Synthesis and Evaluation of Polymer-Based Zwitterionic Stationary

Department of Chemistry, Umeå University, S-901 87 Umeå, Sweden. Anal. Chem. ..... Knut Irgum. Journal of Separation Science 2006 29 (12), 1784-1821...
0 downloads 0 Views 137KB Size
Anal. Chem. 2001, 73, 1993-2003

Synthesis and Evaluation of Polymer-Based Zwitterionic Stationary Phases for Separation of Ionic Species Wen Jiang and Knut Irgum*

Department of Chemistry, Umeå University, S-901 87 Umea˚, Sweden

Three different zwitterionic functional stationary phases for chromatography were synthesized on the basis of 2-hydroxyethyl methacrylate (HEMA) polymeric particles. Two synthesis routes, producing materials designated S300-ECH-DMA-PS or S300-TC-DMA-PS, involved activation of the hydroxyl groups of the HEMA material with epichlorohydrin or thionyl chloride, respectively, followed by dimethylamination and quaternizing 3-sulfopropylation with 1,3-propane sultone. The third route was accomplished by attaching methacrylate moieties to the HEMA through a reaction with methacrylic anhydride, followed by graft photopolymerization of the zwitterionic monomer 3-[N,N-dimethyl-N-(methacryloyloxyethyl)ammonium] propanesulfonate, initiated by benzoin methyl ether under 365-nm light. According to elemental analyses, both the S300-ECH-DMA-PS and S300-TC-DMA-PS materials appeared to have overall charge stoichometries close to unity, whereas the grafted material, S300-MAA-SPE, seemed to carry an excess of anion exchange sites in addition to the zwitterionic groups. Yet all three zwitterionic stationary phases were capable of separating inorganic anions and cations simultaneously and independently using aqueous solutions of perchloric acid or perchlorate salts as eluent, albeit with markedly different selectivities. On the S300-TC-DMA-PS and S300-MAASPE materials, the retention times increased for cations and decreased for anions with increasing eluent concentration, whereas with the S300-ECH-DMA-PS material, the retention times of both anions and cations decreased with increasing eluent concentration. These results demonstrate the importance of choosing appropriate synthesis conditions in order to prepare covalently bonded zwitterionic separation materials with an acceptable charge balance. Separation materials with zwitterionic functionalities have been explored for many years.1 Early experiments with stationary phases containing an intended mix of cation and anion exchange sites were conducted by Knox and Jurand 20 years ago.2 Their materials, based on dynamic modification of reversed-phase (1) Nesterenko, P. N.; Haddad, P. R. Anal. Sci. 2000, 16, 565-574. (2) Knox, J. H.; Jurand, J. (a) J. Chromatogr. 1981, 203, 85-92. (b) J. Chromatogr. 1982, 234, 222-224. (c) J. Chromatogr. 1981, 218, 341-354. (d) J. Chromatogr. 1981, 218, 355-363. 10.1021/ac000933d CCC: $20.00 Published on Web 03/24/2001

© 2001 American Chemical Society

separation materials with long-chain ω-aminocarboxylic acids, had a mixed mode retention mechanism, and the charge characteristics were those of weak/weak zwitterions. The protonation/ dissociation equilibria of the amine/carboxylic groups in weak zwitterionic exchangers imply that an absence of net charge would prevail only at pI, that is, midway between the pKas of the carboxylic and ammonium groups. To overcome the pH-dependent charge state, Yu et al.3 synthesized silica-based materials with covalently bonded zwitterionic exchange sites of both weak/weak and strong/strong types. More recently, Hu and co-workers have exploited the possibilities of using detergents with sulfoalkylbetaine zwitterions dynamically attached to hydrophobic reversed phase columns.4 As opposed to the aminocarboxylic acids, these quaternary ammonium alkylsulfonic acid inner salts, covalently attached to silica3 and applied as dynamic coatings,4 retain their dual charge over practically the entire aqueous pH range. This causes an interesting retention behavior for inorganic and small organic ions, for instance, ion retardation of salts using pure water as eluent, where both the anion and the cation components of a salt contribute to its retention.5 An ideal zwitterionic stationary phase is characterized by having positive and negative charges located in the same functional group. Accordingly, although the surface density of charged moieties is high, the material has in an exact balance and, consequently, no net charge. This close proximity of the opposite charges enables simultaneous separation of cations and anions with water only as eluent. In a recently published paper6 we described the first synthesis of a covalently bonded polymeric zwitterionic stationary phase by a two-step chemical modification of the hydroxyl groups of 2-hydroxyethyl methacrylate (HEMA) polymeric particles. This material could achieve the separation of inorganic anions and cations, independently and simultaneously, using solutions of perchloric acid or perchlorate salts as mobile phases. However, as described in the cited paper, the functionalization reaction was accompanied by a side reaction, which resulted in weak anion exchange groups (estimated to be 1015% of the zwitterionic groups). Although the sorbent showed separation properties characteristic of zwitterionic materials, we were interested in developing a synthesis technique which would (3) Yu, L. W.; Floyd, T. R.; Hartwick, R. A. J. Chromatogr. Sci. 1986, 24, 177182; (b) Yu, L. W.; Hartwick, R. A., J. Chromatogr. Sci. 1989, 27, 176-185. (4) (a) Hu, W. Z.; Haddad, P. R. Trends Anal. Chem. 1998, 17, 73-79. (b) Anal. Commun. 1998, 35, 317-320. (5) Hu, W. Z.; Takeuchi, T.; Haraguchi, H. Anal. Chem. 1993, 65, 2204-2208. (6) Jiang, W.; Irgum, K. Anal. Chem. 1999, 71, 333-344.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001 1993

