Anal. Chem. 1999, 71, 1820-1825
Behavior of Cation-Exchange Materials in Capillary Electrochromatography Maria G. Cikalo, Keith D. Bartle,* and Peter Myers
School of Chemistry, University of Leeds, Leeds, LS2 9JT, UK
The behavior of a strong, cation-exchange material (propanesulfonic acid, SCX) has been studied in capillary electrophoresis (CE) and capillary electrochromatography (CEC) by the use of coated and packed capillaries. In aqueous electrolytes, the SCX-coated capillary showed a far more consistent electroosmotic flow over the pH range 3.6-10.5, compared to untreated fused silica. However, in similar electrolytes containing 80% (v/v) acetonitrile, both coated and untreated capillaries performed similarly, casting doubts upon the stability of the SCX coating. The effect of voltage and mobile-phase parameters such as pH, ionic strength, and organic content was studied in CEC for both 3-µm SCX and C18 packing materials, and the results were compared in terms of linear velocities, currents, and conductivities. Only at pH 5 and below was a higher EOF velocity than expected observed for the SCX column. In accordance with theory, the EOF was seen to increase with decreasing ionic strength for the C18 column. However, for the SCX column, this was not the case: the EOF showed a general reduction as the ionic strength was decreased. The greatest anomaly was observed on changing the acetonitrile composition: the EOF showed a consistent decline with increasing organic, whereas the EOF in both the open capillary and C18 column decreased and then started to rise with acetonitrile contents above 70% (v/v).
Schomburg et al.,3 Wehr,4 and Pesek and Matyska5 have reviewed various capillary modifications. Capillaries that have permanent charged-surface coatings have typically been modified by procedures involving the use of polymers such as poly(2-aminoethylmethacrylate hydrochloride),6 polyamide resins,7 and strong ion exchangers.8,9 These are of particular interest since they are potentially capable of providing a stable EOF over a wide pH range; the dissociation of the charged groups is largely pH independent, and they should reach ionization equilibrium with the buffer system at a faster rate, thus improving migration time reproducibility. With the surge of interest in capillary electrochromatography (CEC), the use of such coatings to gain more control of the EOF and/or provide an alternative separation mechanism should offer a distinct advantage in both packed-column and open-tubular work. Publications in the literature have typically focused on those employing charged particles.10-15 However, the use of permanently modified capillaries,14 monoliths,16,17 and macroporous polyacrylamide matrixes18 has also been reported, and only recently a sulfonic acid-coated capillary has become commercially available.19 The use of strong cation-exchange material for the separation of basic compounds by CEC has become a point of interest on account of the remarkably high peak efficiencies that have been demonstrated on it.10,11 These have ranged from 8 to 40 million plates per meter, but suffer from poor reproducibility and, as yet,
In capillary electrophoresis (CE), the magnitude of the electroosmotic flow (EOF) generated is proportional to the zeta potential developed by the double layer of counterions close to the capillary wall. Since this potential is governed by the surface charge of the capillary and the double-layer thickness, factors influencing these can be used to manipulate the EOF. Practically, this is achieved by adjusting the applied field strength, pH, ionic strength, and viscosity of the running electrolyte. However, these may have adverse effects upon analyte mobility and solubility, while analyte-wall interactions often impair separation efficiency. Dynamic coatings, i.e., buffer additives such as tetradecyltrimethylammonium bromide,1 or permanent coatings2 have often been used to modify the capillary wall; these alter the magnitude and direction of the EOF and may minimize sample adsorption.
