Ÿ 0.64 1.23 1.72 𝑃A1NŸ [108] AMX* LbL AMX** ACS def Ng 0.55 0.4 𝑃 g [119] 1.3 A_
* CMX and AMX are the normal type cation and anion exchange membranes from Astom Corporation, respectively. ** The data is from Ref. [119].
The transport hindrance that bivalent ions experience is imposed by these surface layers of the same fixed charges as bivalent ions, due to stronger electrostatic repulsion compared to monovalent ions. This fractionation of ions is effective in a kinetic manner rather than a thermodynamic manner, as revealed by the investigations of Sata [5]. Firdaous et al. [120] studied the dependence of ionic flux on the ion concentration and the presence of other ions in electrodialysis with mixed salt solutions. The salt solutions contain two or three of Na+, Ca2+ and Mg2+ ions, and the membrane used was the commercial monovalent-cation-selective Neosepta CMS membrane (Table 5). Their results show low sensitivity of divalent cation fluxes to their own concentration and the presence of Na+, however the flux of Na+ is sensitive to the presence of divalent cations and Na+ concentration [120]. These results again imply that monovalent ion selectivity stems from electrostatic repulsion in a kinetic manner of the charged surface layer towards bivalent cations in the solution. Once some portions of bivalent cations enter into the membrane matrix, stronger binding of bivalent cations with sulfonic groups will decrease available sites for Na+ transport. Saracco [121] compared the classical solution-diffusion model, a kinetic model and a combined model to describe the cation and anion transport through CMS and ACS membranes (Table 5), respectively. The solution-diffusion model describes the cation transport data satisfactorily while the kinetic aspects have to be considered for the anion transport [121], which also implies some underlying difference in the transport of these two types of ions. Wiedemann et al. calculated the diffusion coefficients of four vanadium ions (V2+, V3+, VO2+, VO2+) in three types of commercial CEMs based on the Maxwell-Stefan approach to describe the counter-ion diffusion [55]. The three CEMs include CMS, which is monovalent-selective, CMX (Table 5) and another commercial CEM. All diffusion coefficients of the four cations in CMS are substantially smaller than the corresponding values in the other two ordinary CEMs [55]. The transport of ions across the IEM systems involves multiple steps: transport across the diffusion boundary layer in the electrolyte solutions, across the electric double layer at the membrane surface, and across the membrane matrix [70, 122]. These steps change at different rates as response to the change (bias direction and frequency) of the applied voltage [122]. Therefore, electrochemical impedance spectroscopy (EIS) could be used to detect the different relaxation time scales corresponding 29
to the transport steps mentioned above. Recently, our group simulated the dielectric impedance response of a CEM with a positively charged surface layer immersed in single NaCl or CaCl2 solution, and a mixture of the two salts, by direct numerical calculations of the ionic fluxes based on the Nernst-Plank and Poisson equations [123]. The simulation predicted that different ion transport steps possess different characteristic time scales (or frequency) in the impedance spectra [123]. Individual concentration polarization of the two cations with different mobility (Na+ and Ca2+) in the solution boundary or in the positively charged surface layer could be distinguished from each other, based on which it is confirmed that the monovalent ion selectivity of these surface layers is resulted from stronger electrostatic repulsion between multiply-charged ions and the fixed charges compared to monovalent ions [123]. The experimental impedance spectra of a CEM modified with a layer of quaternized poly(2-vinyl pyridine) microgels agreed well with the predictions by the simulation [124]. It was also observed that the characteristic frequency of Na+ diffusion in the solution boundary shifted to lower frequencies due to the competition with Ca2+ [124]. It can be expected that this EIS technique can give more subtle information on ion transport in IEM systems. For CEMs, several positively-charged polymers such as protonated polyaniline (PANI) [5], PEI [5] and quaternized chitosan [125, 126] have been studied. The immobilization of these charged polymers can be realized by either simple immersion adsorption or electro-deposition. Generally, the hindering effect on multivalent ions by surface layers from electrodeposition of polyelectrolytes is more significant than surface layers obtained by immersion adsorption. However the charged layer on an IEM surface formed by these two methods is observed to be instable during the continuous long-term electrodialysis [5]. To create a surface layer that is durable during the long-term implementation, covalent bonds between the oppositely-charged surface polymer and the membrane matrix are preferred. Such kinds of chemical bonds can be created by membrane surface modification methods like plasma-[127] or radiationinduced deposition of the surface layer. Sata et al. succeeded to covalently bond PEI on a PS-DVB copolymer type CEM through acid-amide bond formation between -SO2Cl groups of CEM and amine groups of PEI. The acid-amide bond was tested to be stable under harsh hydrolysis conditions and this kind of monovalent-ion-selective CEM was successfully run in the electrodialytic concentration of NaCl in sea water [128]. Also with the aid of thionyl chloride (-SO2Cl) groups formed on the CEM, small N,N-dimethylethylenediamine molecules were chemically bonded to the surface of the CEM by one end amine groups of the diamine, the remaining amine could be quaternized by methyl iodide or acid as shown in Fig. 13. The resulting membrane with surface cationic groups also showed much improved selectivity towards H+ in the electrodialysis of H+ and Zn2+ metallic cations [129]. A recent work utilizes the same protocol to bind a polyelectrolyte with fixed positive charges onto the surface of a CEM [130]. The polyelectrolyte is prepared from a copolymer of acrylamide and diallyl dimethyl ammonium chloride [130].
30
Figure 13. Surface modification of the cation exchange CMX membrane by chlorosulfonation (Reaction 1), amination (Reaction 2) and quaternization (Reaction 3) [129]. Le et al. described a facile diazonium-induced anchoring process to graft a polyaniline-like thin layer on the surface of the commercial cation exchange Selemion CMV membranes, as shown in Fig. 14. The grafted polyaniline-like film was thin and had little effect on the ion exchange capacity and conductivity of the modified membrane, however the quaternized polyaniline-like film proved to be very effective in NŸ
reducing the penetration of divalent Ni2+ with respect to H+ ions, reducing the 𝑃ž5Ÿ by one order of magnitude from 0.056 to 0.006 for bare and modified membranes, respectively [118].
Figure 14. A surface modification method with the aid of radicals generated by the diazonium salts in solution. A very thin polyaniline-like polyaminophenylene surface layer can be formed in this way [118]. 31
Quaternized chitosan was adopted as a cationic polyelectrolyte material to modify both homogeneous and heterogeneous CEMs by the electrodeposition method [126]. In electrodeposition, the cationic polyelectrolytes were driven under an external electrical field from the bulk solution towards the CEM surface and were deposited there. The composite membrane showed significantly decreased multivalent cation (Zn2+, Al3+) permeation in binary and tertiary mixtures with H+ ions in electrodialysis [126]. This proton-selective transport characteristic of the membranes was observed only when the modified membrane surface faced the anode side in electrodialysis [126]. The chitosan derivatives expand the choice of possible polyelectrolytes, however the surface layer formed by electrodeposition cannot withstand long-term operation as mentioned before. Later, Wang et al. [125] introduced a photo-induced immobilization of azide-functionalized chitosan on the CEM surface to treat Zn2+contaminated wastewater streams. From the electrodialysis experiments of Na+/Mg2+ pairs and H+/Zn2+ pairs based on the modified membranes, the authors concluded that primary and secondary amine groups within the surface layer contribute more to the ion selectivity between monovalent and multivalent metallic cations, and quaternary amine groups are more effective in discriminating between protons and metallic cations [125]. After the work of aniline polymerization by Sata and co-workers, Tan et al. systematically investigated the influence of four different oxidants ((NH4)2S2O8, FeCl3, H2O2 and KIO3) and two different modification methods (one-step versus two-step) for aniline polymerization in/on the Neosepta CMX membrane [131]. The CMX membrane was placed in the middle of a two-compartment Teflon cell. One-step modification means that the oxidant solution and the aniline solution were placed in one compartment of the cell, respectively. Oxidation polymerization of aniline then takes place at the boundary of the oxidants and aniline due to their inter-diffusion process in the membrane. In the two-step modification, the membrane was brought into contact with the aniline and the oxidant solution in the same compartment sequentially, while the other compartment was only filled with water. As shown in Fig. 15, different combinations of oxidants and modification techniques render different distribution and morphology of polyaniline. This work demonstrated that only a very thin but homogeneous PANI layer on the surface of CMX is effectively hindering divalent Zn2+ over H+ [131]. On the other hand, PANI solely distributed within the membrane matrix seems ineffective to block the Zn2+ transport. The electrostatic repulsion between the formed surface PANI layer and Zn2+ rather than the physical barrier property of PANI layer dominates the blocking action towards divalent Zn2+ ions [131]. A new type of CEM based on PVDF and sulfonated PVDF was also used to prepare composite CEMs with polyaniline surface layer by similar methods, and high monovalent-ion-selectivity was observed after the polyaniline layer was doped with acid [132]. The polymerization of aniline on the surface or in the matrix of IEMs can also be done electrochemically [133, 134]. The CEM with a polyaniline surface layer polymerized electrochemically showed also very good blocking effect towards divalent Zn2+ in a mixture of Zn2+ and H+ ions [133].
