Download - Electrokinetic Chromatography Using Thermodynamically Stable Vesicles and Mixed Micelles Formed from Oppositely Charged Surfactants

Electrokinetic Chromatography UsingThermodynamically Stable Vesicles and MixedMicelles Formed from Oppositely ChargedSurfactants

Mei Hong, Brian S. Weekley, Sally J. Grieb, and Joe P. Foley*

Department of Chemistry, Villanova University, Villanova, Pennsylvania 19085

The electrokinetic chromatography (EKC) of a novel mixedsurfactant system consisting of oppositely charged sur-factants, sodium dodecyl sulfate (SDS) and n-dodecyltri-methylammonium bromide (DTAB), was investigated. Thechromatographic characteristics of large liposome-likespontaneous vesicles and rodlike mixed micelles formedfrom the mixture were explored and compared with thoseof SDS micelles. Separations of a series of n-alkylphe-nones showed that the spontaneous vesicles providedabout a 2 times wider elution window than SDS micelles.Both vesicle and mixed micelle systems were found toprovide larger methylene selectivity than SDS. The dif-ferent elution order of a group of nitrotoluene geometricisomers with DTAB/SDS spontaneous vesicles and SDSmicelles pseudostationary phases suggested the possibil-ity of different separation mechanisms with these twosystems. Comparisons of polar group selectivity, reten-tion, and efficiency were made between vesicles, mixedmicelles, and SDS micelles. The correlation between thelogarithms of the retention factors (log k′) and octanol-water partition coefficients (log Pow) for a group of 20neutral compounds was also studied with DTAB/SDSvesicles. Spontaneous vesicles have great potential as apseudostationary phase in electrokinetic chromatography.

Micellar electrokinetic chromatography (MEKC), which wasfirst introduced by Terabe in 1984,1-3 is a very common mode ofcapillary electrophoresis for high-efficiency separations. In MEKC,surfactant monomers that form micelles are added to the bufferto act as a pseudostationary phase. This facilitates the separationof neutral molecules on the basis of their hydrophobicity. Also,the partitioning of analytes between the aqueous and micellarphases increases the selectivity and allows the separation of ionswith very similar electrophoretic mobilities. A wide variety ofsuccessful applications of MEKC, including separation of phar-maceutical compounds,4-8 derivatized amino acids,9-11 water- andfat-soluble vitamins,12-15 and herbicides,16 has shown its significantpromise for the analysis of both charged and neutral analytes.

Among the applications, sodium dodecyl sulfate (SDS) is by farthe most popular micellar system in MEKC due to its negativecharge state, low cost, availability in high purity, and low UVabsorption characteristics. However, a drawback of this popularMEKC system at neutral and alkaline pH is the limited elutionrange during which solute elution occurs. Efforts have been madeto extend the elution window by adjusting buffer pH,17,18 modifyingthe capillary,19,20 adding tetramethylammonium ion21 or organicsolvent22-24 to the running buffer, running in a suppressed-flowenvironment,25,26 or using a mixed micellar system.27

It has been found that the retention behaviors and chemicalselectivity in MEKC are influenced by surfactant type signifi-cantly.28,29 Therefore, mixed surfactant systems have greatpotential to provide different selectivities and chromatographic

(1) Otsuka, K.; Terabe, S.; Ando, T. J. Chromatogr. 1985, 348, 39-47.(2) Otsuka, K.; Terabe, S.; Ando, T. J. Chromatogr. 1985, 332, 219-26.(3) Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A.; Ando, T. Anal. Chem.

1984, 56, 111-113.

(4) Croubels, S.; Baeyens, W.; Dewaele, C.; Vanpeteghem, C. J. Chromatogr. A1994, 673, 267-274.

(5) Lukkari, P.; Siren, H.; Pantsar, M.; Riekkola, M. L. J. Chromatogr. 1993,632, 143-148.

(6) Nishi, H.; Tsumagari, N.; Kakimoto, T.; Terabe, S. J. Chromatogr. 1989,477, 259-70.

(7) Nishi, H.; Terabe, S. J. Pharm. Biomed. Anal.1993, 11, 1277-1287.(8) Ong, C. P.; Ng, C. L.; Lee, H. K.; Li, S. F. Y. J. Chromatogr. 1991, 588,

335-339.(9) Little, E. L.; Foley, J. P. J. Microcolumn Sep. 1992, 4, 145-154.

(10) Castagnola, M.; Rossetti, D. V.; Cassiano, L.; Rabino, R.; Nocca, G.; Giardina,B. J. Chromatogr. 1993, 638, 327-334.

(11) Ong, C. P.; Ng, C. L.; Lee, H. K.; Li, S. F. Y. J. Chromatogr. 1991, 559,537-545.

(12) Fujiwara, S.; Iwase, S.; Honda, S. J. Chromatogr. 1988, 447, 133-40.(13) Nishi, H.; Tsumagari, N.; Kakimoto, T.; Terabe, S. J. Chromatogr. 1989,

465, 331-43.(14) Ong, C. P.; Ng, C. L.; Lee, H. K.; Li, F. Y. J. Chromatogr. 1991, 547, 419-

428.(15) Profumo, A.; Profumo, V.; Vidali, G. Electrophoresis 1996, 17, 1617-1621.(16) Wu, Q.; Claessens, H. A.; Cramers, C. A. Chromatographia 1992, 34, 25-

30.(17) Watzig, H.; Lloyd, D. K. Electrophoresis 1995, 16, 57-63.(18) Otsuka, K.; Terabe, S. J. Microcolumn Sep. 1989, 1, 150-4.(19) Balchunas, A. T.; Sepaniak, M. J. Anal. Chem. 1987, 59, 1466-70.(20) Wu, Q.; Claessens, H. A.; Cramers, C. A. Chromatographia 1992, 33, 303-

308.(21) Nielsen, K. R.; Foley, J. P. J. Microcolumn Sep. 1994, 6, 139-149.(22) Bretnall, A. E.; Clarke, G. S. J. Chromatogr. A 1995, 716, 49-55.(23) Chen, N.; Terabe, S.; Nakagawa, T. Electrophoresis 1995, 16, 1457-1462.(24) Chen, N.; Terabe, S. Electrophoresis 1995, 16, 2100-2103.(25) Janini, G. M.; Muschik, G. M.; Issaq, H. J. Electrophoresis 1996, 17, 1575-

1583.(26) Janini, G. M.; Muschik, G. M.; Issaq, H. J. J. Chromatogr. B 1996, 683,

29-35.(27) Ahuja, E. S.; Little, E. L.; Nielsen, K. R.; Foley, J. P. Anal. Chem. 1995, 67,

26-33.

