Surfactant Gradient Methods Using Mixed Systems of

Anal. Chem. 2006, 78, 8142-8149

Surfactant Gradient Methods Using Mixed Systems of Cetyltrimethylammonium Chloride and Nonionic Surfactants Possessing Polyoxyethylene Chains for Electrokinetic Separation of Benzoate Anions as Model Analytes Yukihiro Esaka,* Mika Sawamura, Hiroya Murakami, and Bunji Uno

Gifu Pharmaceutical University, 5-6-1 Mitahora-higashi, Gifu 502-8585, Japan

Surfactant gradient methods for electrokinetic separation of 10 benzoates as model organic anions were investigated using mixed micellar solutions of cetyltrimethylammonium chloride (CTAC) and nonionic surfactants possessing polyoxyethylene chains, polyoxyethylene sorbitan monolaurate (Tween 20) or polyoxyethylene lauryl ether (Brij 35). Electroosmotic flow (EOF) was eliminated virtually by a coating of the inner wall of the capillaries, and then the benzoates were detected fundamentally in the order of their hydrophobicity. In a pure CTAC system, the synergistic influences of attractive electrostatic and hydrophobic interactions gave rise to quite large retention factors of many of the benzoate anions, resulting in their coelution. Addition of an adequate amount of Tween 20 to the pure CTAC system decreased the electrostatic interaction significantly to give remarkably improved separation of the analytes, but long analysis time was required. A surfactant gradient method would be useful to decrease analysis time and to improve separation simultaneously. Under slight EOF, the micelles in the inlet reservoir can pass through and, thus, interact with all of the analytes before they were detected. In the present system, surfactant gradient separations could be performed simply by changing compositions of the surfactant solutions in the inlet reservoir during a single run. Additionally, we carried out continuous gradient separation using a simple device. Brij 35 gave an effect parallel to that by Tween 20 in migration behavior of the analytes. A practically negligible change in the level of the baseline was observed in a stepwise gradient elution with the CTAC/Brij 35 system because of the small absorbance at the detection wavelength, while that with the CTAC/Tween 20 was considerable. All the benzoates were separated completely within reasonable analysis times using both stepwise and continuous gradient programs for the concentrations of Tween 20 or Brij 35 in the presence of 100 mM CTAC. In the initial studies of micellar electrokinetic chromatography (MEKC), some gradient methods in terms of electric field,1-3 * To whom correspondence should be addressed. E-mail: gifu-pu.ac.jp.

esaka@

8142 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

temperature,3 and solvent composition4-10 were employed. Sepaniak et al. reported the sodium lauryl sulfate (SDS)-MEKC separation of aliphatic amines derived with 7-chrolo-4-nitrobenz2-oxa-1,3-diazole (NBD-Cl), where concentrations of 2-propanol or acetonitrile as additives of running solutions were increased stepwise in their earlier works and linearly in their later works during separations, to realize both improvement of the separations and decrease in separation times simultaneously.4-8 Increase of the concentration of 2-propanol or acetonitrile means a decrease in polarity of the solvent phase to decrease transfer of hydrophobic solutes to the micellar phase, and thus, the solutes were eluted earlier. When the solutes of a very wide range of hydrophobicity are to be separated, solvent gradient methods are expected to be highly effective because change in the solvent composition can result in substantial change in partition behavior of the solutes. Such solvent gradient methods in MEKC would be similar to that in reversed-phase HPLC from the viewpoint of the mechanism of gradient elution. As predicted by Foley in his early report of MEKC,11 surfactant gradient methods also will be useful in separation of solutes possessing a wide range of hydrophobicity, because surfactant gradient methods also, if possible, should enable us to change the partition of solutes remarkably as well as the solvent gradient methods. Furthermore, surfactant gradient methods would involve an interesting MEKC-specific aspect, namely, the volume or the composition of distribution phases possessing a higher-order structure can be changed during a single run. In liquid chromatography also, stationary phases possess a higher-order structure, but the stationary phases cannot be changed gradually during a (1) Tsai, P.; Patel, B.; Lee, C. S. Anal. Chem. 1993, 65, 1439-42. (2) Tsai, P.; Patel, B.; Lee, C. S. Electrophoresis 1994, 15, 1229-33. (3) Djordjevic, N. M.; Houdiere, F.; Lerch, G.; Fitzpatrick, F. J. High Resolut. Chromatogr. 1999, 22, 443-448. (4) Balchunas, A. T.; Sepaniak, M. J. Anal. Chem. 1988, 60, 617-21. (5) Balchunas, A. T.; Swaile, D. F.; Powell. A. C.; Sepaniak, M. J. Sep. Sci. Technol. 1988, 23, 1891-904. (6) Sepaniak, M. J.; Swaile, D. F.; Powell, A. C. J. Chromatogr. 1989, 480, 18596. (7) Powell, A. C.; Sepaniak, M. J. J. Microcolumn Sep. 1990, 2, 278-84. (8) Powell, A. C.; Sepaniak, M. J. Anal. Instrum. 1993, 21, 25-41. (9) Buetehorn, U.; Pyell, U. Chromatographia 1996, 43, 237-141. (10) Kutter, J. P.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1997, 69, 51655171. (11) Foley, J. P. Anal. Chem. 1990, 62, 1302-1308. 10.1021/ac061557l CCC: $33.50