(3) Schomburg, G.; Belder, D.; Gilges, M.; Motsch, S. J. Capillary Electrophor. 1994, 1, 219-230. (4) Wehr, T. LC-GC 1993, 11, 14-20. (5) Pesek, J. J.; Matyska, M. T. Electrophoresis 1997, 18, 2228-2238. (6) Lui, Q. C.; Lin, F. M.; Hartwick, R. A. J. Liq. Chromatogr. 1997, 20, 707718. (7) Burt, H.; Lewis, D. M.; Tapley, K. N. J. Chromatogr., A 1996, 739, 367371. (8) Huang, M.; Yi, G.; Bradshaw, J. S.; Lee, M. L. J. Microcolumn Sep. 1993, 5, 199-205. (9) Kohr, J.; Engelhardt, H. J. Microcolumn Sep. 1991, 3, 491-495. (10) Smith, N. W.; Evans, M. B.Chromatographia 1995, 41, 197-203. (11) Euerby, M. R.; Gilligan, D.; Johnson, C. M.; Roulin, S. C. P.; Myers, P.; Bartle, K. D. J. Microcolumn Sep. 1997, 9, 373-387. (12) Kitagawa, S.; Tsuji, A.; Watanabe, H.; Nakashima, M.; Tsuda, T. J. Microcolumn Sep. 1997, 9, 347-356. (13) Li, D.; Knobel, H. H.; Remcho, V. T. J. Chromatogr., B 1997, 695, 169174. (14) Choudhary, G.; Horva´th, C. J. Chromatogr., A 1997, 781, 161-183. (15) Wei, W.; Luo, G.; Yan, C. Am. Lab. 1998, 30, C20-E20. (16) Liao, J.-L.; Chen, N.; Ericson, C.; Hjerte´n, S. Anal. Chem. 1996, 68, 34683472. (17) Peters, E. C.; Petro, M.; Svec, F.; Fre´chet, J. M. J. Anal. Chem. 1997, 69, 3646-3649. (18) Palm, A.; Novotny, M. V. Anal. Chem. 1997, 69, 4499-4507. (19) Majors, R. E. LC-GC Int. 1998, 11, 416-427.
* Corresponding author: (tel/fax) +44 (0) 113 233-6490; (e-mail) K.D.Bartle@ chem.leeds.ac.uk. (1) Huang, X.; Luckey, J. A.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1989, 61, 766-770. (2) Chiari, M.; Nesi, M.; Sandoval, J. E.; Pesek, J. J. J. Chromatogr., A 1995, 717, 1-13.
1820 Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
10.1021/ac980960s CCC: $18.00
© 1999 American Chemical Society Published on Web 03/19/1999
are not fully understood. To our knowledge, this phenomenon has not been observed on columns packed with either C18 or strong anion-exchange material. The answer to this enigma may well lie in the behavior of the SCX packing material under certain operating conditions. Rathore and Horva´th20 have studied the relevant literature on theories dealing with EOF and found two models that are particularly relevant to the use of packed columns in CEC. Of most interest was that proposed by Dukhin and coworkers, which suggested that an unexpectedly high EOF velocity may be generated in columns packed with conductive particles such as ion exchangers, provided the conductivity is greater than that of the electrophoretic medium. Further work on the behavior of ion-exchange materials under various CEC conditions is essential if the SCX phase is to be characterized with respect to generation and control of EOF. The EOF velocity, νEOF, is best described by the Smoluchowski equation,21
νEOF ) orζE/η
(1)
where o and r are the vacuum and relative permittivities, ζ is the zeta potential, E is the electric field strength, and η is the viscosity. It is best controlled by parameters that influence ζ by way of the surface charge of the capillary and by the double-layer thickness.22 In this paper, we have studied the effect of changing common separation parameters such as pH, ionic strength, and solvent composition of the electrolyte on the migration of a neutral compound believed to be unretained on a SCX column. The results are compared to open capillaries and ODS1 columns and are discussed in terms of the monitored separation current and resultant EOF velocity. EXPERIMENTAL SECTION Reagents and Materials. Electroseparations were carried out on 100-µm-i.d. untreated fused-silica capillaries (Composite Metal Services Ltd., Hallow, UK), either open, coated with a strong cation exchanger (Degussa Organosilane 285), or packed with 3-µm Waters Spherisorb SCX or ODS1 stationary phases (Phase Separations Ltd., Deeside, UK). Thiourea was obtained from Sigma (Poole, UK) and HPLC grade acetonitrile from Merck Ltd. (Lutterworth, UK). Water purified with an Elgastat UHQII (Elga, High Wycombe, UK) was used throughout, and all mobile phases were filtered prior to use through a 0.2-µm Whatman Anotop syringe filter (Phase Separations Ltd.). Aqueous spectrophotometric buffers, of low and constant ionic strength (10 mM), were prepared according to Perrin;23 to cover the pH range of interest, the buffers required were KCl-HCl (pH 2.9), succinate (pH 3.6-6.0), phosphate (pH 6.5-7.5), borate (pH 9.0-9.5), and carbonate (pH 9.6-10.5). For studies utilizing 80 + 20 (untreated vs coated capillaries) or 70 + 30 (CEC) (v/v) acetonitrile-H2O media, the electrolytes were prepared in a similar manner. Thus, the buffer concentrations of the final solutions were identical to that found in analogous aqueous systems. However, the ionic strength was expected to differ from (20) Rathore, A. S.; Horva´th, C. J. Chromatogr., A 1997, 781, 185-195. (21) Knox, J. H. Chromatographia 1988, 26, 329-337. (22) Foret, F.; Kriva´nkova´, L.; Bocˇek, P. Capillary Zone Electrophoresis; VCH Publishers Ltd.: Cambridge, 1993; Chapter 4. (23) Perrin, D. D. Aust. J. Chem. 1963, 16, 572-578.