32
Figure 15. Distribution of polyaniline in/on the cation exchange CMX membrane by different oxidation polymerization methods. A thin and homogeneous positively-charged surface layer is proved to be sufficient for blocking divalent Zn2+ ions [131]. Up to now, creating a thin charged surface layer different from the IEM matrix is mostly done with CEMs. The first reports on AEMs were also from Sata and his co-workers. Adsorption of anionic polyelectrolyte, e.g. poly(styrene sulfonic acid) or the polycondensation products of sodium naphthalene sulfonate and formaldehyde, on the surface of strongly basic AEMs can decrease the transport number of SO42- relative to that of Cl-, also due to stronger electrostatic repulsion between the anionic surface layer and .g .g g also increases (however 𝑃A_ g still SO42- ions [6]. For F- of larger hydrated radius (cf. Table 3), the 𝑃A_ less than 1) due to the electrostatic repulsion between the anionic surface layer and F-[6]. Nagarale et al. [135] studied the oxidative polymerization of aniline by (NH4)2S2O8 on a single surface of interpolymer-type CEMs and AEMs. Besides the observation of PANI layer´s blocking effect towards bivalent cations, a decrease of the total cation transport number in the composite CEMs estimated by the membrane potential was observed. This was explained by the reduction of membrane surface charge density due to the weakly basic nature of PANI [135]. For the composite AEMs with strongly basic quaternary ammonium groups and a surface PANI layer, the total anion transport number through the composite membrane increased with the polymerization time of aniline, and Cl- had a larger transport number than SO42-[135]. 33
Layer-by-Layer (LbL) films Layer-by-Layer (LbL) surface modification is a versatile technique and has been applied in a wide range of research fields. Polyelectrolyte multilayers prepared by the LbL deposition were observed to favor the transport of ionic species with smaller charge [136]. For the surface modification aimed at enhancing the permselectivity of ions, especially ions of the same charge sign and valence, through IEMs in electrodriven processes, the LbL technique has been proved to be effective (Fig. 12 c). Bruening´s research group have been studying pressure-driven ion transport through polyelectrolyte multilayer membranes (PEMs) deposited by the LbL technique [137-141]. Membranes composed of poly(sodium 4-styrene sulfonate)/poly(allylamine hydrochloride) (PSS/PAH) PEMs and porous alumina supports demonstrated extremely high K+/Mg2+ selectivity larger than 350 in diffusion dialysis experiments. No matter whether PEMs were terminated with polycation PAH or polyanion PSS, Mg2+ transference number through the membrane - indicated by the transmembrane potential - was nearly 0 [138]. Recently, the group published some preliminary results about K+/Mg2+ selectivity and ionic fluxes in electrodialysis for functional membranes composed of (PSS/PAH)5 films deposited on a porous alumina support or nanofiltration membranes. Large K+/Mg2+ selectivity values of 100 or more can be maintained under electrodialysis conditions of 0.01 M K+ and 0.01 M Mg2+ mixed cations in chloride, and a current density around 4 mA cm-2 [137]. The cation selectivity depends on the used support and also on the anions of the mixed salt solution. Sulfate reduces the K+/Mg2+ selectivity to around 40, which is explained as a result of the surface charge decrease in the presence of sulfate. A distinct difference between these functional membranes and conventional IEMs lies in the fact that both cations and anions can be transported through these new membranes. In fact, a large portion of the system current is carried by anions during electrodialysis [137]. Nevertheless, these systematic studies on the ion transport characteristics through PEMs serve as the basis for the application of LbL on IEMs. Abdu et al. [108] investigated the alteration of Na+ and Ca2+ ion selectivity in CEMs having hyperbranched poly(ethyleneimine)/poly(sodium 4-styrene sulfonate) (PEI/PSS) multilayer films atop. Ÿ
ž1 The 𝑃A1 NŸ of the modified membranes increases with the increase of PEI/PSS bilayer number, as shown in Fig. 16 [108]. An ‘’odd-even’’ effect was observed for the ion selectivity when the LbL film was terminated with the polycation PEI or the polyanion PSS, and this effect was more profound when the bilayer number was below 3 [108]. With only 6 bilayers, the modified membranes achieved monovalent ion selectivity comparable to that of the commercial Selemion CSO membrane (cf. Table 5) without significant increase of the membrane electrical resistance [108]. The endowment of monovalent-ionselectivity by LbL films was interpreted as a result of two synergetic effects: stronger Donnan exclusion towards bivalent ions through multiple polycation layers and hydrophobization of the membrane surface. The latter reason was confirmed by the increased water contact angles concomitant to the modification [108]. What’s more, when the membranes are characterized by electrochemical impedance spectroscopy in a mixture of mono- and bivalent salts, the fitted capacitance of the electrical
34
double layer on the membrane surface was shown to be qualitatively indicative of the repulsion force towards bivalent ions imposed by the membrane surface [108]. The effect of multiple utilization of the Donnan exclusion in these LbL films to increase the monovalent-ion-selectivity is confirmed with direct numerical simulation results of a proposed (Electrolyte)n-(PolyElectrolyte)n (EnPEn) model [71]. The dynamic version of this model can predict impedance spectra and helps to quantify the resistances experienced by the monovalent and bivalent ion exerted through the modification layer [123, 124].
Figure 16. Layer-by-layer (LbL) deposition of hyperbranched PEI/PSS multilayer on the surface of cation exchange CMX membranes changes the cation selectivity. An “odd-even” effect of the ion selectivity is witnessed when the outermost surface layer is terminated with the polycation or the polyanion [108]. White et al. [141] modified the surface of cation exchange Nafion membranes with 5.5 bilayers of PSS/PAH ((PSS/PAH)5PAH). In electrodialysis (ED) experiments with the modified Nafion membranes, Ÿ
£ remarkable 𝑃¡¢ NŸ values in the range of 22 to more than 1000 were observed [141]. The ED
experiments were performed with a mixture of 0.01M KNO3 and 0.01 M Mg(NO3)2 on the anodic side of a two-compartment cell, and 0.01 M HNO3 on the cathodic side [141]. The current density investigated varied from 0.32 mA cm-2 to 2.54 mA cm-2, and the limiting current density of the system was around 0.6 mA cm-2 [141]. Even though the influence of H+ generated at the anode could not be excluded with the two-compartment cell, the results presented in this work still show remarkable monovalent ion selectivity in the range of several hundreds. The two distinctive features of these polyelectrolyte multilayers formed by PSS/PAH on the surface of Nafion are: (a) very low permeability of bivalent cations (Mg2+ and Ca2+), (b) catalytic water-splitting effect [141]. The ohmic resistance of the modified Nafion membrane was estimated to be likely about twice the resistance of the bare Nafion membrane [141]. This high resistance resulted in the low limiting current density as observed. If the measurements about ion-selectivity could have been done with the four-compartment (or even six-compartment) cell as suggested in Section 3.2.1, to exclude any possible influence of H+ and OH- ions generated by the electrode reactions and to have a quantitative characterization of the water-splitting effect, more quantitative information should be obtained. Nevertheless, this work shows on one hand that 35
polyelectrolyte multilayers formed LbL could achieve remarkable monovalent ion selectivity, on the other hand that the chemistry of LbL films is also very important. LbL assembly was also used for the modification of AEMs [119, 142]. Mulyati et al. [119] used the PSS/PAH polyelectrolyte pair to perform the LbL deposition on one surface of a commercial AEM Neosepta AMX. The ion selectivity between SO42- and Cl- anions was evaluated by the transport number ratio of these two ions in electrodialysis with the modified membrane surface facing the cathode compartment. The experimental conditions were as follows: initial concentrations of NaCl and Na2SO4 in the feed solution were both 0.01 M, current density was constant at 2 mA cm-2 and the temperature was 30 ± 1 oC [119]. As shown in Fig. 17, with the increase of the layer number until it exceeded 15 (PSS de Ng
terminated on the surface), the transport number ratio 𝑃A_gf decreased from 1.3 for the unmodified AMX to a constant value at approximately 0.55 (Fig. 17 a). Interestingly, the surface hydrophilicity did not decrease with the number of LbL layers (PSS terminated). Actually water contact angles of the modified membranes terminated with PSS layers are almost the same as the pristine AMX membrane when the LbL layer numbers are larger than 15 (Fig. 17 b). So the electro-repulsion between surface negative charges and anions would be the possible explanation. The authors interpreted this in light of the increase of excessive negative charges throughout the whole LbL layers with the increase of the LbL film thickness [119].
Figure 17. Layer-by-layer (LbL) deposition of PSS/PAH multilayers on the surface of anion exchange AMX membranes changes the anion selectivity. (a) Water contact angles for AMX membranes terminated with PSS and PAH, respectively; (b) Transport number ratio of SO42- and Cl- in electrodialysis for AMX membranes terminated with PSS layers, as a function of the layer number. This is the ratio at t = 0 of the electrodialysis process (reproduced with permission from [119]).
36
These reports of LbL films on the surface of both CEMs and AEMs to tune the ion selectivity exemplify the versatility of this technique. It is worthwhile noting that in both cases the transport numbers of bivalent ions, Ca2+ and SO42-, generally decrease with the increase of polyelectrolyte layer numbers, with the presence of additional ‘odd-even’ effect in the case of cation transport (Fig. 16). This indicates the important effect of electrostatic repulsion between bivalent ions and similarly-charged polyelectrolytes within the LbL film. Actually, the simulations assuming perfectly layered and well-defined polyelectrolyte layers within the LbL film suggest that only a few numbers (less than 5) of bilayers are sufficient to increase the monovalent ion selectivity significantly [71]. This discrepancy in the effect of number of bilayers results from the challenge of obtaining perfectly layered polyelectrolyte layers covering the whole membrane surface practically, especially for the first few layers. The studies about surface coverage of LbL polyelectrolyte layers on the surface of porous alumina support (pore size 20 nm) show that four or five bilayers of PSS/PAH are required to achieve complete surface coverage of the underlying support by the polyelectrolytes [139]. The report from White et al. about Nafion membranes with deposited PSS/PAH bilayers on the surface reveals that after at least 3.5 bilayers the fluorine elemental content on the membrane surface approaches zero, as shown by XPS (X-ray photoelectron spectroscopy) measurements [141]. These reports suggest the importance and difficulty of obtaining complete surface coverage with a few numbers of bilayers, and this is even more significant for the heterogeneous IEMs. The monovalent ion selectivity endowed by the LbL films has already been adopted in the design of membrane separators for redox flow batteries. Xi et al. reported a composite Nafion membrane with surface polyelectrolyte multilayers as a separator for the vanadium redox flow battery [143]. The composite membrane was fabricated by LbL deposition of polycation poly(diallyldimethylammonium chloride) (PDDA) and polyanion poly(sodium styrene sulfonate) (PSS). Because of the improved H+ selectivity over vanadium ions (V2+, V3+, VO2+, VO2+), the battery with the composite Nafion membrane delivers higher energy efficiency compared with the one having nascent Nafion. Similar LbL films were deposited on the surface and within the pores of a nanofiltration substrate for the vanadium battery application [144]. The composition of the LbL film deposited on IEM surfaces resembles amphoteric IEMs. Though the adsorption of polyelectrolyte is performed in discrete steps during the LbL process, there is still interpenetration of the polycations and the polyanions, which makes the boundary of polycation and polyanion layers fuzzy [145]. Actually, one of the four types of amphoteric IEMs classified by Sata is membranes prepared by the LbL method [30]f. One amphoteric membrane displays remarkable H+ permselectivity relative to Na+ ions in electrodialysis, whose ion transport properties are shown in Fig. 18 [30]f. The amphoteric membranes were prepared by sulfonation of styrene and alkylation of 2methyl-5-vinylpyridine residues in the interpolymer-type membranes. Due to the high H+ selectivity, these membranes are quite attractive for the application as separators for redox flow batteries [30]f. An amphoteric membrane based on inert polymer matrix assembled in an all vanadium redox flow battery leads to high coulombic and energy efficiency because of the high selectivity between H+ and 37
multivalent vanadium ions [146]. The challenge of an amphoteric membrane is the control over the distribution of both cationic and anionic charges within the membrane matrix.