© 2006 American Chemical Society Published on Web 10/28/2006

single separation run practically. This interesting aspect of the present method derives from the fact that micelles are dynamic phases in both senses of having a variable structure based on thermodynamic equilibrium and moving freely in separation capillaries, and this may lead to remarkable increase in separation selectivity. In a previous work, we studied the separation of organic cations using a mixed micellar system of anionic SDS and nonionic polyoxyethylene sorbitan monolaurate (Tween 20).12 Electrokinetic separation using micellar running solutions will be highly effective for not only neutral compounds but also ionic ones, as reported for metal complexes, proteins, peptides, aromatic amines, and inorganic ions, because we can take advantage of additional interactions between the solutes and the micelles to increase separation selectivity.12-34 Using charged surfactants, however, there may be some inconvenience, namely, effects on migration behavior of ionic analytes possessing the same signs of charge to those of the micelles may be too small because of electrostatic repulsion to the analytes or may be excessive because of electrostatic attraction toward analytes of the opposite charge. In the latter cases, the synergistic “attractive” influences of both electrostatic and hydrophobic interactions can cause predominant transfer of the solutes to the micellar phases to result in coelution of most analytes. In the previous work, polyoxyethylene chains of Tween 20m which were inserted into the surface of the SDS micelle, could decrease attractive ionic interaction between the anionic sulfate groups of SDS and the cationic solutes remarkably to resolve the predominant transfer of the cationic solutes to the anionic micellar phase causing the coelution.12 This suppressive effect of polyoxyethylene chains will be useful to optimize the separations of both general organic cations and anions using ionic surfactants of the opposite signs of charges to those of the analytes. Furthermore, surfactant gradient elutions (12) Esaka, Y.; Tanaka, K.; Uno, B.; Goto, M.; Kano, K.; Anal. Chem. 1997, 69, 1332-1338. (13) Shepherd, R. E. Coord. Chem. Rev. 2003, 247, 147-184. (14) Qin, L.; Collins, G. E. Analyst 2001, 126, 429-432. (15) Hilder, E. F.; Macka, M.; Haddad, P. R. Analyst 1998, 123, 2865-2870. (16) Haddad, P. R.; Macka, M.; Hilder, E. F.; Bogan, D. P. J. Chromatogr., A 1997, 780, 329-341. (17) Terabe, S.; Markuszewski, M. J.; Inoue, N.; Otsuka, K.; Nishioka, T. Pure Appl. Chem. 2001, 73, 1563-1572. (18) Liu, W.; Lee, H. K. J. Chromatogr., A 1998, 796, 385-395. (19) Wainright, A. J. Microcolumn Sep. 1990, 2, 166-75. (20) Strege, M. A.; Lagu, A. L. J. Chromatogr., A 1997, 780, 285-296. (21) Strege, M. A.; Lagu, A. L. Anal. Biochem. 1993, 210, 402-10. (22) Hu, S.; Li, P. C. H. J. Chromatogr., A 2000, 876, 183-191. (23) Rodriguez, I.; Lee, H. K.; Li, S. F. Y. Electrophoresis 1999, 20, 1862-1868. (24) Angelino, S.; Prevot, A. Bianco; Gennaro, M. C.; Pramauro, E. J. Chromatogr., A 1999, 845, 257-271. (25) Bjergegaard, C.; Moller, P.; Sorensen, H. J. Chromatogr., A 1995, 717, 40914. (26) Lukkari, P.; Siren, H. J. Chromatogr., A 1995, 717, 211-17. (27) Nishi, H. J. Chromatogr., A 1997, 780, 243-264. (28) Kang, J-W.; De Reymaeker, G.; Van Schepdael, A.; Roets, E.; Hoogmartens, J. Electrophoresis 2001, 22, 1356-1362. (29) Mrestani, Y.; Neubert, R. H. H. J. Pharm. Biom. Anal. 2001, 24, 637-643. (30) Furtos-Matei, A.; Day, R.; St-Pierre, S. A.; St-Pierre, L. G.; Waldron, K. C. Electrophoresis 2000, 21, 715-723. (31) Takayanagi, T.; Fushimi, K.; Motomizu, S. J. Microcolumn Sep. 2000, 12, 107-112. (32) Macka, M.; Haddad, P. R.; Buchberger, W. J. Chromatogr., A 1995, 706, 493-501. (33) Nishi, H; Tsumagari, N; Terabe, S; Anal. Chem. 1989, 61, 2434-2439. (34) Gotti, R.; Pomponio, R.; Bertucci, C.; Cavrini, V. J. Chromatogr., A 2001, 916, 175-183.

