New Gel-Like Polymers as Selective Weak-Base Anion Exchangers

A group of new anion exchangers, based on polyamine podands and of excellent ion-binding capacity, were synthesized. The materials were obtained in reactions between various poly(ethyleneamines) with glycidyl derivatives of cyclotetrasiloxane. The final polymeric, strongly cross-linked materials form gel-like solids. Their structures and interactions with anions adsorbed were studied by spectroscopic methods (CP-MAS NMR, FR-IR, UV-Vis). The sorption isotherms and kinetic parameters were determined for 29 anions. Materials studied show high ion capacity and selectivity towards some important anions, e.g., selenate(VI) or perrhenate.


Introduction
The search for selective polymers or hybrid exchangers or sorbents of anionic species is a fastdeveloping aspect of material chemistry [1,2,3,4], however the number of reports on this topic is much lower than those on the sorption of cations by polymer materials. The anion-binding and anion-exchanging polymers and composites are of interest as materials for harmful anion removal from drinking water [5,6,7], waste-water management [8,9,10], analyte preconcentration prior to anionic-species analysis [11,12], as carriers of anionic pharmaceuticals [13,14,15] or for catalysts recovery [16]. A variety of materials have been tested for anion sorption, including the inorganic sorbents (such as alumina and other metal oxides) [6,17,18], synthetic resins [19,20], chitosan [5,7,21,22,23] and a wide range of inorganic and organic supports grafted with anion binding side-chains [11,12,24,25]. In most cases, the active anion-binding sites are the quaternized amino groups (both exo-and endocyclic), which interact with anions through electrostatic forces. For some oxoanions, more specific systems have been designed, e.g. 1,2-diaminoaryl moieties for Se(IV) separation [12] or organophosphonate-Zr(IV) adducts for binding of tungstate(VI) and molybdate(VI) [26].
In our previous paper we have shown the preparation and application of novel gel-like polyamine resins for removal of heavy metal cations [27]. The synthesis of the polymer involves a new strategy, based on simultaneous polymerization of polysiloxane precursor and its modification by chelating units. We decided to use the same material for preparation of anion-exchange polymers, containing ionized amino groups, i.e. weak-base ion exchangers. This class of ion-exchange materials has some important advantages over strong-base ones (materials containing quaternized ammonium and phosphonium centres). The most important one is the increased selectivity towards some oxyanions. Strong-base exchangers are hydrophobic, therefore the major parameter determining their selectivity towards various anions is the degree of the hydration of separated ions, as long as their charge remains the same. The weak-base exchangers interacts with anions adsorbed not only via the electrostatic forces, but by formation of hydrogen and coordination bonds. This change their selectivity, e.g. increases the affinity to multivalent anions or oxoanions containing larger number of oxygen atoms [28][29][30][31].

Experimental
All reagents used are commercial products. The salts used for anion binding studies had the purity grade p.a. or higher and were purchased from Sigma-Aldrich. They were used without purification. The starting polymers (N2-N6) were obtained according to procedures described earlier [27]. All other compounds were purchased from Sigma-Aldrich and Merck and used as received. The solvents used for measurements were SPECTRANAL grade.

Synthesis of protonated resins
One g of the starting polymer (N2-N6) was ground in a mortar and suspended in acetonitrile (50 ml). To the obtained mixture, fivefold molar excess (calculated on the basis of the nitrogen amount in the starting material) of perchloric or hexafluorophosphoric(V) acid in acetonitrile (5% by weight) was added. The suspensions obtained were mixed overnight, then the solid resins were filtered off, washed three times with acetonitrile and dried in vacuum at room temperature.
CAUTION. Solutions of perchloric acid in organic solvents may be explosive. Amine perchlorates could detonate upon heating or hitting.

Spectroscopy
Nuclear magnetic resonance spectra of the solids ( 13 C, 15 N, 29 Si) were obtained at 298 K, on a Varian VNMR-S 400 MHz spectrometer, operating at frequencies 101.25, 40.84 and 80.00 MHz, respectively. The standard CP-MAS technique was used (tancpx sequence with 1 H decoupling). The 13 C spectra were referred to the methylene signal of external glycine (43.30 ppm), acquisition time-40 ms, cross-polarization contact time-2 ms, sample rotation-5 kHz. The 15 N spectra were referred to the signal of external glycine (-347.60 ppm), acquisition time-35 ms, cross-polarization contact time-2 ms, sample rotation-6 kHz. The 29 Si spectra were referred to external neat TMS (0.00 ppm), the acquisition parameters were the same as for 15 N NMR spectra.
Infrared spectra were recorded on a Bruker IFS 66s spectrometer, using KBr pellets (2 mg of sample in 200 mg of KBr). The complexes for these measurements were prepared by immersion of a polymer sample in 1Á10 -2 M solution of the corresponding sodium salt in water for 10 h, followed by decantation, washing of a solid material with water and methanol and drying in vacuum at room temperature.

