Electro-osmotic and pressure-driven flow properties of frits for packed column capillary electrochromatography prepared from functionalised and bare silica packings

Abstract

Changes in electro-osmotic flow (EOF) induced in an open fused silica capillary by the introduction of a single packing retainer (referred to as a ‘frit’ throughout the article) are investigated. Frits located in a 34.5 cm long capillary close to the detection window, as used commonly in packed column capillary electro-chromatography, were made by sintering silica based packing materials having varying functionalities (bare silica, ODS-silica, a strong cation exchange packing (SCX) and a strong anion exchanger (SAX) at 430 °C for 15 s. The contributions of such frits to the total EOF measured in the capillary are discussed. Despite the high temperature of heating of the packing, sufficient residual functional groups remained on the frit such that distinctively different EOF behaviour was observed for each type of packing. Frits made from materials providing negative surface charges, such as silica, ODS-silica and SCX, increased the magnitude of the cathodic EOF compared to the open capillary. On the other hand, a substantial anodic EOF of −1.88 × 10⁻⁸ m² V⁻¹ s⁻¹ was produced by introduction of a single frit made from the SAX material. An explanation of this behaviour is given, based on the hypothesis that the EOF generated by the frit determines the overall flow in the whole open capillary. The frit is considered to work as a pump and overrides the EOF generated at the capillary wall.

---

Introduction

Capillary electrochromatography (CEC) using packed columns has been an emerging technique over recent years and since suitable instrumentation for CEC has become available commercially, increased interest in this new separation method has resulted. This fact is demonstrated clearly by the steadily growing number of publications within this field.^1–8 The success of CEC can be attributed to several potential advantages of this approach over other liquid phase separation methods, such as high-performance liquid chromatography (HPLC) or capillary zone electrophoresis (CZE). First, the almost plug-like flow profile of the electro-osmotic flow (EOF) helps to minimise the dispersion of analyte zones which occurs in HPLC. Second, the velocity of the electroosmotically generated flow is independent of the capillary diameter, the column length and the particle size of the packing, and this allows the use of microbore capillaries and microparticulate chromatographic supports without the limitation of increased back pressure as occurs with pressure-driven methods such as HPLC. Finally, CEC offers additional separation selectivity because of its combined electrophoretic and chromatographic separation mechanisms.

Since the beginnings of CEC various methods for packing of columns have been described, such as employing high pressure,^9,10 EOF¹¹ or even centripetal forces.¹² Recently, so-called monolithic columns for CEC, prepared by in situ polymerisation of a stationary phase inside the capillary, have been reported.¹³ In the latter case the chromatographic support is immobilised by chemical bonding to the capillary walls, but with all other packing methods a retainer is necessary to keep the particles inside the column. This can be accomplished either by modification of the capillary design using tapered capillaries as described recently¹⁴ or through the use of frits, which can be inserted into the column mechanically,¹⁵ or generated inside the capillary by various chemical or physical means. In the literature published so far, the latter method is still the most widely used and a number of techniques for the preparation of frits have been described.^16–18

There have been several reports on the effects of frits and packings on EOF, but the results have been inconsistent and vary from reporting a reduction of 44% in EOF when two frits were introduced into an open capillary,¹⁹ to an increase in EOF for a capillary packed with an ODS material compared to an equivalent open capillary.^20,21 As there were only minor differences in the column packing procedures reported in these studies, the observed discrepancies might be attributable to differences in the materials and procedures used for making the frits, since these vary considerably between the studies. Furthermore, it is well known that many of the problems associated with CEC column operation (e.g., bubble formation) can be due to the interface between packed and open zones within the capillary column.^21–23 This means that it is desirable that the resistivity of the frits be as similar to either the packed or the open zone as possible, thus avoiding further discontinuities in the column structure.

Therefore, whilst the influence of the frits on the performance of the entire packed column can be regarded as crucial, only a few detailed reports investigating these important components of the CEC column exist.^19,21,24,25 In this paper we investigate the effects on key parameters such as pressure resistance and EOF caused by the introduction into a fused silica capillary of a single frit made of bare or functionalised silica.