can be performed with mixed systems of the ionic surfactants and the nonionic surfactants with polyoxyethylene chains, where the higher concentrations of the nonionic surfactants are used at the former stages to improve separation and then the lower concentrations are used at the later stages to decrease analysis time in a single run. We will report electrokinetic separations of 10 benzoate anions of a wide range of hydrophobicity as model analytes with surfactant gradient methods using mixed surfactant systems of “cationic” cetyltrimethylammonium chloride (CTAC) and Tween 20 or polyoxyethylene lauryl ether (Brij 35) in this paper. Electroosmotic flow (EOF) was suppressed almost completely by the inner-wall coating with poly(N,N-dimethyacrylamide) (PDMA) in the present methods. Under the slight EOF condition, the micelles in the inlet running solution can pass through and, thus, interact with all of the analytes before they are detected. Surfactant gradient elution can be performed simply by decreasing the concentration of Tween 20 or Brij 35 in the presence of a fixed concentration of CTAC in the inlet reservoir during a single run. The stable PDMA coating also enabled us to perform the electrokinetic separation of the benzoates successfully with good reproducibility. A considerable reduction in separation time, maintaining the improved separation obtained by addition of Tween 20, was achieved using stepwise and continuous gradient programs of decrease in the concentration of Tween 20 or Brij 35 in the presence of 100 mM CTAC. EXPERIMENTAL SECTION As model samples, unsubstituted benzoic acid (a) and nine substituted benzoic acids [salicylic acid (b), 4-nitrobenzoic acid (c), 4-toluic acid (d), 4-dimethylaminobenzoic acid (e), 4-acetoxybenzoic acid (f), 2-phthalaldehydic acid (g), 4-phtalaldehydic acid (h), 4-hydroxybenzoic acid (i), and 4-acetamidebenzoic acid (j)] were purchased from Wako Chemical (Tokyo, Japan, for a and b), Nacalai Tesque (Kyoto, Japan, for c-e and g-j), and Tokyo Chemical Industry (Tokyo, Japan, for f). CTAC, Tween 20, and Brij 35 as components of micelles were obtained from Tokyo Chemical Industry, Kishida Chemicals (Osaka, Japan), and Nacalai Tesque, respectively. N,N-Dimethylacrylamide (l), 3-(trimethoxysilyl)propyl methacrylate (m), tetramethylethylenediamine (n), and ammonium peroxodisulfate (o) for inner wall coating of capillaries were obtained from Aldrich Chemical (for l and m) and Nacalai Tesque (for n and o). The other chemicals were of analytical grade. Capillary electrophoresis was performed using a system combined in our laboratory consisting of a Matsusada HCZE30PNO high-voltage supply (Siga, Japan) and a Jasco MD-2010 multiwavelength detector (Tokyo, Japan) equipped with an Epson Pro-720L personal computer (Nagano, Japan). Fused-silica capillaries (0.05-mm i.d. and 0.375-mm o.d.) were purchased from GL Science (Tokyo, Japan). The total column length was 600 mm, in which the effective length (Leff) was 300 mm and the capillaries were coated on their inner walls with PDMA according to the method of Wan et al.35 The coated capillaries were rinsed with distilled water for 30 min using an aspirator at the beginning of daily experiments and further rinsed with water and then running solutions to be used (35) Wan, H.; Ohman, M.; Blomberg, L. G. J. Chromatogr., A 2001, 924, 5970.

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8143

initially for 5 and 2 min, respectively, before each run. Running solutions were 10 mM phosphate buffer solutions (pH 7.2) containing CTAC alone or both CTAC and Tween 20 or Brij 35. The applied voltage was fixed at -10 kV. Each concentration of the analytes in sample solutions was usually ∼0.1 mM. The sample solutions were injected on the anodic side by siphoning at a height of 20 cm for 20 s. Aliquots of the anthracene-saturated methanol solution were added to sample solutions for measurement of the migration velocity of micelles. A house-made device for continuous gradient MEKC is described below in detail. The retention factor (k) of an analyte was evaluated according to the following equation, k ) (V - Vep - Veo)/(Vmc - V), where V, Vep, Veo, and Vmc are the migration velocity of the analyte () Leff./migration time of an analyte), the electrophoretic velocity of the analyte as the free anion, the electroosmotic flow rate, and the migration velocity of the micelle marker () Leff./migration time of anthracene), respectively.36-38 We tried to measure the values of Veo with the present coated capillaries under each condition using formamide as a marker of EOF; however, we could detect the peak of formamide at neither the anodic side nor the cathodic side within at least 3 h from the injections. Thus, we assumed that Veo under the present conditions was negligible compared with V and Vep in estimation of retention factors, namely, k ≈ (V - Vep)/(Vmc - V). The values of V and Vmc were calculated simultaneously under the MEKC mode, while Vep measurements were independently done under the zone electrophoretic mode (without surfactants) by injection of all analytes on the cathodic side under the identical electric field as under the present conditions of MEKC, assuming the value of Vep of the analyte in the presence of the micelle would be practically the same as that in the absence of the micelle as well as in our previous work.12 RESULTS AND DISCUSSION Effect of Increase in Concentration of Tween 20 in a Mixed System of CTAC and Tween 20 for Separation of 10 Benzoate Anions. Figure 1A shows static electrokinetic separation of an unsubstituted benzoate and nine substituted benzoates as model organic anions using a 10 mM phosphate buffer (pH 7.2) containing 100 mM CTAC alone as a running solution (see Figure 1 for abbreviation of the analytes). To suppress EOF, we used capillaries coated on their inner walls with PDMA in the present work, while we employed an acidic condition to suppress EOF in our previous work using the mixed system of SDS and Tween 20.12 PDMA coating has been used for elimination of both EOF and adsorption of proteins and DNAs in CE separations.35,39,40 It was found that the PDMA coating suppressed EOF toward the anodic side also, which was observed commonly using running solutions containing tert-tetraalkylammonium surfactants such as CTAC with bare capillaries. We will be able to choose running buffer solutions of a wide pH range using the present coated capillaries because the EOF-eliminating feature will be almost pHindependent. The ability of the PDMA coating to eliminate EOF (36) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 57, 834-841. (37) Khaledi, M. G.; Smith, S. C.; Strasters, K. Anal. Chem. 1991, 63, 18201830. (38) Straster, K.; Khaledi, M. G. Anal. Chem. 1991, 63, 2503-2508. (39) Madabhushi, R. S. Electrophoresis 1998, 19, 224-230. (40) Li, H.; Wu, H. Y.; Wang, Y.; Sims, C. E.; Allbritton, N. L. J. Chromatogr., B 2001, 757, 79-88.