Elemental analysis, chemical analysis
The contents of C, H and N were determined using a Vario EL III elemental analyser. The content of silicon in the polymers was determined on an EDXRF spectrometer MiniPal 2 (PANalytical) after sample solubilisation with hydrofluoric acid. Iodine and phosphorus content in final polymers, after its solubilisation in 0.5 M NaOH, were determined by the ICP-OES method on a Varian Vista-MPX spectrometer with argon plasma ionization. The perchlorates were determined in the same solution spectrophotometrically, as an ion-pair with Brilliant Green according to the procedure described by Fogg et al [32]. Ion-chromatographic analyses were carried out on Methom Vario system.

Isotherms of adsorption, adsorption kinetics, sorption-regeneration experiments
The adsorption measurements were performed for water solutions of sodium salts (except tellurate, which was used as a potassium salt) of studied ions (F -, Cl -, Br -, I -, NO 3 - ). To make each isotherm, a series of samples containing 0.01 g of polymer and 5 ml of salt solution were used. The adsorption properties at eight concentrations were measured for each system (0.1, 0.5, 1, 2, 3, 4, 5 and 10 mM). The pH of the solutions was adjusted with NaOH and HClO 4 or HPF 6 (depending on the polymer counter ion) to 5.0. The mixtures were equilibrated for 24 h at 298 ± 1 K. The starting (C 0 ) and final, equilibrium (C A ) concentrations of anions were determined with the method dependent on the character of the counter ion in the polymer ( Table 1). The amount of the anion adsorbed (q A ; mmol/g) in equilibrium was calculated from the difference between these concentrations (Eq 1): where m is the mass of the adsorbing polymer (g) and V is the volume of salt solution (dm 3 ). The relations between the sorption properties of the polymers studied and the pH of the solution were studied in various pH ranges, taking into regard the complex pH-dependent equilibriums known for some anions studied (e.g. MoO 4 2or WO 4 2-) or anion instability (e.g. SO 3

2-
). The pH of the solutions was adjusted with NaOH and HClO 4 (or HPF 6 ). To calculate the maximal amount of the metal ions adsorbed by the polymer, the measurements were made for five salt concentrations. Other experimental details were as for the isotherms determination.
The adsorption kinetics was studied for 1 × 10 -2 M salt concentration, using 20 mg of finely ground adsorbent (150-200 mesh) and 20 ml of adsorbate solution at 298 ± 1 K; pH was adjusted at 5.0. This stoichiometry guarantees large excess of the ions over the binding sites, therefore the pseudo second-order kinetic model could be used. The reaction mixtures were magnetically stirred vigorously at a stirring speed of 800 rpm. The same volume aliquots (0.1 ml) were withdrawn periodically and the metal concentration was determined by the methods mentioned above. The first-order rate constants were calculated from the equation (Eq 2): while the second-order rate constants were calculated from the equation (Eq 3): where q t and q A are the amounts of anion adsorbed after time t and at equilibrium, respectively, (in mmol of adsorbate per 1 g of sorbent); k 1 , k 2 are the rate constants. Sorption-regeneration experiments were made for 0.05 g samples of resins. The sample studied was suspended in 5 ml of the appropriate adsorbate solution (1 × 10 -2 M), pH was adjusted to 5.0 and the suspension was stirred vigorously for over 5 h. Then the solid material was filtered off. The amount of exchanged ions was determined in the filtrate by the method indicated in Table 1. The resins were regenerated in two ways. The first method was a two-step process. The solid material was suspended in 10 ml of NaOH solution (1 × 10 -2 M), mixed over 5 h, separated by centrifugation, washed with water and protonated with HClO 4 (or HPF 6 ) as described in section 2.1. The amount of desorbed ions was determined in alkaline supernatant. The second method of resins regeneration consisted of mixing it with 10 ml of 10% solution of NaClO 4 (or NaPF 6 ) and separation of the solid material by centrifugation. The resins were re-used after washing with water. The concentration of desorbed ions was determined in brine.
Ten cycles of ions binding/sorbent regeneration were performed. The efficiency of the anion retention in the n-th cycle was expressed as the retention factor R n (Eq 4): while the ability to ions release in alkaline/brine solution in the n-th cycle was described by the release factor (Eq 5): where n 1 A , n n A are the amounts of adsorbed ions in the first and n-th adsorption cycle (in mmol), n n R -the amount of anions released during the n-th regeneration.

