Anal. Chem. 1998, 70, 4586-4593
Modeling and Optimization of Enantioseparation by Capillary Electrochromatography Yulin Deng,*,†,‡ Jianhua Zhang,§ Takao Tsuda,| Peter H. Yu,† Alan A. Boulton,† and Richard M. Cassidy§
Neuropsychiatry Research Unit, Department of Psychiatry, and Department of Chemistry, University of Saskatchewan, Saskatoon, Saskatchewan, S7N 0W0, Canada, College of Chemical Engineering and Materials, Beijing Institute of Technology, Beijing 100081, China, and Department of Applied Chemistry, Nagoya Institute of Technology, Nagoya 466, Japan
Capillary electrochromatography (CEC) is considered to be a hybrid technique that combines the features of both capillary HPLC and capillary electrophoresis (CE).1 In CEC, a mobile phase is driven through a packed or a open-tubular coating capillary column by electroosmotic flow2-5 and/or pressurized flow.6-9 This allows the analyte to partition between the mobile
and stationary phases. As a high voltage is applied, electrophoretic mobility should also contribute to the chromatographic separation for charged analyses. The ability of CEC to combine electrophoretic mobility with partitioning mechanisms is one of its strongest advantages. For electroosmotically driven capillary electrochromatography (ED-CEC), the resulting flow profile is near pluglike;10 thus, a high column efficiency comparable to that in CE can be obtained.11 Since the velocity of electroosmotic flow is independent of the geometry and size of the packing materials, small particles (typically 3 µm or smaller) or long capillary columns (up to 50 cm) can be used. Additionally, neutral molecules can be separated without micelles or other organic additives, and this makes CEC more amenable to coupling with mass spectrometry (MS). One limitation of ED-CEC is that ionic analytes migrating against the electroosmosis may have such a low velocity that elution will be impossible within a reasonable time period. Using appropriate additives or buffer ions, the manipulation of electroosmosis is possible,12,13 but probably difficult, especially in some complicated samples containing basic, acidic, and neutral compounds. For pressure-driven capillary electrochromatography (PDCEC), although dispersion caused by flow-velocity differences causes the zone broadening, plate numbers are higher than in capillary HPLC due to the contribution of the electric field to total flow rate.10 Unlike ED-CEC, the use of an HPLC pump provides stable flow conditions and thus offers improvements in retention reproducibility, in sample introduction (e.g., split injection), in suppression of bubble formation, and in gradient elution. More importantly, since solvent can be mainly driven by pressurized flow, the change in direction of the electric field is no longer limited, and the separation of mixtures of cationic, anionic, and neutral compounds becomes possible in a single run.
* Corresponding author: (phone) (306)-966-8812; (fax) (306)-966-8830; (email)
[email protected]. † Department of Psychiatry, University of Saskatchewan. ‡ Beijing Institute of Technology. § Department of Chemistry, University of Saskatchewan. | Nagoya Institute of Technology. (1) Tsuda, T. In Electric Field Applications in Chromatography, Industrial and Chemical Processes; Tsuda, T., Ed.; VCH: Weinheim, 1995; pp 1-9. (2) Tsuda, T.; Nomura, K.; Nagakawa, G. J. Chromatogr. 1982, 248, 241-7. (3) Jorgenson, J. W.; Lukacs, K. D. J. Chromatogr. 1981, 218, 209-16. (4) Knox, J. H.; Grant, I. H. Chromatographia 1987, 24, 135. (5) Pfeffer, W. D.; Yeung, E. S. J. Chromatogr. 1991, 557, 125.
(6) Tsuda, T. Anal. Chem. 1987, 59, 521-3. (7) Tsuda, T. Anal. Chem. 1988, 60, 1677-80. (8) Tsuda, T. LC-GC Int. 1992, 5, 26-36. (9) Verheij, E. R.; Tjaden, U. R.; Niessen, W. M. A.; van der Greef, J. J. Chromatogr. 1991, 554, 339-49. (10) Tsuda, T.; Kitagawa, S. In Electric Field Applications in Chromatography, Industrial and Chemical Processes; Tsuda, T., Ed.; VCH: Weinheim, 1995; pp 13-45. (11) Boughtflower, R. J.; Underwood, T.; Paterson, C. J. Chromatographia 1995, 40, 329-35. (12) Tsuda, T. J. Liq. Chromatogr. 1989, 12, 2501. (13) Li, S.; Lloyd, D. K. J. Chromatogr., A 1994, 666, 321.
Both electrophoretic and chromatographic transport mechanisms are combined in electrochromatographic separation. In this paper, we developed a model of enantioselectivity in capillary electrochromatography (CEC) which can be applied in the separation of both neutral and ionic compounds. The overall selectivity in enantioseparation is considered to be made up of two contributions: one is the intrinsic difference in formation constants of a pair of enantiomers, and the other is the conversion efficiency of the intrinsic difference into the apparent difference in the migration velocity. The model was illustrated through the chiral separation of (R)- and (S)-salsolinols. Under a positive electric field, enantioseparation of salsolinols was achieved on an ODS column with β-cyclodextrin as a chiral mobile-phase additive. The experimental results are discussed in relation to the effect of separation parameters, such as the direction and size of electric field and properties of the stationary and mobile phases. It was demonstrated that if both electrophoretic and partitioning mechanisms produce positive effects, high overall selectivity in CEC can be obtained. For pressurizeddriven electrochromatography, although the column efficiency is sacrificed due to the introduction of hydrodynamic flow, the increased selectivity significantly reduced the requirement of large column plate numbers for resolution.