Anal. Chem. 1998, 70, 1394-1403

1394 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998 S0003-2700(97)00730-0 CCC: $15.00 © 1998 American Chemical SocietyPublished on Web 03/04/1998

properties compared to regular (single-component) micelles. Atpresent, most mixed micelles reported have been limited tomixtures of similarly charged surfactants30,31 or to mixtures ofcharged surfactants and “net zero charge” (nonionic or zwitter-ionic) surfactants.27,32 The use of mixed cationic and anionicsurfactants has rarely been examined.21,33

Vesicles are large aggregates (compared to micelles) ofmonomers that have a spherical structure, formed by a bilayerthat surrounds an internal cavity. The hydrophobic part of thebilayer is expected to provide hydrophilic-hydrophobic discrimi-nation power. One type of commonly encountered vesicles isthose formed from phospholipid molecules.34 Often termedliposomes, they have been used as model membranes, drugdelivery devices, and microreactors. A review by Lundhal andYang35 illustrated the use of liposomes for separating biomoleculesby direct ion-exchange with the bilayer membrane and by affinityinteractions using bilayer-immobilized receptors. Recently, it hasbeen suggested that it might be possible to separate other typesof solutes using liposomes as pseudostationary phase in capillaryelectrophoresis (CE).36,37 Roberts et al. demonstrated that themigration time of riboflavin increased in the presence of DiI-dopedliposomes (liposomes with 1,1′-dioctadecyl-3,3,3′,3′-tetramethyl-indodicarbocyanine as cationic membrane probe).36 Zhang et al.also investigated drug- and peptide-liposome interactions byCE.37

Alternatively, spontaneous vesicles can be prepared fromoppositely charged surfactants. It has been found that aqueousmixtures of anionic and cationic surfactants exhibit many uniqueproperties due to the strong electrostatic interactions between theoppositely charged headgroups. Interestingly, microstructures notformed by the pure components, such as vesicles and/or rodlikemicelles, can be observed upon mixing the two surfactantstogether.38,39 Compared to liposomes that are assembled fromphospholipid molecules, the spontaneous vesicles have theadvantages of ease of preparation, controllable size and surfacecharge, thermodynamic stability, and lower cost. The phasebehavior of aqueous mixtures of n-dodecyltrimethylammoniumbromide (DTAB) and SDS was investigated and described byHerrington and Kaler40 in 1993. The phase diagram provided bythe authors indicated that there was a narrow one-phase vesiclelobe (V) in SDS-rich mixtures and large mixed micelle regions atboth sides of the phase diagram. Compared to SDS micelles, the

DTAB/SDS vesicles have a much larger diameter (severalhundred nanometers) and a different surface charge distribution.These characteristics may explain some of the unique chromato-graphic properties of this system. Due to the presence of largeparticles, vesicle solutions appear bluish to the eye.

It is anticipated that the spontaneous vesicles can providepotentially better efficiency and selectivity as well as a longerelution window as a new electrokinetic separation media. In thisstudy, we examined the DTAB and SDS mixed cationic/anionicsurfactant system since these surfactants are commonly encoun-tered and available at relatively low cost. The mixed micelleregion we chose to study was adjacent to the vesicle lobe on thephase diagram. Our purpose was to compare these two systemsthat have very similar compositions but different microstructures.To the best of our knowledge, we report here the first investigationof spontaneous vesicles as an alternative pseudostationary phasein electrokinetic chromatography.

EXPERIMENTAL SECTIONApparatus. A Waters Quanta 4000E capillary electrophoresis

system (Waters Inc., Milford, MA) equipped with fixed-wavelengthUV detection at either 254 or 214 nm was employed for all theseparations performed in this study. Electrokinetic separationswere performed in either a 47.5-cm (total length) × 75-µm-i.d.(366-µm-o.d.) or a 37.5-cm × 50-µm-i.d. (368-µm-o.d.) fused-silicacapillary (Polymicro Technologies, Tucson, AZ), except for thevan Deemter studies, which were performed with a 35-cm × 50-µm-i.d. capillary. Injections were made hydrostatically for 2 s (ata height of 9.8 cm). The data were collected at a rate of 5 Hz andprocessed on an NEC Image 466es computer (Milford, MA) usingMillennium 2000 or 2010 software (Waters, Inc.). A Perkin-ElmerLambda 4B UV/visible spectrophotometer (Perkin-Elmer, Nor-wark, CT) was employed to obtain the UV-visible absorbancespectra. All experiments were performed at ambient temperature(25 °C). The mean vesicle size was determined by laser-assistedphoton correlation spectroscopy (PCS) using a Brookhaven modelBT200SM particle size analyzer and a BI9000 autocorrelator. AHoriba LA-910 laser scattering particle size analyzer was employedto analyze the vesicle size distribution.

Materials. The n-alkylphenone homologous series was pur-chased as a kit from Aldrich (Milwaukee, WI). Sodium dodecylsulfate (SDS) was purchased from Sigma (St. Louis, MO), whilen-dodecyltrimethylammonium bromide (DTAB) was purchasedfrom Lancaster (Windham, NH). All surfactants were used asreceived. The buffers, sodium phosphate and HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid) were purchasedfrom J. T. Baker (Phillipsburg, NJ). Neutral test analytes werepurchased from Aldrich unless otherwise noted. The neutral testmixture consisted of benzene (J. T. Baker), butylbenzene, benzylalcohol (EM Science, Gibbstown, NJ), nitrobenzene, nitroethane,anisole, naphthalene, benzaldehyde, methyl benzoate, p-chloroni-trobenzene, benzophenone, biphenyl (MCB Reagents, Cincinnati,OH), 2-, 3-, and 4-nitrotoluene (Chem Service, West Chester, PA),toluene (J. T. Baker) bromobenzene (Fisher Scientific, Fair Lawn,NJ), and chlorobenzene (Chem Service). HPLC grade water usedfor all solutions was obtained from J. T. Baker.

