An Introduction to Capillary Electrochromatography

KEITH D. BARTLE, MARIA G. CIKALO AND MARK M. ROBSON

1 What is Capillary Electrochromatography?

Capillary electrochromatography (CEC) is a recently developed variant of high-performance liquid chromatography (HPLC) in which the flow of mobile phase is driven through the column by an electric field, a phenomenon known as electroosmosis, rather than by applied pressure. This electroosmotic flow (EOF) is generated by applying a large voltage across the column; positive ions of the added electrolyte accumulate in the electrical double layer of particles of column packing, move towards the cathode and drag the liquid mobile phase with them. As in capillary electrophoresis (CE) and micellar electrokinetic chromatography (MEKC), small diameter (typically 50-100 µm) columns with favourable surface area-to-volume ratio are employed to minimise thermal gradients from ohmic heating, which can have an adverse effect on band widths. CEC differs crucially from CE and MEKC, however, in that the separating principle is partition between the liquid and solid phases (Table 1.1).

Avoiding the use of pressure results in a number of important advantages for CEC over conventional HPLC. Firstly, the pressure-driven flow rate through a packed bed depends directly on the square of the particle diameter and inversely on column length; for practical pressures, generally used particle diameters are seldom less than 3 µm, with column lengths restricted to approximately 25 cm. By contrast the electrically driven flow rate is independent of particle diameter and column length so that, in principle, smaller particles and longer columns can be used. If follows that considerably higher efficiencies can be generated in CEC than in HPLC. A second consequence of employing electrodrive is that the plug-like flow -velocity profile in EOF reduces dispersion of the band of solute as it passes through the column, further increasing column efficiency. The combined effect of reduced particle diameter, increased column length and plug flow leads to CEC efficiencies of typically 200 000 plates per metre, and substantially improved resolution.

Voltages up to 30 kV are applied to generate the electric field usually for solutions of 1–50 mM buffers in aqueous reversed-phase mobile phases, although non-aqueous CEC has also been carried out. The dependence of EOF rate on solvent dielectric constant has been confirmed, but the electrical potential (the zeta potential) of the boundary between the fixed and diffuse layers (the double layer; see pages 43–5 for further discussion) of positive ions at the stationary phase wall (Figure 1.1) is less well understood. The conclusion of an early theoretical study which suggested that flat EOF profiles in a capillary of diameter d would result if d were considerably greater than the double layer thickness, δ, has been confirmed by experiment; for channels between particles, however, the influence of δ is less clear. Current indications are that it should be possible to use monodisperse particles with diameters down to 0.5 µm. Pore sizes of commonly used HPLC particles are too small to give rise to EOF, but larger pore packings show promise. Although CEC has been demonstrated for stationary phases bonded to the walls of open tubes, and in sol–gel derived phases, most work has been carried out on columns packed with HPLC stationary phases; a new generation of packings custom-synthesized for CEC is, however, now beginning to make an impact.

2 History of CEC

Strain first reported the use of the EOF in chromatography; he recognized the difference between electrophoresis and electrochromatography on the one hand, and partitioning of analytes between a mobile and a stationary phase on the other. Electrodrive (electrophoretic mobility and electroosmosis) was used to move the analytes through a separation medium, so that the importance of the EOF was recognized in electrochromatography. Early work in electrical chromatography was either in relatively large diameter columns (>1 mm) or in thin layers, which were used to analyse neutral, basic and acidic molecules by electromigration through a paper matrix.

The separation of polysaccharides using electrodrive through a colloidal membrane is probably the first reported use of EOF to drive a mobile phase through a stationary phase. For thin layers, Kowalczyk quantified the EOF velocity while Hybarger et al. proposed an annular bed system for preparative separations. Cylindrical columns packed with Sephadex were used by a number of groups for protein separation. Gel columns were employed in the separation of high-molecular-weight (>500 kDa) compounds in experiments in which counteracting electrophoretic and hydrodynamic forces were used.

The originators of CEC were, however, the Pretorius group, who reported that if the EOF were used to drive the mobile phase flow, as opposed to the hydrodynamic flow in conventional liquid chromatography, the plate height was reduced. They also pointed out the absence of pressure drop across the column if the EOF were used. Significant progress in CEC began in the 1980s; Jorgenson and Lukacs demonstrated the use of electroosmosis in capillaries and showed the possibilities for low reduced plate heights. Tsuda then showed that CEC was possible in coated open tubular columns and recognized the factors that control the EOF as well as the importance of practical effects, such as bubble formation, in packed columns.

The recent resurgence of CEC dates from the detailed theoretical analysis of Knox and Grant, published in 1987, followed by practical demonstrations by the same group in 1991. Both slurry-packed and draw-packed capillaries were used in a detailed study of factors affecting the EOF. Particle sizes down to 1.5 µm were used, and reduced plate heights near unity were demonstrated in 30–200 µm i.d. columns up to 1 m long. The important observation was made that columns driven electrically show higher efficiencies than the same column with pressure drive (Figure 1.2).