produce a material with a more exact charge balance without isolated cation or anion exchange groups. In the current study, we thus report on the synthesis of three different types of covalently bonded polymeric sulfoalkylbetaine zwitterionic stationary phases based on the same HEMA polymer sorbent used in our previous work. The chromatographic properties of these materials are evaluated and compared to the previously described material.6 EXPERIMENTAL SECTION Reagents and Chemicals. Separon HEMA S300 particles (7µm particle size, 300-Å nominal pore size) were purchased from Tessek (Prague, Czech Republic). The surface area of these particles was 27 m2/g, as measured by N2 sorption using a Micromeritics (Norcross, GA) FlowSorb II 2300 instrument. The zwitterionic monomers 3-[N,N-dimethyl-N-(methacryloyloxyethyl)ammonium] propanesulfonate (SPE) and 3-[N,N-dimethyl-N(methacrylamidopropyl)ammonium] propanesulfonate (SPP) were obtained from Raschig Chemie (Ludwigshafen, Germany). Methacrylic anhydride (94%), 4-(dimethylamino)pyridine (DMAP, 99%), benzoin methyl ether (BME, 99%), nitrobenzene (99%), 1,3-propane sultone (98%), magnesium perchlorate hexahydrate (99%), calcium perchlorate tetrahydrate (99%), and cerium perchlorate (40% solution in water) were obtained from Aldrich (Steinheim, Germany), and dimethylamine (40% in water), epichlorohydrin (98%), thionyl chloride (99%), and potassium perchlorate (99%) were from Fluka (Buchs, Switzerland). Acetonitrile (HPLC), methanol (HPLC), and sodium hydroxide (50% aqueous solution) were J. T. Baker (Deventeer, Holland) brand. Perchloric acid (70%, p.a.) was purchased from Riedel de Hae¨n (Seelze, Germany). Acetone (purum) for washing was from Svenda AB (Stockholm, Sweden), and chloroform (p.a.) was from Prolabo (Fontenay S/Bois, France). Water was purified by a Milli-Q water purification system (Millipore; Bedford, MA). All of the inorganic salts and the preparations of their stock solutions were the same as in our previous paper.6 Synthesis of the Zwitterionic Stationary Phase S300-ECHDMA-PS. Separon 300 HEMA polymer beads were activated with epichlorohydrin according to a procedure described elsewhere.6 Three grams of these activated particles were placed in a 250-mL round-bottom flask to which was added 100 mL of a 40% aqueous solution of dimethylamine. This suspension was then heated to 60 °C under slow stirring and allowed to react under refluxing conditions for 18 h. These dimethylamine functionalized particles were then filtered on a glass filter gooch and washed with large amounts of water until neutral, then with acetone, and were dried overnight at 40 °C in a vacuum. A 2.5-g aliquot of these dry, dimethylaminated polymeric particles was then weighed into a 250 mL round-bottom flask, to which a mixture containing 3 g of 1,3-propane sultone and 2 mL of nitrobenzene in 100 mL of acetonitrile was added. The suspension was brought to reflux (about 70-80 °C) under stirring, and the reaction took place under nitrogen for 48 h. The reacted particles were then washed repeatedly with water, methanol, and finally, acetone. The acetonemoist particles were directly transferred to water and allowed to settle, and the supernatant was decanted twice to remove fines produced in the synthesis procedure and packed following the procedure below. 1994

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

WARNING: Epichlorohydrin and 1,3-propane sultone are recognized carcinogens and should be handled with due care. Synthesis of the Zwitterionic Stationary Phase S300-TCDMA-PS. Separon 300 HEMA polymer particles (3 g) were converted to the chloroethyl form by reaction with 8 mL of thionyl chloride in 20 mL chloroform for 24 h at room temperature. The activated particles were washed with methanol, water, then acetone, followed by reaction with 100 mL of 40% aqueous dimethylamine solution at 60 °C for 18 h. The dimethylaminated particles were washed, dried, quaternized by sulfopropylation with 1,3-propane sultone, and prepared for packing according to the procedure described in the synthesis of S300-ECH-DMA-PS above. Synthesis of the Grafted Zwitterionic Stationary Phase S300-MAA-SPE. Separon 300 HEMA polymer beads (2 g) were suspended in 50 mL of acetonitrile in a 100-mL Erlenmeyer flask to which was subsequently added 8 mmol (1.233 g) methacrylic anhydride and 0.8 mmol (98 mg) 4-(dimethylamino)pyridine. The flask was placed on an orbital shaker (VIBRAX-VXR; Janke & Kunkel, Staufen im Breisgau, Germany), and the reaction was carried out with a moderate swirl at room temperature (21 ( 2 °C) for 24 h. This methacryloylated polymer was washed repeatedly with water, followed by acetone, and dried under suction. The dried particles were then swollen in a mixture containing 10 mL methanol, 900 mg of the zwitterionic monomer SPE, and 18 mg of the photoinitiator BME for 30 min. To this suspension thereafter was slowly added 20 mL of water under slow stirring. After this addition, the mixture was degassed by He purging for 30 min, followed by photopolymerization. The irradiation chamber was a Spectrolinker XL-1500 (Spectronics Corp.; Westbury, NY), intended for DNA cross-linking, in which the six 15 W transparent quartz UV tubes originally present were replaced by fluorescent blacklight tubes (F15T8/BLB; GTE Sylvania) of the same power rating. This produced an UV irradiation predominantly of 365-nm wavelength, which the built-in power measurement facility of the XL-1500 indicated to be of 8-10 mW/cm2 intensity in the polymerization zone; however, this value is not absolute, because the readout was not calibrated for the wavelength used here. This photopolymerization was allowed to proceed for 1 h under slow stirring at room temperature with the reaction vessel sealed. The grafted particles were then filtered on a glass filter gooch and washed repeatedly with NaCl (∼0.5 M), water, followed by methanol, and again in water, from which it was decanted as described above to prepare the particles for packing. A synthesis with SPP as the zwitterionic monomer was carried out in exactly the same way, with the same mass of monomer, resulting in a material termed S300-MAA-SPP. Characterization of the Zwitterionic Materials. The sulfur and nitrogen contents of all intermediates and final materials were determined by elemental analysis at two different accredited laboratories, who reported the results of duplicate and triplicate analyses, respectively. The reported results are the combined values from both laboratories. An ATI Mattson Genesis Series FT-IR instrument was used to obtain the infrared spectra from tablets pressed from a mixture of ground dried polymer and KBr, and the appearance of a peak in the vicinity of 1040 cm-1 was used to verify the existence of the zwitterionic group through the alkylsulfonic acid moiety.6 The ζ-potentials of the unfunctionalized S300 HEMA material and the fully functionalized particles in

Scheme 1. Synthesis Steps Used in the Preparation of the Zwitterionic Stationary Phase S300-ECH-DMA-PS