Both phosphate buffer and HEPES buffer were examined aselectrolytes. Phosphate buffer gave a slightly longer elution range

(28) Yang, S. Y.; Khaledi, M. G. Anal. Chem. 1995, 67, 499-510.(29) Yang, S. Y.; Bumgarner, J. G.; Khaledi, M. G. J. Chromatogr. A 1996, 738,

265-274.(30) Yang, S. Y.; Bumgarner, J. G.; Kruk, L. F. R.; Khaledi, M. G. J. Chromatogr.

A 1996, 721, 323-335.(31) Wallingford, R. A.; Curry, P. D., Jr.; Ewing, A. G. J. Microcolumn Sep. 1989,

1, 23-7.(32) Ahuja, E. S.; Preston, B. P.; Foley, J. P. J. Chromatogr. 1994, 657, 271-

284.(33) Ong, C. P.; Ng, C. L.; Lee, H. K.; Li, S. F. Y. Electrophoresis 1994, 15, 1273-

1275.(34) Fendler, J. Membrane Mimetic Chemistry; Wiley: New York, 1983.(35) Lundahl, P.; Yang, Q. J. Chromatogr. 1991, 544, 283-304.(36) Roberts, M. A.; Locasciobrown, L.; MacCrehan, W. A.; Durst, R. A. Anal.

Chem. 1996, 68, 3434-3440.(37) Zhang, Y. X.; Zhang, R.; Hjerten, S.; Lundahl, P. Electrophoresis 1995, 16,

1519-1523.(38) Huang, J. B.; Zhao, G. X. Colloid Polym. Sci. 1995, 273, 156-164.(39) Kaler, E. W.; Herrington, K. L.; Murthy, A. K. J. Phys. Chem. 1992, 96,

6698-6707.(40) Herrington, K. L.; Kaler, E. W. J. Phys. Chem. 1993, 97, 13792-13802.

Analytical Chemistry, Vol. 70, No. 7, April 1, 1998 1395

(10%-15%) due to the higher ionic strength, which results indouble-layer compression, decreased ú potential, and reducedEOF. But since HEPES is a zwitterionic buffer that has muchlower conductivity, Joule heating is less significant, and it wasused as the electrolyte in all separations reported in this paper.Stock buffer solutions were prepared with HEPES and sodiumhydroxide (NaOH) to give a 50 mM HEPES buffer at pH 7.2. AHEPES buffer concentration of 10 mM was used in all theexperiments. The SDS (175 mM) micelle stock solution wasprepared in distilled water. SDS (35 mM) in HEPES at pH 7.2running buffer was used for MEKC experiments. The stocksolutions of DTAB and SDS for making vesicles or mixed micelleswere prepared in the same way as the SDS micelle stock solution.Vesicle and mixed micelle solutions for electrokinetic separationswere made by pipetting the two stock solutions and buffer stockinto a 25-mL volumetric flask, diluting with distilled water, andthen vortex-mixing for several minutes. The vesicle solution hada total surfactant concentration of 1% w/v and a weight ratio of39.06/60.94 DTAB/SDS. The mixed micelle solution had thesame total surfactant concentration as the vesicle solution but adifferent ratio of 30/70 DTAB/SDS. All vesicle solutions wereequilibrated at least overnight in a 25 °C isothermal water bath.Other solutions were kept at room temperature. Electrolytes werefiltered through either 0.20-µm (SDS micelle and/or mixedmicelle) or 0.45-µm (vesicle) membrane filters obtained fromAlltech Associates, Inc. (Deerfield, IL). Sample solutions weremade up of 50% acetonitrile and 50% HEPES buffer, with soluteconcentrations of approximately 0.2 mg/mL.

Methods. The capillary was activated by purging with 0.1 MNaOH for 20 min, followed by distilled water for 20 min. Thecapillary was then purged for 20 min with the operating buffer.Purges with the buffer without any surfactant were performed aftereach run for 4-8 min to keep the adsorption of surfactant on thecapillary wall to a minimum. Ohm’s law plots were prepared todetermine the linear voltage range where Joule heating would notcause significant nonuniform temperature gradients and subse-quent zone broadening. To keep the analysis time short, thehighest operating voltage within the linear part of the Ohm’s plot,17 kV (75-µm-i.d. capillary) or 21 kV (50-µm-i.d. capillary), waschosen for each set of separation conditions.

Immediately prior to each mean particle size and size distribu-tion measurement, the sample was filtered through a Millex-HV13-µm filter unit with a 0.45-µm nominal pore size.

Calculations. The electroosmotic mobility, µeo, was deter-mined for each system from the relationship

where Ld is the injector-to-detector column length, Lt is the totallength of the capillary, V is the applied voltage, and to is themigration time of acetonitrile.

The tmc value, which represents the elution time of thepseudostationary phase for each separation, was calculated usingalkylphenone homologues with the iterative computation methoddeveloped by Bushey and Jorgensen41 and confirmed by usingdecanophenone as the tmc marker.

The electrophoretic mobility of the pseudostationary phase,µep, was calculated from the following relationship:

where µmc was calculated using eq 1 by substituting tmc for to.

The retention factor, k, was calculated using the equation

where tR is the migration time of the solute.Methylene selectivity (RCH2) was calculated by measuring the

slope of a regression line of log k versus alkyl chain length(number of -CH2- units) of a series of alkylphenones (acetophe-none to hexanophenone).

The plate-count equation based on W0.1 was used to calculatethe efficiency:

Polar group selectivity (RPG) was calculated from the followingrelationship:

where ks is the retention factor of the solute and kb is the retentionfactor of benzene.