The potential of CEC in the analysis of mixtures relevant to the pharmaceutical industry was realized by Smith and Evans in 1994. The capabilities of CEC were demonstrated in high-resolution chromatograms of drug compounds (e.g. Figure 1.3) on a reversed-phase C18 stationary phase. The same group than went on to show how especially strong cation-exchanger stationary phases allowed analyte focusing and very narrow peaks for basic drugs. Some landmarks in the history of CEC are listed in Table 1.2.

Since the re-emergence of CEC, much work has centred on the establishment of reliable column technology (Chapter 3), on identifying suitable mobile phase buffers and on investigating the theory and mechanism of CEC (Chapter 4). Since 1996, however, there has been a very rapid increase in the number of publications and reviews relating to CEC. The numbers of reports of CEC separations of compounds from the environment, and of biomolecules, are growing rapidly, while the complementary relation of CEC separations to generally used HPLC methods has led to much activity in the analysis of pharmaceuticals by CEC (Chapter 7). The extremely low flow rates in CEC (<1 mL min-1) help make coupling to mass spectrometry an especially attractive possibility (Chapter 6), first demonstrated in 1991.

3 Electroosmosis

Electroosmosis is best described as the movement of liquid relative to a stationary charged surface under an applied electric field. Substances tend to acquire a surface charge as a result of ionization of the surface and/or by interaction with ionic species. In a fused silica capillary, the ionization of silanol groups gives rise to a negatively charged surface, which affects the distribution of nearby ions in solution. Ions of opposite charge (counterions) are attracted to the surface to maintain the charge balance whilst ions of like charge (co-ions) are repelled. The double layer of electric charge thus formed (see Figure 1.1) is generally explained by a revised version of the Gouy-Chapman model. Essentially the counterions are arranged in two layers, fixed and diffuse, with a surface of shear at just beyond the interface. The voltage drop between the wall and this surface of shear is known as the zeta potential, ζ. In the diffuse layer, the potential falls exponentially to zero, and the distance over which it falls by e-1 is known as the double layer thickness, δ. When the voltage is applied, the solvated cations in the diffuse layer migrate towards the cathode, dragging the solvent molecules along with them.

The linear velocity of the EOF, ueo, is described by the Smoluchowski equation

ueo = ε0εrζE/ν (1.1)

This shows how the EOF is governed by ζ, the permittivity and viscosity of the mobile phase, εr and ν, and the electric field strength, E.

The flow profile is assumed to be near-plug-like since essentially it originates from the capillary wall, but in reality it depends on the capillary internal diameter, d, and δ. Theoretical studies by Rice and Whitehead proposed that ueo is only independent of the capillary diameter when d [much greater than] δ. As d approaches δ, double layer overlap occurs with a simultaneous reduction in flow velocity, until finally a parabolic flow profile is obtained when d and δ are similar. It has been proposed that the EOF velocity is acceptable when d ≥ 10δ. The use of microscope optics to image flow profiles in narrow capillaries has produced conflicting results. Whilst the plug flow profile predicted from theory has been observed by Taylor and Yeung, Tsuda et al. found a higher EOF at the capillary wall than at the centre. The importance of the EOF profile in CEC necessitates further research in this area.

From equation (1.1) it can be seen that neither the diameter of the stationary phase particle (dp) nor the column length (L) affect the mobile phase velocity. This is in contrast with pressure-driven flow velocity [bar.v] which is described by the Kozeny–Carman equation

[bar.v] = ε2/180(1 - ε)2 d2pδp/ν L (1.2)

where ε is the porosity (approximately 0.4 for a randomly packed column) and δp is the pressure drop.

In packed CEC, both the capillary wall and the column packing carry surface charges that are capable of supporting EOF. To date, most of the work carried out suggests that the column packing is responsible for the generation of EOF; there is a greater number of free silanol groups present since the solid packing has a far larger surface area compared to that of the internal silica wall. If the column is assumed to consist of a closely-packed array of non-porous spherical particles, then the EOF arises from the channels between the particles. The average interparticle channel is estimated to be one-quarter to one-fifth the particle diameter. It has been suggested that the EOF velocity in a CEC column is most likely to be reduced compared to that in an open tube, on account of the tortuosity and porosity of the packed bed. Although there does not appear to be any adverse effect as a result of packing irregularities further investigations have been made on packing structure using electrical conductivity measurements.

Using capillary columns comprising a packed and an open section, it was possible to calculate the voltage drop in each section, and thus to determine the individual contributions to the total EOF. Measurements made at different pH for columns packed to different lengths with reversed-phase octadecyl silica (ODS) and strong cation-exchange (SCX) particles showed that: at extremes of pH the electric field strength is far greater in the packed section; at intermediate pH the field strengths are similar in both sections; at pH 7.5 the length of the packed section has little impact on the total conductance of the column, but the conductance of the open section does scale approximately with its length. At pH 7.5 the contributions of the open and packed sections are roughly equal when the capillary is half packed, but the same effect at pH 10.5 is only apparent when the capillary is about one-quarter packed. This study confirmed that, except at low pH, the EOF in a packed column is generally less than that in an open capillary, and that the greatest influence of length of packed bed on ueo is observed at low or high pH.

