Anal. Chem. 1997, 69, 1332-1338
Sodium Dodecyl Sulfate-Tween 20 Mixed Micellar Electrokinetic Chromatography for Separation of Hydrophobic Cations: Application to Adrenaline and Its Precursors Yukihiro Esaka,* Kazuya Tanaka, Bunji Uno, and Masashi Goto
Gifu Pharmaceutical University, 5-6-1 Mitahora-higashi, Gifu 502, Japan Kenji Kano*
Department of Agricultural Chemistry, Faculty of Agriculture, Kyoto University, Sakyo-ku, Kyoto 606, Japan
Capillary electrophoresis (CE) utilizing micellar systems is a powerful tool for separation of ionic as well as nonionic compounds.1-11 In such micellar systems, ionic analytes are separated under both modes of micellar electrokinetic chromatography (MEKC) and capillary zone electrophoresis (CZE), while
the separation of nonionic compounds is governed by the MEKC mode alone. In the separations of nonionic compounds under the MEKC mode, hydrophobic interaction between analytes and micelles is the major factor to govern the partitioning behavior of analytes. In contrast, in the separation of ionic solutes, ionic interactions between analytes and micelles become important together with hydrophobic interactions. The ionic interactions change partition equilibrium of analytes between the micelle and solution phases and then, in principle, can provide some benefits to MEKC. Kaneta et al. separated inorganic anions using a cationic surfactant system and reported that ionic interactions between the solutes and the micelle changed separation selectivity remarkably.9 However, cooperative ionic interactions with hydrophobic ones, sometimes, cause an extreme shift in the partition equilibrium of analytes to the micelle phase, resulting in elongation of the migration time, broadening of the peak shape, and a decrease in the separation efficiency and selectivity. Such disadvantage is encountered, especially, in the analysis of hydrophobic cations on sodium dodecyl sulfate (SDS)-based MEKC.5,8 Therefore, some strategy would be required to reduce the interaction between solutes and micelle down to a moderate level in strength. On the other hand, mixed surfactant systems in MEKC provide us various analytical advantages, because some additional interactions can be introduced in the separation system to change the selectivity. Chiral compounds incorporated into ionic surfactant micelles grant chiral selectivity to MEKC,12,13 and some surfactants with various hydrophilic groups allow novel selectivity owing to additional electrostatic interaction.14-20 Separation selectivity is easily optimized by a change of mixing ratio of the major surfactant
(1) Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T. Anal. Chem. 1984, 56, 111-113. (2) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 57, 834-841. (3) Walbroehl, Y.; Jorgenson, J. W. Anal. Chem. 1986, 58, 479-481. (4) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1988, 60, 258-263. (5) Wallingford, R. A.; Ewing, A. G. J. Chromatogr. 1988, 441, 299-309. (6) Nishi, H.; Tsumagari, N.; Terabe, S. Anal. Chem. 1989, 61, 2434-2439. (7) Khaledi, M. G.; Smith, S. C.; Strasters, K. Anal. Chem. 1991, 63, 18201830. (8) Strasters, K.; Khaledi, M. G. Anal. Chem. 1991, 63, 2503-2508. (9) Kaneta, T.; Tanaka, S.; Taga, M.; Yoshida, H. Anal. Chem. 1992, 64, 798801. (10) Smith, S. C.; Khaledi, M. G. J. Chromatogr. 1993, 632, 177-184. (11) Quang, C.; Strasters, J. K.; Khaledi, M. G. Anal. Chem. 1994, 66, 16461653.