RESULTS AND DISCUSSIONVesicle size is composition-dependent. When more cationic

surfactant was added to the mixture (approaching the equimolarcomposition), the headgroup repulsion decreased; hence, thecurvature decreased, and, correspondingly, the vesicle sizeincreased.42 At the surfactant ratio used in this study, thediameters of the DTAB/SDS vesicles are expected to be around100 nm, much larger than those of the SDS micelles (severalnanometers). The DTAB/SDS mixed micelles are smaller thanvesicles but larger than SDS micelles.

Optical Characteristics of Surfactant Systems. The absor-bance spectra (1-cm path length) of SDS micelles, DTAB/SDSmixed micelles, and DTAB/SDS vesicles are shown in Figure 1.Due to the light-scattering properties of large molecular ag-gregates, the apparent absorbance of the DTAB/SDS vesicles islarge in both the visible and UV regions of the electromagneticspectrum compared to the SDS micelles and DTAB/SDS mixedmicelles. To the eye, the mixed micelle solution looks clear andcolorless, while the vesicle solution appears bluish. Nevertheless,the DTAB/SDS vesicles proved to be fully compatible with UVabsorbance detection in our study because of the relatively shortpath lengths across the capillary.

Mean Particle Size and Size Distribution. The mean sizeand distribution of the DTAB/SDS vesicles were measured bylaser-assisted photon correlation spectroscopy (PCS) at various

(41) Bushey, M. M.; Jorgenson, J. W. J. Microcolumn Sep. 1989, 1, 125-130.

(42) Kaler, E. W.; Herrington, K. L.; Zasadzinski, J. A. N. Materials ResearchSociety Symposium Proceedings; Materials Research Society: Pittsburgh, PA,1992; pp 3-10.

µeo )Ld Lt

Vto(1)

µep ) µeo - µmc (2)

k )tR - to

to[1 - (tR/tmc)](3)

N ) 18.42(tR/W0.1)2 (4)

RPG ) ks/kb (5)

1396 Analytical Chemistry, Vol. 70, No. 7, April 1, 1998

time intervals. Sometimes, a very small amount of fine crystalswas observed floating in the solution after it had aged forapproximately 1 day; therefore, the samples for mean particle sizeand size distribution analyses were filtered through a 0.45-µmmembrane filter. This pore size was the same as that of themembrane filters used to filter the electrolytes before a separation.

Figure 2 illustrates the time dependence of the vesicle size.It can be seen that, within the first several hours after solutionpreparation, the mean particle size was relatively small andincreased rather quickly (about 80% in the first 8 h). But thegrowth rate of the vesicles gradually slowed: within the first 2

days, the size of the vesicles increased less than 10% to ap-proximately 110-120 nm (day 1 to day 2), and in the next 3 daysit increased about 17% to approximately 140 nm (day 2 to day 5).

Most of the vesicle solutions used in this study were either 1day or 2 days old and, thus, nearly if not completely at equilibrium(Figure 2). Importantly, no significant chromatographic differenceswere observed among such solutions. Moreover, although smallamounts of precipitate were occasionally observed, no changes inthe chromatographic properties were observed upon filtration andreuse.

In terms of vesicle size distribution, no significant differenceswere observed between day 1 and day 2 for a given solution(distribution measurements were not made after 48 h); thedistribution of vesicle sizes (reported as the standard deviation)varied from solution to solution and ranged from 15 to 31 nm.Although a more reproducible distribution of vesicle size couldpossibly be achieved with still greater care in vesicle preparation,for purposes of separation or partitioning studies, the slightpolydispersity in vesicle size is unimportant since, as mentionedearlier, the chomatographic properties (efficiency, selectivity,retention, elution range) of all vesicle solutions aged for at least24 h were essentially the same.

Specific Considerations on Vesicle Preparation. Since thevesicle solution used for the electrokinetic separations was madein buffer instead of pure water, its ionic strength was differentfrom that in the phase behavior study. According to the phasediagram,40 we first made the vesicle solution in the center of thelobe (DTAB/SDS 37.5/62.5 w/v ratio), but the results obtainedwith this system were not reproducible. Probably, the increasein ionic strength caused a slight shift of the vesicle lobe in thephase diagram. The stable vesicle system that was used in thisstudy was found by slightly modifying the surfactant ratio.

Because the size of the vesicles and mixed micelles wascomposition dependent, solutions that are prepared with insuf-ficient mixing produce widely varying local compositions and acorrespondingly wide distribution of initial sizes. In this investiga-tion, the second stock solution was added into the first one slowlywhile vortex-mixing. After the solution was brought up to volume,it was vortex-mixed for several more minutes. This method ofpreparation resulted in satisfactory precision without internalstandards; the relative standard deviations (RSDs) of migration

c

b

a

Figure 1. Room-temperature absorbance spectra of (a) SDSmicelles, (b) DTAB/SDS mixed micelles, and (c) DTAB/SDS vesiclesin water. Solution composition is as described in the ExperimentalSection.

Figure 2. Time dependence of the average size DTAB/SDSvesicles.


times were approximately 1%, while RSDs of absolute peak areaswere less than 3%.

Phase Ratio. In MEKC, the phase ratio (b) is defined as thevolume of the micellar phase divided by the volume of the aqueousphase. It can be written in terms of surfactant concentration[SURF], the critical micelle concentration (cmc), and the partialmolar volume (V) of the surfactant:

By substituting the partial molar volume, 0.246 L/mol, and thecmc, 4.5 × 10-3 M (in buffer) of SDS into the above equation, thephase ratio of 35 mM SDS micellar solution was calculated to be7.56 × 10-3.