Figure 18. Transport number ratio of H+ and Na+ ions in electrodialysis, and the current effciency of cations (cation permselectivity) as a function of the composition of the amphoteric IEMs based on vinylpyridine/sulfonated resin. Circles represent quaternized membranes (N-methyl pyridinium and sulfonic acid groups), triangles represent tertiary amino groups membrane (pyridinium hydrochloride and sulfonic acid groups) (reproduced from [30]f). Dense surface layer formed by neutral polymers Similar to the idea of a surface layer with high cross-linking degree, a dense and neutral polymer layer formed on IEM surfaces (Fig. 12 d) can also change the selectivity between ions. The electron conducting polymer polypyrrole (PPY) having an extremely tight and rigid structure was combined with both CEMs and AEMs to change the transport selectivity of ions [6]. PPY can be easily polymerized from pyrrole by oxidant FeCl3 throughout or solely on the surface of AEMs. The transport properties of anions through PPY composite membranes studied by electrodialysis of mixed salt solutions were found to be independent of the base membrane IEC values and controlled predominantly by PPY [6]. The PPY layer de Ng
distributed solely on the membrane surface was observed to be effective in decreasing 𝑃A_gf and in že g
increasing 𝑃A_g ¤ respectively, since PPY is considered very hydrophobic (cf. 2.3) [6]. Even in the case of composite AEMs with both PPY layer on the surface and PPY polymer within the AEM matrix, the surface PPY layer was suggested to be responsible for the change of the ion selectivity between anions [6]. Pyrrole also has good affinity to CEMs, and can be polymerized within the membrane matrix and on the surface to prepare composite membranes. Pyrrole imbibed in Neosepta CM-1 CEM can be polymerized by FeCl3 across the whole thickness of the membrane, however the membrane becomes too brittle to be used [5]. On the other hand, the CM-1 membrane in Fe3+ form reacted with pyrrole solution can yield flexible composite membranes with controlled penetration of PPY from the membrane surface into the 38
membrane interior simply by the polymerization time [5]. A single PPY layer on one surface of a NŸ
A1 composite CEM was already effective in decreasing 𝑃ž1 Ÿ , especially when the PPY layer was faced to anode in electrodialysis [5]. Even though PPY has secondary amines, the sieving effect of cations by the dense and rigid PPY layer other than the cationic charge of the secondary amines in acidic conditions was concluded as the main reason of the ion selectivity alteration [5].
Gohil et al. [103] examined a different oxidant Na2S2O8 for the polymerization of pyrrole on the surface of interpolymer membranes based on polyethylene, styrene and DVB. To restrict the polymerization solely on the membrane surface, a CEM was ion-exchanged with protonated pyrrole first and then one surface of the membrane was brought into contact with a 1 M Na2S2O8 aqueous solution; for AEMs, anion S2O82- was ion-exchanged into a membrane first and subsequent polymerization of pyrrole occurred on the surface of the membrane when one membrane surface was in contact with a pyrrole solution. The polymerization time was restricted to less than 4 hours [103]. The tight PPY layer on the membrane surface endows CEMs with monovalent ion selectivity, which was confirmed by electrodialysis with mixed salt solutions. The relative transport rate of Na+ compared with bivalent Ca2+, Mg2+ and Cu2+ were observed to be between 5 and 8 [103]. For AEMs with PPY surface layers, the transport number of counter-ions (Cl-) increased because of the blocking effect of the PPY layer towards cations [103]. What is interesting for the surface PPY layer polymerized by Na2S2O8 is that the conductivities of the composite membranes even increase a little after the formation of the PPY layer [103], thus making these monovalent-ion-selective membranes energetically efficient. Another neutral polymer poly(methyl methacrylate) (PMMA) was also utilized by emulsion polymerization of monomers on the surface of heterogeneous CEMs to change the ion selectivity between cations. The composite CEM possessing a PMMA layer on one surface exhibited increased Ÿ
ž1 𝑃81 NŸ derived from the membrane potential, however the composite membrane was not investigated in electrodialysis with mixed salt solutions [147].
A dense and neutral polymer layer on the membrane surface can increase the monovalent ion Ÿ
selectivity to a larger degree than the oppositely-charged surface layer, for example 𝑃¡ž1NŸ between 5 Ÿ
ž1 and 8 for PPY layers [103] and 𝑃A1 NŸ being only 1.72 for the commercial CSO membranes (Table 5). However, the increased monovalent ion selectivity of neutral polymer layers is generally at the cost of substantially increased membrane electrical resistance. Even though the polymerization methods to obtain such surface layers can induce some chemical bonding between the neutral modification layer and the original membrane material, the long-term stability of these neutral layers is still a concern. In general, the interpretation of the transport data to identify the origin of the selectivity change remains a challenge as the properties of the layers are difficult to assess. Theoretical models and new methods such as impedance spectroscopy can help to further build up fundamental understanding on structure property relationship in relation to synthetic modification methods.
39
Other types of surface layers Besides the four types of surface layers illustrated in Fig. 12, there are some newly developed surface layers to change the ion selectivity of IEMs. Thakur et al. [148] introduced a metal (Cu) occupied layer in the surface region of a CEM by in-situ reduction of Cu2+ ions that had been ion-exchanged into the cation exchange sites (sulfonic acid groups) of the membrane [148]. The Cu loading amount in the surface region can be controlled by the reaction time [148]. The resulting composite membranes showed significantly reduced transport of bi-valent cations (Ni2+ and Zn2+) while the Na+ transport decreased only slightly in electrodialysis with mixed salts [148]. The selective complexation between crown ethers and alkali metal ions has been utilized by Sata and co-workers in the study of electrodialytic separation of these ions [5]. Recently Chaudhury et al. [149] investigated the confined loading of free crown ether moiety (dibenzo-21-crown-7, DB21C7) in the surface region of a Nafion 117 membrane. The surface-modified composite membrane showed very selective Cs+ transport over Na+ compared with Nafion membranes imbibed throughout the whole thickness with different loading amount of DB21C7, as confirmed by electrodialysis [149]. The co-valent immobilization of crown ether moiety in the surface layers of IEMs can be expected to enhance the stability of this composite membrane, even though these crown ether moieties have limited solubility in water. 4.2. Novel bulk morphology of membranes Apart from the generally adopted surface modification approaches to increase the selectivity of monovalent ions, IEMs possessing novel bulk morphology may also tune the ion selectivity. One interesting type of membrane bulk morphology was introduced by Zhang et al. [150] to IEMs as an ionomer separator for the all vanadium redox flow battery. The membrane has a porous symmetric cross-sectional structure composed of non-interconnected pores in the range of several micrometers, as shown by the SEM pictures in Fig. 19. The underlying concept is the multiple utilization of the Donnan exclusion effect at the interface between positively charged pore walls and the solution within the hollow pores (Fig. 19, b) [150]. Because the Donnan exclusion towards multiple charged cations (V2+, V3+, VO2+) is stronger compared with H+, across hundreds of times of such interfaces spanning the whole membrane thickness (Fig. 19, a), a high selectivity between H+ and vanadium cations is achieved [150152]. Meanwhile the effective thickness of the membrane is also reduced, so the general conflict of reduced membrane conductivity and increased selectivity is deliberately tackled. Moreover, the membrane structure can be easily obtained by water vapor induced phase separation, a method that can be readily integrated into an existing industrial manufacturing process. It is, however, worthy noting that this multiple utilization of Donnan exclusion to increase ion selectivity is applicable only between H+ and other cations, but also possibly between OH- and other anions, since H+ and OH- are so small that they can leak through both CEMs and AEMs.