(12) Otsuka, K.; Terabe, S. J. Chromatogr. 1990, 515, 221-226. (13) Otsuka, K.; Kashihara, M.; Kawaguchi, Y.; Koike, R.; Hisamitsu, T.; Terabe, S. J. Chromatogr., A 1993, 652, 253-257. (14) Wallingford, R. A.; Curry, P. D.; Ewing, A. G. J. Microcolumn Sep. 1989, 1, 23-27. (15) Ahuja, E. S.; Preston, B. P.; Foley, J. P. J. Chromatogr., B 1994, 657, 271284. (16) Rasmussen, H. T.; Goebel, L. K.; Mcnair, H. M. J. Chromatogr. 1990, 517, 549-555. (17) Rasmussen, H. T.; Goebel, L. K.; Mcnair, H. M. J. High Resolut. Chromatogr. 1991, 14, 25-28. (18) Wu, Q.; Claessens, H. A.; Cramers, C. A. Chromatographia 1992, 34, 2530. (19) Little, E. L.; Foley, J. P. J. Microcolumn Sep. 1992, 4, 145-154. (20) Terabe, S.; Ozaki, H.; Ishihara, H. Bunseki Kagaku 1993, 42, 859-866.
1332 Analytical Chemistry, Vol. 69, No. 7, April 1, 1997
S0003-2700(96)00731-7 CCC: $14.00
Separation of hydrophobic cations in capillary electrophoresis under the mixed micellar system composed of SDS and Tween 20 was investigated from the viewpoints of thermodynamics and practical application. Hydrophobic cations interact strongly with the anionic SDS micelle, and this often lead to predominant dissolution of the analytes into the micellar phase, resulting in poor resolution. The ionic interaction was evaluated to be close in strength to the hydrophobic interaction between benzenes and the SDS micelle. Tween 20 as a component to the mixed micelle was found to weaken the attractive ionic interactions between the cationic solutes and SDS. In addition, the polyether chain of Tween 20 serves also as a hydrogen acceptor to cause the attractive hydrogenbonding interactions with hydrogen-donating analytes. The two different functions of Tween 20 improve the separation of hydrophobic cations remarkably. Separation patterns are well controlled by varying the mixing ratio of the two surfactants. Adrenaline and its six precursors were successfully separated with this mixed micellar system. The improved separation efficiency was not affected by the presence of bovine serum matrix, realizing the direct MEKC analysis of serum sample without deproteination, in which proteins were comigrated with the micelle.
© 1997 American Chemical Society
and additive compounds. A mixed micellar system was also employed in order to modify electroosmotic flow.21 Furthermore improved peak shapes and separation efficiency compared with those of the chiral pure micelle systems were also realized in several mixed surfactant systems such as SDS-zwitterionic surfactant, SDS-Brij 35, and SDS-bile salt systems.15,19,22 Our previous paper described the significance of mixed surfactant systems composed of SDS and Brij 35 or Tween 20 in separations of nonionic compounds.23 Incorporation of the nonionic surfactants possessing polyether chains into the SDS system was found to cause hydrogen-bonding interactions between hydrogen-donating analytes and the polyether moieties as hydrogen acceptors, resulting in novel selectivity. On the other hand, polyethers such as poly(ethylene glycols) are frequently used as coating materials in CE to prevent analytes from adsorption on capillary inner walls.24,25 The adsorption phenomena would be attributable in part to ionic interactions of analytes with the deprotonated silanol groups on inner walls of capillaries. Thus, we may expect that this property of polyether compounds can be utilized to weaken strong and attractive interactions of cationic (anionic) solutes with anionic (cationic) micelles. In this paper, we assessed thermodynamically the ionic interactions between hydrophobic cations and the anionic SDS micelle in MEKC using simple substituted benzenes as model analytes. The ionic interaction energy was revealed to be the same order in strength as the hydrophobic interaction energy between benzenes and the SDS micelle. We further focused our attention to a mixed micellar system composed of SDS and Tween 20 in order to reduce the strong ionic interaction as well as to allow the hydrogen-bonding interaction. The significance of the mixed micelle-based MEKC is described in its application to the separation of adrenaline and its six precursors in model biological samples. EXPERIMENTAL SECTION Apparatus. CE experiments were performed on a Jasco CE800 system (Tokyo, Japan) with a Jasco 807-IT integrator. A capillary with 0.05 mm i.d. and 0.375 mm o.d. was supplied from GL Science (Japan, Tokyo). The total column length of the capillaries was 500 mm. In MEKC separation, the effective length of the