When mixtures of two or more surfactants produce mixedmicelles, vesicles, or other aggregate structures from the surfac-tant monomers, the volume of the aggregated phase is the sumof the volumes occupied by the individual surfactants. The phaseratio may, therefore, be estimated from

where øi is the overall mole fraction of the ith surfactant, CAC isthe critical aggregation concentration (total surfactant concentra-tion above which aggregates are observed), and the mole fractionsof surfactants in the monomer and aggregated states are assumedto be approximately if not exactly the same. The CAC turnedout to be the cmc of the predominant pure component40 (whichwas SDS). Since the molar ratio of DTAB and SDS was 1:1.67,the cmcis were estimated as cmcdtab ≈ cmctotal {[DTAB]/([DTAB]+ [SDS])} and cmcsds ≈ cmctotal {[SDS]/[DTAB] + [SDS])}instead of using the cmcs of pure DTAB and SDS, which wouldbe inappropriate for the vesicle mixture. The partial molar volumeof DTAB, 0.325 L/mol, was measured by dissolving a certainamount of surfactant in distilled water and then determining theincreased volume from the weight of the expanded amount ofwater. Using these approaches, the phase ratio of the vesicularsolution was 8.14 × 10-3, slightly higher than that of the SDSmicelle solution. The phase ratio of DTAB/SDS mixed micellesolution was calculated similarly after changing the DTAB/SDSmolar ratio to 1:2.49, and was found to be 7.98 × 10-3.

Elution Range and Methylene Selectivity. The separationof a series of alkylphenones (acetophenone to hexanophenone)using DTAB/SDS vesicles in HEPES buffer is shown in Figure3c. For comparison, a SDS solution (35 mM) and a solution fromthe SDS-rich mixed micelle region which is very close to thevesicle lobe of the phase diagram (30/70 DTAB/SDS w/v) wereprepared. Both solutions were clear and colorless and had thesame total concentration of surfactant(s) as the vesicle solution.Electropherograms for the separation of the same alkylphenoneseries are shown in Figure 3a and b.

On examining the electropherograms, it is immediately obviousthat the vesicle solution provides a significantly larger elutionrange, as exemplified by the much larger resolution betweenpeaks 4 and 5 compared to that in either micellar solution.

Interestingly, the electroosmotic flow was approximately the samein all systems, but the migration time of the most hydrophobiccompound was much longer with the vesicle solution. The peakshapes were generally very good, except that hydrophobic peakswere broader for the vesicle solution. The major part of this

â )V([SURF] - cmc)

1 - V([SURF] - cmc)(6)

â ) ∑Vsurf,i(Csurf,i - øiCAC)

1 - ∑Vsurf,i(Csurf,i - øiCAC)(7)

c

b

a

Figure 3. Comparison of alkylphenone separation using (a) SDSmicelles, (b) DTAB/SDS mixed micelles, and (c) DTAB/SDS vesicles.Conditions: sample and running buffers, 10 mM pH 7.2 HEPES;applied voltage, 17 kV (33-39 µA); capillary dimensions (bare fusedsilica), 75 µm i.d. × 40 cm (effective length). Solute identification:(1) acetophenone, (2) propiophenone, (3) butyrophenone, (4) vale-rophenone, and (5) hexanophenone.


broadening was due to the lower migration velocity of thehydrophobic compounds across the detector window. By dividingthe width of each peak by its migration time,43 it was found thatthe normalized peak width of the hydrophobic compounds onlyincreased by 20%-30% per methylene unit. The separationabilities of the three systems were further compared in terms ofelution range tmc/to, methylene selectivity RCH2, µeo, µep, andefficiency N. The results are shown in Table 1.

Under the investigation conditions, all three aggregates arenegatively charged and migrate against the electroosmotic flow(EOF). The electroosmotic mobilities of each solution are verysimilar indicating that the viscosities and ú potentials at the doublelayer of the mixed cationic/anionic solutions are similar to thoseof the SDS micelle solution. Although the vesicles have muchlarger size than the micelles, their surface-to-volume ratio (in-versely proportional to the radius) is much smaller than that ofthe micelles. Therefore, the frictional drag per unit volume isactually smaller for these vesicles than for micelles. Furthermore,due to their bilayer structures, the vesicles may have a highercharge density than typical micelles. Hence, the DTAB/SDSvesicles have larger electrophoretic mobilities in the oppositedirection of EOF. This gives the vesicle solution a migrationwindow that is about 2 times longer than that of the micellarsolutions.

Surprisingly, the efficiency of the vesicle solution is somewhatbetter than that of the micellar solutions. In contrast, a decreasein efficiency was observed with liposomes formed from lipids usedas the pseudostationary phase in EKC.36,37 In this study, thehighest efficiency achieved was about 260 000 plates usingvesicles, without optimization of the sample buffer and runningbuffer composition, ionic strength, and the detector slit width.

Since peak capacity nc ) 1 + N1/2/4 ln(tmc/to), the longerelution range and the higher efficiency give the vesicle solutiona significantly larger peak capacity than for either micellar system.The maximum attainable resolution (Rs)max is also related to theefficiency and elution range,44 as shown in the equations below,applicable to neutral solutes:

Assuming an efficiency (N) and selectivity (R) of 150 000 and 1.03,respectively, the maximum attainable resolution of the DTAB/SDS vesicle solution (tmc/to ) 6.74) would be 1.25, whereas the(Rs)max for the SDS micelle solution (tmc/to ) 3.12) would be 0.78.

The higher methylene selectivity of the vesicles and the mixedmicelles could possibly be explained by “the degree of waterpenetration” model that is under further investigation in ourresearch group. This model was conceived after our observationthat zwitterionic buffers such as HEPES provided larger methyleneselectivity than anionic buffers such as phosphate at the same

pH and ionic strength. This may be due to interactions of thehydrophobic and positively charged regions of HEPES with thehydrophobic and negatively charged regions of SDS, resulting ininsertion of the HEPES molecules into the micelle and partiallyoccupying the space between SDS molecules. As a result, fewerwater molecules could penetrate into the micelle, and thehydrophobicity of the interior of the micelle increased. Since thehydrophobic interaction between DTAB and SDS molecules isstronger than that between the HEPES and SDS molecules, andthe DTAB concentration was higher than the HEPES concentra-tion in the vesicle and mixed micelle solutions, it is more difficultfor the water molecules to access the centers of these microstruc-tures. Presumably, it is due to the structures of vesicles, andmixed micelles would be more rigidly packed than SDS micellesin the same buffer. The lower degree of water penetration andhigher inherent hydrophobicity of the vesicles and the mixedmicelles provided a significantly higher hydrophobic selectivitythan that of SDS micelles.