Scheme 2. Synthesis Steps Used in the Preparation of the Zwitterionic Stationary Phase S300-TC-DMA-PS

different acid and salts were measured by photon correlation spectroscopy using a Malvern (Malvern, U.K.) Zetasizer 4 instrument. The concentrations of the test solutions are indicated in the tables. Samples were prepared by suspending 50 mg of material in 30 mL of water, and final test solutions were prepared by adding 1 mL of this suspension to 5 mL of the appropriate solution of acid or salt. The materials were suspended in the test solutions, thoroughly mixed on a Heidolph REAX Control vortex mixer, and immediately transferred to the measurement cell. Reported values are mean ( s.d. for three repeated measurements, made in rapid succession, on the same sample. To determine whether a charge hysteresis was present following treatment in acid or base, 50 mg aliquots of S300-ECH-DMA-PS and S300-TCDMA-PS were treated separately with 30 mL of 100 mM HCl and NaOH for 30 min, in each instance followed by washing with two liters of water on a membrane filter. The washed particles were subjected to ζ-potential measurements as described above. Slurry Packing and Chromatographic Evaluation. All zwitterionic stationary phases were slurry-packed into 150-mmlong by 4-mm-i.d. poly(ether-ether-ketone) (PEEK) column blanks (Upchurch Scientific, Oak Harbor, WA), using Milli-Q water as the packing solvent. The packing was accomplished using a highflow pump of the pneumatic amplifier type (Knauer, Berlin, Germany) by applying a gradual, rather fast (≈ 10 s) pressure increase to the final value of 60 MPa. The chromatographic system used for evaluation was essentially identical to that used in our previous paper,6 with the exceptions that the samples were injected through a Rheodyne (Berkeley, CA) 9010 PEEK sample injection valve using a 20-µL PEEK injection loop. Chromatograms were recorded on a Star Chromatography Workstation (Varian, Palo Alto, CA). A SAMS/ CARS continuously regenerated membrane suppressor system from SeQuant AB (Umeå, Sweden) was used for the suppressed conductivity separations. All of the chromatographic evaluations were carried out at room temperature (21 ( 2 °C). RESULTS AND DISCUSSION Synthesis of the Zwitterionic Stationary Phases S300ECH-DMA-PS and S300-TC-DMA-PS. The synthesis of the

covalently bonded zwitterionic phase (S300-ECH-DMAES) which we recently described6 is based on activation of the hydroxyls of a HEMA sorbent with epichlorohydrin, followed by a one-step introduction of the zwitterionic functionality by reaction of the epoxide groups with dimethylaminoethanesulfonic acid inner salt. Because we found that the material had a net anion exchange capacity due to loss of sulfonic acid groups in the synthetic procedure, we wanted to evaluate a two-step procedure in which a tertiary amine was first introduced on the activated HEMA material, followed by quaternizing sulfoalkylation using an alkyl sultone. The synthesis route used for preparing the S300-ECHDMA-PS sorbent is schematically described in Scheme 1 and involved activation of the HEMA hydroxyls with epichlorohydrin, further reaction of the thus introduced epoxide groups with dimethylamine, forming a (2-hydroxy)propyl-linked dimethylamino group, which was finally sulfopropylated in a quaternizing reaction with 1,3-propane sultone. The intended functionalization was the (2-hydroxy-4-oxo-hexyl)-3-sulfopropyl-dimethyl-ammonio poly(methacrylate) inner salt. Nucleophiles react with 1,3-propane sultone in a ring-opening reaction, and the reagent has been used extensively as a quaternizing agent for tertiary amines in the synthesis of zwitterionic monomers and polymers.7-10 However, another use of 1,3-propane sultone is as a sulfopropylation agent for preparing strong cation exchangers, based on HEMA materials, through reaction with hydroxyl groups on the polymer surface.11 It can, therefore, be anticipated that some sulfopropyl functionality is introduced on the β-hydroxyl groups resulting from ring opening of the epoxide groups, leading to formation of cation exchange sites. We, therefore, ran a parallel synthesis route for the material S300-TC-DMA-PS, shown in Scheme 2, which differs from the S300-ECH-DMA-PS synthesis by activation of the Separon 300 HEMA particles with thionyl chloride instead of epichlorohydrin, thereby avoiding the formation of a β-hydroxyl group on (7) Hart, R.; Timmerman, D. J. Polym. Sci. 1958, 28, 638-640. (8) Galin, J. C.; Monroy Soto, V. M. Polymer 1983, 25, 121-128. (9) Davidson, N. S.; Jetters, L. J.; Funk, W. G.; Graessley, W. W.; Hadjichristidis, N. Macromolecules 1988, 21, 112-121. (10) Laschewsky, A.; Zerbe, I. Polymer 1991, 32, 2070-2080. (11) Kahovec, J.; Coupek, J. React. Polym. 1988, 8, 105-111.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

1995

Table 1. Elemental Analysis of Nitrogen and Sulfur for the Different Zwitterionic Stationary Phases

c

material

% NE.A.a

% SE.A.a

% Scalcb

% SE.A.:% Scalcc

S300-ECH-DMAESd S300-ECH-DMA-PS S300-MAA-SPE S300-MAA-SPP S300-TC-DMA-PS

1.06 ( 0.08 1.58 ( 0.02 1.57 ( 0.03 2.22 ( 0.13 2.52 ( 0.10

2.14 ( 0.14 3.68 ( 0.18 2.77 ( 0.10 1.99 ( 0.11 5.94 ( 0.59

2.43 3.63 3.60 2.54 5.76

0.83 1.01 0.77 0.78 1.03

a % N and % S determined by elemental analysis. b Stoichiometric % S calculated from N E.A. according to the molar weight ratio of 32.06:14.01. Amount of S incorporated in relation to the stoichiometric ratio. d Data from ref 6, included for reference.