8144 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

Figure 1. Electrokinetic separations of 10 benzoates using (A) a pure 100 mM CTAC system and (B) a mixed system of 100 mM CTAC and 40 mM Tween 20. Conditions: PDMA-coated capillary, 0.05 mm i.d. × 600 mm (effective length, 300 mm); applied potential (current), -10 kV [(A) 17 and (B) 19 µA.], runnning solution, 10 mM phosphate buffer containing the surfactants (pH 7.2); injection, anodic side; detection, 210 nm. Peak assignment: 0, anthracene (micelle maker); 1, salicylate (2OH-BA); 2, 4-nitrobenzoate (4NO2-BA); 3, 4-toluate (4CH3-BA); 4, 4-dimethylaminobenzoate [4(CH3)2N-BA]; 5, 4-acetoxybenzoate (4CH3COO-BA); 6, benzoate (BA); 7, 4-phthalaldehydate (4CHO-BA); 8, 2-phthalaldehydate (2CHO-BA); 9, 4-hydroxybenzoate (4OH-BA); 10, 4-acetamidebenzoate (4CH3CONH-BA).

of both normal and reversed directions will be useful to perform the surfactant gradient elution using both anionic and cationic surfactants. Under slight EOF, cationic micelles will migrate to the cathodic side faster than most of the solutes, and thus, micelle markers should be detected at first followed by all of the other solutes. With the pure 100 mM CTAC system, only two benzoates possessing relatively hydrophilic substituents, 4OH-BA and 4CH3CONH-BA, were eluted as the single component and the others were coeluted with some of the others right after the micelle marker as shown in Figure 1A. This migration behavior of the benzoates would suggest that a large part of the benzoates transferred to the pure CTAC micellar phase predominantly, due to synergistic influences of electrostatic and hydrophobic interactions between the cationic micelle and the anionic benzoates, because free anions of the benzoates migrate to the anodic side, being opposite to the detector in the absence of the CTAC micelles. The relatively hydrophobic benzoates, which could not be separated with the pure 100 mM CTAC system, were separated almost completely using a mixed system of 100 mM CTAC and

Figure 2. Dependence of ln k on C for the 10 benzoates and anisole as a neutral compound. Symbols: B, 2OH-BA; O, 4NO2-BA; 2, 4CH3BA; 4, 4(CH3)2N-BA; 9, 4CH3COO-BA; 0, BA; [, 4CHO-BA; ], 2CHO-BA; +, 4OH-BA; ×, 4CH3CONH-BA; -, anisole.

40 mM Tween 20 as indicated in Figure 1B, although marked increase in migration times of the benzoates occurred simultaneously. The retention factor (k) of all the benzoates was decreased markedly with increase in the analytical concentration of Tween 20 in the running solutions (C), while that of anisole as a nonionic reference was not practically changed as shown in Figure 2. This means that the decrease in the k values of the benzoates will be responsible in large part for the suppression of the attractive electrostatic interaction between CTAC and the benzoates by the addition of Tween 20. In our previous study, a similar strong suppressing effect of Tween 20 on attractive ionic interaction between anionic SDS micelles and aromatic ammoniums was observed in the separation of the organic cations. The polyoxyethylene chains of Tween 20 inserted into the surfaces of the present CTAC micelle and the previous SDS micelle would prevent the ionic analytes of the opposite sign of charge to those of the micelles from approaching the ionic groups of the surfactants in the micelles. The velocity of anthracene as the micelle marker (Vmc) in the mixed system of 100 mM CTAC and 40 mM Tween 20 was decreased to about half of that in the pure 100 mM CTAC system as shown in Figure 1. The decrease in Vmc with increase in C would be ascribed to increase in the micelle size, decrease in the amount of the charge of one aggregation by inclusion of Tween 20, or both and would extend the intervals of migration times of analytes. Therefore, the improved separation shown in Figure 1B should be attributed to both suppression of excessive transfer of the benzoates to the micellar phase and decrease in mobility of micelles by addition of Tween 20 to the CTAC system. The former, however, had much larger effects for the improvement of separation than the latter as follows. We calculated tentative t - tmc of the benzoates, where t and tmc mean migration time of the analytes and the micelle marker, using the tmc at each C and the k values at C ) 0 mM assuming their k values had not been changed with increase in C as well as the case of anisole. The calculated values of t - tmc of the benzoates at C ) 40 mM were about one-third of the real values from Figure 1B. This means considerably larger contribution of the suppression of the ionic interaction to the