columns was 200 mm in separations at pH 2.5 or 300 mm at pH 7.8. In CZE separation, the effective length was 300 mm. The detection wavelength was set at 210 nm. Chemicals. Tween 20 (polyethylene sorbitan monolaurate) and SDS were purchased from Kishida Chemical (Osaka, Japan) and Nacalai Tesque (Kyoto, Japan), respectively, and used as received. As analytes, phenol, 4-aminophenol, 4-(2-aminoethyl)phenol hydrochloride (tyramine hydrochloride), benzaldehyde, L-noradrenaline, L-dopa, dopamine, and L-phenylalanine were obtained from Nacalai Tesque, L-3-methoxytyrosine and adrenaline were obtained from Sigma (St. Louis, MO), 4-ethylphenol and aniline were obtained from Wako Chemical (Tokyo, Japan), and L-tyrosine was purchased from Tokyo Kasei Kogyo (Tokyo, Japan). Bovine serum was purchased from Nacalai Tesque. Oil Yellow (21) Ahuja, E. S.; Little, E. L.; Nielsen, K. R.; Foley, J. P. Anal. Chem. 1995, 67, 26. (22) Terabe, S.; Shibata, M.; Miyashita, Y. J. Chromatogr. 1989, 480, 403-411. (23) Esaka, Y.; Kobayashi, M.; Ikeda, T.; Kano, K. J. Chromatogr., A 1996, 736, 273-280. (24) Mizuno, M.; Tochigi, K.; Taki, M. Bunseki Kagaku 1992, 41, 485-489. (25) Wang, T.; Hartwich, R. A. J. Chromatogr. 1992, 594, 325-334.
AB and formamide used as markers of micelle migration and electroosmotic flow, respectively, were purchased from Nacalai Tesque. All other chemicals were of analytical reagent grade. Procedure. Electrolyte solutions were 10 mM phosphate buffers of pH 2.5 or 7.8 in the presence or absence of the surfactants. The applied voltage was fixed to be 18 kV, where the currents were kept at 21-22 µA at pH 2.5 or at 19-20 µA at pH 7.8. At the beginning of daily experiments, the capillary was rinsed using an aspirator with 0.1 M sodium hydroxide for 10 min and then distilled water for 5 min. The capillary was further rinsed for 5 min with the electrolyte solutions to be used before each run. The concentrations of the analytes in sample solutions were usually 1-2 mM. When MEKC experiments were performed at pH 7.8, the sample solutions were injected on the anodic side by siphoning at a height of 10 cm for 10 s. Here if necessary, aliquots of Oil Yellow AB26-saturated methanol solution and aqueous formaldehyde solution were added to the sample solutions for simultaneous measurements of the migration velocity of the micelle (Vmc) and the electroosmotic velocity (Veo), respectively. At pH 2.5, however, the anionic micelles and all analytes used here migrated in the direction of the anode because of very small Veo (compared with the electrophoretic velocity of the micelle with the opposite sign) and also attractive interactions between the analytes and the micelles. Therefore, the sample solutions were injected on the cathodic side. For simultaneous measurements of Vmc and Veo, Oil Yellow AB was added to the sample solution, while the formamide solution was separately injected on the anodic side (15 cm, 2 s) after the sample injection (10 cm, 20 s). As a marker of the electroosmotic flow in our MEKC experiments (where the marker peak was broadened owing to small Veo, especially at pH 2.5), formaldehyde was superior to methanol, which is frequently used because of larger absorption coefficient at 210 nm. Our preliminary experiments verified that the migration time of formaldehyde is identical with that of methanol. In spiking experiments, 200 µL of a sample solution containing adrenaline and its six precursors as well as Oil Yellow AB was mixed with 240 mL of 5-fold-diluted bovine serum and the resultant mixed solution was directly injected without any pretreatment. The capacity factor (k′) was evaluated according to the following equation,2,7,8
k′ )
V - (Vep + Veo) Vmc - V
(1)
where V and Vep are the net migration velocity and the electrophoretic velocity of an analyte, respectively. Values of V, Veo, and Vmc were simultaneously measured under the MEKC mode as described above, while Vep measurements were independently performed under the CZE mode (without micelles) by injection of all analytes and markers on the anodic side under the identical electric field as in MEKC. In this work, ion-pair formation between the cationic solutes and the SDS monomer was reasonably ruled out, because Vep was independent of the analytical concentration of SDS at SDS concentrations lower than the critical micellar concentration (cmc) (data not shown), which would be ∼4 mM under our experimental conditions.8 (26) Otsuka, K.; Terabe, S.; Ando, T. J. Chromatogr. 1985, 332, 219-226.
