Oil in water microemulsions as mobile phases in liquid

Oct 1, 1992 - Oil in water microemulsions as mobile phases in liquid chromatography. Alain. Berthod and Malua. De Carvalho. Anal. Chem. , 1992, 64 (19...
0 downloads 0 Views 656KB Size
2267

Anal. Chem. 1002, 64, 2207-2272

Oil-in-Water Microemulsions as Mobile Phases in Liquid Chromatography Alain Berthed* and Malua De Carvalho Laboratoire des Sciences Analytiques, UA CNRS 435, Universit6 de Lyon 1, 69622 Villeurbanne Cedex, France

Heptans-sodlum dodecyl sulfate (SDS)-pentanol-water ollIn-water mkroemuldons were Investigatedas a moblle phase HquM chromatography. The massratb SDS/ In e rvpentanol was kept constant, 112, for all compodlons. The retention and peak efflclency of alkylbenzenes, toluene, butyEenzene,heptylbenzene,and decylbenzene, were measured on an octadecyl bonded 15-m column. Water-rlch mkroemulslo~(>87% w/w water) of the L1 oll-ln-water structure were able to separate the alkylbenzene homologues with a linear k'vs nc relatlonshlp, nc k the atom carbon number of the alkyl chaln. The dopes and Interceptsof the k'vs nc llnes depend on the relative vdume of organic phase. Heptane and the active blend SDS pentanol play a similar rob on selectlvlty. They have an oppodte effect on efflclency. A higher content of the active blend SDS pentanol produced a higher dfklency for the alkylbenzene peaks. Addition of heptane to any SDS-pentanol-water phases produced a dramatic decrease In etllclency, specialty for the long-chain homokgueci. TheoolutemasrtranderdecreasesexpamntWy with the alkyl chaln length. Efficiency studles can be related to physlcochemlcalchanges of the mlcroemuldon structure.

+

+

Micellar solutions were first used as mobile phases in thinlayer chromatography by Armstrong and Terrill in the late seventies.' The use of such phases were extended to liquid ~hromatography.~~~ Micellar liquid chromatography (MLC) has been the subject of numerous review In most MLC applications, micellar solutions, Le. aqueous solutions of a surfactant, were used as the mobile phase. An important problem in MLC is the low efficiency. The observed peaks are broad. This is a major drawback of the technique. It was shown that any surfactant adsorbs on the stationary phase728 The low efficiency is mainly due to a slow mass transfer between the surfactant adsorbedlayer and the micellar mobile phase.+" The addition of small amounts of 1-propanoP or l-pentanoli3to the mobile phase greatly improves efficiencies by reducing the adsorbed amount of surfactant. (1) Armstrong, D. W.; Terrill, Q. Anal. Chem. 1979, 51, 2160-2163. (2) Armstrong, D. W.; Henry, S. J. J.Liq. Chromatogr. 1980,3,657672. (3) Armstrong, D. W. Sep. Purif. Methods 1985,14, 213-304. (4) Cline Love, L. J.; Habarta, J. G.; Dorsey, J. G. Anal. Chem. 1984, 56,1132A-1148A. (5) Berthod, A,; Dorsey, J. G.Analusis 1988, 16, 75-89. (6) Armstrong, D.W.; Hinze, W. L. Use of Ordered Media in Chemical Separations; ACS Symposium Series 342; American Chemical Society: Washington, DC, 1987. (7) Berthod, A.; Girard, 1,; Gonnet, C. Anal. Chem. 1986,58, 1 3 5 6 1358. (8) Berthod, A.; Girard, L; Gonnet, C. Anal. Chem. 1986, 58, 13621367. (9) Armstrong, D. W.; Ward, T. J.; Berthod, A. Anal. Chem. 1986,58, 579-583. (10) Borgerding, M. F.;Hinze, W. L.; Stafford, L. D.; Fulp, G. W.; Hamlin, W. C. Anal. Chem. 1989,61, 1353-1358. (11) Berthod, A.;Borgerding,M.F.;Hinze,W. L.J. Chromatogr. 1991, 556.263-276. - - - 7 - - -