remarkably improved separation shown in Figure 1B, compared with that of decrease in Vmc. The detection order of the benzoates was not changed with increase in C in the present micellar system, while marked improvement in separation was attained with the increase in C. In contrast, in the previous mixed system of SDS and Tween 20, elution order of analytes can be changed remarkably by change in composition of the surfactants.12,41 This situation would be interpreted substantially by the hydrogen-bonding basicity between a pure CTAC micelle and a pure SDS micelle. The linear solvation energy relationship (LSER) method enables us to characterize a micelle as a partitioning phase mainly in terms of its hydrophobicity and hydrogen-bonding ability. The LSER method is one of linear regression methods based on the equation; logSP ) c + a∑R2 + b∑β2 + sπ2 + rR2 + v(Vx/100), where SP means a solubility-related property such as a distribution coefficient in a partitioning system and a, b, s, r, and v are values corresponding to difference of the physical properties between a partitioning phase, a micelle in this case, and an aqueous phase as another partitioning phase in the system, while ∑R2, ∑β2, π2, R2, and Vx are inherent values of a solute related to its physical properties. A term in the LSER equation, except to c of a systemdependent constant, corresponds to a stabilization energy based on an interaction such as hydrogen bond and hydrophobic interaction caused by transfer of the solute from the water phase to the micelle. The two terms, a∑R2 and b∑β2, are those related to hydrogen-bonding interactions, and the values, a and b, reflect hydrogen-bonding basicity and acidity of the micelle, respectively. A larger a value in the LSER method means larger hydrogenbonding basicity of the corresponding micelle. Published values of a for a cethyltrimetylammonium bromide (CTAB) system and an SDS system were -0.18 and 1.02, respectively.42,43 Assuming the present CTAC system has an a value similar to that of the reported CTAB system, the present pure CTAC micelle should possess much larger hydrogen-bonding basicity compared with the previous pure SDS micelle. Thus, addition of Tween 20 possessing the “hydrogen-accepting” polyoxyethylene chains would change slightly the hydrogen-bonding basicity of the present CTAC micelle in contrast with the case of the previous SDS micelle. Stepwise Gradient Method. As shown in Figure 1, analysis time increased markedly with increase in C, because of both the decrease in k and the increase in tmc. Such a time-consuming aspect can be a serious disadvantage in the present mixed surfactant system in practical use. Surfactant gradient methods, however, can be performed in the present system as a solution to the problem. In a gradient elution in MEKC, changes in composition, volume ratios of partitioning phases, or both have to occur in areas in which sample zones exist. Namely, the partitioning phases, changing the composition, the phase ratios, or both gradually, have to come across the sample zones in a capillary during separation. As mentioned above, some solvent gradient methods were successfully introduced in MEKC,4-10 and the methods were (41) Esaka, Y.; Kobayashi, M.; Ikeda, T.; Kano, K. J. Chromatogr., A 1996, 736, 273-280. (42) Yang, S; Khaledi, M. G. Anal. Chem. 1995, 67, 499-510. (43) Quina. F. H.; Alono, E. O.; Farah, J. P. S. J. Phys. Chem. 1995, 99, 1170811714.