Water uptake
The swelling of the resins was studied in water. The polymers used for this study were vacuum dried at 373 K and the particles of 20-50 mesh were used. The dry, weighted material (about 250 mg) was allowed to swell in water for 24 h at 298 K, filtered off, the excess of water was removed by blotting and the sample was weighted again. The water uptake was calculated from equation (Eq 6): where m 0 and m s are the masses of dry and swelled samples respectively.

Thermal analysis
Derivative scanning calorimetry and thermogravimetric measurements were performed on a Setoram TGA thermoanalyser in nitrogen atmosphere at the constant heating rate of 10 K min -1 . The samples were dried in vacuum (1 mmHg) at 353 K for 5 h before analysis. Due to violent decomposition of perchlorate-containing resins, only hexafluorophosphate(V) derivatives were studied.

Results and Discussion
Polymer structure The new polymeric ion adsorbents with polyamine podand arms were synthesized according to the synthetic route presented in Fig 1. After formation of cross-linked polyamine/polysiloxane resin, the nitrogen atoms were quaternalized by protonation with perchloric or hexafluorophosphoric acid. These acids were chosen due to their chemical properties. Both are strong, monoprotonic acids (pK a < -5), therefore they do not undergo anionic hydrolysis, which may influence the sorption properties in more acidic solution. Perchlorate anion forms strong hydrogen bonds with polyamine net of the studied materials and is effectively bonded by them. Contrary, PF 6 anion is weakly adsorbed and easily exchanged. The use of these anions as counter-ions in the resins studied permits the studies and comparison of the sorption of both weakly adsorbed anions (e.g. halides) as well as strongly bonded ones (e.g. perrhenate, selenate (VI)). Moreover, they are stable in oxidizing and reducing conditions, therefore they do not  N + -molar ratio of protonated nitrogen atoms to total nitrogen; WA-water uptake elemental analysis and N 1 :N 2 :N 3 ratio in parent polymers. As shown for all studied systems, the cationized nitrogen atoms constitute over 90% of the total number of nitrogen atoms. Water uptake by the resins studied was high. The values of this parameter were distinctly higher than that for the parent, non-ionic polymers [27]. This is a result of higher affinity of water to cationic N-centres than to amino groups as well as high hydration rates of perchlorate and hexafluorophosphate(V) ions. For the resins studied, the water uptake increases with increasing polyamine chain length, which is a result of an increasing number of water binding groups (N atoms and counter ions) as well as loosening of the polymer structure due to elongation of the cross-linking chains.
The materials obtained are stable in acidic conditions-i.e. the polymer matrix does not show any signs of decomposition after stirring for 10 days in 1 M HCl or H 2 SO 4 . The IR and NMR spectra of so treated resins are identical as those of the parent ones (except changes resulting from ion-exchange). The treatment of the exchangers with strong, concentrated base solutions (e.g. 1 M NaOH or KOH) leads to fragmentation of poly(siloxane) chains and decomposition of the polymers studied. They are however stable in less concentrated hydroxide solutions (< 0.05 M).

CP-MAS spectra of polymers and their complexes
The structures of the polymers obtained are fully confirmed by NMR measurements. The 13 C spectra are very similar to those of parent resins. The signals of non-polyamine part of the molecules were observed at the same chemical shifts in both protonated and neutral polymers [27]. The siloxane parts of the molecules gave the characteristic signals at -1.7 ppm (SiCH 3 , cyclic siloxane), -0.1 ppm (SiCH 3 , linear siloxane), 13.6 ppm (SiCH 2 ), 23.5 ppm (CCH 2 C), 68.8 ppm (CHOH) and 74.0 ppm (CH 2 O). The only difference was the greater line widths observed for the cationic forms. The ionization affects strongly the signals of the carbon atoms bonded to nitrogen atoms. Protonation resulted in the 3-7 ppm upfield shift of the signals assigned to the nitrogen-bonded C atoms. The 29 Si NMR spectra of the polymers studied were not affected by protonation of polyamine part of the molecule, i.e. the signals appeared at -21.8 (linear polysiloxane) and -16.5 ppm (cyclotetrasiloxane). The 15 N NMR spectra of protonated polymers show the signals at ca. -336, -338 and -325 ppm (traces), corresponding to R 2 NH 2 + , RNH 3 + and R 3 NH + centres. Anion exchange does not affect the 13 C, 15 N and 29 Si NMR spectra, i.e. the observed shifts are smaller than the measurements accuracy. As the signals of the studied resins are rather broad, shifts smaller than 2 ppm are not detectable. For the complexes with CrO 4 2-, a large increase in the NMR line widths was detected, caused by paramagnetic relaxation induced by Cr 3+ ion, formed by reduction of chromate(VI) anions. The NMR data are collected in Table 4.