Scheme 3. Synthesis Steps Used in the Preparation of the Zwitterionic Stationary Phase S300-MAA-SPE

reaction with DMA. The nitrogen and sulfur elemental analyses in Table 1 show that the N:S stoichiometries were close to unity for both the S300-ECH-DMA-PS and the S300-TC-DMA-PS materials. Consequently, it can be tentatively concluded that both of these materials have balanced charges within the experimental errors of the elemental analyses, which is a significant improvement over the previously described S300-ECH-DMAES material.6 These pathways via sulfopropylation of a tertiary amine, thus, appear to yield a better gross charge balance than the reaction of dimethylaminoethanesulfonic acid inner salt onto the activated HEMA sorbents in a quaternizing reaction; however, the ζ-potential measurements (see below) revealed that these two materials had quite different charge properties in different salts and pH. Synthesis of the Grafted Zwitterionic Stationary Phase S300-MAA-SPE. As an alternative to the conversion of surface hydroxyls into zwitterionic groups, we furthermore wanted to evaluate a procedure in which zwitterions are not formed in a reaction with the surface functional groups but, rather, are introduced in a grafting procedure using a zwitterionic monomer. This should conceptually provide an exact charge balance, because the monomer is relatively simple to purify and characterize. Apart from the conceived advantage of incorporating zwitterionic groups with exact stoichiometry, we were also interested in evaluating the potential for higher functional group loading and the effect of having the zwitterionic groups situated at regular intervals on flexible grafted chains instead of randomly distributed in the surface layer of the cross-linked sorbent. An intriguing property of grafted chains is their “tentacle” properties,12 which we are interested in evaluating for the separation of biological macromolecules.13,14 The schematic synthesis procedure for S300-MAA(12) Mu ¨ ller, W. J. Chromatogr. 1990, 510, 133-140. (13) Viklund, C.; Irgum, K. Macromolecules 2000, 33, 2539-2544. (14) Viklund, C.; Irgum, K. Anal. Chem. 2001, 073, 444-452.

1996

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 1. Structure of the metastable intermediate formed by hydrolysis and activation of HEMA with DMAP.

SPE is shown in Scheme 3. The first step is designed to introduce methacrylic groups onto the HEMA particles by reaction of methacrylic anhydride with the surface hydroxyls, and it was necessary to use a catalyst (DMAP) to enhance the reaction efficiency.15 The methacrylic groups thus introduced were subsequently intended as anchoring groups for the grafting of the zwitterionic monomer onto the HEMA particle surface. In the grafting procedure, the methacryloylated HEMA particles were first suspended and stirred in a mixture containing the zwitterionic monomer SPE and the photoinitiator BME in methanol. This procedure should cause swelling at least in the porous surface layer, whereby the monomer and initiator are brought into contact with the activated surface. Water was used as a dispersion solvent in this study. It was added to the mixture slowly and under stirring in order to prevent BME from precipitating in solution, because this reagent is insoluble in water. By this procedure, most of the BME should adsorb onto the particle surface rather than being precipitated, and we expected the polymerization to start mainly at the particle surface instead of in solution when the mixture was irradiated under the UV light. Our hypothesis was that this method should produce a zwitterionic stationary phase with close to perfect charge balance. However, the elemental analyses (Table 1) indicated that the grafting with SPE produced a sorbent with an unexpected excess of nitrogen-containing groups. Because elemental analysis of the (15) Scriven, E. F. V. Chem. Soc. Rev., 1978, 129-161.

Figure 2. Effect of eluent concentration on the retention time of inorganic anions and cations with perchloric acid at 1 mL/min as eluent on (a) S300, (b) S300-ECH-DMA-PS, (c) S300-MAA-SPE, and (d) S300-TC-DMA-PS.

SPE monomer showed the intended 1:1 ratio between N and S before grafting and the contents of S and N in the S300 base material were below the detection limit of the elemental analysis instrumentation, we ran a control experiment with SPP as graft monomer instead of SPE. This resulted in a material with the same apparent stoichiometry, which made us suspect a systematic error. We, therefore, subjected the methacrylic-anhydride-activated S300 to elemental analysis, and found that this intermediate material contained 0.77 ( 0.02% N (n ) 3), which can only originate from DMAP. As a hypernucleophilic acylation catalyst, DMAP is capable of hydrolyzing the HEMA ester bond, forming relatively stable cationic intermediates,15 as depicted in Figure 1. These intermediates will have anion exchange properties and are subject to nucleophilic substitution with hydroxide ion. Hydrolysis is, thus, expected to consume at least parts of these anion exchange sites as the materials are exposed to water. This complicates the evaluation, because carboxylic acid groups which may form this way have weak cation exchange properties. Because the concentration of residual groups in the surface layer is low and most of

the spectral features of these groups are similar to the base material, it is difficult to characterize the material accurately.6 Effect of Mobile Phase Concentration on the Separation with Different Zwitterionic Stationary Phases. Individual solutions containing NaCl, RbCl, Na2SO4, KBr, NaNO3, CaCl2, and Ca(NO3)2 were injected on S300, S300-ECH-DMA-PS, S300-TC-DMAPS, and S300-MAA-SPE columns using varying concentration of perchloric acid6,16 as eluents. The relationships between the retention times of each individual ion and the eluent concentrations are plotted in Figure 2. When evaluated with perchloric acid as eluent, the HEMA base material S300 showed a low retention for sodium (Figure 2a), which is likely due to residual carboxylic sites, as discussed below. There was also a slight latent retention for anions. As expected, this retention was most notable for the chaotropic ions, and it was only marginally affected by the eluent concentration. This indicates that the interaction was of hydrophobic rather than electrostatic nature. On the S300-ECH-DMA(16) Hu, W. Z.; Haddad, P. R.; Tanakar, K.; Hasebe, K. Analyst 2000, 125, 241244.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

1997

Figure 3. Breakthrough curves measured as electrolytic conductivity, resulting from the frontal chromatography experiment using step changes from pure water to 5 mM perchloric acid pumped at 1 mL/ min.