4586 Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
10.1021/ac980366i CCC: $15.00
© 1998 American Chemical Society Published on Web 10/07/1998
The first electrochromatographic experiments were done in early 1974 by Pretorius,14 who applied an electric field across a packed column (50 cm × 1 mm i.d.). More recent pioneer work on CEC was reported by the Tsuda and Jorgenson groups, in which the mobile phase was driven through a open-tubular octadecylsilane capillary column (i.d. 30 µm) 2 and a packed capillary column3 by electroosmotic flow or mainly pressurized flow.8 During the past few years, interest in CEC has increased due to the improvement in the preparation of capillary columns11,15,16 and in the stability and efficiency of separations.17-21 Recent interviews22 with several active scientists in separation science reveal that CEC has a great potential in separation technology, but a more fundamental understanding of separation mechanism is needed. Chiral separation in capillary electrophoresis is usually achieved by the addition of chiral complexing agents to form in situ diastereometric complexes between the enantiomers and the chiral complexing agent. Many of the chiral selectors successfully used in HPLC23 can also be applied in CE, and thus the experience from both HPLC and CE can be transferred to CEC. For chiral separation in CEC, just a few were achieved with chiral packed or coated capillary column, but enantioseparation by the use of a chiral mobile phase was not reported. The aim of this study is to develop models for separation selectivity in CEC, especially in the chiral CEC separations. Several groups developed some theoretical models for chiral separation in CE.24-27 Lelievre et al.28,29 defined an intrinsic selectivity as the ratio of the affinity factors of both analytes with a separating agent (such as stationary phase, pseudophase, or complexing agents), but it is difficult to relate intrinsic selectivity to resolution. CEC is a more complicated system than CE and HPLC due to the combination of both electrophoretic and chromatographic transport mechanisms. The model developed here can illustrate the effect of separation parameters on the selectivity and can be applied in the CEC separation of both neutral and ionic compounds. Also we are interested in exploring the potential advantages offered by capillary electrochromatography and, in particular, its practical utility for enantioseparation. A chiral separation of the (R)- and (S)-enantiomers of the neurotoxin salsolinol, which is considered to induce Parkinson’s disease and (14) Pretorius, V.; Hopkins, B. J.; Schieke, J. D. J. Chromatogr. 1974, 99, 2330. (15) Kitagawa, S.; Inagaki, M.; Tsuda, T. Chromatography (Kuromatoguraphi) 1993, 14, 39R-43R. (16) Yan, C. U.S. Patent 5453163, 1995. (17) Knox, J. H.; Grant, I. H. Chromatographia 1991, 32, 317-28. (18) Smith, N. W.; Evans, M. B. Chromatographia 1994, 38, 649-57. (19) Dittmann, M. M.; Rozing G. P. J. Chromatogr., A. 1996, 744, 63-74. (20) Taloy, M. R.; Teale, P.; Westwood, S. A.; Perrett, D. Anal. Chem. 1997, 69, 2554-8. (21) Wu, J.; Huang, P.; Li, M. X.; Qian, M. G.; Lubman, D. M. Anal. Chem. 1997, 69, 320-6. (22) Majors, R. E. LC-GC 1998, 16, 12-3. (23) Deng, Y.; Maruyama, W.; Kawai, M.; Dostert, P.; Naoi, M. In Multielectrode EC Array Application to Trace Analysis of Neurotransmitters, Pharamaceuticals and Environmental Markers, Progress in HPLC and HPCE; VSP: Utrecht, 1997; Vol. 6, pp 301-38. (24) Wren, S. A. C.; Rowe, R. C. J. Chromatogr. 1992, 603, 235-41. (25) Wren, S. A. C.; Rowe, R. C. J. Chromatogr. 1992, 609, 363-7. (26) Wren, S. A. C.; Rowe, R. C. J. Chromatogr. 1993, 635, 113-8. (27) Britz McKibbin, P.; Chen, D. D. Y. J. Chromatogr., A 1997, 781, 23-34. (28) Lelievre, F.; Gareil, P.; Jardy, A. Anal. Chem. 1997, 69, 385-92. (29) Lelievre, F.; Gareil, P.; Bahaddi, Y.; Galons, H. Anal. Chem. 1997, 69, 393401.
occurs in human brain,30,31 was used to illustrate the features of an PD-CEC system. THEORETICAL MODEL Resolution and Selectivity in CEC. For CEC systems (including pressurized flow), the apparent mean linear flow velocity of an analyte, vapp, is
vapp ) vpres + veo + vep
(1)
with veo ) µeoE, and vep ) µepE, where E is electric field strength, vpres, veo, and vep are the mean linear velocity of pressurized, electroosmotic, and electrophoretic flow, respectively and µeo and µep are electroosmotic and electrophoretic mobilities, respectively. Values of veo and vep or µeo and µep can be positive or negative. Elution time (te) can be calculated by the following equation: 10
te ) L(1 + k′)/vapp
(2)
where L is column length and k′ is the capacity factor. Since the capacity factor is a definition only for the partition between stationary and mobile phases, it is necessary to find an analogical parameter that is representative of the contribution of both partition and electrophoresis. For this, we propose to define a migration factor (ke) as the ratio of the elution time (te) of analyte to the elution time (t0) of an unretained neutral compound (i.e., usually the elution time of the solvent peak); i.e., t0 ) L/(vpres + veo) ) L/v0. By substituting t0 into the definition equation, it gives
ke )
te v0(1 + k′) ) t0 (v0 + vep)
(3)
As in the case of electrophoresis,32,33 an apparent selectivity (Rapp) in CEC can be defined as the ratio of the elution times or the migration factors of a pair of analytes.