10 mM due to a change in the dissociation constant of the buffer components: dissociation constants (Ka) of weak acids decrease with a decrease in the dielectric constant of the solvent.24 Throughout the text, the electrolyte pH refers to the pH of the aqueous buffer solution it is based on, and not the “apparent” pH of the mixed solvent system, which was not measured. Electrolytes of different ionic strength were prepared from the same stock solutions by varying the relative proportions of each. The EOF was determined by measuring the mobility of a neutral and unretained marker, i.e., 0.5 mM thiourea in H2O and 1-2 mM thiourea in 60 + 40 (v/v) CH3CN-H2O for CE and packed-column CEC, respectively. While Lelie`vre et al. suggested that thiourea can be retained on an ODS phase,25 it has been assumed that any retention effects on either phase should be small due to the short length of column utilized. However, for further work in this area it would be advisable to investigate the suitability of other marker compounds. Throughout the experimentation, it has been assumed that there are no effects on pH, ionic strength, and buffer capacity of the electrolyte on either dilution or varying acetonitrile content. Capillary Surface Modification. In the general reaction scheme, the SCX functionalities, provided by an organosilane reagent (Si-R), are covalently bonded to the silica surface through a siloxane (Si-O-Si-C-) linkage. Initially, the untreated fusedsilica capillary was treated with HCl (20% v/v, 1 capillary volume) and then distilled water (10 capillary volumes) to give a uniform internal surface for coating. The capillary was then connected to a GC oven and heated at 50 °C for 1 h under pressure to eliminate residual water, followed by 1 h at 200 °C to remove contaminants from the surface. The ion-exchange reagent ([3-(trihydroxysilyl)propane]sulfonic acid) was forced through sections of the capillary using nitrogen (1 bar) for ∼5 min, after which excess liquid was removed using 3-4 bar. After the capillary ends had been sealed with a flame, the capillaries were heated for 16 h overnight in a GC oven at 80 °C. Following cooling, the capillaries were flushed with water and 60 + 40 (v/v) acetone-water to remove any remaining reagent prior to use. Preparation of Packed Columns for CEC. Sections of capillaries (typically 50-cm length) were packed using a Shandon HPLC column packer (Hypersil, Runcorn, UK) at 300 bar, against a Valco union (Phase Separations Ltd.) containing a metal screen (2-µm pores). The reservoir, containing a slurry of the stationary phase in acetone (80-100 mg mL-1) was sonicated throughout, and acetone was used as the packing solvent. The capillary was typically packed within the first few minutes, after which the pressure was maintained for a further 15 min to ensure a firmly packed bed. The packed capillary was then cut into ∼17-cm lengths for conversion into short columns of packed length 8.5 cm. Each section was connected to a Valco union fitted with a metal screen (2-µm pores) to prevent the material being expelled from the capillary. Initially the section was conditioned on an HPLC pump with 80 + 20 (v/v) CH3CN-H2O (1 h at 35 bar) followed by H2O (1 h at 35-70 bar). A hot filament device was used to manufacture the first frit near the union, after which the union was removed and the column flushed for a further 1 h to (24) Robinson, R. A.; Stokes, R. H. Electrolyte Solutions; Butterworth: London, 1965; p 351. (25) Lelie`vre, F.; Yan, C.; Zare, R. N.; Gareil, P. J. Chromatogr., A 1996, 723, 145-156.
Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
1821
remove excess packing. Finally, the second frit was burnt at the required distance from the first. The total length of column was then cut down to size (12-13 cm) and connected to a separate section of fused-silica capillary (100-µm i.d.) via a PTFE sleeve (10-15 mm of 1/16-in. PTFE tubing, drilled to 340-µm i.d.). CEC columns were conditioned from storage by flushing on an HPLC pump (1.25 h at 35-55 bar) with the relevant electrolyte and then installing in the instrument for voltage conditioning (10 kV) with pressurization (2 bar on each vial) until a steady current was obtained. For subsequent changes in electrolyte, voltage conditioning at 10 kV for 30 min with pressurization was employed. Instrumentation and Operating Procedure. Both CE and CEC were performed on a HP 3DCE system (Hewlett-Packard Ltd., Cheadle Heath, UK) capable of pressurization, using HP ChemStations software for system control, data acquisition, and data analysis. CE experiments were performed with capillaries of total length 33 cm (length to detector 25 cm), which had been conditioned as follows: for the untreated capillary, 0.1 M NaOH (5 min), H2O (10 min), and electrolyte (20 min for mixed-solvent systems and 30 min for aqueous); and for the coated capillary, H2O (10 min), and electrolyte (20-30 min as before). Prior to each injection the capillary was rinsed with electrolyte (1.5 min). Thiourea was loaded by a 2-s pressure injection (10 mbar) at the anode and separated under normal polarity conditions using a voltage of 10 kV and a ramp of 0.01 min. CEC separations were carried out on coupled capillaries of packed length 8.5 cm and total length ∼33 cm (length to detector 8.5 cm). Thiourea was injected electrokinetically (5 s at 5 kV) at the anode and separated using reversed-polarity mode. The applied voltage was either 10 or 20 kV with a ramp of 0.10 min, and 2 bar pressure was applied to the electrolyte vials throughout the run. For both CE and CEC, the external temperature of the capillary/column was thermostated at 20 °C and thiourea peaks were detected at 240 and 210 nm with a detector response time of 0.3 s. For practical purposes, detection in CEC was performed through the packing material by way of the frit; in the absence of a detector window after the frit, the column lifetime should be increased because there is one less weak point. Data presented typically represent an average of the results obtained on two separate columns unless otherwise indicated. RESULTS AND DISCUSSION Effect of pH on EOF for Untreated and Coated Capillaries. EOF profiles were obtained for the untreated and SCX-coated capillaries in aqueous media. Since the pKa value of sulfonic acid is -2.0,26 in the absence of free silanol groups a permanent charge, and hence consistent EOF in the normal direction, should result. Figure 1 illustrates that the observed trend was generally in accordance with theory: the SCX-coated capillary (Figure 1b) showed a more consistent EOF over the pH range studied compared to the untreated silica (Figure 1a). Profiles obtained using 80 + 20 (v/v) acetonitrile-H2O media are also shown; CEC separations are typically performed using a mobile phase with a high organic solvent content.27 Assuming no shift in pH when acetonitrile is incorporated into the electrolyte, the untreated and (26) Hirokawa, T.; Nishino, M.; Aoki, N.; Kiso, Y.; Sawamoto, Y.; Yagi, T.; Akiyama, J.-I. J. Chromatogr. 1983, 271, D1-D106. (27) Robson, M. M.; Cikalo, M. G.; Myers, P.; Euerby, M. R.; Bartle, K. D. J. Microcolumn Sep. 1997, 9, 357-372.
1822 Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
Figure 1. Comparison of EOF profiles in CE obtained for (a) untreated fused silica capillary and (b) SCX-coated capillary in aqueous (0) and mixed-solvent (+) electrolytes. Conditions: capillary, 33 cm (25-cm effective length) 100-µm i.d.; temperature, 20 °C; preinjection rinse, 1.5 min electrolyte; pressure injection, 2 s at 10 mbar; applied voltage, 10 kV, ramp 0.01 min; detection, UV absorbance at 240 and 210 nm with a detector response time of 0.3 s; electrolytes (mobile phases) of varying pH, aqueous systems 10 mM ionic strength, and similar solutions containing 80% (v/v) acetonitrile; analyte concentration, 0.5 mM thiourea in H2O.
