Anal. Chem. 2000, 72, 4413-4419
Quaternized Trimethylaminated Polystyrene-Coated Zirconia as a Strong Anion Exchange Material for HPLC Jianhong Zhao† and Peter. W. Carr*
Department of Chemistry, University of Minnesota, 207 Pleasant St. SE., Minneapolis, Minneapolis 55414
The synthesis and characterization of a new, base-stable, strong anion exchange phase by amination of polystyrenecoated zirconia (PS-ZrO2) are described. Even though the ion exchange capacity of the quaternized trimethylaminated PSZrO2 (QTMA-PS-ZrO2) is only 0.07 mequiv/g, it is able to separate various inorganic anions, benzoic acid derivatives, and nucleotides in their deprotonated states. The effects of ionic strength, eluent pH, and counterion type are discussed. In the presence of both phosphate and fluoride ions in the eluent, band broadening caused by Lewis acid/base interactions between zirconia and analytes is greatly suppressed. The mixed retention modes (ion exchange, hydrophobic interaction, and Lewis acid/base interactions) on QTMA-PS-ZrO2 offer a different selectivity toward various anionic analytes than do other zirconia- and nonzirconia-based ion exchangers. Ion exchange chromatography (IEC) is an important chromatographic mode in the analysis of biological samples, pharmaceuticals, inorganic salts, and organometallic ions.1-10 Organic polymers, such as polystyrene-divinylbenzene (PS-DVB), are still the predominant support materials due to their excellent stability toward extreme pH conditions;11,12 however, silica-based ion exchangers generally exhibit better chromatographic efficiency13-16 * To whom correspondence should be addressed (
[email protected]). † Current address: Analytical Research & Development, Pharmaceutical Science, Pfizer Global Research & Development, Pfizer Inc., Groton, CT 06340. (1) Clausen, A. M.; Subramanian, A.; Carr, P. W. J. Chromatogr., A 1999, 831, 63. (2) Hanai, T.; Miyazaki, R.; Kinoshita, T. Anal. Chim. Acta 1999, 378, 77. (3) Henderson, I. K.; Saari-Nordhaus, R. J. Chromatogr. 1992, 602, 149. (4) Paster, A.; Alcacer, E.; Forcada, C.; Garcera, M. D.; Martinex, R. J. Chromatogr., A 1997, 789, 379. (5) Perrett, D.; Bhusate, L.; Patel, J.; Herbert, K. Biomed. Chromatogr. 1991, 5, 207. (6) Slingsby, R. W.; Rey, M. J. Liq. Chromatogr. 1990, 13, 107. (7) Tsai, E. W.; Chamberlin, S. D.; Forsyth, R. J.; Bell, C.; Ip, D. P.; Brooks, M. A. J. Pharm. Biomed. Anal. 1994, 12, 983. (8) Regnier, F. E. Methods Enzymol. 1984, 104, 170. (9) Small, H. J. Chromatogr. 1991, 546, 3. (10) Snyder, L. R.; Glajch, J. L.; Kirkland, J. J. Practical HPLC Method Development, 2nd ed.; Wiley-Interscience: New York, 1996. (11) Hellferich, F. Ion Exchange; McGraw-Hill: New York, 1962. (12) Weiss, J. Ion Chromatography, Second ed.; VCH: Weinheim, 1995. (13) Chicz, R. M.; Shi, Z.; Regnier, F. E. J. Chromatogr. 1986, 339, 121. (14) Haddad, P. R.; Foley, R. C. J. Chromatogr. 1990, 500, 301. (15) Kennedy, L. A.; Kopaciewicz, W.; Regnier, F. E. J. Chromatogr. 1986, 359, 73. (16) Matsushita, S.; Tada, Y.; Baba, N.; Hosako, K. J. Chromatogr. 1983, 259, 459. 10.1021/ac000398h CCC: $19.00 Published on Web 08/18/2000
© 2000 American Chemical Society