Analytical Chemistry, Vol. 69, No. 7, April 1, 1997
1333
Table 1. Capacity Factors (k′) of the Five Substituted Benzenes in 10 mM Phosphate Buffers (pH 7.8 and 2.5) Containing 25 mM SDS
analytes
capacity factor (k′) pH pH 7.8 2.5
4-aminophenol (pKa ) 5.82) 0.0596 4.71 aniline (pKa ) 4.7) 0.195 16.6 phenol 0.244 0.200 4-ethylphenol 1.55 1.22 tyramine 14.6 15.0 a
Figure 1. MEKC separations of five mono- or disubstituted benzenes in the absence (A, B) and the presence (C) of 5 mM Tween 20 in electrolyte solution containing 25 mM SDS. Conditions: applied voltage, 18 kV; operating current, 20-21 µA; capillary length, 500 mm [effective length, 200 mm in (A) and 300 mm in (B) and (C)]; electrolyte solutions; 10 mM phosphate buffer containing surfactants [at pH 7.8 in (A) and at pH 2.5 in (B) and (C)]. Peaks: 0, formamide (a marker of electroosmotic flow); 1, 4-aminophenol; 2, aniline; 3, phenol; 4, 4-ethylphenol; 5, tyramine; 6, Oil Yellow AB (a marker of micelle migration); 7, benzaldehyde [a reference compound in (B) and (C)].
RESULTS AND DISCUSSION Estimation of Ionic Interactions in Dissolution of Solutes to the SDS Micelle. As pointed out in some articles,5,8 combination of hydrophobic and ionic interactions between hydrophobic cations and ionic micelles enhances the dissolution of these solutes into the micelles phase drastically. We attempted first to estimate the strength of the ionic interactions between hydrophobic cations and the SDS micelle using substituted benzenes as model samples. Figure 1 shows an example of the separation of 4-aminophenol, aniline, phenol, 4-ethylphenol, and tyramine [4-(2-aminoethyl)phenol] in 10 mM phosphate buffers containing 25 mM SDS at pH 7.8 (A) and 2.5 (B). Oil Yellow AB (peak 6) as a marker of the micelle migration was detected last at pH 7.8 (with the cathodic detection) because of the electrophoresis of the SDS micelle competitive in direction to the electroosmotic flow, while it was detected first at pH 2.5 (with the anodic detection) because of a very small Veo compared with the electrophoretic flow of SDS in the direction of the anode. On the pH change from 7.8 to 2.5, the peaks of 4-aminophenol (peak 1) and aniline (peak 2) were brought too close to the peak of Oil Yellow AB. The pH effect can be more clearly expressed by the capacity factors (k′), which can be estimated from electropherograms A and B in Figure 1 on the basis of eq 1. The results are summarized in Table 1 along with the total charge of the analytes at pH 7.8 and 2.5 judged from pKa values of the corresponding conjugate acids. The k′ values of 4-aminophenol and aniline increased ∼80 times with the pH change from 7.8 to 2.5 during which the amino groups are protonated. As calculated from the k′ values, the fractions of 4-aminophenol and aniline in the SDS micelle phase were drastically increased on the protonation from 6 to 82% and 16 to 94%, respectively. On the other hand, the k′ values of the others without charge change (phenol, 4-ethylphenol, tyramine) were 1334
Analytical Chemistry, Vol. 69, No. 7, April 1, 1997
charge ∆∆µ°a (kJ mol-1, pH pH 293 K) 7.8 2.5 -10.7 -10.8 0.484 0.583 -0.066
0 +1 0 +1 0 0 0 0 +1 +1
∆∆µ° ) ∆µ°(pH 2.5) - ∆µ°(pH 7.8); see text.