Rapp )
te2 ke2 (1 + k′2) (v0 + vep1) ) ) te1 ke1 (1 + k′1) (v0 + vep2)
(4)
where the subscripts 1 and 2 indicate a pair of analytes. From eq 4, it is very clear that the selectivity in CEC is determined by the difference between k′1 and k′2 and the difference between vep1 and vep2. The defined selectivity takes into account all of the separation parameters, such as electrophoretic mobility and the nature of both stationary and mobile phases. However, in eq 4, the effectiveness of k′ or vep on selectivity is determined by some nonselective factors, such as the flow rate of the mobile phase, v0, which is given by (vpres + veo). (30) Deng, Y.; Maruyama, W.; Dostert, P.; Takahashi, T.; Kawaai, M.; Naoi, M. J. Chromatogr., B 1995, 670, 47-54. (31) Deng, Y.; Maruyama, W.; Kawai, M.; Dostert, P.; Yamamura, H.; Takahashi, T.; Naoi, M. J. Chromatogr., B 1997, 689, 313-20. (32) Baumy, P.; Morin, P.; Dreux, M.; Viaud, M. C.; Boye, S.; Guillaumet, G. J. Chromatogr., A 1995, 707, 311-26. (33) Chankvetadze, B.; Endresz, G.; Blaschke, G. Electrophoresis 1994, 15, 8047.
Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
4587
It is difficult to define an effective selectivity (separation factor) as in the case of general chromatography or general electrophoresis because the selectivity in CEC is produced from a cooperative interaction rather than the simple addition of partition and electrophoresis interaction. To better illustrate the interactions that control selectivity, one can define a relative selectivity, Rr, as the relative difference in the elution times or the migration factors between a pair of analytes.
Rr ) ∆te/te2 ) ∆ke/ke2
(5)
with ∆te ) te2 - te1, ∆ke ) ke2 - ke1. By combining eqs 3 and 5 and replacing vep ) µepE, it gives
Rr )
(k′2 - k′1)v0 + (µep1 - µep2)E + (µep1k′2 - µep2k′1)E (1 + k′2)(v0 + µep1E)
(6)
Equation 6 shows that the numerator of the overall selectivity in CEC is made up of three terms: a capacity term given by (k′2 k′1) v0, a mobility term by (µep1 - µep2)E, and an union term by (µep1k′2 - µep2k′1)E. The first two terms are the simple differences in capacity factors and in electrophoretic mobility, respectively, and the last term indicates the cooperative interaction between partition and electrophoresis. Generally, an analyte with a larger capacity factor (k′) in a reversed-phase system has a lower electrophoretic mobility (µep). Thus if the electrophoretic mobility acts in the same direction as v0, all of three terms in eq 6 should be positive and result in higher selectivity than capillary CE or HPLC. For pressure-driven electrochromatography, it is easy to manipulate the direction of electrophoretic mobility just by changing the direction of the electric field, while for electroosmotically driven electrochromatography, the direction of electroosmotic mobility can be manipulated using appropriate additives. From the basic definition of resolution in chromatography and the definition of N(16te2/W2), the resolution equation (Res) for a pair of analytes can be rearranged as follows:
2(te2 - te1) xN te2 - te1 xN Res ) R ) ) W2 + W 1 4 te2 4 r
(7)
Res )
xN ∆µepE 4 vapp
and when a pair of analytes are neutral (µep1 ) µep2 ) 0), the following expression is obtained from eq 8 if the same definition of R as in chromatography is used in this case:
Res )
k′2 xN k′2 - k′1 xN (R - 1) ) 4 1 + k′2 4 R (1 + k′2)
Res )
xN (k′2 - k′1)v0 + (µep1 - µep2)E + (µep1k′2 - µep2k′1)E 4 (1 + k′2)(v0 + µep1E) (8) In two particular situations, unretained compounds and neutral compounds, eq 8 can be further simplified. When the analytes are unretained on the stationary phase (k′1 ) k′2 ) 0), eq 8 becomes 4588 Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
(10)
Note that eq 9 is the same expression of resolution as in general electrophoresis and eq 10 is the same equation as in general chromatography. Enantioseparation by CEC with a Chiral Column. In this case, a chiral complexing agent is bonded on the stationary phase. The distribution of an analyte onto the stationary phase is characterized by complex formation of the analyte (Q) with a chiral complexing agent (C), and is considered as a simple situation where an analyte interacts with a single complexing agent:
Qm + Cs h QCs Qm h Qs
Kf ) [QC]s/[Q]m[C]s
(11)
K ) [Q]s/[Q]m
(12)
where the subscripts “m” and “s” indicate the mobile and stationary phases, respectively. Cs is the stationary phase chiral site and thus [C]s can be considered as 1. Kf is the formation constant of complex, and the difference in the formation constants between a pair of enantiomers is required for the chiral recognition. Equation 12 refers to nonstereospecific interactions, and K is the equilibrium constant of free analyte between the mobile and stationary phases. It is also possible to describe such a system by an apparently partitioning equilibrium.