40
Figure 19. Morphology of sponge-like AEMs. (a) The symmetric cross-sectional SEM images. (b) Enlarged view of membrane cross section depicting the non-interconnected pores in micrometer range (reproduced with permission from [150]). 4.3. Blends of polymers Blending is a convenient method to modify the properties of polymer systems. Blends from ion exchange materials and other polymers are also studied for the purpose of tuning ion selectivity. Balster et al. prepared monovalent-ion-selective CEMs from blends of SPEEK and poly(ether sulfone) (PES), and compared the ion selectivity between H+ and Ca2+ of these blend membranes with commercial CEMs [153]. The results show that the Ca2+ transport rate increases with increasing membrane conductivity and increasing charge density [153]. Blending hydrophobic PES with SPEEK can decrease the water uptake as well as the conductivity of membranes [153]. With appropriate combinations of the sulfonation degree of the SPEEK polymer and the PES blend amount, it is possible to attain high H+/Ca2+ selectivity while maintaining reasonably high membrane conductivity. In this work, it is also found that the CMS membrane with a positively charged surface layer (Table 5) shows decreased H+/Ca2+ selectivity at high current densities (30 mA cm-2) [153]. This is possibly due to the effect of substantially declined H+ concentration in the solution boundary layer at high current densities, as discussed in Section 2.2.2. It is also demonstrated for the CMS membrane and SPEEK-PES CEMs, the Ca2+ flux through a membrane increases with the current density and/or Ca2+ concentration [153]. The blend of sulfonated polysulfone and sulfonated poly(ether ether ketone) was also studied [154]. Composite CEMs that made from blends of PVDF and sulfonated PVDF were found to be, however, more permeable to bivalent ions (Ca2+, Mg2+) in the presence of Na+ ions [155, 156]. In the example of SPEEK-PES CEMs, the change of membrane water content through the introduction of hydrophobic PES polymer is considered to be the reason for monovalent ion selectivity. Besides, a stronger interaction between the ion exchange materials and the blending polymer can be employed. Ge et al. introduced the acid - base interaction into IEMs to prepare H+ selective membranes [157]. An IEM was obtained by polymerizing vinyl imidazole monomers in sulfonated PPO (SPPO) solution for the in-situ formation of poly(vinyl imidazole) (PVI) - SPPO network (Fig. 20), and then the casting of the polymer solution [157]. With the increase of PVI content, the IEMs become increasingly compact and show a decreasing trend in H+ ion exchange capacity and in water uptake [157]. Compared with the pure SPPO membrane, the acid - base membrane with 1:1 molar ratio of sulfonic acid groups and imidazole NŸ groups shows 𝑃 “$Ÿ as low as 0.003 [157]. What´s more interesting, the electric area resistance of the 1:1 acid-base membrane is comparable to that of Nafion 117, and 24 % lower than that of the pure SPPO membrane [157]. The enhancement of the H+ selectivity is only accompanied by slight decrease of H+ flux in electrodialysis. The significant enhancement of the H+ selectivity (without the penalty of membrane resistance and H+ flux) is explained by the authors to be the result of pore-size sieving and H+ 41
transport through the acid - base network by a mechanism analogous to the Grotthuss transport (Fig. 20) [157]. This method creates compact membrane structure and also H+ transport channels; it can be possibly also applicable to OH- transport due to the small radii of these two ions and their special transport mechanisms (Fig. 3), but not for ions of larger radius. This acid - base type H+ selective IEM can also be considered as the extension of amphoteric layer from the surface of membranes to the whole membrane bulk.
Figure 20. Acid - base type IEMs formed in-situ from SPPO and poly(vinyl imidazole) (PVI) for the selective hydroxonium ion (H+) transport in a mixture of H+ and Zn2+ ions (Reproduced with permission from [157]). 4.4. Inorganic-organic hybrid ion exchange membranes An inorganic-organic hybrid IEM has two intrinsically different components in the same matrix. Membrane properties can benefit from these two components if they are combined in a synergetic way [21, 104]. Since the hybridization with inorganic particles or copolymers can have an impact on the membrane void/pore volume, water uptake and segmental mobility of polymer chains, the resultant change of the membrane bulk property can also influence the permselectivity of IEMs. Poly(vinyl alcohol)-silica (PVA-silica) membranes with ion exchange groups are typical hybrid IEMs of such kind [158-160]. In such IEMs, ion exchange groups (-SO3H) stay at the inorganic silica phase, and the links between the silica phase and the PVA phase are chemical bonds formed during the sol - gel process [21]. Kumar et al. [158] studied the transport of several ions (Na+, Ca2+, Mg2+ and Fe3+) through a 42
series of PVA-silica CEMs with different ion exchange capacities (IECs). The hybrid CEMs showed interesting selective transport of Na+ ions in electrodialysis with mixed salt solutions [158]. A polyaniline layer formed on the surface of the hybrid CEM by oxidative polymerization further improved the Na+ selectivity over Zn2+ and Al3+ in electrodialysis [161]. Later, a different study shows that the increase of PVA crystallinity by thermal annealing at elevated temperatures following the sol - gel membrane preparation can effectively decrease the water uptake from around 138 % for a pristine membrane to around 13 % for a membrane having crystallinity of 43.9 %, while the membrane IECs are almost maintained [159]. The electrodialysis experiments with mixed ZnSO4 and H2SO4 solutions demonstrated NŸ 𝑃 “$Ÿ values as low as 0.007 [159]. However, the high crystallinity not only renders the membranes enhanced H+ selectivity but also significantly increased electric resistance [159]. The ion transport channels formed in the inorganic phase of these hybrid membranes are responsible for the selective transport, and this may add a new dimension for the construction of selective IEMs. There is another type of hybrid IEMs composed of fillers dispersed in the matrix of a heterogeneous IEM. The fillers that have been investigated include Fe2NiO4 nanoparticles [162], carbon nanotubes [163] and activated carbon [164]. The hybrid IEMs were prepared by solvent evaporation of the mixture containing ion exchange resin powders, fillers, inert polymer binder and solvent. These hybrid IEMs show some permselectivity improvement when the filler amount is not too high (generally ~1 - 2 wt. %), and show also slightly increased monovalent ion selectivity [162-165]. When the filler amount is further increased, the hybrid membranes become again more selective to bivalent ions [162-165]. Even hybrid membranes with only inert polymer binder and Cu3(PO4)2/Ni3(PO4)2 particles are reported to show some ion permselectivity [166]. The permselectivity improvement and the slight monovalent ion selectivity endowed by these fillers are, however, limited compared with IEMs modified with oppositely charged surface layers. 4.5. Different ion exchange groups There is only a limited number of different cation exchange groups: mostly sulfonic acid, phosphoric acid, A1 NŸ carboxylic acid and phenolic groups are available. Sata et al. studied 𝑃ž1 Ÿ of CEMs bearing boric acid groups, and little difference was observed compared with CEMs having the conventional sulfonic acid groups [5]. Compared to the situation in AEMs, fixed ion exchange groups in CEMs with different chemistry have a more significant effect on the counter-ion transport [5, 6]. The mobility decline of bivalent alkaline earth metal cations (Mg2+, Ca2+, Sr2+, Ba2+) in CEMs having phosphoric acid groups results in reduced transport number ratio of bivalent cations compared with that in CEMs having sulfonic acid groups, despite higher ion exchange equilibrium constants of these bivalent cations in phosphoric acid membranes being attained [5]. The study of Nagarale et al. sugessted that CEMs with fixed phosphoric acid groups, compared to fixed sulfonic acid and carboxylic acid groups, were more suitable for the fractionation of cations with identical charges [167]. On the other hand, there might be more options in the substitute groups of quaternary ammonium anion exchange sites. The systematic work of anion permselectivity through AEMs with ion exchange sites based on quaternary ammonium groups leads to a conclusion that the anion permselectivity is mainly dependent on the affinity of specific anions to an AEM, and partially on the change in their mobility through the membrane [6] (Section 2.3 and Table 3). The affinity of specific anions to the membrane refers to the hydration energy of anions and hydrophobic/hydrophilic atmosphere around ion exchange sites within the membrane. Therefore, different anion exchange groups will surely have a significant impact on the membrane permselectivity. Positively-charged quaternary phosphonium 43
groups [168], immidazolium groups [26] and the recently developed ruthenium-cation based anion exchange groups [169] are those whose ion selectivity information is still lacking. 5.
Conclusions and Perspectives
Ion exchange membranes of finely tuned ion selectivity - not only the permselectivity between counter- and co-ions but also the selectivity between counter-ions of different valence - are important for the success of a wide spectrum of technical processes like redox flow batteries, ion exchange membrane bioreactor, microbial fuel cells etc. Further understanding of ion transport selectivity and economic membrane preparation methods are pursued to enable wider employment of ion exchange membranes in technical processes for sustainable development. This work reviews the development of the ion exchange membrane permselectivity. In the first part, membrane microstructure and possible ion transport mechanisms, influence of the solution boundary layers are discussed. The generalized cation and anion selectivity orders are summarized. In the second part, the two types of experimental methods to determine ion permselectivity are compared. In the last part, interesting membrane preparation methods and the surface modification of ion exchange membranes are classified and discussed. The transport mechanisms of ions in IEM matrix include migration, diffusion, surface site hopping and convection, among which convection is believed to be of minor contribution due to the dense nature of IEMs. H+ and OH- ions have extraordinary structure diffusion mechanism in the membrane that is similar to their behavior in the solution. The present understanding of the micro-phase separation of IEM matrix stresses the importance of water content on ion transport in several ways: the water content influences the fixed charge concentration, therefore the migration and hopping of counter ions; the water content also induces the evolution of micro-phase separated membrane morphology, in this way influence the sorption and transport of co-ions, and also the interaction of fixed charges and mobile ions. Solution boundary layers developed at the two solution - membrane interfaces have an influence on the ion permselectivity. The selectivity of ions in the diffusion boundary layer is determined by the diffusivity of ions in the solution. Ions with larger diffusivity will gain transport selectivity with the increase of the relative thickness of boundary layers compared with the membrane thickness, which A1 NŸ means for example 𝑃ž1 Ÿ will decrease with the increase of the current density as a result of the boundary layer development. For co-ion transport through the IEMs, the permselectivity can gain in the depleted solution boundary layer due to the decreased local co-ion concentration, however lose in the enriched solution boundary layer as a result of the increased local concentration. The overall change of the permselectivity depends on the relative thickness of the two boundary layers. It is verified both theoretically and experimentally that the ion selectivity is not only determined by the membrane itself but also by the boundary conditions, however the experimental verifications have been done only on cations till now.