PS material, the retention times for both anions and cations decreased with increasing eluent concentration (see Figure 2b). This elution pattern is more characteristic of conventional ion exchange elution, and consequently, the S300-ECH-DMA-PS material did not behave as a zwitterionic separation material, despite the apparent balanced stoichiometry. However, the cations were not eluted with a “conventional” sensitivity to eluent strength, because plots of log(k′) vs log[HClO4] produced values of -0.61, -0.57, and -0.55 for Na+, K+, and Rb+, respectively (r2 g 0.99). The retention, therefore, appears to result from a mix of conventional cation and anion exchange, and zwitterionic interaction components. We interpret this to be caused by a charge patchiness due to spatially separated anion and cation exchange sites, rationalized by the ability of 1,3-propane sultone to sulfonate β-hydroxyls introduced by the epichlorohydrin. The S300-MAA-SPE (Figure 2c) displayed elution characteristics similar to the previously reported S300-ECH-DMAES material;6 that is, the retention of anions decreased, whereas that of cations increased as the perchloric acid concentration was increased. The column packed with the S300-TC-DMA-PS material (Figure 2d) also showed a retention pattern that was increasing with eluent concentration for cations and decreasing for anions, but these retention differences were less affected by ionic strength for both anions and cations than they were on the S300-MAA-SPE column. As we described in the above synthesis section, the elemental analyses in Table 1 indicated that the materials S300-TC-DMA-PS and S300ECH-DMA-PS had relatively good charge balances, whereas the material S300-MAA-SPE had an excess of nitrogen, which ought to provide for an excess of anion exchange sites. When the surface has an excess of anion or cation exchange sites, these ionic sites with local net charge will most probably overshadow the weak retention power of the chargeless zwitterionic groups and become the factor determining the overall retention. The similar retention pattern for S300-MAA-SPE and the previously described S300ECH-DMAES material corroborates our previous conclusion6 that the retention pattern for S300-ECH-DMAES was due to excess anion exchange groups. Breakthrough experiments with perchloric acid are shown in Figure 3. The S300-TC-DMA-PS and S300-ECH-DMA-PS materials 1998

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

had practically identical breakthrough capacity, indicating that these two materials have a similar ability to bind perchloric acid,6 despite the considerably higher degree of functionalization for the S300-TC-DMA-PS (Table 1). A possible cause can be a better opportunity for the polarizable perchlorate ion to interact with the quaternary ammonium group in the S300-ECH-DMA-PS, where the zwitterionic group is situated on a hydrophilic spacer. Improved sorption capacity in flexible layers is further supported by the grafted S300-MAA-SPE material having nearly three times as high a sorption capacity for perchloric acid as the surfacefunctionalized materials, despite the elemental analysis indicating a functionalization on par with the S300-ECH-DMA-PS. It should be pointed out that the degree of functionalization of all the materials presented in this work exceeds monolayer coverage. On the basis of the measured surface area of 27 m2/g for the S300 base material and assuming that the surface packing density of the zwitterionic groups would be in the same range as the maximum packing density of trimethylsilyl groups bonded onto the pore surface of porous silica (4.4 µmol/m2 for 250 Å silica),17 a surface monolayer coverage would amount to approximately 0.17% N. When comparing this value to the functional group loading in Table 1, it is apparent that the surface-modified materials have been functionalized beyond the surface. The grafted material S300-MAA-SPE also has a higher functionalization than can be accounted for by a surface reaction, indicating that a polymeric coating is actually attached to the surface. Simultaneous Separation of Anions and Cations. To compare the chromatographic properties of these three newly synthesized zwitterionic materials to the S300-ECH-DMAES material in the former paper, we carried out separations using individual salts and test mixtures that were identical to those that were used before. We also used 1 mM perchloric acid as eluent, except with the S300-TC-DMA-PS column, where this eluent caused the peaks of sodium or potassium to appear in the middle of the chromatograms, coeluting with the peaks of some anions. In Figure 4a, we can see how the cations have a longer retention on the S300-ECHDMA-PS, as compared to the S300-ECH-DMAES material that we presented in our previous paper,6 and Ca2+and Mg2+ were retained for >30 min. A possible mechanism could be the presence of isolated cation exchange sites produced through sulfopropylation of the β-hydroxy groups, as discussed above. Although this cannot be rationalized in view of the elemental analysis data in Table 1, the consistently negative ζ-potentials (see below) reveal that cation exchange properties are dominant in the S300-ECH-DMA-PS material. The high retention for cations also enabled separation of the monovalent ions Na+ and Rb+, which coeluted on the S300ECH-DMAES material. Figure 4b shows a simultaneous separation of anions and cations on the grafted S300-MAA-SPE material. The retention properties for the cations were markedly different from the S300-ECH-DMA-PS column, with the monovalent cations coeluted at a low retention and Ca2+ eluted in just about 2 min. The system peak appeared at a retention volume nearly three times that of the S300-ECH-DMA-PS column, corresponding well with the larger capacity of the S300-MAA-SPE for perchloric acid, as shown in Figure 3. We furthermore noted that I- could not be detected within 30 min on the S300-MAA-SPE columns. We (17) Poole C. F., Poole, S. K. Chromatography Today; Elsevier: Amsterdam, 1991; p 332.

Figure 4. Chromatogram from the separation of mixtures containing (a) 1 mM RbCl, NaNO3, and NaBr on column S300-ECH-DMA-PS using 1 mM perchloric acid as eluent at a flow rate of 1 mL/min; (b) 2 mM NaCl, 2 mM KBr, and 1 mM Ca(NO3)2 on column S300-MAA-SPE, using 1 mM perchloric acid as eluent; (c) 1 mM each Na2SO4, CaCl2, Ca(NO3)2, and 2 mM KBr on column S300-TC-DMA-PS, using 1.5 mM magnesium perchlorate as eluent at a flow rate of 1 mL/min. Direct conductivity detection was used in all of the chromatograms.