SCX capillaries both displayed a general reduction in the EOF compared to the aqueous electrolyte systems. This was attributed to changes in the dielectric constant and viscosity of the running electrolyte; the dependence of the EOF on these variables has already been shown in eq 1. However, since the µEOF values and characteristics exhibited by both capillaries were similar, it was questionable whether the acetonitrile has a dominating effect on EOF in the SCX capillary or the coating has been destroyed. Until further studies have been performed to determine the optimum coating procedures and the coating stability, it has been considered inadvisable to coat capillary walls prior to packing with the SCX phase. Effect of pH on EOF in Packed-Column CEC. EOF profiles for SCX and ODS1 columns, measured using electrolytes containing 70% (v/v) acetonitrile, are shown in Figure 2; a change in solvent content was necessary since buffer precipitation was becoming increasingly evident at the 80% acetonitrile level. The
Table 1. Comparison of the Conductivity of the Open and Packed Sections of SCX and ODS1 Columns av conductivity at 20 °C/10-9 Sm pH
open
ODS1
SCX
10.5 9.5 7.5 6.0 5.0 4.0 2.9
0.33 0.41 0.29 0.41 0.49 0.60 0.92
0.11 0.24 0.30 0.19 0.19 0.18 0.14
0.11 0.24 0.26 0.16 0.27 0.41 0.15
resistance, R, is Figure 2. Comparison of EOF profiles obtained in CE (untreated fused silica, [) and for CEC utilizing ODS1 (O) and SCX (4) columns. CEC conditions: coupled capillaries, packed length (and length to detector) 8.5 cm, total length 33 cm, 100-µm i.d.; electrokinetic injection, 5 s at 5 kV; reverse polarity; applied voltage, 10 kV, ramp 0.10 min; run pressure, 2 bar on both electrolyte vials; detection through the frit; electrolytes of varying pH, ∼10 mM ionic strength containing 70% (v/v) acetonitrile; sample concentration, 1-2 mM thiourea in 60 + 40 (v/v) CH3CN-H2O. CE and all other conditions as for Figure 1.
ODS1 columns follow a trend similar to that obtained in capillary electrophoresis, although surprisingly the EOF was greater than that found in CE utilizing fused-silica capillaries. This is in contrast to the suggestion by Knox and Grant28 that a lower EOF should result on account of the tortuous nature of the packed bed. As expected, the SCX column exhibits a far more consistent EOF over the pH range studied; the differences between the highest and lowest velocities obtained were approximately 4 × 10-4 and 6 × 10-4 m s-1 for the SCX and ODS1 columns, respectively. Below pH 5, the EOF for the SCX phase was greater than that for the ODS1 phase. At the lower pH values, the EOF would be expected to be higher than that for ODS1, because the sulfonic acid groups are predominantly ionized. However, any residual silanol groups would be un-ionized and that should reduce the effect; during this set of experiments, results at “pH 2.9” for the open capillary and ODS1 column were not routinely obtained. As mentioned previously, the model proposed by Dukhin and coworkers could account for higher than expected EOF velocities generated in columns packed with conductive particles. The main condition necessary to satisfy the model was that the conductivity of the particles had to be greater than that of the electrophoretic medium. A typical CEC column comprises a packed section and an open section, each having different conductances. Since the current is conserved across the column, this leads to different field strengths in each section. By treating the individual segments as resistors in series, the conductance and hence conductivity of the packed section can be calculated. The resistance of an open capillary containing the same electrolyte as the column is used to provide the resistance of the open section. From this and the overall resistance of the column, the resistance of the packed section can be calculated. Applying a voltage, V, across an open capillary of length gives a current, I, which can be measured. From Ohm’s law, the (28) Knox, J. H.; Grant, I. H. Chromatographia 1991, 32, 317-327.