44
In this work, a generalized selectivity order for common cations transported through normal CEMs with fixed sulfonic acid groups is provided: Ba2+ > Sr2+ > Ca2+ > Mg2+ > H+ > (Cu2+ ~ Zn2+ ~ Ni2+) > K+ > Na+ > Li+ > Fe3+. This order is concluded from electrodialysis experiments with binary salts when the current density is well below the overlimiting current density of the system. This order coincides with the ion exchange sequence as a result of the determining role of ion exchange (solubility) for cation transport when the solution boundary layers do not come into play. The generalized anion selectivity order is the same as the anion exchange sequence and also the Hofmeister series: I- > (NO3- ~ Br-) > NO2- > Cl- > OH- > SO42- > F-. There are generally two classes of experimental methods to determine the counter-ion transport number through an IEM (permselectivity) and the ion selectivity between two counter-ions in the same system. One class of the methods relies on the electrodialysis of a single salt solution for permselectivity, or mixed binary salts of the same co-ions for counter-ion selectivity. Another group of methods are based on the membrane potentials generated when the two surfaces of an IEM are respectively in contact with the same electrolyte of different concentrations (concentration potential) for permselectivity, and when the two membrane surfaces are in contact with two solutions of different counter-ions but the same co-ions (bi-ionic potential) for the ion selectivity between two counter-ions. The membrane potential method is simple and easy, and serves as a quick estimation of the ion selectivity. The counter-ion transport numbers obtained by the concentration potential method are smaller compared with those obtained by the electrodialysis method due to the water transfer in electrodialysis. The transport numbers of two counter-ions obtained by the bi-ionic membrane potential method have not been compared with those from the electrodialysis experiments with mixed binary salts, but the trend should be also the same as a single counter ion resulting from the water transfer. To tune the ion selectivity of IEMs, several methods could be employed. The concentration of fixed charges and the water content of membranes control the counter- ion transport in the membrane phase, the high concentration of fixed charges favours the counter-ion transport. The ion selectivity between counter-ions of different valence could be changed by the surface modification of IEMs. Among the four major types of surface modification methods, a thin oppositely-charged layer of high surface coverage is the most effective one to increase the monovalent-ion-selectivity. The electrostatic repulsion between the counter-ions and the similarly-charged surface layer blocks the multivalent counter-ions more effectively than the monovalent ones, therefore rendering the IEMs monovalent-ion-selectivity. The present understanding of the effect of LbL films for monovalent-ion-selectivity is also due to the stronger electrostatic repulsion of multivalent ions in the similarly-charged polyelectrolyte layers of the LbL film. The challenge of transferring the extremely high monovalent-ion-selectivity obtained by LbL films on porous supports might be the control over the polyelectrolyte assembly on IEM surface during the first few steps in LbL adsorption. The most important bulk properties of IEMs for different counter-ion transport are also water content and fixed charge concentration. Blending ion exchange polymers with a hydrophobic polymer or the hybridization with inorganic components can also change the relative transport rate of mono- and multivalent counter-ions. For improved selectivity between H+ (or OH-) and other cations (or anions), acid - base interaction in the membrane matrix, sponge-like membrane morphology of noninterconnected macro-pores and amphoteric IEMs could be considered. The change of ion exchange groups in the membrane matrix will have a determinant role in the change of ion transport characteristics. Normally the possible choices of ion exchange groups are limited to the few types, combinations of different ones and also geometrically confined distribution (e.g. a surface layer) could provide the interplay space for specific requirements of ion selectivity. The possibility for high ion 45
selectivity between counter-ions of the same charge lies in the specific interaction between a certain ion and selective chelating groups, such as the well-known crown ether and alkali metal ions, and lies probably also in the possibility of polyelectrolyte complexes [170]. The challenge of using crown ethers is that the loss in ion mobility negates the gain in solubility of ions in the membrane phase [171]. Until today, combining crown ether units and ion exchange groups to realize efficient fractionation of alkali metal ions in potential-driven transport process is still more like a concept. The ion selectivity requirement of IEMs for the application in microbial fuel cells remains a challenge with IEMs.
6.
Acknowledgements
Tao Luo acknowledges the financial support of a CSC scholarship (201306240054). Matthias Wessling appreciates the support through the Alexander-von-Humboldt Foundation. The authors would like to thank Robert Femmer for the discussion about his simulations, and proof-reading of Section 2.2.2; thank Ben Scheitler for the double-checking of the generalized cation selectivity order. Tao Luo would also give special thanks to Prof. V. V. Nikonenko from Kuban State University, Russia for providing the English version of his book and the discussion about the surface hydrophobicity of membranes.
7.
References
[1] H. Strathmann, Electrodialysis, a mature technology with a multitude of new applications, Desalination, 264 (2010) 268-288. [2] J.R. Varcoe, P. Atanassov, D.R. Dekel, A.M. Herring, M.A. Hickner, P.A. Kohl, A.R. Kucernak, W.E. Mustain, K. Nijmeijer, K. Scott, Anion-exchange membranes in electrochemical energy systems, Energy & environmental science, 7 (2014) 3135-3191. [3] G. Merle, M. Wessling, K. Nijmeijer, Anion exchange membranes for alkaline fuel cells: A review, Journal of Membrane Science, 377 (2011) 1-35. [4] F.G. Donnan, Theorie der Membrangleichgewichte und Membranpotentiale bei Vorhandensein von nicht dialysierenden Elektrolyten. Ein Beitrag zur physikalisch-chemischen Physiologie., Zeitschrift für Elektrochemie und angewandte physikalische Chemie, 17 (1911) 572-581. [5] T. Sata, T. Sata, W. Yang, Studies on cation-exchange membranes having permselectivity between cations in electrodialysis, Journal of Membrane Science, 206 (2002) 31–60. [6] T. Sata, Studies on anion exchange membranes having permselectivity for specific anions in electrodialysis — effect of hydrophilicity of anion exchange membranes on permselectivity of anions, Journal of Membrane Science, 167 (2000) 1–31. [7] C. Ponce de León, A. Frías-Ferrer, J. González-García, D.A. Szánto, F.C. Walsh, Redox flow cells for energy conversion, Journal of Power Sources, 160 (2006) 716-732. [8] J. Luo, C. Wu, T. Xu, Y. Wu, Diffusion dialysis-concept, principle and applications, Journal of Membrane Science, 366 (2011) 1-16. [9] P.M. Biesheuvel, A. van der Wal, Membrane capacitive deionization, Journal of Membrane Science, 346 (2010) 256-262. [10] Y.-J. Kim, J.-H. Choi, Enhanced desalination efficiency in capacitive deionization with an ionselective membrane, Separation and Purification Technology, 71 (2010) 70-75. [11] E. Güler, W. van Baak, M. Saakes, K. Nijmeijer, Monovalent-ion-selective membranes for reverse electrodialysis, Journal of Membrane Science, 455 (2014) 254-270. [12] B.E. Logan, M. Elimelech, Membrane-based processes for sustainable power generation using water, Nature, 488 (2012) 313-319. 46
[13] D.A. Vermaas, J. Veerman, M. Saakes, K. Nijmeijer, Influence of multivalent ions on renewable energy generation in reverse electrodialysis, Energy & environmental science, 7 (2014) 1434-1445. [14] G.M. Geise, D.R. Paul, B.D. Freeman, Fundamental water and salt transport properties of polymeric materials, Progress in Polymer Science, (2013). [15] H.B. Park, B.D. Freeman, Z.B. Zhang, M. Sankir, J.E. McGrath, Highly chlorine-tolerant polymers for desalination, Angew Chem Int Ed Engl, 47 (2008) 6019-6024. [16] B.E. Logan, K. Rabaey, Conversion of Wastes into Bioelectricity and Chemicals by Using Microbial Electrochemical Technologies, Science, 337 (2012) 686-690. [17] F. Harnisch, U. Schröder, Selectivity versus Mobility: Separation of Anode and Cathode in Microbial Bioelectrochemical Systems, ChemSusChem, 2 (2009) 921-926. [18] C.T. Matos, R. Fortunato, S. Velizarov, M.A.M. Reis, J.G. Crespo, Removal of mono-valent oxyanions from water in an ion exchange membrane bioreactor: Influence of membrane permselectivity, Water Research, 42 (2008) 1785-1795. [19] S. Zhang, C. Yin, D. Xing, D. Yang, X. Jian, Preparation of chloromethylated/quaternized poly(phthalazinone ether ketone) anion exchange membrane materials for vanadium redox flow battery applications, Journal of Membrane Science, (2010) 243-249. [20] C.T. Matos, S. Velizarov, J.G. Crespo, M.A.M. Reis, Simultaneous removal of perchlorate and nitrate from drinking water using the ion exchange membrane bioreactor concept, Water Research, 40 (2006) 231-240. [21] T. Xu, Ion exchange membranes: State of their development and perspective, Journal of Membrane Science, 263 (2005) 1-29. [22] J. Stodollick, R. Femmer, M. Gloede, T. Melin, M. Wessling, Electrodialysis of itaconic acid: A shortcut model quantifying the electrical resistance in the overlimiting current density region, Journal of Membrane Science, 453 (2014) 275-281. [23] J. Balster, D. Stamatialis, M. Wessling, Electro-catalytic membrane reactors and the development of bipolar membrane technology, Chemical Engineering and Processing: Process Intensification, 43 (2004) 1115-1127. [24] Z. Twardowski, H.L. Yeager, B. O'Dell, A Comparison of Perfluorinated Carboxylate and Sulfonate Ion Exchange Polymes, Journal of Electrochemical Society: Electrochemical Science and Technology, 129 (1982) 328-332. [25] M. Wang, C.