Qm h Q′s
K′ ) [Q′]s/[Q]m
(13)
with [Q′]s ) [Q]s + [QC]s. Combining eqs 11-13, the apparent capacity factor (k′app) can be expressed as
k′app ) φK′app ) φ(Kf + K) where W1 and W2 are the bandwidths of a pair analytes and N is the theoretical plate number. Equation 7 shows that the overall resolution is determined by a column efficiency term and a relative selectivity term. Substituting eq 6 into eq 7 gives a more complete expression for the resolution:
(9)
(14)
where φ is the phase ratio of column. Substituting eq 14 into eq 6 and considering µep1 ) µep2 and K1 ) K2 for a pair of enantiomers (noted as subscripts 1 and 2), an expression of the relative selectivity for enantioseparation by CEC is obtained:
Rr )
φ(Kf2 - Kf1) 1 + φK2 + φKf2
(15)
Equation 15 indicates that the chiral selectivity in CEC with a chiral column is dependent on the difference in the formation constants of enantiomers with chiral complexing agents. Thus, in this case, the electrophoresis mechanism does not influence the enantioselectivity and the electric field only plays a role in driving the mobile phase.
Enantioseparation in CEC with Chiral Additives. In the presence of a chiral complexing agent, such as cyclodextrin, the following equilibrium take place when the complex between an analyte and a chiral additive is assumed to be 1:1 stoichiometry, which is usually appropriate if the complexing agent is cyclodextrin.
Qm + Cm h QCm
Kf
) [QC]m/[Q]m[C]m (16)
Qm h Qs
K
) [Q]s/[Q]m
(17)
Cm h Cs
KC ) [C]s/[C]m
(18)
KQC ) [QC]s/[QC]m
(19)
QCm h QCs
where K, KC, and KQC are the partitioning constants of free analyte, free complexing agent, and complex compound between the mobile and stationary phases, respectively. Usually the large excess of the complexing agent is used in the mobile phase, and thus eqs 18 and 19 can be considered to be negligible according to Walhagen.34 An apparent partitioning equilibrium, K′app, can be also described by
Q′m h Qs
K′ ) [Q]s/[Q′]m
(20)
with [Q′]m ) [Q]m + [QC]m. Thus the apparent capacity ratio (k′app) based on eqs 16, 17, and 20 can be expressed as
k′app ) φK′app )
φK 1 + Kf[C]m
(21)
The apparent electrophoretic mobility of the analyte will be related to both the time when analyte is free and when it is complexed24 and, therefore,
µep )
µf + µcKf[C]m 1 + Kf[C]m
(22)
where µf and µc are the electrophoretic mobilities of the free analyte and the analyte-chiral selector complex, respectively. For a pair of enantiomers, noted as subscripts 1 and 2, the partitioning constants and the electrophoretic mobilities should be essentially the same between their free forms and between their complexed forms; i.e., K1 ) K2, µf1 ) µf2, and µc1 ) µc2. By combining eqs 21 and 22 with eq 6, an expression for the enantioselectivity in CEC with chiral mobile phase is obtained:
Rr )
(Kf1 - Kf2)[φKvc + (µc - µf)E][C]m (1 + φK + Kf2[C]m)(vf + vcKf1[C]m)
(23)
with vf ) vpres + veo + µfE and vc ) vpres + veo + µcE, where vf and vc are the apparent flow velocity of the free analyte and the complexed analyte, respectively. In CEC, as in HPLC, the enantiomer migration order for a given chiral selector as a mobilephase additive will be opposite to that for the same chiral selector (34) Walhagen, A.; Edholm, L. E. Chromatographia 1991, 32, 215.