R ) V/I
(2)
The conductance, G, is the reciprocal of resistance and when multiplied by the corresponding length of capillary, l, gives the conductivity; the cross-sectional area has been ignored since it is constant throughout this study. Assuming that R is proportional to length, then for a packed capillary the resistance of the open section, Ro, is given by
Ro ) R(lo/l)
(3)
where lo and l are the lengths of the open section and open capillary, respectively. Knowing the applied voltage, VA, and resultant current, IT, when this column is used gives the total resistance of the packed column, Ro+p, from which the resistance in the packed section, Rp, can be determined:
Ro+p ) VA/IT ) Ro + Rp
(4)
The conductivity of each section is then calculated by dividing the length of the section by its resistance. Conductivity values for the open and packed sections of the ODS1 and SCX packed columns at varying electrolyte pH are shown in Table 1. Although the packed columns generally showed a higher EOF than open capillaries under identical conditions (Figure 2), from the conductivity values there would appear to be no evidence at this stage to support the model proposed by Dukhin. In addition, the conductivity suggests that there is little difference between the two types of packing material. Effect of Acetonitrile Content. The organic solvent composition of the electrolyte was varied in an attempt to alter the polarization of the SCX particles as a result of changes in solvent polarity. The acetonitrile content of a phosphate electrolyte (pH 7.5, 10 mM ionic strength when prepared in H2O) was studied over the range 20-80% (v/v) for both open and packed capillaries; experiments at higher percentages of acetonitrile were not performed owing to phosphate precipitation. The open capillary and ODS1 column showed a similar trend of decreasing and then increasing EOF with increasing acetonitrile content, as observed by Wright et al.,29 whereas the SCX columns showed a consistent (29) Wright, P. B.; Lister, A. S.; Dorsey, J. G. Anal. Chem. 1997, 69, 32513259.
Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
1823
Figure 3. Effect of acetonitrile content on linear velocities obtained in CE (untreated fused silica, [) and for CEC utilizing ODS1 (O) and SCX (4) columns. Electrolytes: “pH 7.5” phosphate, ∼10 mM ionic strength, containing 20-80% (v/v) acetonitrile. Linear velocities measured at 10 kV. The data point for CE using 70% acetonitrile represents an average of the results obtained on one capillary only. All other conditions as for Figures 1 and 2.
Figure 4. Effect of ionic strength on linear velocities obtained in CE (untreated fused silica, () and for CEC utilizing ODS1 (O) and SCX (4) columns. Electrolytes: “pH 7.5” phosphate, ∼1-20 mM ionic strength, containing 70% (v/v) acetonitrile. Linear velocities measured at 10 kV. All other conditions as for Figures 1 and 2.
Table 2. Effect of Changing the Acetonitrile Content of the Electrolyte on the Measured Current and Calculated Conductivity for Open and Packed Capillaries av current/µA
av conductivity at 20 °C/10-9 Sm
% CH3CN
open
ODS1
SCX
open
ODS1
SCX
20 40 60 70 80
14.6 13.6 11.0 7.0 4.0
9.5 8.7 7.4 6.4 3.0
8.0 7.6 7.2 6.8 4.2
0.48 0.45 0.37 0.23 0.13
0.16 0.15 0.13 0.18 0.06
0.12 0.11 0.12 0.22 0.81
decrease in EOF (Figure 3). This behavior could not be explained by the measured currents presented in Table 2, since the data behaved similarly for all capillaries. However, while the conductivity of the open and packed sections for the ODS1 column showed a general reduction with increasing solvent, for the SCX packing, the conductivity was stable up to 60% acetonitrile and then it began to rise. This appears to be a reversal of the trend observed for the EOF velocity and once again in contrast to Dukhin’s model: although the conductivity of the packed section was greater than the electrophoretic medium, it did not result in a higher EOF. Effect of Ionic Strength. The effect of ionic strength was studied over the range 1-20 mM phosphate in “pH 7.5” electrolytes containing 70% (v/v) acetonitrile. As the ionic strength is reduced, the EOF is expected to increase unless double-layer overlap occurs. Since the packing material has a pore size of 8 nm, double-layer overlap in the pores is expected to occur at ionic strengths of 2.5 mM and below where the double layer is calculated to be at least 4 nm. Figure 4 illustrates that the increases in EOF for CE and CEC are generally similar at the higher ionic strengths, but at lower ionic strengths it starts to level off and drop slightly in CEC, whereas in CE it continues to rise. For the ODS1 column, this occurs below 5 mM; this supports the theory of double-layer overlap. However, for SCX columns the reduction in EOF is apparent at ionic strengths of