Experimental

Instrumentation

Experiments were performed using an HP^3D CE system (Hewlett-Packard, Waldbronn, Germany), equipped with a diode array detector and connected to an HP 3D-CE Chemstation (Hewlett-Packard) for data processing. The temperature of the column was maintained at 25 °C during all separations. Samples were injected hydrodynamically by applying a pressure of 20 mbar (open capillary) and 2 bar (capillary with frit) for 3 s.

Materials and reagents

Fused silica capillaries (75 μm or 100 μm id × 360 μm od) obtained from Polymicro Technologies Inc. (Phoenix, AZ, USA) were used throughout this work. All capillaries were flushed with 0.1 M NaOH (at least 20 capillary volumes) before use, and flushed with BGE (at least 5 capillary volumes) between runs.

The following packing materials (all from XTec Consultants Ltd., Clwyd, UK) were used: 3 μm silica (190 m² g⁻¹, 8 nm pore size), 3 μm silica-based reversed phase material (OSD1, 6% carbon, 8 nm pore size) 3 μm silica-based strong anion exchange material (SAX, capacity approx. 200 μequiv. g⁻¹, 2.5% carbon, 8 nm pore size) 3 μm silica-based strong cation exchange material (SCX, capacity approx. 120 μequiv. g⁻¹, 1% carbon, 8 nm pore size). Additionally, frits were prepared from Hypersil ODS 3 μm silica (170 m² g⁻¹, 10% carbon, 12 nm pore size).

All chemicals used were of analytical reagent grade. Water was purified using a Milli-Q water (Millipore, Bedford, MA, USA) system.

Electrolytes were prepared from a 500 mM Tris stock solution, titrated to pH 8.0 with hydrochloric acid and diluted with a mixture of water and acetonitrile or pure water. All electrolytes were filtered through a 0.45 μm membrane filter (Millipore) and degassed before use. Acetone was used as a marker for the determination of the EOF.

Procedures

Preparation of the frits.. Pretreated fused silica capillaries were packed using a slurry packing technique similar to that described by Hilder et al.²⁶ Frits were generated by heating a narrow band of the packing 25 cm from the inlet end of the capillary at 430 °C for 15 s (unless otherwise stated) using a device similar to the one described by van den Bosch et al.¹⁶ consisting of a heating element in the shape of a ribbon 0.3 mm thick with a 0.5 mm hole in it. During the heating, the capillary was flushed with water at 300 bar. After formation of the frit, the excess packing was flushed out of the capillary and the length of the formed frit was measured using a microscope. A detection window was fabricated 10 mm behind the frit using hot sulfuric acid with the frit protected from the heating element during this process. Finally, the capillary was cut to a total length of 34.5 cm, mounted in the Hewlett-Packard capillary cartridge and conditioned with mobile phase before use.

Measurement of electro-osmotic and pressure-driven flow properties of open capillaries and capillaries with frits.. This was accomplished using a procedure similar to that previously described by Dittmann et al.²¹ First, the magnitude of the EOF at +20 kV (−20 kV in the case of the SAX packing) as well as the pressure resistance was measured for each empty capillary (75 or 100 μm id) using a pH 8.0 electrolyte consisting of 50 mM Tris–acetonitrile (1∶4) or 10 mM Tris and employing acetone as a neutral marker. After the formation of frits using the procedure described above, the EOF and the pressure resistance were measured again for each frit. Finally, to include the influence of the length of the packed bed into these considerations, the measured difference in EOF (ΔEOF) was normalised to a 1 mm frit length by dividing the ΔEOF value by the frit length in mm.