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8145

based on the fact that EOF has the largest net mobility compared with all solutes under the usual conditions of CE. In such situations, it is possible that a solvent phase, of which the solvent composition is that of the running solutions in the inlet reservoir, passes through sample zones migrating in the capillary. Thus, solvent gradient elution can be performed by gradual change in solvent composition of the running solution in the inlet reservoir during separation. In the published works,4-10 concentrations of organic solvents such as 2-propanol and acetonitrile in inlet running solutions were increased gradually during separation, and solvent gradient elutions were performed to improve separation of analytes possessing a wide range of hydrophobicity, which could not be separated sufficiently within permissible periods with isocratic methods. Those solvent gradient methods would be based on a mechanism similar to that in reversed-phase HPLC. In MEKC, a micelle as another partitioning phase against a solvent phase is movable and possesses a dynamic structure governed by thermodynamic equilibrium. We can change the physical property of micellar phases, volume ratios between micellar phases and solvent phases, or both by changing the composition and concentration of surfactants in running solutions. Therefore, surfactant gradient elution is promising in MEKC as well as solvent gradient elution. However, surfactant gradient elution will be difficult practically under the usual conditions of CE in which EOF of large velocity is present. Under such conditions, micelles would have the smallest net mobility compared with all solutes generally, and thus, micelles in reservoirs cannot pass through the sample zones. Although surfactant gradient elution may be performed using a capillary filled with micellar solutions of different composition, concentration of surfactants, or both before a run, it seems to be a much impracticable method. One answer to realize surfactant gradient elution will be suppression of EOF. Under slight EOF, micelles can migrate faster than most solutes, and thus, micelles in the inlet reservoir will pass through all sample zones in the presence of an almost immobilized solvent phase. Therefore, surfactant gradient elution will be performed by simply changing the composition, concentrations, or both of surfactants in the running solution of the inlet reservoir. As mentioned above, we could suppress EOF almost completely using the capillaries with the inner walls coated with PDMA for the present mixed system of CTAC and Tween 20. We attempted surfactant gradient elution for electrokinetic separation of the 10 benzoate anions. As indicated in Figure 1, increase in C caused both remarkable improvement in separation of the relatively hydrophilic solutes and marked increase in migration time of the relatively hydrophilic ones. Achievement of both a complete separation and a reasonable analysis time was expected with a surfactant gradient method, namely, starting with a CTAC solution containing Tween 20 of a relatively large concentration sufficient for complete separation of the relatively hydrophobic analytes followed by gradual decreasing in the concentration of Tween 20 in the running solutions of the inlet reservoir (Cinlet) to hasten migration of the relatively hydrophilic ones. Here, the more hydrophobic benzoates exist for longer time in the area containing higher concentration of Tween 20. This fact enables us to improve separation of the relatively hydrophobic ones, decreasing the detection times of the 8146 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

Figure 3. Electrokinetic separation of the 10 benzoates with a stepwise gradient program of Cinlet. Conditions: PDMA-coated capillary, 0.05 mm i.d. × 600 mm (effective length, 300 mm); applied potential (current), -10 kV (19-27 µA), runnning solution, 10 mM phosphate buffer solutions (pH 7.2) containing 100 mM CTAC and 4-40 mM Tween 20. The concentration of Tween 20 in the inlet reservoir (Cinlet) was changed from 40 (initial) to 30 (at 2 min), 20 (at 4 min), 10 (at 6 min), and 4 mM (at 8 min), while that in the outlet reservoir was fixed to 40 mM; injection, anodic side; detection, 210 nm. Peak assignment: same as in Figure 1.

relating hydrophilic ones simultaneously. Stepwise change in Cinlet was achieved by immediate replacement of the inlet reservoir containing surfactant solutions of different compositions within 1 s without switching off the HV supply. In this quick changing action, air intrusion to the capillary seldom occurred. The anodic electrode in the inlet reservoirs was grounded to facilitate changing the reservoirs during applying voltage. Figure 3 shows a micellar electrokinetic separation of the 10 benzoate anions employing a stepwise gradient program of Cinlet. The Cinlet was changed stepwise from 40 to 30, 20, 10, and 4 mM at 2, 4, 6, and 8 min, while the concentration of Tween 20 in the outlet reservoir (Coutlet) was maintained to be 40 mM and the concentration of CTAC in running solutions of both reservoirs was fixed at 100 mM. Obvious improvement in both separation and analysis time compared with the isocratic methods in Figure 1 was realized as shown in Figure 3; namely, the 10 benzoates were separated almost completely within 27 min. The stepwise decrease in the baseline height from 20 to 26 min in Figure 3 indicates decrease in C at the position of the detector, because Tween 20 has some absorption at the detection wavelength (210 nm). This stepwise decrease in the baseline height strongly suggests formation of the expected gradient of C in the separation area in the capillary to result in the successful separation. Reproducibility in migration time of the benzoates using the stepwise gradient program was estimated to 0.31-0.77% as RSD (n ) 5) values. The PDMA coating seems much more stable compared with normal polyacrylamide coating to give both good peak shapes and a sufficient effect in suppression of EOF. We could use the capillaries coated with PDMA in more than 50 runs without serious troubles, when we rinsed them with water adequately between runs and stored them in a refrigerator at 4 °C after daily measurements. The stability of the PDMA coating would contribute significantly to the good reproducibility of the present gradient methods.

Figure 4. Schematic description of the continuous gradient method.