attribute this retention pattern to an excess of anion exchange sites, most likely introduced through the DMAP catalyst, as discussed above. This is supported by the sulfur content on this column being only 77% of the expected. The same tendency is apparent in the retention vs eluent strength curves in Figure 2, and plots of log k′ vs log[HClO4] yielded values for SO42-, Cl-, Br-, and NO3-, which were -1.74, -0.95, -0.89, and -0.85 (r2 g 0.995), respectively. Slopes this close to the charge of the ion being separated with a monovalent eluent are strong indicators of an anion exchange mechanism. Separation of a salt mixture containing 1 mM each of Na2SO4, CaCl2, Ca(NO3), and 2 mM KBr was attempted on column S300-TC-DMA-PS using perchloric acid as eluent; however, as mentioned above, a different eluent had to be used, and the choice fell on 1.5 mM magnesium perchlorate, which separated this salt mixture well. The resulting chromatogram is shown in Figure 4c. The retention for cations on the S300TC-DMA-PS material is obviously intermediate between the two other materials. Effect of Different Perchlorate Salts on the Separation with S300-TC-DMA-PS. As rationalized in our previous paper,6 we assume that the highly polarizable perchlorate ion is capable of strong interactions with the quaternary ammonium layer,18-20 a process that will tether the positive charges and induce a net negative charge on the material surface.19 Although the eluent cation has a substantially lower influence than the anion on the retention in zwitterionic chromatography,21 we still wanted to investigate the influence of cations on the retention pattern by using a perchlorate salt with cations of different charge, namely, potassium, magnesium, and cerium perchlorates, as the eluents. The resulting retentions for the ionic species and the system peak are shown in Figure 5. It is evident that both the system peak and the polarizable anions were more retained when the eluent cation was changed from K+ through Mg2+ to Ce3+. A reasonable (18) Brochsztain, S.; Berci, P.; Toscano, V. G.; Chaimovich, H.; Politi, M. J. J. Phys. Chem. 1990, 94, 6781-6785. (19) Di Profio, P.; Germani, R.; Savelli, G.; Cerichelli, G.; Chiarini, M.; Mancini, G.; Bunton, C. A.; Gillitt, N. D. Langmuir 1998, 14, 2662-2669. (20) Berlinova, I. V.; Dimitrov, I. V.; Kalinova, R. G.; Vladimirov N. G. Polymer 2000, 41, 831-837. (21) Patil, J. M.; Okada, T. Anal. Commun. 1999, 36, 9-11.

explanation is that the induced negative charge becomes compensated by the polyvalent eluent cations, which can bind more strongly to the sulfonic acid layer of the zwitterions moiety. Multiply charged eluent cations also shortened the cation retention, following the same argument. The ζ-potential measurements (see below) corroborate these findings. It should be noted that these results are in accordance with Hu and Haddad,4b who saw little effect on weakly retained anions and a minor change in the retention for iodide when chloride salts of differently charged cations were used as eluents on silicas dynamically modified with zwitterionic detergents. ζ-Potential Measurements. The ζ-potentials of the HEMA base material S300 and the materials functionalized with zwitterionic groups were measured at two different concentration levels, relevant for the elution conditions under which the chromatographic properties of the materials were evaluated. Four different acids, hydrochloric, sulfuric, nitric, and perchloric, were chosen for testing. As shown in our previous paper,6 perchloric acid has a uniquely high binding to the zwitterionic material presented in that paper, and a comparison of this acid to other strong acids whose anions have a lower retention on the current group of materials was, therefore, interesting. The sodium salts corresponding to these acids were included to evaluate the hypothesis that anions bind to the material as a function of their chaotropic properties, producing a net negative charge that controls the ion exchange properties of the zwitterion exchanger.6,22 Two additional sodium salts were included: NaOH to determine the surface charge at alkaline pH and NaF to determine if the surface charge densities seen with NaOH were due to the basicity or to the “hardness” of the hydroxide ion. Finally, to evaluate the hypothesis formulated above, that polyvalent cations bind to anionic groups rendering the surface more positive, we also included the potassium, magnesium, and cerium(III) perchlorates among the salts. The acids were prepared with the aim of maintaining the proton concentration, whereas the effect of the cations with perchlorate salts were performed with the concentration of the perchlorate ion kept constant. This way, we could determine how the anion affected the surface charge under a constant protonation (22) Iso, K.; Okada, T. Langmuir. 2000, 16, 9199-9204.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

1999

Figure 5. Effect of perchlorate salt on the retention time of inorganic anions (left) and cations (right) at 1 mL/min with columns packed with the zwitterionic stationary phases synthesized in this work. Legend: O, Na+; 0, system peak; ), Ca2+; (, SO42-; 2, Cl-; 9, Br-; 1, I-; b, NO3-; 3, SCN-.

pressure on the sulfonic acid group, while in the studies of the surface charge as a function of different cations, the concentration of the anions assumed to interact primarily with the quaternary ammonium group was maintained at a constant level. The results of the ζ-potential measurements are presented in Table 2. The surface charge of the unfunctionalized HEMA material S300 was close to zero in all of the acids that were tested, whereas a weak negative charge without significant preference for particular anions was evident when the ζ-potentials were measured in the presence of salts at low concentration. The higher of the salt concentrations tested appeared to shield the charges effectively, which indicates a low surface charge density. In sodium hydroxide, the base material acquired a negative potential that was higher at the lower of the two tested concentrations. Because the surface charge is very close to zero in the corresponding acids and considerably more negative in the sodium hydroxide solutions than in the neutral salts, we attribute this negative potential of S300 to residual carboxylic groups. The ζ-potentials for the grafted material S300-MAA-SPE were remarkably similar to those measured for the S300 base material, both in contact with salts with different anions and in the presence of acids and base. This indicates that the overall grafting procedure did not introduce sites capable of producing a net charge when the material is brought into contact with salts, acids, or bases. In other words, we conclude that the grafting did not result in dissociable groups other than those present on the material from the start. Neither did the surface potential change radically as the cation was changed from K+ through Mg2+ to Ce3+, indicating a low net surface charge; yet, this grafted material had a substantially higher tenacity for perchloric acid, as evident from Figure 3. This means that the preferential interaction with perchloric acid demonstrated in our previous paper6 also existed for S300-MAA-SPE, despite its surface potentials being more or less identical to the unfunctionalized S300 HEMA material. 2000

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

The first observation that can be made when comparing the surface charges in the presence of acids, is that they all resulted in ζ-potentials that were more positive and generally closer to zero, as compared to their corresponding salts. With HCl and H2SO4 at the high concentration level, the ζ-potentials of S300-MAA-SPE were not signifycantly different from that of the S300 starting material and were very close to zero net surface charge. After changing to perchloric acid, the surface charge of the S300 starting material remained at zero, whereas a small but significant negative charge was induced in the grafted material S300-MAA-SPE. The results are more difficult to interpret at the low acid concentration levels, but the general tendency is that potentials were slightly higher. The exception is the S300 base material, which instead showed a weak negative charge in the weak perchloric acid solution. Therefore, the grafted material appeared to contain some positively charged groups that could neutralize the induced negative charge. This would explain the preferential retention of anions, as compared to the S300-TC-DMA-PS, as seen in Figure 2c,d. The S300-TC-DMA-PS had several interesting surface charge properties. First of all, there was a large difference (>20 mV) between the ζ-potentials in sodium hydroxide and sodium fluoride, indicating the presence of functional groups that could either be deprotonated or dissociated in base. The ζ-potentials of this material also showed a marked sensitivity to the type of salt, with a ζ-potential difference between the chaotropic perchlorate ion and the cosmotropic fluoride ion of 20 mV at the lower concentration level and 14 mV at the higher of the tested concentrations. Trends such as this were not seen for the other materials, including the base material S300. The sensitivity to the type of anion also prevailed in the acids series, in particular at the lower concentration where there was a 27 mV difference in ζ-potential between the HCl and HClO4 solutions. A similar but weaker trend was seen at the higher concentration level. These observations,