compared with that shown by organic polymers. Nonetheless, a new family of zirconia-based ion exchange supports17-23 has great potential for replacing polymer and silica based ion exchangers because they overcome some of the weaknesses of both polymer- and silica-based ion exchangers. In addition to the high column efficiency afforded by silica-based exchangers, zirconia-based materials can be stable over the entire pH range from 1 to 1424-26 and at high temperatures up to 200 °C.26-28 For example, anion exchangers made by coating zirconia with polyethyleneimine20-22 show excellent alkaline (pH ) 13), acidic (pH ) 1), and thermal stability (90 °C); have reasonable column efficiency (60 000 plates/m); and are suitable for bioseparations. The work presented here shows how a polystyrene-coated zirconia can be converted to a strong anion exchanger by introducing quaternized trimethylamine groups into the polymer coating. Ion exchange characteristics, such as the ionic strength and pH effects on the retention of ionic compounds, were characterized. Additionally, we studied the effects of adding phosphate and fluoride to the eluent on the efficiency and peak symmetry of acidic solutes. Furthermore, the base stability was evaluated to ensure that the new phase maintained the excellent pH stability of bare and polystyrene-coated zirconia. Finally, we compared the selectivity of various anionic compounds on the new anion exchanger and quaternized polyethyleneimine-coated zirconia (QPEI-ZrO2). This new ion exchange phase will provide the pharmaceutical and biotechnological industries with an additional separation strategy. EXPERIMENTAL SECTION Reagents. All reagents were obtained from commercial sources and were reagent grade or better. Trimethylamine, p-chlorobenzoic acid, p-hydroxybenzoic acid, p-propoxybenzoic (17) Blackwell, J. A.; Carr, P. W. J. Chromatogr. 1991, 549, 59. (18) Clausen, A. M.; Carr, P. W. Anal. Chem. 1998, 70, 378. (19) Hu, Y.; Carr, P. W. Anal. Chem. 1998, 70, 1934. (20) McNeff, C.; Zhao, Q. H.; Carr, P. W. J. Chromatogr., A 1994, 684, 201. (21) McNeff, C.; Carr, P. W. Anal. Chem. 1995, 67, 3886. (22) McNeff, C.; Carr, P. W. Anal. Chem. 1995, 67, 2350. (23) Schafer, W. A.; Carr, P. W. J. Chromatogr. 1991, 587, 149. (24) McNeff, C. Ph. D. Thesis, University of Minnesota, Minneapolis, 1996. (25) Rigney, M. P.; Weber, T. P.; Carr, P. W. J. Chromatogr. 1989, 484, 273. (26) Zhao, J.; Carr, P. W. Anal. Chem. 1999, 72, 5217. (27) Li, J. W.; Hu, Y.; Carr, P. W. Anal. Chem. 1997, 69, 3884. (28) Yan, B.; Zhao, J.; Brown, J. S.; Blackwell, J.; Carr, P. W. Anal. Chem. 2000, 72, 1253.
Analytical Chemistry, Vol. 72, No. 18, September 15, 2000 4413
Table 1. Elemental Analysis of PS-ZrO2 and QTMA-PS-ZrO2 phase PS-ZrO2 QTMA-PS-ZrO2
percent weight %C %N %Cl 1.49 1.49
d 0.090
0.30 0.090
surface coverage (µmol/m2)a C N Cl 41.4 41.4
d 2.14
2.82 0.85
ratio of elementsb C/N C/Cl d 19.3
14.7 48.7
ion exchange capacity (µeq/g)c d 70
a Calculated on the basis of the number of micromoles of the element per square meter on zirconia and a surface area of 30 m2/g of porous zirconia. b Calculated by mole ratio of a pair of elements. c Based on the amount of nitrogen. d Not measured.