practically independent of pH. These results clearly indicate the enhanced dissolution of 4-aminophenol and aniline into the SDS micelle owing predominantly to the ionic interaction between the ammonium group and the sulfate group. Here one may consider that the transfer free energy of solutes in pure SDS micellar solutions (∆µ°) is governed additively by the hydrophobic and ionic interactions between the solutes and SDS micelles. Then, ∆µ° would be expressed by
∆µ° ) ∆µ°HP + ∆µ°IO
(2)
where HP and IO denote the hydrophobic and ionic interactions, respectively. It is important to assess ∆µ°IO and compare it with ∆µ°HP in order to understand SDS-based MEKC analysis of hydrophobic cations. The capacity factor in MEKC for an analyte is related to ∆µ° in a micellar solution by
-RT ln k′ ) ∆µ° + RT ln Φ
(3)
where Φ is the phase factor characteristic of the micellar system alone, independent of solutes, and expressed by Φ ) Cwater,aq/ {Csf,MνM(Csf - cmc)}, Cwater,aq and Csf,M being the molar concentrations of water in aqueous phase and of the surfactants in micellar phase and νM and Csf being the partial specific volume of the micelle and the analytical concentration of the surfactant, respectively.23 Therefore, the term involving Φ may be canceled out by comparison of two k′ values: a ratio of k′ between two states of a given analyte or between two analytes yields difference in the transfer energy between them. Values of ∆µ°(pH 2.5) - ∆µ°(pH 7.8) for 4-aminophenol and aniline are almost identical with each other and are calculated to be ∼-11 kJ mol-1, while those of the others are negligibly small (Table 1). Taking account of the protonation of 4-aminophenol and aniline on the pH change from 7.8 to 2.5, the ∆µ°(pH 2.5) ∆µ°(pH 7.8) values can be attributed to the difference of ∆µ° between the ammonium and the amino substituents. A value of ∆µ° may be calculated to be 3.43 kJ mol-1 from the k′ ratio between 4-aminophenol and phenol at pH 7.8 within the first approximation. As a result, the ∆µ° value of the ammonium substituent of protonated 4-aminophenol and aniline is evaluated to be ∼-7 kJ mol-1. On the other hand, the difference in ∆µ° between protonated 4-aminophenol and phenol (at pH 2.5) is evaluated as -7.70 kJ mol-1. Considering the difference in the structure of the two analytes, the value can be ascribed to ∆µ° of the ammonium substituent of protonated 4-aminophenol. A
similar argument is applied to the comparison of k′ between protonated tyramine and 4-ethylphenol, yielding -6.11 kJ mol-1 as a difference in ∆µ° at pH 2.5 and is almost equal to that at pH 7.8 (-5.46 kJ mol-1). These values are regarded as the ∆µ° of the ammonium substituent of protonated tyramine. All these ∆µ° values for the ammonium substituents are close to each other and are ∼-7 kJ mol-1. Because the ammonium substituents are hydrophilic and then the corresponding ∆µ°HP would be positive, the ∆µ°IO owing to the ionic interaction between the positive charge (of the ammonium substituents) and the negative charge of the SDS micelle is considered to have a more negative value than -7 kJ mol-1 on the basis of eq 2. On the other hand, values of ∆µ° for nonionic phenol, toluene, and 2-naphthol, for example, in the pure SDS system were reported to be -11, -15, and -17 kJ mol-1, respectively.2 These values are regarded as the ∆µ°HP of the corresponding