4 Classification of CEC

Separation in CE is a consequence of differential migration of charged and neutral species (Figure 1.4). CEC may be compared with CE and classified in the hierarchy of separation methods employing liquids by the unified description proposed by Rathore and Horvath. The differential migration of solute bands can be divided into components that are separative (selective interactions with a stationary phase, or differences in electrophoretic migration velocities), and components that are non-separative (migration not contributing directly to separation). The concept of 'virtual migration lengths' then allows the CEC retention factor, kc to be defined. Noting that the separative (HPLC) retention factor, k, is given by

k = tR - t0/t0 (1.3)

and that the CE velocity factor, ke, is, correspondingly

ke = teo - tm/tm (1.4)

where tR and tm are elution and migration times respectively, t0 is the retention time of an unretained marker, and teo is the retention time of a solute moved only by the EOF, then,

kc = k + kke + ke (1.5)

The product kke is the consequence of simultaneous chromatography and electrophoresis. If ke = 0 then kc simply equals k, and only HPLC operates. Correspondingly, if k = 0, then kc = ke and the only process is CE.

In HPLC, and in CEC of neutral species (case A in Figure 1.5), the solute and he mobile phase move in the same direction, and the sample components merge in order of retardation by the stationary phase. However, if the solute is charged, there are three operational modes depending on the direction and magnitude of electrophoretic migration with respect to the direction of the EOF Figure 1.5):

B. co-directional, where the migration velocities of charged species are always greater than that of the EOF marker. The components emerge before the EOF marker;

C. counter-directional, where the EOF velocity is greater than the electrophoretic velocity of a charged component, which emerges after the EOF marker;

D. counter-directional, where the EOF velocity is less than the electrophoretic velocity, and detection of a charged component is only possible with instrument polarity reversed. In this case, the EOF marker is not detected.

The order of emergence of different mixture components relative to each other depends on the combination of their different retardation and migration velocities, in accordance with equation (1.5).

CEC offers the substantial advantage over MEKC of a time window for separation which is, in principle, unlimited. In MEKC, however, all electrically neutral compounds have migration times between t0 and tmc, that of the micelle.

A hybrid of MEKC and CEC has been proposed by Knox. In colloidal sol electrochromatography, a colloidal sol is used as the moving fluid. If the colloidal particles are charged they move relative to the eluent, and partition between two phases occurs resulting in separation.

5 Band-broadening in CEC

In column chromatography, a number of processes bring about the broadening of solute bands: (a) eddy diffusion, originating from the variety of flow paths through the packed bed; (b) axial molecular diffusion; (c) resistance to mass transfer in the mobile and stationary phases and (d) system effects, such as those arising from dead volumes. The smaller theoretical plate heights (H) and hence greater plate numbers N (=L/H) in CEC in comparison with conventional pressure-driven HPLC arises from reduced contributions to H from factors (a) and (c) above.

Figure 1.6 illustrates the differences in low velocity profiles in the packed bed. Clearly, the plug flow profile of CEC substantially reduces the eddy diffusion (or multipath) term in comparison with the parabolic flow profile of HPLC. Since this term is also proportional to the column particle diameter, the use of smaller particles should further reduce the contribution to H; the contribution from slow mass transfer in the mobile phase, Cm[bar.u] where [bar.u] is average linear mobile phase velocity, is proportional to d2p.

The use of EOF to drive the mobile phase flow gives rise (Figure 1.6) to a plug flow profile in the channels between the particles. This also reduces the mass transfer contribution to H by a factor which can be estimated by considering such a channel as an open tube, of diameter dc, for which Cm is given by the Golay equation:

Cm = ƒ (k)d2c/ Dm (1.6)

where Dm is the diffusion coefficient of the solute in the mobile phase. For parabolic flow, as in HPLC

ƒ(k) = 1 + 6k + 11k2/96(1 + k)2 (1.7)

but for plug flow, in CEC

ƒ(k) = k2/16(1 + k)2 (1.8)

It follows that, for a given k, and the same dc and Dm, the contribution to H from this source for CEC is about half that in capillary HPLC.

The model developed by Horvath and Lin to describe band broadening in HPLC was applied to CEC by Dittman et al. An improvement in efficiency over HPLC of approximately 100% was predicted, so that minimum reduced plate heights Hr (=H/dp) near unity are predicted; this has generally been borne out by experiment for CEC for columns packed with particles with dp ≥ 3 µm (see for example Figure 1.7). The small increase in H with increasing [bar.u]opt, the mobile phase velocity at the minimum value of H, suggests that efficient CEC should be possible at high bar.u, thus shortening analysis times. Knox and Grant calculated that the minimum value of dp which can be used without affecting the EOF velocity is 0.5 µm, but, perhaps because columns are more difficult to pack with small particles (Chapter 3) their reported applications have been comparatively few. Lüdtke and co-workers have packed 0.5 µm silica beads, but observed minimum reduced plate heights of ~ 3.5–4.00.