Thermal analysis
The thermograms of the selected polymers studied are presented in Fig 6. The perchlorate forms of the synthesized ionic resins were not studied by TG and DSC because of the violent, explosive decomposition above 250°C. Contrary, PF 6 salts decompose less rapidly, however at lower temperatures than that of the parent, neutral polymers which decompose in the temperature range 325-525°C [27]. Decomposition of ammonium polymers studied consists of several steps. Firstly, between 120 and 150°C, an endothermic process accompanied by an insignificant mass loss (ca. 5-10%) is observed. It is assigned to desorption of water adsorbed in the resin. The major decomposition is observed in the temperature range 200-350°C with two peaks on DSC curve, the first at 270°C (mass loss ca. 25-40%) and the second at 310°C (mass loss of 60-65%). Over 350°C the final decomposition step is observed, resulting in the next substantial loss of mass (up to 75%). It ends at ca. 400°C. As the low-temperature processes should be assigned to pyrolysis of organic part of the polymer, the high temperature ones corresponds to evolution of PF 5 , HF and other products of PF 6 decomposition. As for N2-N6 polymers, the mass loss of ionic resins increases with the number of ethyleneamine units. The residual mass varies from 45% (N2HP) to 32% (N6HP). The residual material contains mainly Si, P, O and traces of F and C. All processes observed on heating of the samples are endothermic (Fig 7). None of the observed processes is reversible, therefore the studied resins do not undergo a glass-transition in the studied temperature ranges. The observed differences in thermal stability between the parent neutral polymers and its protonated derivatives are a result of thermal instability of ammonium salts which undergo elimination reaction, with production of R 3 NH + and C = C double bond.

Anion exchange
All polymers studied are able to exchange anions. The binding capacity of the resins depends on its structure and the anion character. Although the Misak model (Eq 7) is much more adequate for describing the ion exchange [33,34], for the anions originating from weak, polyprotic acids, the determination of the major ionic species adsorbed on a resin is difficult.
where C A , C B are the equilibrium concentrations of the adsorbate (A) and counter ion (B) released to the solution, q A -the amount of ions bonded at the concentration C A , Q maximum amount of ions A bonded, x, y-valence of the ions A and B, K-equilibrium constant of the exchange reaction. Therefore, the maximum amount of the ions bonded (Q, mmol of anion per 1 g of polymer) and the binding equilibrium constant K, were calculated assuming that the adsorption process was described by the Langmuir model (with Boyd modification; Eq 8), which is widely used in New Gel-Like Anion Exchangers ion exchange studies [35]: Calculated K and Q values and are given in Table 6. For the studied weak base anion exchange resins, the affinity of monovalent halogen anions to polymers is very low. All of them interact with the resins studied weaker than PF 6 and This is a result of several cooperative effects. The first is that at pH = 5.0 they exist in a neutral form, so they do not bind to the protonated resin via electrostatic interaction. Moreover, the oxygen atoms of arsenate(III) acid are poorer hydrogen bond acceptors than those of perchlorate anion. Comparing the ions containing the same central atom, it is clearly seen that the affinity to the resin increases with increasing oxidation state of the central atom, e.g. the K values for SO 3 2- . This is related to increasing acid strength and, in consequence, the domination of multiple charged species in the solution. At pH = 5.0 the selenous or sulfurous acids exists mainly as monovalent, HXO 3 anions. They make weaker electrostatic interactions with N + centres than the multiple charged ions. Moreover, the increasing number of oxygen atoms in the anions containing central atom of high ) were not investigated. The kinetics of the reaction was measured for four resins, two perchlorate ones (N2HCl and N6HCl) and their PF 6 analogues (N2HP, N6HP). The exemplary kinetic curves are presented in Fig 9. The kinetic parameters are collected in Table 7. The equilibrium of the reaction is reached quite quickly, in less than 60 min., depending on the anion adsorbed. The adsorption process was analysed using two kinetic models, i.e. pseudo-first order and pseudo-second order [36]. The experimental data were much better reproduced by the second one. Although this simplified approach does not take the sorption mechanism into account, it is widely used for description of the sorption kinetics. Therefore, the presented results should be used only for the comparison of the sorption kinetics properties of the studied materials with literature data and have no physical meaning. The exchange was distinctly slower for N2-derived resins than for N6-derived ones. This is a consequence of the slower diffusion of ions in the polymer matrix cross-linked with shorter polyamine chains. The differences observed are more distinct for large ions (e.g. ReO 4 -) than for those of small radii (Cl -), whose diffusion is less hindered by the close distances between the polysiloxane chains. There is no significant Table 6. Parameters of anions adsorption by the selected polymers studied in water at 298 K (pH = 5.0).