acid, p-butoxybenzoic acid, and p-toluenesulfonic acid monohydrate were purchased from Aldrich (Milwaukee, WI). Sodium hydroxide solution (50% (w/w)), sodium fluoride, potassium iodide and potassium bromide were obtained from Fisher Scientific (Fair Lawn, NJ). Sodium chloride, potassium phosphate dibasic, sodium phosphate monobasic, sodium nitrite, sodium nitrate, benzoic acid, and solvents (dichloromethane, acetonitrile, tetrahydrofuran, and 2-propanol) were obtained from Mallinckrodt (Paris, KY). pEthoxybenzoic acid, p-nitrobenzoic acid, and p-cyanobenzoic acid were from Eastman (Rochester, NY). All 5′-nucleotides were purchased from Sigma (St. Louis, MO). Deuterium oxide was from Cambridge Isotope Laboratory (Moburn, MA). HPLC water was produced by treatment of house deionized water with a Barnsted Nanopure cartridge and a 0.2-µm filter and was further boiled to remove carbon dioxide before use. Microparticulate zirconia particles (batch PICA-7) were produced by the polymerization-induced colloid aggregation method (PICA) in this laboratory and were used throughout the study. The particle size is 2.5 µm (by SEM). Preparation of Quaternized Trimethylaminated PS-ZrO2 (QTMA-PS-ZrO2). The synthetic procedure for making polystyrene-coated zirconia (PS-ZrO2) was described in detail elsewhere.29 The subsequent amination of PS-ZrO2 was performed using a method similar to that for a polystyrene-divinylbenzene resin,12 and the detailed procedure is described below. The actual polymer coated on the zirconia is a copolymer of chloromethylstyrene and vinylmethylsiloxane. Even after the cross-linking of the copolymer by self-condensation of chloromethyl and phenyl groups at high temperature, at least half of the chloromethyl groups remain in the copolymer network. Thus, these chloromethyl groups were reacted with an alcoholic solution of trimethylamine directly at room temperature.30 A 50-mL aliquot of 2-propanol was slowly sparged with gaseous trimethylamine for 15 min to give a nearly saturated solution. The amine solution was then added to a suspension of PS-ZrO2 in 50 mL of 2-propanol, which was previously sonicated under vacuum. The reaction mixture was capped with a rubber septum and then stirred at low speed at room temperature for 24 h. During this period, trimethylamine was bubbled into the reaction slurry five times, and bubbling was conducted for 10 min each time. The reaction slurry was filtered, washed three times with 2-propanol (3 × 100 mL), and dried under vacuum at room temperature for 8 h to remove any excess trimethylamine. The elemental analyses of carbon, nitrogen, and chlorine for PS-ZrO2 and QTMA-PS-ZrO2 were performed by Micro-Analysis, Inc. (Wilmington, DE). Chromatographic Conditions. Chromatographic measurements were performed on a Hewlett-Packard 1090 liquid chro(29) Zhao, J.; Carr, P. W. Anal. Chem. 1998, 70, 3619. (30) Collie; Schryver J. Chem. Soc. 1890, 57, 778.
4414 Analytical Chemistry, Vol. 72, No. 18, September 15, 2000
matograph with an autosampler, a temperature controller, a photodiode-array UV detector, and a computer-based HP ChemStation (Hewlett-Packard, Wilmington, DE). The QTMA-PS-ZrO2 particles were packed into a 50 × 4.6 mm (i.d.) stainless steel column (Alltech Associates Inc., Deerfield, IL) using the upward stirred slurry method at 5000 psi. 2-Propanol was used as the pushing and slurry solvent. Retention data were collected at a flow rate of 0.8 mL/min and a column temperature of 30 °C with UV detection at a wavelength of 240 nm, unless otherwise noted. The system dead volume used in the calculation of retention factors was determined by injection of deuterium oxide (D2O). The analytes were dissolved in water at a concentration of 1-2 mg/mL. The alkaline stability of QTMA-PS-ZrO2 was tested in an eluent of 0.1 M sodium hydroxide solution (pH ) 12.9) using an Altex pump (model 110A, Altex Scientific, Inc.). A precolumn packed with similar particles was used to avoid blocking of frit by particles degraded from the pump seal at this high pH. The retention data for selected solutes were collected on the HP 1090 system, periodically, using an eluent containing 50 mM phosphate buffer (pH ) 7.0), 100 mM sodium fluoride, and 100 mM sodium chloride. RESULTS AND DISCUSSIONS Ion Exchange Capacity. The elemental analyses of carbon, nitrogen, and chlorine for PS-ZrO2 and QTMA-PS-ZrO2 are given in Table 1. On the basis of the nitrogen content, the ion exchange capacity of the new anion exchanger is estimated to be 70 µequiv/ g. This number is relatively low compared with those of other commercially available high-performance ion exchangers. For example, a Hamilton PRP-X100 (trimethylammonium polystyrenedivinylbenzene) has an exchange capacity of 190 µequiv/g, and a Alltech anion HC column has an exchange capacity of 300 µequiv/ g. However, due to the low polymer loading of PS-ZrO2 (C% ) 1.5%), the ion exchange capacity of QTMA-PS-ZrO2 is necessarily greatly limited. On the basis of the chlorine content of PS-ZrO2 in Table 1, the theoretical maximal exchange capacity would be 90 µequiv/g, assuming that all the chloride groups are substituted with the trimethylamine groups. However, one must also consider the fact that zirconia is much denser than silica and organic polymers. Thus, the ion exchange capacity per unit bed volume is not so disparate. Ion Exchange Behavior. The ion exchange behavior of QTMA-PS-ZrO2 was tested with respect to ionic strength, pH, and counterion effects. To confirm that the separation ability of QTMAPS-ZrO2 toward ionic compounds is due to the quaternary amine groups, a PS-ZrO2 column was also used as a control to separate nitrite and nitrate ions. In the same eluent, nitrite and nitrate coeluted at the dead time of the PS-ZrO2 column while they were retained and separated on QTMA-PS-ZrO2.