as a stationary phase; complex formation with the chiral selector in the mobile phase accelerates the migration of analyte, while complex formation with the chiral selector on the stationary phase decelerates it. Therefore, generally in eq 15, Kf2 > Kf1, while in eq 23, Kf1 > Kf2. As expected, eq 23 shows that the enantioselectivity is determined predominantly by the difference in the formation constants and the equilibrium concentration of a chiral selector, and to some extent also by the difference in the electrophoretic mobility between the free and complexed enantiomers. Unlike electrophoresis,24 even if µc ) µf, the resolution or the selectivity in CEC is not zero, indicating that the chiral separation is still possible. However in this case, the electric field will not affect the chiral selectivity and eq 23 becomes
Rr )
φK(Kf1 - Kf2)[C]m (1 + φK + Kf2[C]m)(1 + Kf1[C]m)
(24)
In eq 23, if K ) 0 (unretained on the stationary phase), an equation similar to that derived for electrophoresis by Wren and Rowe24 is obtained:
Rr )
(Kf1 - Kf2)(µc - µf)E[C]m (1 + Kf2[C]m)(vf + vcKf1[C]m)
(25)
In this case, it is essential for chiral separation that the electrophoretic mobility be different between the free and complexed enantiomers (µf * µc). As for other separation systems, eq 23 shows that the enantioselectivity will be zero if the equilibrium concentration of the chiral selector is zero or very large, indicating that there must exist a maximum selectivity at the optimal concentration of chiral selector. The optimal concentration ([C]opt) occurs when dRr/ d[C]m ) 0, that is, (Kf1 - Kf2) [φKvc + (µc - µf)E][(1 + φK)vf vcKf1Kf2[C]m2] ) 0, and thus the resulting expression for [C]opt is
[C]opt ) x(1 + φK)/Kf1Kf2
(26)
In eq 26, it was assumed that the ratio of the apparent flow velocities of the free and complexed analytes is approximately 1 (vf/vc ≈ 1). It is interesting to note that eq 26 is a little different from the one that was derived for electrophoresis by Wren and Rowe24 and that the optimal concentration in eq 26 is not only dependent on the formation constants (Kf1, Kf2) of enantiomers with the chiral selector but also on properties of the column (φ and K). EXPERIMENTAL SECTION Materials. β-Cyclodextrin (CyD) was purchased from Nacalai Tesque (Kyoto, Japan). The standards of optically pure (R)- and (S)-salsolinols were donated by Dr. Makoto Naoi (Bioscience Department of Nagoya Institute of Technology, Nagoya, Japan). Racemic salsolinol (1-methyl-6,7-dihydroxy-1,2,3,4-tetrahydroisoquinoline) and sodium 1-heptanesulfonate (SHS) were obtained from Sigma (St. Louis, MO). All other chemicals were of analytical grade and organic solvents were of HPLC grade from Nacalai Tesque. Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
4589
Aqueous standard solutions of salsolinol enantiomers, prepared at a concentration of 10 mM, were stable for at least 6 months when stored at -20 °C. The 100 µm standard solution was prepared fresh daily from the 10 mM solution. Salsolinol is a neurotoxin specific to dopamine neurons. Although it has been considered not to be transported into brain through the bloodbrain barrier, precaution for the use of salsolinol is necessary (e.g., wearing gloves). Pressure-Driven CEC System. The schematic diagram of the CEC instrumentation is the same as described in the ref 8. The pressure-driven chromatographic system consists of a pump (LC-6A, Shimadzu, Kyoto, Japan), an injector (0.5 mL, Rheodyne 7410, Cotati, CA), a homemade column (which was packed with ODS-silica), a UV detector (CE-970, Jasco, Hachiooji, Tokyo, Japan), and a high-voltage power supply (maximum 30 kV, Type HCZE-30p, Matsusada Precision, Ltd., Kusano, Shiga, Japan). Split injection was employed, and a fused-silica capillary (50-µm i.d., G-L Science, Tokyo, Japan) was used as a resistance tube. The high voltage was applied as follows: the body of the splitter (inlet) was grounded, and a high voltage (5-10 kV) was applied to the outlet of the capillary by means of a Pt electrode; the sign given to the applied voltage refers to the voltage at the detection end. For the column packing, an end frit 0.1-0.5 mm in length was first prepared at 5 cm from the outlet for the ∼25-cm-long capillary (75-mm i.d., 365-mm o.d.). This was accomplished by heating the H2O/K2SiO3 solution (containing approximately 27-29% SiO2, Wako Pure Chemical Ltd., Osaka, Japan). Silica gels modified with ODS (ODS 2, 5 µm, Rainin Instrument Co. Inc., Woburn, MA) were slurried in ethanol/water (80/20) and homogenized in an ultrasonic bath for 10 min. The slurry was then placed into a stainless steel reservoir (15 cm × 0.8 mm i.d.) and pumped into the frit-treated capillary at 150 kg/cm2 using a Shimadzu pump (LC-6A). During this process, both the reservoir containing packing materials and the capillary were immersed in an ultrasonic water bath. Chiral Separation Using CyD as a Chiral Mobile Additive. The mobile phase for the enantiomeric separation of salsolinol was 20 mM sodium phosphate buffer (pH 3.0) containing 12 mM β-cyclodextrin/5 mM sodium 1-heptanesulfonate. Other mobilephase compositions used for the study on the effect of separation parameters on chiral separation are described in the figure legends and in the Results and Discussion section. The pump was operated at a constant-pressure mode of 100 kg/cm2. On-column detection was measured by a spectrophotometer at 210 nm. Our previously described method35 was used to estimate the apparent formation constants of (R)- and (S)-salsolinols with β-cyclodextrin. The retention of salsolinol enantiomers on the ODS capillary column was examined at various concentrations of β-cyclodextrin in the mobile phase of 20 mM sodium phosphate buffer (pH 3.0) containing 4 mM sodium 1-heptanesulfonate without an electric field. The reciprocal of the capacity ratios was plotted against the concentration of β-cyclodextrin, and the ratio of slope over intercept provides directly a value for Kf. The retention time of the first peak (solvent peak) was used as the holdup time (t0). Data represent the average of four or more analyses. (35) Deng. Y.; Maruyama, W.; Dostert, p.; Yamamura, H.; Kawai, M.; Naoi, M. Anal. Chem. 1996, 68, 2826-31.
4590 Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
Figure 1. Electrochromatogram of salsolinol enantiomers on a packed capillary column: column, ODS-C18, 29 cm (23-cm effective length) × 75 µm i.d.; applied electric field strength, ∼250 V/cm; mobile phase, 20 mM sodium phosphate buffer (pH 3.0) containing 12 mM β-cyclodextrin/5 mM sodium 1-heptanesulfonate. The pump was set at the constant pressure of 100 kg/cm2.