Anion
Adsorption parameters: q m (mmol g -1 ; ± 0.02) and K (± 0.05) difference in the process kinetics between the perchlorate and hexafluorophosphate forms of the exchangers. The pH dependence of the ion-exchange capacity (IEC) of the studied resins is shown in Fig 10. As expected, IEC is almost constant below pH = 5.5. Above this value, IEC drops and reaches 0 at pH > 8. This is a result of deprotonation of NH + centres of the resin. Above pH = 8, the resins exist in neutral, amine form. Any binding of anions in the pH range 8-14 is a physisorption or a result of NHO and OHO hydrogen bonding between resin heteroatoms and anion oxygen atoms. The influence of pH of the solution on the maximum amount of ions bonded (Q) is presented in Fig 11. At low pH values (pH < 3) the anions of strong acids (sulphate, selenate, perchlorate etc.) are preferably adsorbed. Anions of strong acids exist in ionic form even at low pH. The uptake of these anions rise insignificantly with increasing pH, these changes are more distinct for multivalent anions. As the second and subsequent dissociation constants of polyprotonic acids are smaller than the first one, the increasing pH value results in the rise of the population of double (and triple) ionized species, which interact stronger with the protonated resin. Contrary, the anions of weak acids show high increase in the affinity to the resin with decreasing solution acidity. At low pH, these acids exist as non-ionized molecules or the molar ratios of ionic forms are very low. The increase in pH results in ionisation of weak acids and formation of anionic species, which show affinity to the resin. The anion of the weakest acid studied (AsO 2 -) adsorbs at pH > 8 and the formed adduct is stabilised by hydrogen bonds.    as an anion sources, not to the form of the anions existing in solution or bonded to the polymer after sorption, which may be different due to protolytic equilibria). The results of these experiments are shown in Fig 12 and 13. Fig 12 presents the efficiency of two-step regeneration, i.e. desorption of bonded ion in NaOH solution followed by re-protonation of the polymer material. Synthesized material shows excellent reversibility of the exchange properties and high stability. Small (ca. 5%) decrease in the sorption efficiency (R) is shown in the first sorptionregeneration cycle, but the further amounts of exchanged ions are the same up to the tenth cycle. Over 97% of the ions adsorbed are released during treatment with NaOH. The regeneration with concentrated salt (NaClO 4 or NaPF 6 ) solutions gives somewhat different results (Fig 13). Since the regeneration with NaClO 4 almost restores the starting sorption ability of the materials studied, the regeneration with NaPF 6 is much less effective. The recovery of oxoanions of strong acids, showing high affinity to the resins studied, in this regeneration method is ca. 60-70%. This is a result of much stronger bonding of these ions than that of a PF 6 anion, which does not form hydrogen bonds with the polymers studied. For simple anions (e.g. Br -), having lower affinity to the resin than PF 6 -, the D n parameter is over 0.9.

Conclusions
It has been shown that the studied synthetic weak-base anion exchangers show excellent ionexchange properties. They have high ion exchange capacity (2-7 mmol g -1 ), much higher than those of the most commercially used anion-exchange resins (0.5-2 mmol g -1 ) and comparable to those reported for modern, recently reported adsorbents [28,37]. The materials studied are thermally stable and could be regenerated many times without significant loss of activity. They show high affinity to strong acids-related anions, so they could be used for separation of ions containing the same atom at different oxidation states (e.g. arsenate(III) from arsenate(V)). Their excellent adsorption properties to metal oxoions make them interesting materials for separation and pre-concentration of valuable metals (e.g. vanadium or rhenium) in hydrometallurgical processes or for removal of toxic metalloids (selenium, arsenic) from drinking water.

Author Contributions
Conceived and designed the experiments: BG MZ. Performed the experiments: BG MC. Analyzed the data: BG MC MZ. Contributed reagents/materials/analysis tools: BG. Wrote the paper: BG.