Reduction in Analysis Time. Analysis time is anotherimportant parameter to evaluate a separation system. To shortenthe analysis time of vesicle-mediated separations, we also used a50-µm-i.d. × 30-cm (injector to detector) capillary in place of the75-µm-i.d. × 40-cm capillary. Since Joule heating is proportionalto the square of the capillary diameter, it will be significantly lowerwith the smaller inner diameter capillaries; thus, a higher electricfield can be employed. Figure 4 shows the separation of the samealkylphenone series as Figure 2 but with a 50-µm-i.d. column. Theapplied voltage was 21 kV. All of the five analytes eluted within7.5 min, whereas it took about 18 min for all the peaks to elutefrom the 75-µm-i.d. capillary. The major shortcoming of the smallerinner diameter column is that the signal-to-noise ratio and thelimit of detection are somewhat lower due to the shorter opticalpath length. However, this can be alleviated by the use ofextended path (“bubble”) capillaries or “Z” flow cells that are nowcommercially available.

Polar Group Selectivity. The polar group selectivity is theratio of the solute retention factor to the retention factor ofbenzene between the vesicle and the micellar solutions. Table 2lists the retention factors and polar group selectivities for a numberof substituted benzenes. By examining the results in Table 2, itcan be shown that, for the relatively hydrophilic compounds (i.e.,benzaldehyde, benzyl alcohol, and nitrobenzene), the vesiclesolution provided the lowest degree of retention, whereas anequimolar SDS micellar solution provided the highest. Con-versely, hydrophobic compounds such as toluene and chloroben-

(43) Hjerten, S.; Elenbring, K.; Kilar, F.; Liao, J.-L.; Chen, A. J. C.; Siebert, C. J.;Zhu, M.-D. J. Chromatogr. 1987, 403, 47-61.

(44) Foley, J. P.; Ahuja, E. S. In Pharmaceutical and Biomedical Applications ofCapillary Electrophoresis; Lunte, S. M., Radzik, D. M., Eds.; PergamonPress: Tarrytown, NY, 1996; Vol. 2, pp 81-178.

Rs )xN

4 (R - 1R )( k

1 + k)( 1 - to/tmc

1 + k(to/tmc)) (8)

kopt(max Rs) ) xtmc/to (9)

Table 1. Comparison of Electrokinetic andChromatographic Properties of SDS Micelles, DTAB/SDS Mixed Micelles, and DTAB/SDS Vesicles

SDSmicelle

DTAB/SDSmixed micelle

DTAB/SDSvesicle

µeo (cm2/(V‚s)) 6.49 × 10-4 6.63 × 10-4 6.65 × 10-4

µep (cm2/(V‚s)) -4.20 × 10-4 -4.15 × 10-4 -6.53 × 10-4

tmc/to 3.12 2.97 6.74Npropiophenone 66 600 32 600 105 000Npropiophenone/m 167 000 81 500 263 000Nbutyrophenone 41 200 27 500 59 700Nbutyrophenone/m 103 000 68 800 149 000RCH2 2.18 2.98 3.04


zene had the largest retention factors with the vesicles and thelowest with SDS micelles as the pseudostationary phase. It isnot yet clear if the separation mechanism differs significantlybetween the vesicle- and micelle-mediated electrokinetic chroma-tography systems, or if differences in chromatographic propertiessuch as selectivity merely reflect the significant difference in theaggregated structures of vesicles, mixed micelles, and micelles.

To make the comparison more explicit, a group of substitutedbenzenes was separated with the three solutions. The electro-pherograms are shown in Figures 5. The superior selectivity ofthe DTAB/SDS vesicle solution is immediately obvious, since theresolution of all eight compounds was achieved. In contrast,compounds 3-5 (methyl benzoate, p-nitrotoluene, and p-nitro-chlorobenzene) and compounds 4 and 5 completely coeluted withSDS micelles and DTAB/SDS mixed micelles, respectively.

“Shape Selectivity” for Geometrical Isomers. The abilityof a stationary phase to separate closely related compounds orstructural isomers has been investigated extensively in HPLC (seeref 45 for a review) but much less so in MEKC. In the presentstudy, a set of geometrical isomers (p-, o-, m-nitrotoluene) wasemployed as probes to compare the shape selectivity amongvesicle and micellar solutions; the chromatograms are shown in

Figure 6. Using DTAB/SDS vesicles as the pseudostationaryphase, the three geometric isomers were almost baseline sepa-rated, whereas two of them were only partially separated usingSDS micelle solution. Using DTAB/SDS mixed micelles, onlytwo partially separated peaks were observed. It is very interestingthat the elution order of o-nitrotoluene and p-nitrotoluene wasreversed when using the vesicle solution compared to the SDSmicelles. Using the DTAB/SDS vesicle, p-nitrotoluene eluted firstfollowed by o-nitrotoluene, whereas with the SDS micellar solution,o-nitrotoluene eluted first, followed by p-nitrotoluene. The elutionorder for the DTAB/SDS mixed micelles was the same as thatfor the vesicle solution, but o- and m-nitrotoluene coeluted andwere only partially separated from the p-nitrotoluene peak. Thechange of elution order between the DTAB/SDS vesicle and SDSmicelle could be due to the different surface charge of thecationic/anionic mixtures. This implies that there may be somedifferent separation mechanism other than hydrophobic interac-tions, i.e., electrostatic interactions. If this is true, uniqueselectivities could be expected with the vesicle pseudostationaryphase.