Interestingly, sharp peaks of narrow width could be observed in areas between zones of different C, indicated by sudden changes in background absorbance (data not shown). This phenomenon will be interpreted as a sweeping effect,44,45 because the following zone of smaller C gives larger retention factors of the benzoates. We may be able to make use of this partial sweeping effect to improve sensitivity of an analyte somewhat. It should be mentioned that Balchunas and Sepaniak used Triton-X of a nonionic surfactant, as an additional additive of the running solution, together with 2-propanol in SDS-MEKC separation of derivatives of organic amines by NBD-Cl.4,5 The concentration of Triton-X in the running solution of the inlet reservoir was increased stepwise together with 2-propanol. Formation of mixed micelles of SDS with Triton-X caused an increase in net velocity of the micelles because of decrease in electrophoretic velocity of the micelles. However, their measurements were performed in the presence of EOF of large velocity, and thus, the micelles migrated more slowly than all analytes in their system. Under such conditions, micelles in the inlet reservoir never pass through sample zones, and thus, change in composition of surfactants in the inlet reservoir would not be practically reflected in partition of solutes between the micellar phase and the solvent phase. Therefore, the marked improvement of the separation in their work should be attributable predominantly to solvent gradient elution with change in the concentration of 2-propanol. Continuous Gradient Method. Continuous change in C in the separation area will enable us to design more elaborate gradient programs than stepwise programs. This should be realized by changing Cinlet continuously. We will perform the continuous gradient MEKC using a simple device with substantially small requirement of running solutions. Figure 4 indicates the device and the procedure schematically. A micro magnetic stirrer (2.0 mm o.d. × 2.0 mm) is equipped in the inlet reservoir. (44) Quirino, J.; Terabe, S. Anal. Chem. 1991, 71, 638-1644. (45) Quirino, J.; Kim, J.; Terabe, S. J. Chromatogr., A 2002, 965, 357-373.

Figure 5. Electrokinetic separation of the 10 benzoates with a contimuous gradient program of Cinlet using the device illustrated in Figure 4. Conditions: PDMA-coated capillary, 0.05 mm i.d. × 600 mm (effective length, 300 mm); applied potential (current), -10 kV (19-27 µA), running solution, 10 mM phosphate buffer solutions (pH 7.2) containing 100 mM CTAC and 40-3.6 mM Tween 20. Cinlet was fixed to 40 mM (0-2 min), then changed continuously from 40 to 3.6 mM (2-12 min), and fixed again to the end at 3.6 mM, while that in the outlet reservoir was fixed to 40 mM; injection, anodic side; detection, 210 nm. Peak assignment: same as in Figure 1.

A small amount, 0.10 mL in the present work, of an initial running solution containing a composition of surfactants is set in the inlet reservoir, and then a mixing solution containing the other composition of surfactants is added to the inlet reservoir continuously by gravity during separation. Adding rate of the mixing solution can be controlled easily with both height of the addition reservoir containing the mixing solution and the internal diameter and the length of the regulation tube equipped at a lower position of the addition reservoir as indicated in Figure 4, making desired slopes of the surfactant concentrations. Figure 5 shows a continuous surfactant gradient electrokinetic separation of the 10 benzoates using the device, where the initial Cinlet was 40 mM in the presence of 100 mM CTAC, and at 2 min after the start of the measurement, the pure 100 mM CTAC solution as the mixing solution was added for 10 min with the fixed rate of 0.10 mL/min, and then, the addition was stopped at 12 min. In this measurement, the height of the addition reservoir, the length of the regulation tube, and its i.d. were set to 20 cm, 5.0 cm, and 0.2 mm, respectively. Total added volume of the mixing solution was 1.00 mL, namely, the final volume in the inlet reservoir was 1.10 mL, thus the final Cinlet was 3.6 mM, while Coutlet in the presence of 100 mM CTAC was fixed at 40 mM during the separation. Continuous change in C in the separation area would be verified by the change in the baseline level in the chart. Reproducibility in migration time of the benzoates with this continuous gradient program became 0.41-1.71% as RSD values (n ) 5). These RSD values are somewhat larger than those with the stepwise method mentioned above. It will be mainly attributable to variation of the adding rate of the mixing solution. Judging from the good peak shapes and the separation behavior shown in Figure 5, the formation of mixed micelles of various compositions between CTAC and Tween 20 would be achieved immediately after addition of the mixing solution in the inlet reservoir. This fact will be essential to perform continuous Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8147