Figure 1. Plots of log k′ as a function of log [Cl-] for different anions on QTMA-PS-ZrO2. Conditions: see Table 2. The open symbols are for inorganic anions (O, nitrite; 0, nitrate) and the filled ones are for organic anions (b, o-phthalic acid; 9, p-hydroxybenzoic acid; (, benzoic acid).
Ionic Strength Effects. First, we investigated the effect of chloride concentration on the retention of five ionic compounds, nitrite, nitrate, benzoic acid, o-phthalic acid, and p-hydroxybenzoic acid. Note that the concentration of phosphate buffer was held constant as the chloride concentration was varied. As shown in Figure 1, the logarithmic retention factor (log k′) is linearly correlated to the logarithmic concentration of chloride. Theoretically, the slope should be equal to 1 for monovalent analytes and 2 for the divalent analytes when a monovalent eluent ion is used.11 As shown in Table 2, we observe that the slope for o-phthalic acid, which has two negative charges in this eluent (pH ) 7), is about twice as large as those for p-hydroxybenzoic acid and benzoic acid, which have only one negative charge. Furthermore, for monovalent analytes, nitrate and nitrite have slightly larger slopes than do p-hydroxybenzoic acid and benzoic acid. However, the experimental slopes are all much less than the theoretical slopes. It is likely that phosphate ions (H2PO4-, HPO42-, and PO43-) also act as displacing agents and cause the dependence of log k′ on the concentration of chloride to be less than the theoretical value. We then performed an additional experiment excluding phosphate ions from the eluent. As shown in Table 2, in the absence of phosphate, the dependencies of log k′ of nitrite and nitrate ions on the concentration of chloride are still less than 1. Furthermore, the dependencies of log k′ of these two monovalent ions on the concentration of a divalent displacing ion (sulfate) are less than the theoretical value of 0.5. Since QTMA-PS-ZrO2 is made from a relatively hydrophobic reversed-phase material, PS-ZrO2, we suspect that the retention mechanism on QTMA-PS-ZrO2 is not purely ion exchange but is due to a mixed-mode, that is both ion exchange and hydrophobic interactions. Additional evidence for hydrophobic interaction is that divalent o-phthalic acid is less retained than monovalent p-hydroxybenzoic acid or benzoic acid in the eluents we studied (see Figure 1). The retention order of these three acids follows the order of their polarity: o-phthalic acid > p-hydroxybenzoic acid > benzoic acid. The hydrophobic interaction was clearly observed when a homologous series of p-alkoxybenzoic acids was studied (see Figure 2). Table 3 compares the free energy of transfer per methylene group from the mobile phase to the stationary phase o o (∆GCH ) for different materials.21,39 We observed that ∆GCH for 2 2 QTMA-PS-ZrO2 (-712 cal/mol) is much more negative than that
of QPEI-ZrO2 (-479 cal/mol) due to the fact that the phenyl groups in QTMA-PS-ZrO2 are more hydrophobic than the ethyleneimine groups in QPEI-ZrO2. pH Effect. The influence of pH on the retention of inorganic anions and organic anions is shown in Figure 3. As expected, the retention of inorganic anions, nitrate, and nitrite is not affected by variations in pH. On the other hand, the retention of organic anions is greatly affected by pH. For example, the pKas of o-phthalic acid are 2.95 and 5.41; and as a result, there is a drastic change in the retention of the analyte as pH is changed from 5 to 7. For p-hydroxybenzoic acid, we see two steep changes as pH is changed from 5 to 10; one change at low pH corresponds to the deprotonation of the carboxylic group and the other at high pH corresponds to the deprotonation of the hydroxyl group. For benzoic acid, only one steep change is observed around its pKa (4.20). Consequently, we observe a change in the elution order of three acidic solutes as pH is changed from 4 to 7. It is surprising to note that retention