analytes because no ionic interaction occurs (∆µ°IO ) 0). Therefore, we can reasonably conclude that the ionic interaction between cationic solutes and the anionic SDS micelle is comparable in strength with the hydrophobic interaction between benzenes and the SDS micelle. Such large stabilization energy owing to the ionic interactions might be sometimes inconvenient in MEKC, but it would become useful when the strength can be controlled. Separation in a Mixed Surfactant System. Our attention was directed to the SDS and Tween 20 mixed micelle system by expecting the function of the polyether moiety of Tween 20 to reduce ionic interactions between cationic analytes and charged micelle surface, as an analogy of the reported function of poly(ethylene glycols) to weaken the ionic adsorption.24,25 Figure 1C shows an example of SDS (25 mM)/Tween 20 (5 mM)-based MEKC separation of 4-aminophenol, aniline, phenol, 4-ethylphenol, and tyramine in 10 mM phosphate buffers of pH 2.5. It is noteworthy that the difference in the migration time of tyramine (peak 5), aniline (peak 2), and 4-aminophenol (peak 1) from Oil Yellow AB (peak 6) increased on the addition of Tween 20 as compared with the electropherogram in the absence of Tween 20 (Figure 1B). As a result, complete separation of tyramine and aniline from the Oil Yellow AB was realized. Considering that these three analytes have a positive charge at pH 2.5 and give large k′ values in the pure SDS micelle system (Table 1), this phenomenon means that Tween 20 reduces the strong ionic interaction between the cationic analytes and the anionic micelles. On the other hand, phenol (peak 3) and 4-ethylphenol (peak 4) were detected at decreased migration time on the addition of Tween 20 (Figure 1C and B). These two phenolic analytes are attracted to the mixed micelle by hydrogenbonding interactions with polyether chains of Tween 20 as well as hydrophobic interactions.23,27 As a result, the order and rate of the migration were changed remarkably and then the separation selectivity and efficiency were improved by addition of Tween 20 to the pure SDS system. As an analytical example, adrenaline and its six precursors (Lphenylalanine, L-tyrosine, L-3-methoxytyrosine, L-dopa, dopamine, and L-noradrenaline) were selected as separation targets, where all of them have a positive charge at pH 2.5. Figure 2 shows separations of the analytes in the absence (A) and the presence of 12.5 mM Tween 20 (B) in 10 mM phosphate buffer (pH 2.5) (27) Marangoni, D. G.; Rodenhiser, A. P.; Thomas, J. M.; Kwak, C. T. In Mixed Surfactant Systems; Holland, P. M., Rubingh, D. N., Eds.; ACS Symposium Series 501; American Chemical Society: Washington, DC, 1991; Chapter 11.
Figure 2. MEKC separations of adrenaline and its six precursors in the absence (A) and the presence (B) of 12.5 mM Tween 20 in electrolyte solution containing 25 mM SDS. Conditions: applied voltage, 18 kV; operating current, 19-20 µA (at pH 7.8) and 21-22 µA (at pH 2.5); capillary length, 500 mm (effective length, 200 mm); electrolyte solution; 10 mM phosphate buffer containing surfactants (pH 2.5). Peaks: 0, Oil Yellow AB (a marker of micelle migration); 1, L-phenylalanine; 2, dopamine; 3, L-3-methoxytyrosine; 4, L-tyrosine; 5, L-dopa; 6, adrenaline; 7, L-noradrenaline; 8, benzaldehyde (a reference); 9, formamide (a marker of electroosmotic flow).