6.7 mM -1.7 ( 0.2 -5.8 ( 0.4 -6.1 ( 0.6 -2.2 ( 0.3 20 mM 20 mM -10.0 ( 0.6 -5.4 ( 0.3 -34.3 ( 1.0 -13.3 ( 0.3 -51.6 ( 3.3 -35.3 ( 3.9 -11.5 ( 0.4 -5.1 ( 1.0 20 mM 10 mM 20 mM 20 mM S300 1.2 ( 0.6 0.9 ( 0.7 1.9 ( 1.6 -0.3 ( 0.5 S300-TC-DMA-PS -5.0 ( 0.2 -9.6 ( 0.5 -11.1 ( 0.8 -12.3 ( 0.4 S300-ECH-DMA-PS -5.3 ( 1.6 -11.7 ( 0.6 -15.2 ( 2.8 -25.0 ( 1.9 S300-MAA-SPE 1.7 ( 0.5 0.5 ( 0.3 -0.5 ( 0.4 -6.8 ( 0.6

20 mM 10 mM 20 mM 20 mM 20 mM 10 mM -2.3 ( 0.6 -2.5 ( 0.6 -3.3 ( 0.9 -4.8 ( 0.4 -5.7 ( 0.3 -3.7 ( 0.6 -15.3 ( 0.5 -13.0 ( 0.3 -21.5 ( 1.3 -27.3 ( 0.2 -30.9 ( 0.4 -10.9 ( 0.9 -32.0 ( 0.1 -29.8 ( 3.0 -34.2 ( 0.9 -35.2 ( 1.9 -35.5 ( 0.7 -18.9 ( 0.5 -4.4 ( 0.9 -3.8 ( 1.2 -4.7 ( 0.6 -7.8 ( 0.6 -9.9 ( 0.5 -6.8 ( 0.4

2 mM 1 mM 0.67 mM -11.3 ( 0.6 -7.6 ( 0.9 -3.9 ( 0.5 -34.0 ( 0.9 -19.5 ( 0.1 -11.7 ( 0.0 -41.9 ( 5.3 -21.6 ( 5.3 -9.8 ( 2.5 -11.0 ( 0.9 -7.2 ( 0.5 -1.0 ( 0.8 2 mM -13.2 ( 0.3 -34.3 ( 0.3 -45.5 ( 2.9 -13.4 ( 0.2 1 mM 2 mM -13.4 ( 1.6 -9.2 ( 1.0 -21.6 ( 1.2 -21.7 ( 2.8 -43.6 ( 6.3 -42.9 ( 2.5 -14.3 ( 2.2 -8.6 ( 1.6 2 mM -10.2 ( 1.3 -19.8 ( 1.5 -43.0 ( 5.9 -9.0 ( 1.3 2 mM -12.0 ( 0.3 -14.0 ( 0.8 -44.8 ( 4.4 -12.5 ( 0.2 2 mM -19.0 ( 2.2 -38.1 ( 0.3 -47.7 ( 0.9 -23.2 ( 6.8 2 mM 1 mM 2 mM 2 mM S300 2.1 ( 0.5 1.5 ( 0.3 -0.1 ( 0.2 -4.4 ( 0.6 S300-TC-DMA-PS 3.0 ( 0.4 -0.5 ( 0.7 -7.3 ( 1.0 -24.1 ( 0.8 S300-ECH-DMA-PS -5.7 ( 7.5 -16.5 ( 1.3 -14.3 ( 7.2 -27.7 ( 7.4 S300-MAA-SPE 4.8 ( 0.8 3.3 ( 0.0 3.7 ( 0.3 -2.2 ( 1.4

Ce(ClO4)3 Mg(ClO4)2 KClO4 NaClO4 NaNO3 Na2SO4 NaCl NaF NaOH HClO4 HNO3 H2SO4 HCl material

Table 2. ζ-Potentials of Unmodified Spheron 300 HEMA Sorbent and the Different Material Zwitterionic Stationary Phases Derived from This Material

along with the low effect of eluent concentration on retention (Figure 2d), make us conclude that the S300-TC-DMA-PS material has the most pronounced zwitterionic character of the materials synthesized in this work. This is supported by the large anioninduced negative surface potentials that were measured by Iso and Okada22 when zwitterionic micelles were challenged with various salts. The S300-ECH-DMA-PS was the most negatively charged of all of the materials in all of the test solutions. The difference in surface charge in the presence of base was small at low base concentration, but attained a higher value at the high concentration level. The ζ-potentials were, furthermore, strongly negative and essentially unaffected by the type of anion when present as sodium salts, whereas a clear trend was seen toward a more negative surface charge in acids with chaotropic anions. Among perchloric salts with cations of different valence, the ζ-potentials increased (got closer to zero) as the charge of the cation increased. This increase in ζ-potential verifies the hypothesis formulated above and is a result of association between the cations and the sulfonic acid part of the zwitterionic group. The effect was larger on the surface modified S300-TC-DMA-PS and S300-ECH-DMA-PS materials than on the grafted S300-MAA-SPE material, which in turn was almost comparable to the S300 starting material. To ascertain that the differences in ζ-potential were not due to variations in the degree of protonation and dissociation of the particle surfaces in the unbuffered salt solutions, measurements were made after treating the materials with acid or base, followed by thorough washing with water before the materials were suspended in the salt solutions for measurement. The nongrafted materials were selected for this test, because these materials showed the most significant potential differences in different salt solutions and also had different ζ-potential response in base. As seen in Table 3, the ζ-potentials were identical or close to identical for the materials that had been treated in acid and base. There was, consequently, no charge hysteresis, and because it appears to be possible to arrive at the same surface charge after treatment with strong alkali and acid simply by washing with water, the materials seem to have the ability of “self-elution” that is characteristic of zwitterion exchangers. Chromatographic Evaluation with Eluents Other Than Perchlorates. Alongside the development of novel synthesis routes to polymeric zwitterionic separation materials, we are also actively searching for new eluents for these and other zwitterionic separation materials.6 So far we have evaluated more than 40 different substances as eluents, including acids such as nitric, sulfuric, and oxalic acids, zwitterionic compounds such as the monomer SPE, glycine, lysine, 3-cyclohexylamino-1-propanesulfonic acid, 2-aminoethanesulfonic acid, and salts such as potassium phthalate, tetraethylammonium perchlorate, and sodium bicarbonate, to mention a few. We have still not found a group of salts that can accomplish the simultaneous separation of inorganic anions and cations better than perchloric acid or its salts. However, suppressible eluents23 are attractive, because they offer lower detection limits due to lower background and higher analyte signals. These eluents are characterized by being anions (23) Haddad, P. R.; Jackson, P. E. Ion Chromatography-Principles and Applications, Elsevier: Amsterdam, 1990; pp 106-131.