surfactant gradient methods. So we will be able to perform more complicated gradient programs, if we need, using a two-pump system equipped with an interface device including a micromixer as in the system of Sepaniak et al.,6-8 though it will become an expensive system and will consume much larger amount of running solutions practically. Elimination of Variation in Background Absorbance Using Brij 35 in Place of Tween 20. Marked stepwise changes of baselines such as that in Figure 3 can become troublesome in quantification of peak components, because sudden or discontinuous change of the background absorbance such as those seen between the steps in Figure 3 is apt to cause errors in measurement of peak heights and areas. The continuous gradient method mentioned above and longer wavelengths for detection will reduce the quantitative errors generally giving continuous baselines and smaller difference in the height of baseline steps, respectively, although the latter can spoil the universality of detection. Indeed, employing 254 nm in place of 210 nm, we could repress the change in absorbance almost completely without practically affecting sensitivity for peaks 2, 4, 5, 7, 8, 9, and 10, although peaks 1, 3, and 6 almost disappeared (data not shown). On the other hand, alternatives to Tween 20 will be preferable, which have lower absorbance than Tween 20 at the present detection wavelength, 210 nm, representing an effect similar to that of Tween 20 on transfer of organic anions simultaneously, because the marked stepwise change in Figure 4 was attributed to the absorption of Tween 20. Here, we examined Brij 35 as an alternative to Tween 20, which has a polyether chain necessary to shield the positive charge of CTAC in the mixed micelles with CTAC and which was expected to have slight absorption at detection wavelengths of a wide range, because Brij 35 has a simple structure possessing no meaningful groups giving a large absorption, while Tween 20 has carbonyl groups. A complete separation very similar to Figure 1B was obtained with 30 mM as the concentration of Brij 35 in the running solution (CBrij 35) in the presence of 100 mM CTAC, and it took 47 min under the same conditions as those for Figure 1B except for the surfactant system (data not shown). Figure 6 shows a gradient separation of the 10 benzoates with a stepwise gradient program of the concentration of Brij 35 in the inlet reservoir (Cinlet.Brij 35) under the same conditions as those of Figure 3 except for employment of Brij 35 and the gradient program. The Cinlet. Brij 35 was changed stepwise from 30 to 18 mM at 3.0 min, and from 18 to 1.5 mM at 6.5 min, while the concentration of Brij 35 in the outlet reservoir was maintained to be 30 mM and the concentration of CTAC in running solutions of both reservoirs was fixed at 100 mM. All of the benzoates were separated completely within 27 min. As might have been expected, no stepwise change in the baseline was observed with detection at 210 nm (B). On the other hand, stepwise change in the baseline can be seen in the electropherogram detected at 195 nm (A), and it would represent an anticipated gradient change in CBrij 35 in the separation area of the capillary during the separation. The reproducibility of this stepwise gradient method using Brij 35 was 0.38 - 0.51% as RSD (n ) 5), and this result stands in comparison with the case using Tween 20 as mentioned above. Brij 35 will become an good alternative to Tween 20 from the viewpoint of quantitative analysis. 8148 Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

Figure 6. Electrokinetic separation of the 10 benzoates with a stepwise gradient program of Cinlet, Brij 35. Conditions: PDMA-coated capillary, 0.05 mm i.d. × 600 mm (effective length, 300 mm); applied potential (current), -10 kV (17-25 µA), runnning solution, 10 mM phosphate buffer solutions (pH 7.2) containing 100 mM CTAC and 30-1.5 mM Brij 35. The concentration of Brij 35 in the inlet reservoir (Cinlet,Brij 35) was changed from 30 (initial) to 18 (at 3 min) and 1.5 mM (at 6.5 min), while that in the outlet reservoir was fixed to 30 mM; injection, anodic side; detection, (A)195 nm and (B) 210 nm. Peak assignment: same as in Figure 1.

CONCLUSION Between a water phase and a CTAC micellar phase, predominant transfer of organic anions to the micellar phase can be reduced remarkably by addition of some nonionic surfactants possessing polyether chains, such as Tween 20 and Brij 35, which will shield the positive charge of CTAC. Using both the reducing effect and the fact that we can change the composition of surfactants one after another in the separation area in capillaries simply by changing the composition of surfactants in inlet reservoir under slight EOF, we developed surfactant gradient methods of MEKC in both stepwise and continuous ways and applied the gradient methods to separation of an unsubstituted benzoate and substituted benzoates as model analytes. We can accomplish both considerable reduction in separation time and improved separations. The present surfactant gradient method will be extended to separation of organic cations such as biological amines simply using anionic surfactant systems like the mixed systems of SDS and Tween 20.12 Strictly speaking, in the present work, the separations were governed in large part by the MEKC mode based on partition of analytes between ionic micellar phases and solvent phases and, in part, by zone electrophoresis mode simultaneously. But, in the original MEKC also, similar surfactant gradient

methods would be potent ones in the separation of neutral analytes with a large range of hydrophobicity. Related studies are in progress. Solvent gradient methods and surfactant gradient methods may look as if they provide very similar benefits from the viewpoint of controlling hydrophobicity of the partitioning the phases to optimize MEKC separation. However, there can be significant difference between separation selectivity induced by changing the compositions of solvent phases and that provided by changing the compositions of micellar phases. The orientation of solute molecules in the partitioning phases should be important in recognition of solutes and the orientation shown in micellar phases, which possess a higher-order structure, can be much stronger than those in solvent phases. Surfactant gradient methods can involve remarkable change in orientation of the solutes in micellar phases during a single run. Such situation would give us

novel selectivities in separation being different from those afforded by solvent gradient methods. ACKNOWLEDGMENT Y.E. thanks to M. Sekido and C. Sato for their support in experiments. He thanks also Prof. K. Kano in Kyoto University for helpful discussion. Additionally, he thanks Dr. M. Torimura in AIST for his kind loan of the multichannel detector. This work was supported in part by Grant-in-Aid for Scientific Research to Y.E. (13771358 and 17590035) from Japan Society for the Promotion of Science and the Special grant in 2004 from Gifu Pharmaceutical University. Received for review August 21, 2006. Accepted September 25, 2006. AC061557L

Analytical Chemistry, Vol. 78, No. 23, December 1, 2006

8149