RESULTS AND DISCUSSION SECTION Enantioseparation of Salsolinols. Salsolinol, which has been considered to be an endogenous neurotoxin in human brain, has an asymmetric center at C1 and exists as (R)- and (S)-enantiomers. Several reports demonstrated that the chiral properties of salsolinol and its derivatives may play an important role in their neurotoxicity, which is specific to dopamine neurons.36,37 Therefore, it is quite important to determine the enantiomeric composition of salsolinol in biological samples, such as brain tissues. Figure 1 shows that the separation of racemic salsolinol is achieved by the use of CEC with 12 mM β-cyclodextrin as the mobile-phase additive in 20 mM sodium phosphate buffer (pH 3.0) containing 5 mM sodium 1-heptanesulfonate. The applied electric field strength is ∼+250 V/cm. Salsolinol is a hydrophilic amine and exists as a cation under acidic conditions. Usually salsolinol has small k′ values on the reversed stationary phases and thus sodium 1-heptanesulfonate is used as a counterion to improve the retention.35 The enantiomer migration order was identified by the authentic standards of (R)- and (S)-salsolinols, indicating that the (R)-enantiomer of salsolinol with the shorter elution time has the greater affinity for β-CyD. Effect of the β-CyD Concentrations on Enantioselectivity. As predicted by eq 23, the experimental results in Figure 2 shows that the overall enantioselectivity for salsolinol enantiomers in pressure-driven electrochromatography is dependent on the CyD concentration and passes through a maximum value. Due to the limited solubility of β-CyD, the examined concentrations are only up to 15 mM, and the maximum concentration has already passed the optimal value for enantioseparation of salsolinol enantiomers. (36) Takahashi, T.; Deng, Y.; Maruyama, W.; Dostert, P.; Kawai, M.; Naoi, M. J. Neural Transm. 1994, 98, 107-18. (37) Naoi, M.; Maruyama, W.; Dostert, P.; Hashizume, Y.; Nakahara, D.; Takahashi, T.; Ota, M. Brain Res. 1996, 709, 285-95.
Figure 2. Dependence of the overall enantioselectivity for salsolinol enantiomers upon the equilibrium concentration of β-cyclodextrin. Solid line represents the experimental data, and the separation conditions are as follows: column, ODS-C18, 20 cm (17 cm effective length) × 75 µm i.d.; applied electric field strength, ∼250 V/cm; mobile phase, 20 mM sodium phosphate buffer (pH 3.0) containing 12 mM β-cyclodextrin/4 mM sodium 1-heptanesulfonate. Dashed line represents the theoretical predication that was calculated from eq 23 by substituting φK ) 11.76, µc ) -6.7 × 10-5 cm2 V-1 s-1, µf ) -8.1 × 10-4 cm2 V-1 s-1, Kf1 ) 189.2, and Kf2 ) 165.7.
The experimental data for the relative selectivity of salsolinol enantiomers are in reasonable agreement with the derived mathematical model, as shown in Figure 2. The theoretical curve was generated by the substitution of some plausible values for the parameters in eq 23. The apparent formation constants of salsolinol enantiomers with β-cyclodextrin were estimated under the same conditions as used in Figure 2, and the values are 189.2 ( 17.1 and 165.7 ( 9.8 for (R)- and (S)-salsolinols in 20 mM phosphate buffer containing 4 mM sodium 1-heptanesulfonate. The value of φK, which is only related to the distribution of free salsolinol, was calculated to be 11.76 according to eq 21 by the substitution of the experimental data when the applied electric field strength was zero. The electrophoretic mobilities and apparent flow velocities are -8.1 × 10-4 cm2 V-1 s-1 and 0.343 cm/s for the free salsolinol, and -6.7 × 10-5 cm2 V-1 s-1 and 0.528 cm/s for the complexed salsolinol, respectively; these were obtained from the experimental results when an electric field strength of + 250 V/cm was applied. Effect of Electric Field Strength on Enantioseparation. The dependence of the chiral separation of salsolinol enantiomers on the electric field strength was examined from 0 to +7.0 kV, as shown in Figure 3. As a positive electric field was applied (positive at detection end), the enantioseparation was improved. It is clearly shown that the separation at 5 kV in Figure 3 is much better than that at 0 kV (the corresponding capillary LC separation). The difference in the separation between Figures 1 and 3 (at 5 kV) was caused by the use of two different homemade columns and the use of a lower concentration of sodium 1-heptanesulfonate in the mobile phase in Figure 3. As mentioned above, the selectivity in CEC is determined by both partition and electrophoresis, and the electric field either increases the selectivity or decreases the selectivity. Figure 4 shows the effect of an electric field on the enantioselectivity of salsolinols. The theoretical curve (dash line in Figure 4) for the enantioselectivity of salsolinols under positive and negative electric
Figure 3. Chromatograms of salsolinol enantiomers obtained by pressure-driven electrochromatography at different voltages: column, ODS-C18, 20 cm (17 cm effective length) × 75 µm i.d.; mobile phase, 20 mM sodium phosphate buffer (pH 3.0) containing 12 mM β-cyclodextrin/4 mM sodium 1-heptanesulfonate. The pump was set at the constant pressure of 100 kg/cm2.