Van Deemter Studies of DTAB/SDS Vesicle System. Thedependence of plate height on electric field strength is shown inFigure 7 for six test solutes. Generally, relatively hydrophiliccompounds (i.e., acetophenone, benzaldehyde, and propiophe-none) had better efficiencies (lower plate heights) over the entirefield strength range. The optimum field strength for hydrophiliccompounds was around 400 V/cm, whereas it was around 600V/cm for hydrophobic compounds (butyrophenone, valerophe-none, and benzophenone). The reason for this apparent trend isnot clear and is being investigated. Given power levels of 0.70,1.89, and 3.76 W/m at field strengths of 400, 600, and 800 V/cm,respectively, and the onset of nonlinearity in the Ohm’s law plotfor these vesicles at about 1.2 W/m (50-µm-i.d. capillaries), wewould expect Joule heating to play a role, along with resistanceto mass transfer. However, it is not possible to deconvolute theseeffects under the present experimental conditions.

Estimating the Hydrophobicity of Neutral Compounds.Hydrophobic interactions play a critical role in many kinds oftransmembrane processes. Therefore, a physical-chemical methodthat can quantitatively measure the hydrophobicity of biologicallyactive substances is of great importance for drug design, toxicol-ogy, and other related fields.

The octanol-water partition coefficient (Pow), which was firstproposed by Fujita et al. in the early 1960s,46 has become the(45) Sander, L. C.; Wise, S. A. J. Chromatogr. 1993, 656, 335-351.

Table 2. Comparison of Retention, Relative Retention, and Polar Group Selectivity of Substituted Benzenesbetween SDS Micelle, DTAB/SDS Mixed Micelle, and Vesicle Solutions


DTAB/SDSvesicle

substitutedbenzenes

SDSmicelle

k′ k′k′MM/k′SDS k′

k′V/k′SDS

SDSmicelle

RPG


RPG

DTAB/SDSvesicle

RPG

benzyl alcohol 0.33 0.25 0.76 0.17 0.52 0.51 0.23 0.18benzaldehyde 0.61 0.45 0.73 0.32 0.52 0.94 0.41 0.34benzene 0.65 1.08 1.66 0.94 1.45nitrobenzene 0.82 1.08 1.32 0.90 1.10 1.27 1.00 0.96toluene 1.76 1.54 0.88 3.21 1.82 2.72 1.43 3.42chlorobenzene 2.32 4.95 2.13 5.74 2.47 3.58 4.60 6.12p-nitrotoluene 2.40 1.54 0.64 2.61 1.09 3.70 1.43 2.78p-chloronitrobenzene 2.40 1.54 0.64 2.71 1.13 3.70 1.43 2.88

Figure 4. Reduction in the time required to separate a series ofalkylphenones using DTAB/SDS vesicles via a reduction in capillarydimensions and an increase in the applied voltage. A 50-µm-i.d. ×30-cm (effective length) bare fused silica capillary was used. Appliedvoltage, 21 kV; current, 27 µA; other conditions are as in Figure 3.


standard. The direct determination of Po/w values by the shake-flask technique is very time-consuming and inconvenient. Many

attempts have been made to measure the hydrophobicity in afaster, less costly, and theoretically more correct way.47-50

(46) Fujita, T.; Iwasa, J.; Hansch, C. J. Am. Chem. Soc. 1964, 86, 5175-5180.(47) Medinahernandez, M. J.; Sagrado, S. J. Chromatogr. A 1995, 718, 273-

282.

c

b

a

Figure 5. Comparison of polar group selectivity for the separationof substituted benzenes using (a) SDS micelles, (b) DTAB/SDS mixedmicelles, and (c) DTAB/SDS vesicles. Solute identification: (1)benzaldehyde, (2) nitrobenzene, (3) methyl benzoate, (4) p-nitrotolu-ene, (5) p-chloronitrobenzene, (6) m-nitrotoluene, (7) benzophenone,and (8) biphenyl. Other conditions are as in Figure 3.

c

b

a

Figure 6. Comparison of shape selectivity via the separation of thegeometrical isomers of nitrotoluene using (a) SDS micelles, (b) DTAB/SDS mixed micelles, and (c) DTAB/SDS vesicles. Solute identifica-tion: (1) p-nitrotoluene, (2) o-nitrotoluene, and (3) m-nitrotoluene.Other conditions are as in Figure 3.


Although using log Po/w values for explicating the biologicalpartitioning has been suggested,48 there are remarkable mecha-nism differences between the two processes. Biological mem-branes consist of phospholipids arranged in anisotropic bilayerstructures, and the physical properties vary with distance fromthe interface. In contrast, for bulk-phase hydrocarbon-waterpartitioning, the physical properties within each phase are uniform.Studies have shown that membrane-water partitioning is de-pendent not only on the hydrophobic interaction but also on thesurface density of the bilayer chains.51 The higher the surfacedensity, the lower the partitioning through the membrane.Therefore, liposomes, cells, and micelles have been proposed asalternative models. MEKC has been evaluated as a method forthe estimation of hydrophobicities52 and a model for biopartition-ing.53

In this study, the feasibility of vesicular electrokinetic chro-matography (VEKC) as an improved method for the estimationof hydrophobicities was explored. Compared to micelles, spon-taneous vesicles, which have a bilayer structure, more closelymodel biomembranes. The surface density of hydrocarbon chainsfor these vesicles is variable by selecting different surfactants and/or changing the composition, and, like in MEKC, the experimentalconditions (pH, ionic strength, temperature) can be carefullycontrolled to mimic physiological conditions.

The logarithms of retention factors for a group of neutralsolutes were measured and calculated with the DTAB/SDS vesicleas a pseudostationary phase. Since the linear range of a correla-tion curve is always an important parameter to consider inevaluating a new method, compounds with a log Po/w value rangingfrom 0.18 to 4.44 were analyzed. The log Po/w data were takenfrom literature.54 The log k′ and log Po/w data for the 20 solutesare listed in Table 3, and the correlation plot is shown in Figure

8. Linear regression gives the equation

To further examine the validity of this correlation withstructurally unrelated compounds, 13 noncongeneric solutes outof the 20 were selected and plotted (Figure 8). The linear least-squares fit equation is

Obviously, more compounds could have been included in orderto increase the breadth and significance of the correlation. Webelieve, however, that the results obtained with the 20 compoundsof the present study (13 noncongeners) are sufficient to illustratethe potential of vesicles as a model for the estimation of thehydrophobicity and/or biopartitioning of neutral compounds.