Results

Properties of frits made from Hypersil ODS

Table 1 summarises the experimentally obtained linear velocities characterising the electro-osmotic and pressure-driven flow properties of open capillaries and the same capillaries with frits made from Hypersil ODS. The following points can be made from these results. First, the pressure resistance data for the frits show that only low frit-to-frit reproducibility could be achieved although the length of the sintered zone was kept within a narrow range (1.8–2.1 mm). Second, for the electro-osmotically driven flow, very similar linear velocities were encountered for all three frits. This leads to the conclusion that within certain constraints (i.e., as long as no significant blockage of the capillary occurs) the correlation between the pressure resistance of a frit and its influence on the EOF is likely to be rather low or insignificant. Third, an increased EOF could be observed for all capillaries with a frit compared to the open capillaries. It is reasonable to suggest that the surface charge on the open capillary wall and inside the frit channels will predetermine the EOF in each region, but it is somewhat difficult to explain why for example a frit prepared from an ODS material (as in Table 1) which is likely to have a relatively low surface charge density also increases the EOF velocity compared to an open fused silica capillary. One possible explanation is that substantial additional ionisable silanol groups are created during the heating used for the sintering process. It might also be considered that the differences in EOF between open capillaries and the same capillaries containing frits may be in part due to the differences in the nature of the electro-osmotically driven flow depending on whether it is generated only on the walls of an open capillary or in a multitude of channels in a frit. In the latter case the frit could work as an EOF pump which could push the liquid through the whole cross-section of the capillary, as discussed in greater detail later.

Table 1 Electro-osmotic and pressure-driven flow properties of 3 open capillaries compared with the same capillaries containing frits prepared from Hypersil ODS^a

Linear flow velocity/mm s^−1^b
Open capillary		Capillary with frit
Pressure-driven (20 mbar)	Voltage-driven (20 kV)	Pressure-driven (1500 mbar)	Voltage-driven (20 kV)
a Conditions: capillary, 100 μm id; electrolyte, 50 mM Tris–acetonitrile (1∶4), pH 8.0; for other conditions see Experimental.b Each data point in the Table is the mean (± standard deviation) of at least three determinations.
7.55 ± 0.03	1.75 ± 0.01	1.50 ± 0.06	2.45 ± 0.01
7.98 ± 0.03	1.86 ± 0.01	2.79 ± 0.08	2.40 ± 0.01
7.76 ± 0.05	2.02 ± 0.02	1.82 ± 0.02	2.44 ± 0.01

---

Influence of the packing functionality on the EOF

The contribution of EOF of frits made from different packings, all obtained by functionalisation of the same silica base material, was determined by measuring the EOF velocity before and after the frit was created. Table 2 shows the total EOF values obtained for the open capillaries as well as the contribution to EOF of the frit (i.e., the total EOF minus the EOF obtained for the empty capillary), standardised to a frit length of 1 mm. As can be seen from these data, ODS1, silica and SCX stationary phases increased the magnitude of the EOF, with the increase being in proportion to the amount of negatively charged sites available in the corresponding packing material. The fact that the ODS material provided the smallest and the SCX the largest change demonstrates that the functional groups of these packing materials persisted after the high temperature treatment applied during the frit making process. This is in agreement with the observations of Dittmann et al. for an ODS packing heated by an element at 580 °C and moved at up to 2 mm s⁻¹.²¹ At this point it should be noted that the temperature given as the heating temperature is measured at the outside of the capillary and it can be expected that the inside temperature will be somewhat lower, but it is difficult to judge the significance of this difference and no estimations of this error have been offered in the CEC literature.

Table 2 EOF obtained for capillaries equipped with frits made using packing materials with different functionalites^a

Material	EOF/10^−8 m^2 V^−1 s^−1^b	ΔEOF/10^−8 m^2 V^−1 s^−1 per mm frit length
a Conditions: capillary, 75 μm id; other conditions as in Table 1.b Each data point in the Table is the mean (± standard deviation) of at least five determinations.
No frit	3.40 ± 0.04	—
Silica	4.26 ± 0.02	0.16
ODS1	3.59 ± 0.04	0.08
SCX	4.68 ± 0.02	0.32
SAX	−1.88 ± 0.02	−1.32