containing 25 mM SDS. In the absence of Tween 20, Lphenylalanine (peak 1), dopamine (peak 2), L-3-methoxytyrosine (peak 3), and L-tyrosine (peak 4) were comigrated near the migration time of the SDS micelle (peak 0). L-Dopa (peak 5) and adrenaline (peak 6) also overlapped each other. Addition of Tween 20 improved the separation of these seven analytes remarkably, and a base line separation of all the analytes was achieved at [Tween 20] ) 12.5 mM. The successful separation is attributable primarily to a decrease in the strong and attractive ionic interactions between the cationic analytes and SDS by the polyether chains of Tween 20 and secondarily to the additional weak hydrogen-bonding interactions between the phenolic analytes and the polyether chains of the mixed micelle.23 The relatively short time required for the separation in Figure 2B (13 min) results from very slow electroosmotic flow at pH 2.5 and the moderately strong interactions between the analytes and the micelle as well as the detection on the anodic side. Tween 20 has long polyether chains as hydrophilic groups and the polyether chains will be stuck out the surface of the mixed micelles as depicted in Figure 3.28 The stuck-out polyether moiety may in part draw apart cationic solutes from the sulfate group for the micelle, resulting in the reduction of the ionic interaction. In contrast, the polyether moiety can work also as the proton acceptor toward the proton-donating solutes to form hydrogenbonding complexes. This hydrogen-bonding interaction is attractive but weak in strength. The two additional interactions incorporated by Tween 20 are selective to analytes in strength and in direction (the sign of the interaction energy), resulting in the improved selectivity. Catecholamines and their analogues may be cationic, anionic, zwitterionic, and, in some cases, nonionic. Thus, it would be difficult to separate all of them under a simple CZE mode. CZE methods assisted by complex formation of catechols with boric acid was reported.5,29 Complex formation with borate ion was effective to separate catechols and catecholamines by CZE, and 10 catecholamines were almost separated within 23 min by CZE (28) Ogino, K.; Abe, M. In Mixed Surfactant Systems; Holland, P. M., Rubingh, D. N., Eds.; ACS Symposium Series 501; American Chemical Society: Washington, DC, 1991; Chapter 8. (29) Kaneta, T.; Tanaka, S.; Yoshida, H. J. Chromatogr. 1991, 538, 385-391.
Analytical Chemistry, Vol. 69, No. 7, April 1, 1997
1335
in borate buffer (pH 9.1).29 MEKC would be the alternative to the borate-assisted CZE,5,30 but some of the positively charged amines interacted too strongly to migrate within appropriate periods. Borate complex formation was also utilized in MEKC, and two catechols and five catecholamines were completely separated within ∼10 min by MEKC in borate/phosphate buffer (pH 7), while over 40 min was required to separate them by MEKC in a phosphate buffer (pH 7).5 In a qualitative sense, our SDS/Tween 20-based MEKC method at low pH would be comparable with or superior to borate-assisted CE in the separation efficiency. In our method, the ionic interactions between solutes and micelles are controlled by modification of the micelle surface, while those are controlled by modifying the apparent charge of solutes with the borate complex formation in the borateassisted MEKC. Partitioning Behavior of Solutes in the Mixed Surfactant System. In the optimization of the mixed surfactant-based MEKC, the mixing ratio is an important factor. Therefore, it is valuable to discuss the effects of the concentration of Tween 20 as the micelle modifier on the partitioning behavior of solutes. As reported in the previous paper,23 nonionic surfactants with polyether moieties such as Tween 20 and Brij 35 serve as proton acceptors to attract hydrogen-donating solutes to the mixed micelle. Therefore, the transfer free energy of a solute in the SDS/Tween 20 mixed surfactant system (∆µ°M) is governed by the hydrophobic, ionic, and hydrogen-bonding interactions and then ∆µ° is expressed by
∆µ°M ) ∆µ°HP,M + ∆µ°IO,M + ∆µ°HB,M
(4)
where M denotes the mixed micellar system and ∆µ°HB,M is the hydrogen-bonding interaction energy. On the other hand, the additional interaction energy related to nonionic Tween 20 would be proportional to the analytical concentration of Tween 20 (Cn) at low Cn. Therefore, ∆µ°M can be written as
∆µ°M ) ∆µ°HP + ∆µ°IO + kCn
Figure 3. Schematic diagram of expected effect of polyether chains in the mixed micelle composed of SDS and Tween 20 on the interactions between cationic solutes and the mixed micelle.