Figure 4. Dependence of the overall enantioselectivity for salsolinol enantiomers upon electric field strength. Dashed line represents the calculated values from eq 23 by substituting the same values as used in Figure 2. Solid line represents the experimental data, and the separation conditions are the same as in Figure 3. Open circles (calculated) and filled circles (experimental) indicate that a positive electric field was applied, and open diamonds (calculated) and filled diamonds (experimental) indicate that a negative electric field was applied.
fields was obtained by the substitution of the same values into the parameters in eq 23 as used in Figure 2. The experimental and calculated results show that increasing the positive voltage improved the selectivity for salsolinol enantiomers, while increasing the negative voltage reduced the enantioselectivity. As shown in Figure 4, there is a bigger discrepancy between the theoretical and experimental lines for the negative voltage. In reality, enantioseparation of salsolinols was not obtained with negative electric field strength, and the experimental data for negative voltage were obtained by injection of the individual enantiomers of salsolinol. Thus, the direction of the electric field is an important parameter in CEC, and we cannot conclude from Figure 4 that a positive electric field is always beneficial to the overall Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
4591
Table 1. Relationship between the Field Strength and the Electrophoretic Mobility for Getting High Enantioselectivity in CEC direction of µep µf
µc
+ + + -
+ + +
size relationship (absolute value) µ f < µc µf > µc a µf < µc µf > µc a
direction of E + + +
a The selection of direction of electric field is not influenced by size relationship in absolute values between the electrophoretic mobility of the free and complexed analytes.
enantioselectivity in CEC. According to eq 23, enantioselectivity is dependent both on the direction of the electric field and on the direction and size of electrophoretic mobility between the free and complexed analyte. In Figure 4, the electrophoretic mobilities are assumed to be -8.1 × 10-4 and -6.7 × 10-5 cm2 V-1 s-1 for the free and complexed salsolinol, respectively, i.e., |µf| > |µc|, and hence in eq 23, (µc - µf)E > 0 when a positive field strength is applied. For many other situation, it is possible that |µf| < |µc|, and negative electric field will result in positive effect on the enantioselectivity. Table 1 summarizes the relationship between the direction of field strength and the electrophoretic mobility of the free and complexed analytes. For PD-CEC, solvent is mainly driven by pressurized flow, and there is no limitation to change the direction of the electric field. However, for ED-CEC, the direction of the electric field is limited because the mobile phase is driven electroosmotically. As mentioned above, a possible but probably difficult approach is that appropriate additives or buffer ions in the mobile phase can be applied to change the direction of electroosmotic flow. Contribution of OK to Enantioselectivity. The value of φK for a given analyte is determined by the separation column and mobile phase. The parameter K is the equilibrium constant of free analyte between the stationary and mobile phases. While φK is nonselective for two enantiomers, when a chiral selector is added in the mobile phase, the formation of a transient complex between analyte and chiral selector probably differentiates the apparent capacity ratios (k′app) between two enantiomers as shown in eq 21, k′app ) φK/(1 + Kf[C]m), and thus contributes to the enantioselectivity. The calculated dependence of the optimal concentration of β-CyD and the corresponding maximum selectivity on φK is shown in Figure 5 which is generated from eq 23 by substituting µc ) -6.7 × 10-5 cm2 V-1 s-1 and µf ) -8.1 × 10-4 cm2 V-1 s-1, and Kf1 ) 189.2 and Kf2 ) 165.7. With the increase of φK, the optimal concentration of β-CyD becomes higher, while the maximum enantioselectivity reaches a plateau. Most theoretical models of chiral stationary phases suggest that any nonstereospecific interaction with the stationary phase is detrimental to the chiral separation. Comparing eq 15 to eq 23, it can be seen that this conclusion from chiral stationary phases does not contradict our model here. For enantioseparation with a chiral stationary phase, nonstereospecific interactions, expressed as φK, contribute only to the denominator and thus reduce the enantioselectivity as shown in eq 15. However, for separation with 4592 Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
Figure 5. Dependence of the optimal concentration of β-CyD and the corresponding maximum selectivity upon φK. The results were generated from eqs 23 and 26 by substituting µc ) -6.7 × 10-5 cm2 V-1 s-1, µf ) -8.1 × 10-4 cm2 V-1 s-1, Kf1 ) 189.2, and Kf2 ) 165.7. Table 2. Optimal Concentration of β-CyD and the Correspondingly Required Plate Number at Various Values of OK φK
Copta
Rr(max)b
N requiredc
0 2.5 5.0 7.5 10.0 12.5 15.0
4.6 8.5 11.2 13.3 15.1 16.8 18.3
0.0142 0.0514 0.0648 0.0724 0.0776 0.0814 0.0844
178464 13611 8575 6859 5976 5430 5055
a Calculated from eq 26 by substituting K ) 189.2 and K ) 165.7. f1 f2 Calculated from eq 23 by substituting µc ) -6.7 × 10-5 cm2 V-1 s-1, -4 2 -1 -1 µf ) -8.1 × 10 cm V s , Kf1 ) 189.2, and Kf2 ) 165.7. c For the enantioseparation with Res ) 1.5.