---

The most dramatic effect was observed in the case of the frit made from the SAX material, which provided a reversed EOF. Previously there have been two reports of a reversed (anodic) total EOF in a capillary packed with at least 25 cm of a SAX packing^26,27 but there has been no report of significant alteration or reversal of the total EOF by a frit or short length of packing material housed in a much longer open capillary. Additionally, there has been no explanation offered for the effects on EOF of different segments of the capillary having opposite surface charge and consequently different local EOF values. A possible rationalisation of the observed results is depicted in Fig. 1 and is based on the assumption that there still exists a cathodic EOF at the negatively charged open capillary wall. However, the bulk liquid is drawn through the capillary towards the anode by the high anodic EOF generated in the multitude of channels in the frit (which have positive charge arising from the residual quaternary ammonium functionalities) through the whole cross-section of the capillary. That is, the EOF is governed by a pumping effect arising from the frit. The flow vector of the cathodic EOF generated by the wall becomes smaller with increasing distance from the capillary wall and turns into an anodic EOF further away from the wall. The resulting flow profile can be expected to adopt a parabolic shape away from the frit (i.e., either before or after the frit) because the flow of the bulk liquid is effectively a pressure-driven flow caused by the frit working as an ‘EOF-pump’. This is a reasonable model since it is a well established fact that when an EOF and a higher pressure-driven flow are superimposed in a capillary, the pressure-driven flow moves easily through the open capillary regardless of the direction of the EOF at the wall.²⁸ That suggests that the frit, even when very short compared with the length of the entire column, is the element which defines the total observed EOF by overriding the EOF generated by the wall.

Fig. 1 Schematic representation of electro-osmotically generated flow in an open capillary containing a segment of frit constructed from SAX material. It is assumed that the frit contains residual positively charged functional groups. The shaded area represents the frit. For explanation see Results and Discussion.

One method for verifying the above model is to examine the effect on peak shape caused by the flow profile in the capillary. Fig. 2 shows a comparison of the peak shapes obtained for benzoate in an open capillary using an electrolyte containing 10 mM Tris (titrated to pH 8 with hydrochloric acid) and 0.5 mM tetradecyltrimethylammonium bromide (TTAB) for EOF reversal [Fig. 2(a)], and in a capillary equipped with a frit made from the SAX material [Fig. 2(b)] using the same Tris electrolyte but without TTAB. EOF reversal in the latter case was caused by the frit alone. Since the frit was positioned on the outlet side of the detection window, at the point of detection there will be no dispersion of the analyte band caused by passage through the frit and the peak shape will reflect the shape of the flow profile carrying the analyte to the detector. As can be seen from Fig. 2, the peak shape obtained with the plug-like EOF profile of the open capillary was superior to that in the capillary with the frit, suggesting that there is parabolic flow induced by the SAX frit, as indicated in Fig. 1.

Fig. 2 Peak shapes obtained for benzoate in an open capillary (a) and a capillary equipped with a SAX frit situated directly behind the detection window on the outlet side (b). Conditions: capillary, 75 μm id; electrolyte, (a): 10 mM Tris, 0.5 mM TTAB pH 8; (b) 10 mM Tris, pH 8; for other conditions see Experimental.

Influence of the heating time

As the frits prepared from the SAX material cause the most dramatic change in the EOF flow, the SAX packing was chosen for experiments in which the heating time during manufacture of the frit was varied in order to estimate the progress of the decomposition of the functional groups of the packing. Sintering times between 12 and 20 s were used since heating times below 12 s did not result in a durable frit. When a sintering time less than 12 s was used it was likely that the temperature (particularly in the centre of the capillary) was not sufficiently high to fuse the silica particles. Fig. 3 depicts the contribution of the frit to the total EOF, plotted as the standardised ΔEOF values and shows that only frits generated using heating times of less that 15 s still carried sufficient residual positively charged functional sites to produce a reversed EOF. Similar behaviour can be expected for other functionalised silicas, although the rate of the packing decomposition may differ.

Fig. 3 Dependence of the contribution to the EOF obtained in a fused silica capillary (standardised to a frit length of 1 mm) of a frit made from SAX packing material on the heating time during the frit making process. Conditions: capillary, 75 μm id; electrolyte, 10 mM Tris, pH 8; for other conditions see Experimental.