of the ionizable analytes studied here decreases as pH increases. This is opposite to what is normally observed for an anion exchange phase, where the retention process is an ideal ion exchange. However, this phenomenon has been reported on a strong anion exchanger based on polystyrene-divinylbenzene.31 Previously we showed that QTMA-PS-ZrO2 is rather hydrophobic. Both electrostatic interactions and hydrophobic interactions are very likely to be responsible for retention of organic anions on QTMA-PS-ZrO2. Counterion Effects. As is well-known in ion chromatography, a divalent ion (SO42-) is a stronger displacing agent than a monovalent (Cl-). However, different counterions with the same charge also have different displacing strengths. Here we studied the effect of counterion on the retention of nitrite and nitrate. As shown in Figure 4, the anions are most retained in the fluoride eluent and are least retained in the bromide eluent, indicating that the displacing strength of these counterions increases in the order of F- < Cl- < Br-. This is due to fact that large ions are more attracted to the stationary phase.12 Effect of Lewis Base Additives on Retention and Efficiency of Acidic Solutes. Hard Lewis acid sites on the zirconia surface are known to be rather problematic in the separation of analytes containing hard Lewis base moieties, such as, benzoic acid, nucleotides, peptides, and proteins. Polymers coated on zirconia’s surface (PS-ZrO2, PBD-ZrO2, and PEI-ZrO2) do not completely block these Lewis acid sites.20,21,26,32,33 However, using hard Lewis bases, such as phosphate or fluoride, as eluent additives improves the resolution and peak symmetry of troublesome solutes20,34-37 on polymer-coated zirconia. Here we look into the effect of phosphate and fluoride on the separation of some acidic solutes on QTMA-PS-ZrO2. Phosphate as a Lewis Base Additive. As shown in Figure 5A, it is very puzzling that the concentration of phosphate has only a slight effect on the retention of the solutes tested. On the other hand, it has a substantial influence on the peak width of acidic solutes (see Figure 5B); the plate counts for o-phthalic acid, (31) Caude, M.; Rosset, R. J. Chromatogr. Sci. 1977, 15, 405. (32) Li, J. W.; Carr, P. W. Anal. Chem. 1996, 68, 2857. (33) Sun, L.; McCormick, A. V.; Carr, P. W. J. Chromatogr., A 1994, 658, 465. (34) Hu, Y. Ph. D Thesis, University of Minnesota, Minneapolis, 1998. (35) Sun, L. Ph. D. Thesis, University of Minnesota, Minneapolis, 1994. (36) Sun, L.; Carr, P. W. Anal. Chem. 1995, 67, 2517. (37) Sun, L.; Carr, P. W. Anal. Chem. 1995, 67, 3717.
Analytical Chemistry, Vol. 72, No. 18, September 15, 2000
4415
Table 2. Linear Regression Results of Logarithmic Retention Factor versus Logarithmic Concentration of Counterion for Ionic Analytes on QTMA-PS-ZrO2 counterion chloride (with phosphate buffer)c
chloride (without phosphate buffer)d sulfate (without phosphate buffer)e
analyte
slope
intercept
R2a
o-phthalic acid p-hydroxybenzoic acid benzoic acid nitrite nitrate nitrite nitrate
-1.01 ( 0.03b -0.48 ( 0.02
-1.07 ( 0.03 -0.30 ( 0.02
0.9976 0.9963
-0.46 ( 0.02 -0.63 ( 0.02 -0.65 ( 0.01 -0.87 ( 0.03 -0.78 ( 0.02
-0.05 ( 0.02 -1.24 ( 0.02 -0.83 ( 0.01 -1.45 ( 0.03 -0.89 ( 0.02
0.9957 0.9984 0.9999 0.9943 0.9972
-0.274 ( 0.003 -0.286 ( 0.005
-0.579 ( 0.005 -0.173 ( 0.008
0.9995 0.9990
nitrite nitrate
a Correlation coefficient of the linear regression. b Standard error. c Chromatographic conditions: temperature, 30 °C; UV detection, 240 nm; mobile phase, 50 mM phosphate buffer at pH ) 7.0, and various concentrations of sodium chloride (50, 100, 200, and 400 mM). d Chromatographic conditions: temperature, 30 °C; UV detection, 240 nm; mobile phase, various concentrations of sodium chloride (25, 50, 100, 200, 300, and 400 mM). e Chromatographic conditions: temperature, 30 °C; UV detection, 240 nm; mobile phase, various concentrations of sodium sulfate (10, 20, 50, 100, and 200 mM).