-j. Gao, The State-of-The-Art of the Cation Exchange Membrane Having Monovalent Ion Selectivity - A Patent Review, Recent Patents on Chemical Engineering, 4 (2011). [26] J. Ran, L. Wu, Y. He, Z. Yang, Y. Wang, C. Jiang, L. Ge, E. Bakangura, T. Xu, Ion exchange membranes: New developments and applications, Journal of Membrane Science, 522 (2017) 267-291. [27] V.A. Shaposhnik, K. Kesore, An early history of electrodialysis with permselective membranes, Journal of Membrane Science, 136 (1997) 35-39. [28] L. Bazinet, D. Ippersiel, D. Montpetit, B. Mahdavi, J. Amiot, F. Lamarche, Effect of membrane permselectivity on the fouling of cationic membranes during skim milk electroacidification, Journal of Membrane Science, 174 (2000) 97-110. [29] R.W. Baker, Membrane Technology and Applications, 2nd ed., John Wiley & Sons, Ltd., 2004. [30] T. Sata, Ion Exchange Membranes: Preparation, Characterization, Modification and Application, The Royal Society of Chemistry, Cambridge, 2004. [31] H. Strathmann, Ion-exchange membrane separation processes, Elsevier B.V., 2004. [32] K. Schmidt-Rohr, Q. Chen, Parallel cylindrical water nanochannels in Nafion fuel-cell membranes, Nature Materials, 7 (2008) 75-83. [33] F.G. Donnan, The theory of membrane equilibria, Chemical Reviews, 1 (1924) 73-90. [34] E. Glueckauf, A new approach to ion exchange polymers, in: Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, The Royal Society, 1962, pp. 350-370. [35] J. Petropoulos, D. Tsimboukis, K. Kouzeli, Non-equipotential volume membrane models. Relation between the Glueckauf and equipotential surface models, Journal of Membrane Science, 16 (1983) 379389. [36] J. Kamcev, M. Galizia, F.M. Benedetti, E.-S. Jang, D.R. Paul, B.D. Freeman, G.S. Manning, Partitioning of mobile ions between ion exchange polymers and aqueous salt solutions: importance of counter-ion condensation, Physical Chemistry Chemical Physics, 18 (2016) 6021-6031. [37] J. Kamcev, D.R. Paul, B.D. Freeman, Ion activity coefficients in ion exchange polymers: applicability of Manning’s counterion condensation theory, Macromolecules, 48 (2015) 8011-8024. 47
[38] J. Kamcev, B.D. Freeman, Charged Polymer Membranes for Environmental/Energy Applications, Annual review of chemical and biomolecular engineering, 7 (2016) 111-133. [39] G.S. Manning, Limiting laws and counterion condensation in polyelectrolyte solutions I. Colligative properties, The journal of chemical Physics, 51 (1969) 924-933. [40] N.P. Gnusin, N.P. Berezina, N.A. Kononenko, O.A. Dyomina, Transport structural parameters to characterize ion exchange membranes, Journal of Membrane Science, 243 (2004) 301-310. [41] V.I. Zabolotsky, V.V. Nikonenko, Effect of structural membrane inhomogeneity on transport properties, Journal of Membrane Science, 79 (1993) 181-198. [42] K.-D. Kreuer, Ion Conducting Membranes for Fuel Cells and other Electrochemical Devices, Chemistry of Materials, 26 (2014) 361-380. [43] G.M. Geise, L.P. Falcon, B.D. Freeman, D.R. Paul, Sodium chloride sorption in sulfonated polymers for membrane applications, Journal of Membrane Science, (2012) 195-208. [44] G.M. Geise, M.A. Hickner, B.E. Logan, Ionic resistance and permselectivity tradeoffs in anion exchange membranes, ACS applied materials & interfaces, 5 (2013) 10294-10301. [45] K.-D. Kreuer, S.J. Paddison, E. Spohr, M. Schuster, Transport in Proton Conductors for Fuel-Cell Applications: Simulations, Elementary Reactions, and Phenomenology, Chemical Reviews, 104 (2004) 4637-4678. [46] K.N. Grew, W.K.S. Chiu, A Dusty Fluid Model for Predicting Hydroxyl Anion Conductivity in Alkaline Anion Exchange Membranes, Journal of the Electrochemical Society, 157 (2010) B327. [47] T. Thampan, S. Malhotra, H. Tang, R. Dattaa, Modeling of Conductive Transport in ProtonExchange Membranes for Fuel Cells, Journal of The Electrochemical Society, 147 (2000) 3242-3250. [48] M. Saito, N. Arimura, K. Hayamizu, T. Okada, Mechanisms of Ion and Water Transport in Perfluorosulfonated Ionomer Membranes for Fuel Cells, Journal of Physical Chemistry B, (2004) 606416070. [49] V.V. Nikonenko, N.D. Pismenskaya, E.I. Belova, P. Sistat, P. Huguet, G. Pourcelly, C. Larchet, Intensive current transfer in membrane systems: Modelling, mechanisms and application in electrodialysis, Advances in Colloid and Interface Science, 160 (2010) 101-123. [50] A. Elattar, A. Elmidaoui, N. Pismenskaia, C. Gavach, G. Pourcelly, Comparison of transport properties of monovalent anions through anion-exchange membranes, Journal of Membrane Science, 143 (1998) 249-261. [51] R.P. Buck, Kinetics of bulk and interfacial ionic motion: microscopic bases and limits for the nernst— planck equation applied to membrane systems, Journal of Membrane Science, 17 (1984) 1-62. [52] R. Krishna, J.A. Wesselingh, The Maxwell - Stefan approach to mass transfer, Chemical Engineering Science, 52 (1997) 861-911. [53] J. Wesselingh, P. Vonk, G. Kraaijeveld, Exploring the Maxwell-Stefan description of ion exchange, The Chemical Engineering Journal and The Biochemical Engineering Journal, 57 (1995) 75-89. [54] E. Wiedemann, A. Heintz, R.N. Lichtenthaler, Sorption isotherms of vanadium with H3O+ ions in cation exchange membranes, Journal of Membrane Science, 141 (1998) 207-213. [55] E. Wiedemann, A. Heintz, R.N. Lichtenthaler, Transport properties of vanadium ions in cation exchange membranes: Determination of diffusion coefficients using a dialysis cell, Journal of Membrane Science, 141 (1998) 215-221. [56] A. Heintz, E. Wiedemann, J. Ziegler, Ion exchange diffusion in electromembranes and its description using the Maxwell-Stefan formalism, Journal of Membrane Science, 137 (1997) 121-132. [57] N. LAKSHMINARAYANAIAH, TRANSPORT PHENOMENA IN ARTIFICIAL MEMBRANES, CHEMICAL REVIEWS, 65 (1965) 491-565. [58] P. Meares, Transport in Ion-exchange Polymers, in: J. Crank, G.S. Park (Eds.) Diffusion in Polymers, Academic Press Inc., London and New York, 1968, pp. 373-428. [59] S.A. Rice, M. Nagaswa, Polyelectrolyte Solutions, Academic Press, New York, 1961. [60] A.B. Yaroslavtsev, V.V. Nikonenko, V.I. Zabolotsky, Ion transfer in ion-exchange and membrane materials, Russian Chemical Reviews, 72 (2003) 393-421. [61] S. SUBRAMONIAN, D. CLIFFORD, Monovalent/divalent selectivity and the charge separation concept, Reactive Polymers, 9 (1988) 195-209. [62] R. Netz, Polyelectrolytes in electric fields, The Journal of Physical Chemistry B, 107 (2003) 82088217. [63] J. Kamcev, D.R. Paul, B.D. Freeman, Effect of fixed charge group concentration on equilibrium ion sorption in ion exchange membranes, Journal of Materials Chemistry A, 5 (2017) 4638-4650. 48
[64] J. Kamcev, D.R. Paul, G.S. Manning, B.D. Freeman, Predicting salt permeability coefficients in highly swollen, highly charged ion exchange membranes, ACS applied materials & interfaces, 9 (2017) 4044-4056. [65] J. Kamcev, D.R. Paul, G.S. Manning, B.D. Freeman, Accounting for frame of reference and thermodynamic non-idealities when calculating salt diffusion coefficients in ion exchange membranes, Journal of Membrane Science, 537 (2017) 396-406. [66] J. Mackie, P. Meares, The diffusion of electrolytes in a cation-exchange resin membrane. I. Theoretical, in: Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, The Royal Society, 1955, pp. 498-509. [67] F. Helfferich, Ion Exchange, Dover Publications, New York, 1995. [68] N. Agmon, The Grotthuss mechanism, Chemical Physics Letters, 244 (1995) 456-462. [69] M.E. Tuckerman, D. Marx, M. Parrinello, The nature and transport mechanism of hydrated hydroide ions in aqueous solution, Nature, 417 (2002) 925-929. [70] P. Długołęcki, B. Anet, S.J. Metz, K. Nijmeijer, M. Wessling, Transport limitations in ion exchange membranes at low salt concentrations, Journal of Membrane Science, 346 (2010) 163-171. [71] R. Femmer, A. Mani, M. Wessling, Ion transport through electrolyte/polyelectrolyte multi-layers, Scientific Reports, 5 (2015) 11583. [72] R. abu-Rjal, V. Chinaryan, M.Z. Bazant, I. Rubinstein, B. Zaltzman, Effect of concentration polarization on permselectivity, Physical Review E, 89 (2014). [73] Y. Oren, A. Litan, The state of the solution-membrane interface during ion transport across an ionexchange membrane, The Journal of Physical Chemistry, 78 (1974) 1805-1811. [74] V.I. Zabolotsky, J.A. Manzanares, V.V. Nikonenko, K.A. Lebedev, E.G. Lovtsov, Space charge effect on competitive ion transport through ion-exchange membranes, Desalination, 147 (2002) 387-392. [75] Y. Kim, W.S. Walker, D.F. Lawler, Competitive separation of di- vs. mono-valent cations in electrodialysis: effects of the boundary layer properties, Water Res, 46 (2012) 2042-2056. [76] I. Rubinstein, Theory of Concentration Polarization Effects in Electrodialysis on Counter-ion Selectivity of Ion-exchange Membranes with Differing Counter-ion Distribution Coefficients, Journal