Conclusions

The results presented in this paper demonstrate the following features for frits prepared by sintering a short section of functionalised silica based packing. (i) A frit 4 mm in length which is short compared to the overall length of the capillary (34.5 cm) can significantly alter or even reverse the total observed EOF. An explanation is offered in which the frit is considered to be the element that defines the total observed EOF. (ii) Under appropriate conditions (heating at 430 °C for 15 s), sufficient residual functional groups remain on the packing to significantly influence the total EOF observed in a 35 cm long capillary using a moderately basic carrier electrolye. (iii) The amount of residual functional groups remaining on the packing decreases with increasing heating time and for the SAX packing used, heating times longer than 15 s led to insufficient residual positive charge to produce reversal of EOF.

Acknowledgements

The authors wish to thank Professor Richard Cassidy and Professor Wolfgang Buchberger for comments on Fig. 1. Christian Klampfl would like to acknowledge support from the Austrian Science Fund (FWF) project number J 1622 CHE.

References

1. J. H. Knox and I. H. Grant, Chromatographia, 1987, 24, 135.
2. A. L. Crego, A. Gonzalez and M. L. Marina, Crit. Rev. Anal. Chem., 1996, 26, 261.
3. M. M. Dittmann and G. P. Rozing, J. Chromatogr. A, 1996, 744, 63.
4. L. A. Colon, Y. Guo and A. Fermier, Anal. Chem., 1997, 69, A461.
5. L. A. Colon, K. J. Reynolds, R. Aliceamaldonado and A. M. Fermier, Electrophoresis, 1997, 18, 2162.
6. K. D. Altria, N. W. Smith and C. H. Turnbull, Chromatographia, 1997, 46, 664.
7. K. D. Altria, N. W. Smith and C. H. Turnbull, J. Chromatogr. B, 1998, 717, 341.
8. M. G. Cikalo, K. D. Bartle, M. M. Robson, P. Myers and M. R. Euerby, Analyst, 1998, 123, 87R.
9. R. J. Boughtflower, T. Underwood and C. J. Paterson, Chromatographia, 1995, 40, 329.
10. R. J. Boughtflower, T. Underwood and J. Maddin, Chromatographia, 1995, 41, 398.
11. C. Yan, R. Dadoo, H. Zhao, R. N. Zare and D. J. Rakestraw, Anal. Chem., 1995, 67, 2026.
12. A. M. Fermier and L. A. Colon, J. Microcolumn Sep., 1998, 10, 439.
13. E. C. Peters, M. Petro, F. Svec and J. M. J. Frechet, Anal. Chem., 1997, 69, 3646.
14. M. Mayer, E. Rapp, C. Marck and G. J. M. Bruin, Electrophoresis, 1999, 20, 43.
15. J. N. Alexander, J. B. Polli and K. E. Markides, Anal. Chem., 1999, 71, 2398.
16. S. E. van den Bosch, S. Heemstra, J. C. Kraak and H. Poppe, J. Chromatogr. A, 1996, 755, 165.
17. B. Behnke, E. Grohm and E. Bayer, J. Chromatogr. A, 1995, 716, 207.
18. M. G. Schmid, F. Bauml, A. P. Kohne and T. Welsch, J. High Resolut. Chromatogr., 1999, 22, 438.
19. C. M. Yang and Z. El Rassi, Electrophoresis, 1999, 20, 18.
20. M. G. Cikalo, K. D. Bartle and P. Myers, Anal. Chem., 1999, 71, 1820.
21. M. M. Dittmann, G. P. Rozing, G. Ross, T. Adam and K. K. Unger, J. Cap. Elec., 1997, 5, 201.
22. C. M. Yang and Z. El Rassi, Electrophoresis, 1998, 19, 2061.
23. G. Choudhary and C. Horvath, J. Chromatogr. A, 1997, 781, 161.
24. B. Behnke, J. Johansson, S. Zhang, E. Bayer and S. Nilsson, J. Chromatogr. A, 1998, 818, 257.
25. R. A. Carney, M. M. Robson, K. D. Bartle and P. Myers, J. High Resolut. Chromatogr., 1999, 22, 29.
26. E. F. Hilder, M. Macka and P. R. Haddad, Anal. Commun., 1999, 36, 299.
27. D. Li, H. H. Knobel and V. T. Remcho, J. Chromatogr. B, 1997, 695, 169.
28. S. Hjerten, Chromatogr. Rev., 1967, 9, 122.

---