(48) Lambert, W. J. J Chromatogr 1993, 656, 469-484.(49) Chen, N.; Zhang, Y. K.; Terabe, S.; Nakagawa, T. J. Chromatogr. A 1994,

678, 327-332.(50) Kaliszan, R. Quantitative Structure-Activity Chromatographic Retention

Relationships; Wiley: New York, 1987.(51) Rosen, M. J. Surfactants and Interfacial Phenomena; John Wiley and Sons:

New York, 1989.(52) Herbert, B. J.; Dorsey, J. G. Anal. Chem. 1995, 67, 744-749.(53) Woodrow, B. N.; Dorsey, J. G. Environ. Sci. Technol. 1997, 31, 2812-

2820.(54) Hansch, C.; Leo, A. Exploring QSAR: Fundamentals and Applications in

Chemistry and Biology; American Chemical Society: Washington, DC, 1995.

Figure 7. Van Deemter study of efficiency in vesicular electrokineticchromatography (VEKC) using six neutral compounds with DTAB/SDS vesicles.

Table 3. Data for log k′-log Pow Correlation for 20Neutral Compounds

compoundalogPb

logk′ compounda

logPb

logk′

nitroethanea 0.18 -1.33 1,4-chloronitrobenzene 2.39 0.431-nitropropane 0.87 -1.00 1,3-chloronitrobenzenea 2.46 0.63benzyl alcohola 0.87 -0.77 toluene 2.73 0.51benzaldehydea 1.48 -0.49 butyrophenone 2.77 0.52acetophenonea 1.58 -0.39 chlorobenzenea 2.89 0.76nitrobenzenea 1.85 -0.035 bromobenzenea 2.99 0.90anisolea 2.11 0.13 benzophenonea 3.18 1.02benzene 2.13 -0.018 naphthalenea 3.37 1.32propiophenone 2.19 0.098 biphenyla 3.95 1.594-nitrotoluene 2.37 0.42 butylbenzenea 4.44 1.89

a Noncongeneric compounds. b Reference 54.

Figure 8. Correlation of retention factors in vesicular electrokineticchromatography (VEKC) with DTAB/SDS vesicles and octanol-waterpartition coefficients for 20 neutral test compounds, including 13noncongeneric compounds (squares). Trendline shown is for all 20compounds. Correlation coefficients (r2) are greater than 0.98 for allcompounds and noncongeneric subset as described in text.

log Po/w ) 0.8019 log k′ - 1.5672

r2 ) 0.9823, n ) 20

log Po/w ) 0.7920 log k′ - 1.5081

r2 ) 0.9890, n ) 13


Optimization of Retention. In EKC with finite elution range,the optimization of retention can be viewed from two perspec-tives: (i) the width of the optimum range of retention,44 i.e., what“polarity range” of analytes can be separated optimally with a givenpseudostationary phase under a given set of conditions, and (ii)the ease with which retention can be varied with a givenpseudostationary phase (e.g., by changing surfactant concentrationor organic modifier composition).

The optimum range of retention for the best Rs may be definedas the retention factor range over which the resolution is withinan arbitrary percentage (e.g., 80% or 90% of its maximum value).As shown in Figure 10 of ref 44, the optimum retention rangedepends on the elution range, which is about 2.2 times larger forthe DTAB/SDS vesicles than for SDS micelles, as dicussed earlier(Table 1).

As far as varying retention, SDS micelles are much moreadaptable than the DTAB/SDS vesicles. Since the pure vesiclelobe of DTAB/SDS mixture on the phase diagram is quite small,adjusting retention by changing the concentration of the surfac-tants is difficult due to the restriction on the vesicle phasecomposition. Also, in limited experiments with organic solvents,which are useful in decreasing retention, precipitation wasobserved at relatively low percentages (>10%).

It has been determined that the large difference in the lengthsor branches of the surfactants’ hydrophobic chains helps tostabilize vesicle aggregates.55 The reason for this is that theasymmetric tails cannot pack efficiently into crystalline precipitatesthat are predominant when oppositely charged surfactants withsame length chains are mixed together. Currently, we arestudying a cetyltrimethylammonium bromide (CTAB) and sodiumoctyl sulfate (SOS) mixture that has a much larger vesicle regionon the phase diagram.55 Due to the larger region of the vesicle

phase, varying retention by changing total surfactant concentrationor surfactant ratio or the addition of organic solvent should bepossible.

CONCLUSIONThe potential of spontaneous vesicles prepared from oppositely

charged surfactants as a pseudostationary phase in electrokineticchromatography has been demonstrated. The chromatographicproperties of the vesicle solution were compared with those ofthe SDS micelle solution as well as the mixed micelle solution.The spontaneous vesicles had the largest elution window and thegreatest selectivity compared to those of both SDS micelles andthe mixed micelles. In addition, higher efficiency and larger peakcapacity facilitated the resolution of eight substituted benzenesand three geometrical isomers. However, since the pure vesiclelobe of the DTAB/SDS mixture on the phase diagram is quitesmall, it is somewhat tricky to use. Adjusting retention bychanging the total concentration or relative proportions of thesurfactants is not feasible. As mentioned before, since the DTAB/SDS vesicle consists of two of the most commonly used surfac-tants, it was convenient and inexpensive to use for initialinvestigations. The drawbacks related to the restriction on thephase composition can be overcome by switching to vesiclesystems formed from other combinations of cationic and anionicsurfactants.

ACKNOWLEDGMENTThe authors thank Waters Inc. for loaning the Waters 4000E

CE instrument, and Rhone-Poulenc Rorer for partial financialsupport.

Received for review July 7, 1997. Accepted January 23,1998.

AC970730Y(55) Yatcilla, M. T.; Herrington, K. L.; Brasher, L. L.; Kaler, E. W. J. Phys. Chem.

1996, 100, 5874-5879.