of Chemical Society, Faraday Transactions, 86 (1990) 1857-1861. [77] V.V. Kotov, O.V. Peregonchaya, V.F. Selemenev, Electrodialysis of Binary Electrolyte Mixtures with Membranes Modified by Organic Species, Russian Journal of Electrochemistry, 38 (2002) 927-929. [78] F. Helfferich, Ion-Exchange, McGraw-Hill, New York, 1962. [79] B. Tansel, Significance of thermodynamic and physical characteristics on permeation of ions during membrane separation: Hydrated radius, hydration free energy and viscous effects, Separation and purification technology, 86 (2012) 119-126. [80] J. E. R. Nightingale, Phenomenological theory of ion solvation. Effective radii of hydrated ions., The Journal of Physical Chemistry, 63 (1959) 1381-1387. [81] Y. Marcus, Ionic radii in aqueous solutions, CHEMICAL REVIEWS, 88 (1988) 1475-1498. [82] T. Xu, K. Hu, A simple determination of counter-ionic permselectivity in an ion exchange membrane from bi-ionic membrane potential measurements: permselectivity of anionic species in a novel anion exchange membrane, Separation and Purification Technology, 40 (2004) 231-236. [83] W. Kunz, P.L. Nostro, B.W. Ninham, The present state of affairs with Hofmeister effects, Current Opinion in Colloid & Interface Science, 9 (2004) 1-18. [84] Y. Zhang, P.S. Cremer, Interactions between macromolecules and ions: The Hofmeister series, Curr Opin Chem Biol, 10 (2006) 658-663. [85] J. Song, T.H. Kang, M.W. Kim, S. Han, Ion specific effects: decoupling ion–ion and ion–water interactions, Physical Chemistry Chemical Physics, 17 (2015) 8306-8322. [86] Y. Marcus, Thermodynamics of solvation of ions. Part 5.—Gibbs free energy of hydration at 298.15 K, Journal of the Chemical Society, Faraday Transactions, 87 (1991) 2995-2999. [87] P. Atkins, J.d. Paula, Physical Chemistry, 8 ed., Oxford University Press, New York, 2006. [88] Y.-H. Li, S. Gregory, Diffusion of ions in sea water and in deep-sea sediments, Geochimica et Cosmochimica Acta, 38 (1974) 703-714. [89] W.M. Haynes, CRC handbook of chemistry and physics, CRC press, 2014. [90] Y. Li, T. Xu, Permselectivities of monovalent anions through pyridine-modified anion-exchange membranes, Separation and Purification Technology, 61 (2008) 430-435. [91] R. Tandon, P. Pintauro, Divalent/monovalent cation uptake selectivity in a Nafion cation-exchange membrane: Experimental and modeling studies, Journal of Membrane Science, 136 (1997) 207-219. 49
[92] J. Bontha, P.N. Pintauro, Water orientation and ion solvation effects during multicomponent salt partitioning in a Nafion cation exchange membrane, Chemical Engineering Science, 49 (1994) 38353851. [93] J. Bontha, P. Pintauro, Prediction of ion solvation free energies in a polarizable dielectric continuum, The Journal of Physical Chemistry, 96 (1992) 7778-7782. [94] A.E. Yaroshchuk, Dielectric exclusion of ions from membranes, Advances in colloid and interface science, 85 (2000) 193-230. [95] B. Saliha, F. Patrick, S. Anthony, Investigating nanofiltration of multi-ionic solutions using the steric, electric and dielectric exclusion model, Chemical Engineering Science, 64 (2009) 3789-3798. [96] T. Malewitz, P.N. Pintauro, D. Rear, Multicomponent absorption of anions in commercial anionexchange membranes, Journal of Membrane Science, 301 (2007) 171-179. [97] S. Sodaye, G. Suresh, A.K. Pandey, A. Goswami, Studies on diffusional mobility and selectivity of I− ion in plasticized anion-exchange membrane using radiotracer, Radiochimica Acta, 94 (2006). [98] N. Vlachy, B. Jagoda-Cwiklik, R. Vácha, D. Touraud, P. Jungwirth, W. Kunz, Hofmeister series and specific interactions of charged headgroups with aqueous ions, Advances in colloid and interface science, 146 (2009) 42-47. [99] K.D. Collins, Ions from the Hofmeister series and osmolytes: effects on proteins in solution and in the crystallization process, Methods, 34 (2004) 300-311. [100] K.D. Collins, G.W. Neilson, J.E. Enderby, Ions in water: characterizing the forces that control chemical processes and biological structure, Biophysical chemistry, 128 (2007) 95-104. [101] T.T. Duignan, D.F. Parsons, B.W. Ninham, Collins’s rule, Hofmeister effects and ionic dispersion interactions, Chemical Physics Letters, 608 (2014) 55-59. [102] H. Strathmann, A. Grabowski, G. Eigenberger, Ion-Exchange Membranes in the Chemical Process Industry, Industrial & Engineering Chemistry Research, 52 (2013) 10364-10379. [103] G.S. Gohil, V.V. Binsu, V.K. Shahi, Preparation and characterization of mono-valent ion selective polypyrrole composite ion-exchange membranes, Journal of Membrane Science, 280 (2006) 210-218. [104] R.K. Nagarale, G.S. Gohil, V.K. Shahi, Recent developments on ion-exchange membranes and electro-membrane processes, Advances in Colloid and Interface Science, 119 (2006) 97-130. [105] G.M. Geise, H.J. Cassady, D.R. Paul, B.E. Logan, M.A. Hickner, Specific ion effects on membrane potential and the permselectivity of ion exchange membranes, Physical Chemistry Chemical Physics, 16 (2014) 21673-21681. [106] P. Dlugolecki, K. Nymeijer, S. Metz, M. Wessling, Current status of ion exchange membranes for power generation from salinity gradients, Journal of Membrane Science, 319 (2008) 214-222. [107] Y. Ji, G.M. Geise, The role of experimental factors in membrane permselectivity measurements, Industrial & Engineering Chemistry Research, (2017). [108] S. Abdu, M.-C. Martí-Calatayud, J.E. Wong, M. García-Gabaldón, M. Wessling, Layer-by-Layer Modification of Cation Exchange Membranes Controls Ion Selectivity and Water Splitting, ACS Applied Materials & Interfaces, 6 (2014) 1843-1854. [109] V.I. Zabolotsky, V.V.Nikonenko, Ion Transport in Membranes, Nauka, Moscow, 1996. [110] J. Krol, M. Wessling, H. Strathmann, Chronopotentiometry and overlimiting ion transport through monopolar ion exchange membranes, Journal of Membrane Science, 162 (1999) 155-164. [111] T. Okada, H. Satou, M. Okuno, M. Yuasa, Ion and water transport characteristics of perfluorosulfonated ionomer membranes with H+ and alkali metal cations, Journal of Physical Chemistry B, 106 (2002) 1267-1273. [112] T. Sata, Y. Yamane, K. Matsusaki, Preparation and Properties of Anion Exchange Membranes Having Pyridinium or Pyridinium Derivatives as Anion Exchange Groups, Journal of Polymer Science: Part A: Polymer Chemistry, 36 (1998) 49-58. [113] Y. Toyoshima, H. Nozaki, Bi-ionic Potential across Charged Membranes, The Journal of Physical Chemistry, 74 (1970) 2704-2710. [114] D.K. Hale, K.P. Govindan, Bi-ionic Potentials with Sulfonic and Phosphonic Acid Cation-Exchange Membranes, Journal of Electrochemical Society: Electrochemical Science and Technology, 116 (1969) 1373-1381. [115] F. Helfferich, Bi-ionic Potentials, Discussions of the Faraday Society, (1956) 83-94. [116] M. Ottøy, T. Førland, S.K. Ratkje, S. Møller-Holst, Membrane transference numbers from a new emf method, Journal of Membrane Science, 74 (1992) 1-8. [117] T. Okada, S. Møller-Holst, O. Gorsethb, S. Kjelstrup, Transport and equilibrium properties of Nafion® membranes with H+ and Na+ ions, Journal of Electroanalytical Chemistry, 442 (1998) 137-145. 50
[118] X.T. Le, P. Viel, P. Jégou, A. Garcia, T. Berthelot, T.H. Bui, S. Palacin, Diazonium-induced anchoring process: an application to improve the monovalent selectivity of cation exchange membranes, Journal of Materials Chemistry, 20 (2010) 3750. [119] S. Mulyati, R. Takagi, A. Fujii, Y. Ohmukai, H. Matsuyama, Simultaneous improvement of the monovalent anion selectivity and antifouling properties of an anion exchange membrane in an electrodialysis process, using polyelectrolyte multilayer deposition, Journal of Membrane Science, 431 (2013) 113-120. [120] L. Firdaous, J.P. Malériat, J.P. Schlumpf, F. Quéméneur, Transfer of Monovalent and Divalent Cations in Salt Solutions by Electrodialysis, Separation Science and Technology, 42 (2007) 931-948. [121] G. Saracco, Transport properties of monovalent-ion-permselective membranes, Chemical Engineering Science, 52 (1997) 3019-3031. [122] P. Długołęcki, P. Ogonowski, S.J. Metz, M. Saakes, K. Nijmeijer, M. Wessling, On the resistances of membrane, diffusion boundary layer and double layer in ion exchange membrane transport, Journal of Membrane Science, 349 (2010) 369-379. [123] R. Femmer, M. Martí-Calatayud, M. Wessling, Mechanistic modeling of the dielectric impedance of layered membrane architectures, Journal of Membrane Science, 520 (2016) 29-36. [124] F. Roghmans, M. Martí-Calatayud, S. Abdu, R. Femmer, R. Tiwari, A. Walther, M. Wessling, Electrochemical impedance spectroscopy fingerprints the ion selectivity of microgel functionalized ionexchange membranes, Electrochemistry Communications, 72 (2016) 113-117. [125] M. Wang, Y.-x. Jia, T.-t. Yao, K.-k. Wang, The endowment of monovalent selectivity to cation exchange membrane by photo-induced covalent immobilization and self-crosslinking of chitosan, Journal of Membrane Science, 442 (2013) 39-47. [126] Y. Hu, M. Wang, D. Wang, X. Gao, C. Gao, Feasibility study on surface modification of cation exchange membranes by quaternized chitosan for improving its selectivity, Journal of Membrane Science, 319 (2008) 5-9. [127] C. Vallois, Separation of H+/Cu2+ cations by electrodialysis using modified proton conducting membranes, Journal of Membrane Science, 216 (2003) 13-25. [128] T. Sata, R. Izuo, Modification of Transport Properties of Ion Exchange Membranes, X. Formation of Polyethyleneimine Layer on Cation Exchange Membrane by Acid-Amide Bond, Die Angewandte Makromolekulare Chemie, 171 (1989) 101-117. [129] G. Chamoulaud, D. Belanger, Chemical Modification of the Surface of a Sulfonated Membrane by Formation of a Sulfonamide Bond, Langmuir, 20 (2004) 4989-4995. [130] J. Li, M.