b
a chiral mobile phase, φK appears in both numerator and denominator in eq 23. The partitioning mechanism is important for differentiating the migration of enantiomers even though the partitioning process to achiral stationary phase itself is nonstereospecific to enantiomers. As shown in Figure 5, a suitable φK is advantageous to improvement of enantioselectivity. Table 2 represents the optimal concentration of β-CyD and the correspondingly required plate number for the resolution of 1.5 at various values of φK. The value of zero for φK corresponds to conventional capillary electrophoresis. It is interesting to compare the enantioselectivity in the case of φK ) 0 with that in the case of φK * 0. For the chiral separation of salsolinols using β-CyD as a chiral selector, if conventional capillary electrophoresis is used, a plate number of 178 464 will be required for the resolution of 1.5, while if φK ) 10, i.e., in the case of CEC, the required plate number will be only 5976 for the same resolution. For PDCEC, some column plate number is sacrificed due to the introduction of hydrodynamic flow, but the increased selectivity markedly reduces the requirement for the column efficiency. Effect of Formation Constants on Enantioselectivity. Substantially, chiral recognition of enantiomers is directly the result of the transient formation of diastereomeric complex between enantiomeric analyte and chiral complexing agent, i.e., the difference in formation constants of transient complexes between a pair of enantiomers with the chiral agents. Both electric field and achiral column themselves are nonselective for enantiomers, but their importance lies in the fact that these exogenous
constants between a pair of enantiomers, the larger the relative enantioselectivity, as shown in Figure 6A. If the ratio of formation constants between a pair of enantiomers is kept while varying the formation constants, as in the case of Figure 6B, it was found that the resulting maximum selectivity is the same and the lower optimal concentration of chiral selector corresponds to the larger formation constants. By substituting eq 26 into eq 23, an expression of dependence of the relative selectivity upon the ratio Kf1/Kf2 rather than the absolute values of formation constants can be obtained. The important result indicates that the maximum selectivity corresponding to the optimal concentration of chiral selector is dependent on the difference in formation constants rather than their absolute values. However, the large formation constants are also practically significant in CEC since increasing the formation constants is of great advantage for decreasing the optimal concentration of chiral selector.
Figure 6. Dependence of the overall enantioselectivity upon the equilibrium concentration of chiral selector at various formation constants. The theoretical curves were generated from eq 23 by substituting φK ) 11.76, µc ) -6.7 × 10-5 cm2 V-1 s-1, and µf ) -8.1 × 10-4 cm2 V-1 s-1. (A) the difference in formation constants between a pair of enantiomers with chiral selector is increased from 160/150 (Kf1/Kf2) to 180/150; (B) the formation constants are varied at a constant ratio of Kf1 to Kf2.
factors can convert the intrinsic difference in formation constants of a pair of enantiomers into the apparent difference in their migration velocities along the column. Therefore, the overall selectivity in chiral separation can be considered to be made up of two contributions: one is the intrinsic difference (intrinsic selectivity, represented by the first term in the numerator of eq 23) in formation constants of a pair of enantiomers, and the other is the conversion efficiency (exogenous selectivity, represented by the second term in the numerator of eq 23) of the intrinsic difference into the apparent difference in the migration velocity. Perhaps it would be argued that the conversion efficiency should be considered to be included in the column efficiency. It is wellknown that the column efficiency is a concept in relation to the band broadening. It is clear that the concept of column efficiency is difficult to elucidate some phenomena, such as the marked effect of the direction of electric field on the chiral separation in CEC. Our experimental results described above have demonstrated that enantioselectivity was influenced by the direction and size of the electric field and by the properties of the stationary and mobile phases. As shown in eq 23, the difference in formation constants between a pair of enantiomers is critically important for the chiral separation. Figure 6 generated from eq 23 by substituting φK ) 11.76, and µc ) -6.7 × 10-5 cm2 V-1 s-1 and µf ) -8.1 × 10-4 cm2 V-1 s-1, represents the dependence of the overall enantioselectivity upon the equilibrium concentration of β-CyD at various formation constants. The larger the difference in formation
CONCLUSION A theoretical model in relation to selectivity and resolution was developed for CEC and further applied for the chiral separation. The overall selectivity in chiral separation is determined by the intrinsic difference in formation constants and the conversion efficiency of the intrinsic difference into the apparent difference in the migration velocity. The intrinsic difference in formation constants is critical, but the experimental factors, such as electric field or the stationary and mobile phases, can also contribute to the improvement of the overall enantioselectivity by increasing the conversion efficiency. When both electrophoretic and partitioning mechanisms acted positively, high overall enantioselectivity for (R)- and (S)-salsolinols in CEC was obtained. An optimal β-CyD concentration for the separation of salsolinol enantiomers was demonstrated, and the calculated results from the theoretical model show that the maximum selectivity corresponding to the optimal concentration is dependent on the difference in formation constants rather than their absolute values. However, the large formation constant is also important for obtaining the low optimal concentration of chiral selector. The theoretical model shows clearly that the benefit of combining electrophoresis and partitioning mechanisms in CEC is the increase of selectivity for the separation. For PD-CEC, although the column efficiency is sacrificed due to the introduction of hydrodynamic flow, the increased selectivity significantly reduces more markedly the requirement of high column plate numbers for peak resolution. In addition, an HPLC pump provides the stable flow conditions and thus provides the possible improvements in reproducibility, in sample introduction, and in overcoming the formation of air bubbles. ACKNOWLEDGMENT We are grateful to the Natural Science and Engineering Research Council of Canada, Health Services Utilization and Research Commission of Saskatchewan and the Education Ministry of China (JL98-06) for financial support on certain portions of this research.
Received for review April 1, 1998. Accepted August 25, 1998. AC980366I Analytical Chemistry, Vol. 70, No. 21, November 1, 1998
4593