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2001

Table 3. Zeta Potentials of Nongrafted Zwitterionic Stationary Phases Following Treatment with Acid and Basea 2 mM NaCl

2 mM NaNO3

2 mM NaClO4

material

HCl

NaOH

HCl

NaOH

HCl

NaOH

S300-TC-DMA-PS S300-ECH-DMA-PS

-24.5 ( 2.5 -50.5 ( 3.0

-25.5 ( 3.3 -53.4 ( 2.0

-21.4 ( 1.8 -52.1 ( 1.3

-29.3 ( 2.0 -51.1 ( 4.8

-35.7 ( 0.5 -53.3 ( 5.1

-36.7 ( 2.0 -55.8 ( 4.2

20 mM NaCl

S300-TC-DMA-PS S300-ECH-DMA-PS

20 mM NaNO3

20 mM NaClO4

HCl

NaOH

HCl

NaOH

HCl

NaOH

-19.9 ( 0.9 -37.3 ( 2.8

-15.8 ( 0.9 -35.4 ( 3.3

-20.0 ( 0.9 -38.9 ( 2.3

-23.1 ( 0.3 -38.1 ( 0.7

n/a n/a

n/a n/a

a Materials were equilibriated with 0.1 M HCl or NaOH for 1 h or more, as indicated, and thereafter, thoroughly washed with large amounts of deionized water before being subjected to ζ-potential measurements, as described in the Experimental Section. Triplicate measurements were made for each datum point.

Figure 6. Chromatogram from the separation of a mixture of 0.05 mM Na2SO4, 0.1 mM NaCl, 0.2 mM NaNO2 and NaBr, and 0.4 mM NaNO3 and NaI on column S300-TC-DMA-PS using 10 mM sodium bicarbonate as eluent at a flow rate of 1 mL/min and suppressed conductivity detection.

of relatively weak acids that are weakly dissociated and, hence, of low conductivity after the eluent salt has been converted into the corresponding acid. Figure 6 shows the separation of a mixture of inorganic ions on the S300-TC-DMA-PS material using a sodium bicarbonate eluent with suppressed conductivity detection. It was observed that all of the anions tested, SO42-, Cl-, Br-, and NO3-, have longer retention times, as compared to the perchlorate eluent. This follows the normal elution rules for ion chromatography that the perchlorate ion has a higher eluting ability than the bicarbonate ion. We also found that the retention time for anions decreased with increasing sodium bicarbonate eluent concentration from 5 to 20 mM, but the sensitivity to the eluent concentration appears to be substantially less than in conventional ion exchange elution, as shown in Figure 7. This is different from the results obtained by others with hydrophobic silica dynamically modified with zwitterionic detergent4b in which the anion retention time has been shown to increase with bicarbonate eluent concentration in the interval 0-10 mM, with marginal changes in retention above that range. We cannot provide an explanation for 2002 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 7. Effect of eluent concentration on the retention time of inorganic anions and cations with sodium bicarbonate as eluent at 1 mL/min on column S300-TC-DMA-PS. Legend: (, SO42-; 2, Cl-; 4, NO2-; 9, Br-; b, NO3-; 1, I-.

this difference between the S300-TC-DMA-PS and the silica-based material. Although the elemental analyses indicate a unit ratio between the quaternary ammonium and sulfonic acid groups, it is not unlikely that this retention pattern can be ascribed to a deviation from exact charge balance in the polymeric materials, because determination of the exact charge stoichiometry in these materials appears to be a difficult task.6 CONCLUSION The three different synthesis routes to polymeric zwitterionic stationary phases described in this paper provided materials that were capable of simultaneously separating inorganic cations and anions using aqueous solutions of perchloric acid or perchlorate salts as mobile phases. The anion elution order was SO42- < Cl< Br- < NO3- for all three stationary phases, but the retention order for cations was Na+ < K+ < Rb+, with an unmeasurably high retention for Ca2+ on the S300-ECH-DMA-PS and S300-TCDMA-PS, unless magnesium perchlorate was used as eluent. The monovalent cation retention order was the same on the S300-MAASPE, with the difference that Ca2+ could be eluted with low

retention, using a perchloric acid eluent. The general tendency with perchloric acid eluents of increasing concentration was that the retention time decreased for anions and increased for cations for the S300-TC-DMA-PS and the S300-MAA-SPE. However, for the S300-ECH-DMA-PS, the retention time for cations decreased with increasing concentration of perchloric acid. Considering the difficulties of characterizing this kind of materials,6 our best explanation of these differences in retention pattern is that there are difficulties establishing an exact charge balance ratio between quaternary ammonium and sulfonic acid groups. It is clear, however, that by varying and controlling the synthesis procedures, it will be possible to produce materials with vastly different retention patterns, providing a way to model selectivities to an

extent far greater than is possible for conventional ion exchange materials. ACKNOWLEDGMENT This work was supported by The Swedish Natural Science Research Council, The Magnus Bergwall Foundation, NUTEK, and Teknikbrostiftelserna. Thanks are also due to Peter Ko¨berle at Raschig Chemie for providing the SPE monomer used in this work. Received for review August 7, 2000. Accepted January 23, 2001. AC000933D

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2003