Table 3. Comparison of Hydrophobic Interactiona on QTMA-PS-ZrO2 and QPEI-ZrO2 phase QTMA-PS-ZrO2b QPEI-ZrO2c C8-SiO2d
o ∆GCH (cal/mol) 2
(R2)e
-712 ( 21f -479 ( 18 -827
0.999 0.998 g
a Described by the free energy of transfer per methylene group (∆ o GCH ) which was obtained by -RTln(k′n+1/k′n) where R is the gas 2 constant, T is the temperature in K, n is the number of methylene groups. b Chromatographic conditions: temperature, 40 °C; UV detection, 240 nm; mobile phase, 100 Mm K2HPO4 and 400 mM NaCl), pH ) 7.3 adjusted by adding HCl. Three para-substituted alkoxybenzoic acids (p-ethoxybenzoic acid, p-propoxybenzoic acid, and p-butoxybenzoic acids) were used. c Obtained from ref 21, conditions are the same as the above except that the temperature was 35 °C and four parasubstituted alkoxybenzoic acids (p-ethoxybenzoic acid, p-propoxybenzoic acid, p-butoxybenzoic acid, and p-hexoxybenzoic acid) were used. d Extrapolated from the ∆Go CH2 values in 20 and 30% (v/v) acetonitrile in water using data from ref 39. e Correlation coefficient of the linear regression of ln k′ vs the number of methylene groups. f Standard error. g Not applicable.
Figure 2. A plot of ln k′ versus number of methylene groups for p-alkoxybenzoic acids on QTMA-PS-ZrO2. Conditions: temperature, 40 °C; UV detection, 240 nm; mobile phase, 100 mM K2HPO4 and 400 mM NaCl, pH ) 7.3 adjusted by adding HCl. The line denotes the least-squares line. The linear regression results: slope , 1.16 ( 0.03; intercept, -0.79 ( 0.10; R2, 0.999.
benzoic acid, and p-hydroxybenzoic acid all increase at least 30% as the concentration of phosphate was increased from 20 to 200 mM. As expected, the plate counts observed for nitrite and nitrate do not improve much, but they do change. Fluoride as Lewis Base Additive. Hu34 found that using fluoride and phosphate together as additives dramatically improved the peak shapes and reduced the band broadening of acidic solutes on PBD-ZrO2. We further confirmed this finding when we added fluoride to an eluent containing phosphate. Table 4 compares the plates and peak symmetry of acidic solutes with and without fluoride as an additive. With the fluoride in the eluent, the plate count for p-hydroxybenzoic acid was more than tripled and the plate count for benzoic acid was more than doubled; the plate count for o-phthalic acid was also improved significantly. Furthermore, the peaks became much more symmetric in eluents with fluoride than in those without fluoride. 4416 Analytical Chemistry, Vol. 72, No. 18, September 15, 2000
Base Stability of QTMA-PS-ZrO2. One of the major advantages of zirconia-based stationary phases is their excellent alkaline stability. This is sometimes very crucial for the separation of biomolecules where a high pH is required to optimize the separation. The alkaline stability of the QTMA-PS-ZrO2 phase was tested in a 0.1 M sodium hydroxide solution over more than 2700 column volumes. Some 5′-nucleotides and para-substituted benzoic acids were used as probes to measure their retention times in a separate buffered eluent. We observed a decrease of retention factors during the first 500 column volumes of base treatment, and then the retention remained essentially constant throughout the rest of the measurements (see Figure 6). We calculated the percent change in retention factor per column volume for each tested analyte. Note that, in these calculations, we discarded the initial retention data measured before the base treatment. We observed that, for the nucleotides, the deviations are random and very small (