-l. Zhou, J.-y. Lin, W.-y. Ye, Y.-q. Xu, J.-n. Shen, C.-j. Gao, B. Van der Bruggen, Monovalent cation selective membranes for electrodialysis by introducing polyquaternium-7 in a commercial cation exchange membrane, Journal of Membrane Science, 486 (2015) 89-96. [131] S. Tan, J.H. Tieu, D. Belanger, Chemical Polymerization of Aniline on a Poly(styrene sulfonic acid) Membrane: Controlling the Polymerization Site Using Different Oxidants, Journal of Physical Chemistry B, (2005) 14085-14092. [132] H. Farrokhzad, M. Moghbeli, T. Van Gerven, B. Van der Bruggen, Surface modification of composite ion exchange membranes by polyaniline, Reactive and Functional Polymers, 86 (2015) 161167. [133] P. Sivaraman, J. Chavan, A. Thakur, V. Hande, A. Samui, Electrochemical modification of cation exchange membrane with polyaniline for improvement in permselectivity, Electrochimica acta, 52 (2007) 5046-5052. [134] A.M. Rojas, Y.O. Maldonado, L.M.T. Rodríguez, An easy method to modify the exchange membranes of electrodialysis with electrosynthetized polyaniline, Journal of Membrane Science, 300 (2007) 2-5. [135] R.K. Nagarale, G.S. Gohil, V.K. Shahi, G.S. Trivedi, R. Rangarajan, Preparation and electrochemical characterization of cation- and anion-exchange/polyaniline composite membranes, Journal of Colloid and Interface Science, 277 (2004) 162-171. [136] T.R. Farhat, J.B. Schlenoff, Ion transport and equilibria in polyelectrolyte multilayers, Langmuir, (2001) 1184-1192. [137] C. Cheng, N. White, H. Shi, M. Robson, M.L. Bruening, Cation separations in electrodialysis through membranes coated with polyelectrolyte multilayers, Polymer, 55 (2014) 1397-1403. [138] C. Cheng, A. Yaroshchuk, M.L. Bruening, Fundamentals of Selective Ion Transport through Multilayer Polyelectrolyte Membranes, Langmuir, 29 (2013) 1885-1892. 51
[139] M.L. Bruening, D.M. Sullivan, Enhancing the Ion-Transport Selectivity of Multilayer Polyelectrolyte Membranes, Chemistry - A European Journal, 8 (2002) 3832-3837. [140] J.L. Stair, J.J. Harris, M.L. Bruening, Enhancement of the Ion-Transport Selectivity of Layered Polyelectrolyte Membranes through Cross-Linking and Hybridization, Chemistry of Materials, 13 (2001) 2641-2648. [141] N. White, M. Misovich, A. Yaroshchuk, M.L. Bruening, Coating of Nafion Membranes with Polyelectrolyte Multilayers to Achieve High Monovalent/Divalent Cation Electrodialysis Selectivities, ACS Applied Materials & Interfaces, 7 (2015) 6620-6628. [142] M. Wang, X.-l. Wang, Y.-x. Jia, X. Liu, An attempt for improving electrodialytic transport properties of a heterogeneous anion exchange membrane, Desalination, 351 (2014) 163-170. [143] J. Xi, Z. Wu, X. Teng, Y. Zhao, L. Chen, X. Qiu, Self-assembled polyelectrolyte multilayer modified Nafion membrane with suppressed vanadium ion crossover for vanadium redox flow batteries, Journal of Materials Chemistry, 18 (2008) 1232. [144] W. Xu, X. Li, J. Cao, H. Zhang, H. Zhang, Membranes with well-defined ions transport channels fabricated via solvent-responsive layer-by-layer assembly method for vanadium flow battery, Sci Rep, 4 (2014) 4016. [145] G. Decher, Fuzzy Nanoassemblies: Toward Layered Polymeric Multicomposites, Science, 277 (1997) 1232-1237. [146] J. Qiu, M. Zhai, J. Chen, Y. Wang, J. Peng, L. Xu, J. Li, G. Wei, Performance of vanadium redox flow battery with a novel amphoteric ion exchange membrane synthesized by two-step grafting method, Journal of Membrane Science, 342 (2009) 215-220. [147] S.M. Hosseini, S.S. Madaeni, H. Asiani, A.R. Heidari, Preparation and Electrochemical Characterization of Monovalent Ion Selective Poly (Vinyl Chloride)-Blend-Poly (Styrene-Co-Butadiene) Heterogeneous Cation Exchange Membrane Coated with Poly (Methyl Methacrylate), Separation Science and Technology, 47 (2012) 1443-1454. [148] A.K. Thakur, M. Manohar, V.K. Shahi, Controlled metal loading on poly(2-acrylamido-2-methylpropane-sulfonic acid) membranes by an ion-exchange process to improve electrodialytic separation performance for mono-/bi-valent ions, Journal of Materials Chemistry A, 3 (2015) 18279-18288. [149] S. Chaudhury, A. Bhattacharyya, A. Goswami, Electrodriven Ion Transport through Crown Ether– Nafion Composite Membrane: Enhanced Selectivity of Cs+over Na+by Ion Gating at the Surface, Industrial & Engineering Chemistry Research, 53 (2014) 8804-8809. [150] H. Zhang, H. Zhang, F. Zhang, X. Li, Y. Li, I. Vankelecom, Advanced charged membranes with highly symmetric spongy structures for vanadium flow battery application, Energy & Environmental Science, 6 (2013) 776. [151] T. Luo, O. David, Y. Gendel, M. Wessling, Porous poly (benzimidazole) membrane for all vanadium redox flow battery, Journal of Power Sources, 312 (2016) 45-54. [152] Z. Yuan, Y. Duan, H. Zhang, X. Li, H. Zhang, I. Vankelecom, Advanced porous membranes with ultra-high selectivity and stability for vanadium flow batteries, Energy & environmental science, 9 (2016) 441-447. [153] J. Balster, O. Krupenko, I. Punt, D. Stamatialis, M. Wessling, Preparation and characterisation of monovalent ion selective cation exchange membranes based on sulphonated poly(ether ether ketone), Journal of Membrane Science, 263 (2005) 137-145. [154] G. Gohil, R. Nagarale, V. Binsu, V.K. Shahi, Preparation and characterization of monovalent cation selective sulfonated poly (ether ether ketone) and poly (ether sulfone) composite membranes, Journal of colloid and interface science, 298 (2006) 845-853. [155] H. Farrokhzad, T. Kikhavani, F. Monnaie, S. Ashrafizadeh, G. Koeckelberghs, T. Van Gerven, B. Van der Bruggen, Novel composite cation exchange films based on sulfonated PVDF for electromembrane separations, Journal of Membrane Science, 474 (2015) 167-174. [156] H. Farrokhzad, S. Darvishmanesh, G. Genduso, T. Van Gerven, B. Van der Bruggen, Development of bivalent cation selective ion exchange membranes by varying molecular weight of polyaniline, Electrochimica acta, 158 (2015) 64-72. [157] L. Ge, X. Liu, G. Wang, B. Wu, L. Wu, E. Bakangura, T. Xu, Preparation of proton selective membranes through constructing H+ transfer channels by acid–base pairs, Journal of Membrane Science, 475 (2015) 273-280. [158] M. Kumar, B.P. Tripathi, V.K. Shahi, Ionic transport phenomenon across sol–gel derived organic– inorganic composite mono-valent cation selective membranes, Journal of Membrane Science, 340 (2009) 52-61. 52
[159] L. Ge, L. Wu, B. Wu, G. Wang, T. Xu, Preparation of monovalent cation selective membranes through annealing treatment, Journal of Membrane Science, 459 (2014) 217-222. [160] Y. Wu, C. Wu, Y. Li, T. Xu, Y. Fu, PVA–silica anion-exchange hybrid membranes prepared through a copolymer crosslinking agent, Journal of Membrane Science, 350 (2010) 322-332. [161] M. Kumar, M.A. Khan, Z.A. AlOthman, M.R. Siddiqui, Polyaniline modified organic–inorganic hybrid cation-exchange membranes for the separation of monovalent and multivalent ions, Desalination, 325 (2013) 95-103. [162] S.M. Hosseini, S.S. Madaeni, A.R. Heidari, A. Amirimehr, Preparation and characterization of ionselective polyvinyl chloride based heterogeneous cation exchange membrane modified by magnetic iron–nickel oxide nanoparticles, Desalination, 284 (2012) 191-199. [163] S. Hosseini, S. Madaeni, A. Khodabakhshi, Preparation and characterization of PC/SBR heterogeneous cation exchange membrane filled with carbon nano-tubes, Journal of Membrane Science, 362 (2010) 550-559. [164] S. Hosseini, S. Madaeni, A. Khodabakhshi, Preparation and characterization of ABS/HIPS heterogeneous anion exchange membrane filled with activated carbon, Journal of applied polymer science, 118 (2010) 3371-3383. [165] S.M. Hosseini, A. Gholami, P. Koranian, M. Nemati, S.S. Madaeni, A.R. Moghadassi, Electrochemical characterization of mixed matrix heterogeneous cation exchange membrane modified by aluminum oxide nanoparticles: Mono/bivalent ionic transportation, Journal of the Taiwan Institute of Chemical Engineers, 45 (2014) 1241-1248. [166] M. Arsalan, M.M.A. Khan, Rafiuddin, A comparative study of theoretical, electrochemical and ionic transport through PVC based Cu3(PO4)2 and polystyrene supported Ni3(PO4)2 composite ion exchange porous membranes, Desalination, 318 (2013) 97-106. [167] R. Nagarale, G. Gohil, V.K. Shahi, R. Rangarajan, Preparation and electrochemical characterizations of cation-exchange membranes with different functional groups, Colloids and Surfaces A: Physicochemical and Engineering Aspects, 251 (2004) 133-140. [168] X. Yan, S. Gu, G. He, X. Wu, W. Zheng, X. Ruan, Quaternary phosphonium-functionalized poly(ether ether ketone) as highly conductive and alkali-stable hydroxide exchange membrane for fuel cells, Journal of Membrane Science, 466 (2014) 220-228. [169] Y. Zha, M.L. Disabb-Miller, Z.D. Johnson, M.A. Hickner, G.N. Tew, Metal-Cation-Based Anion Exchange Membranes, Journal of the American Chemical Society, 134 (2012) 4493-4496. [170] C. Sheng, S. Wijeratne, C. Cheng, G.L. Baker, M.L. Bruening, Facilitated ion transport through polyelectrolyte multilayer films containing metal-binding ligands, Journal of Membrane Science, 459 (2014) 169-176. [171] S. Tas, Bio-inspired ion selective crown-ether polymer membranes, in: Membrane Technology Group, Department of Engineering, University of Twente, Enshede, the Netherlands, 2016.
53