(5)
Figure 4. Dependence of -RT ln(k′/k′o) of the five substituted benzenes on the concentration of Tween 20. Conditions as in Figure 1. Key: O, aniline; 0, tyramine; 4, 4-aminophenol; b, 4-ethylphenol; 9, phenol.
a given analyte against that of a certain reference compound (k′o), since Φ is independent of solutes as described above. Thus we get
-RT ln(k′/k′o) ) (∆µ°HP - ∆µ°HP,o) + where k is the proportional constant representing the additional energy related to Tween 20. Considering our proposed function of Tween 20, it provides shielding and attractive effects in terms of the ionic and hydrogenbonding interactions, respectively, and then k would be expressed as k ) katt + kshi, where katt (with the negative sign) and kshi (with the positive sign) are constants representing the attractive and shielding effects. At low Cn, the hydrophobic interaction is assumed to be practically independent of Cn (∆µ°HP,M = ∆µ°HP). Therefore, one can consider that ∆µ°IO,M ) ∆µ°IO + kshiCn and that ∆µ°HB,M ) kattCn. Substitution of ∆µ°M in eq 5 into ∆µ° in eq 3 yields the following equation describing the relation between k′ and Cn.
-RT ln k′ ) ∆µ°HP + ∆µ°IO + kCn + RT ln Φ
(6)
Here the values of Φ is also a function of Cn. In order to eliminate the dependence of Φ on Cn, we will offer a relative value of k′ of (30) Ong, C. P.; Pang, S. F.; Low, S. P.; Lee, H. K.; Li, S. F. Y. J. Chromatogr. 1991, 559, 529-536.
1336
Analytical Chemistry, Vol. 69, No. 7, April 1, 1997
(∆µ°IO - ∆µ°IO,o) + (k - ko)Cn (7)
where ∆µ°HP,o, ∆µ°IO,o, and ko denote the corresponding values of the reference compound. Figure 4 shows -RT ln(k′/k′o) values as functions of the analytical concentration of Tween 20 ([Tween 20]) for phenol, 4-ethylphenol, 4-aminophenol, tyramine, and aniline, where benzaldehyde was used as a reference compound because it is nonionic and has no hydrogen-donating activity, and the interactions of CHO groups with polyethers were found to be very weak as described in our previous works.23,31,32 Tween 20 was added at concentrations from 1 to 7.5 mM to an electrolyte solution of 10 mM phosphate buffer (pH 2.5) containing 25 mM SDS. The values of -RT ln (k′/k′o) go down or up linearly with [Tween 20] in its lower range. After that they appear to level off at increased [Tween 20]. These linear relationships are well described by eq (31) Esaka, Y.; Yamaguchi, Y.; Goto, M.; Haraguchi, H.; Takahashi, J.; Kano, K. Anal. Chem. 1994, 66, 2441-2445. (32) Esaka, Y.; Goto, M.; Haraguchi, H.; Ikeda, T.; Kano, K. J. Chromatogr., A 1995, 711, 305-311.
Table 2. Slopes of -RT ln(k′/k′o) vs [Tween 20] Plots for the Five Substituted Benzenesa analytes
slope (kJ L mol-2)
NH3+
OH
aniline tyramine 4-aminophenol 4-ethylphenol phenol
296 ( 78 150 ( 40 90 ( 31 -281 ( 41 -373 ( 64
+1 +1 +1 0 0
0 1 1 1 1
a Conditions: applied voltage, 18 kV; operating current, 20-21 µA; capillary length, 500 mm (effective length 200 mm); electrolyte solution; 10 mM phosphate buffer containing surfactants (pH 2.5); [SDS] ) 25 mM (fixed); [Tween 20] ) 0-2.5 mM.
Table 3. Slopes of -RT ln(k′/k′o) vs [Tween 20] Plots for Adrenaline and Its Six Precursorsa analytes L-phenylalanine L-3-methoxytyrosine L-tyrosine L-dopa
dopamine adrenaline L-noradrenaline
slopes (kJ L mol-2)
no. of OH groups
charge
194 ( 36 275 ( 30 241 ( 24 226 ( 20 175 ( 26 168 ( 22 59 ( 6
0 1 1 2 2 3 3
+ (