(12) Dorsey, J. G.;DeEtchegaray, M. T.; Landy, J. S. Anal. Chem. 1983,55, 924-929. (13) Berthod, A.; R o w e l , A. J. Chromatogr. 1988,449, 349-360. 0003-2700/92/0364-2267$03.00/0

The addition of medium-chain linear alcohols (1-butanol or 1-pentanol) to micellar solutions is the first step toward microemulsion formation.14J5 Microemulsions are liquid disperse systems containing an oil, water, a surfactant, and a medium-chain alcohol acting as a cosurfactant. Microemulsions appear as clear liquid with a relatively low viscosity. These liquids present some inhomogeneous character at the microscopic down to the molecular level. The polarity of an oily microdomain is very different from that of an aqueous microdomain. Microemulsions form spontaneously and are very stable. It is a dynamic structure. The microdomains are in a continuous formation-destruction process. The average lifetime of a given microdomain (=droplet) is in the microsecond range. The microemulsion thermodynamic structure is very complicated and the object of intense in~estigati0n.l~ The L1, L2, and bicontinuous classification is very convenient. An L1 microemulsion is an oil-in-water system. Oil microdropleta, enclosed in a surfactant-cosurfactant layer, are dispersed in an aqueous continuous phase. L2 microemulsions are water-in-oil systems. In a bicontinuous structure, it is not possible to tell which phase is dispersed in which. Water and oil microdomains overlap in each other with a sponge structure.l6 It was recently shown that L2 microemulsions could be used as normal LC mobile phases.17J8 Water molecules were trapped into reverse micelles. The sensitivity of the silica stationary phase to water molecules was greatly decreaeed.17 Interesting selectivities along with poor efficiencies were obtained.18 It must be noted that the L2 microemulsions used did not contain any alcohol cosurfactant. In this work, L1 microemulsionscontaining1-pentanolwere used as mobile phases for reversed-phaseLC. Sodium dodecylsulfate (SDS) was chosen for the surfactant because it is the most commonly studied anionic surfactant. It was shown that the 112 SDS/ pentanol mass ratio was optimal for microemulsion formation with several linear alcanes.14J5The heptane-water-sodium dodecyl sulfate-pentanol system was selected because it showed on its mass-phase diagram two well-separated composition domains: the oil-in-water (Ll) microemulsion composition domain and the bicontinuous and water-in-oil (L2) domain. L1 compositions of this system were investigated with an octadecyl ((218) bonded stationary phase. Investigation of the retention mechanism and efficiency evolution was done using the classical alkylbenzene homologous series. EXPERIMENTAL SECTION

Chemicals. Heptane was obtained from BDH, Peypin, France. Sodium dodecyl sulfate (SDS) was provided by Merck (14) Mittal, K.L. Surfactants in Solution; Plenum Press: New York, 1989; Vol. 10. (15) Rosano, H.L.; Clauese, M. Microemulaions Systems; Surfactant Science Series 24; Dekker: New York, 1987. (16) Kaler, E. W.; Prager, S. J. Colloid Interface Sci. 1982,86, 359369. (17) Hernandez-Torres, M.A.;Landy,J. S.;Dorsey, J. G. Anal. Chem. 1986.58. 744-141. (18)Berthod, A.; Nicolas, 0.; Porthault, M. Anal. Chem. 1990, 62, 1402-1407. 0 1992 American Chemical Society

2268

ANALYTICAL CHEMISTRY, VOL. 64, NO. 19, OCTOBER 1, 1992

uene-propylbenzene) with a baseline resolution. To investigate the microemulsion retention mechanism, we chose to use toluene, butylbenzene, heptylbenzene, and decylbenzene. The three methylene increment between the solutes allowed us to obtain baseline resolution with some microemuleion compositions. It was still necessary to inject the solutes separately for a reliable efficiency study. Selectivity. Table I lists the compositions of a selected number of microemulsions along with the corresponding capacity factors, k’, of the alkylbenzene solutes. The first important observation is that the linear relationship k’vs ne, the side chain carbon number, was obtained for all compositions studied. The slopes, a,intercept, 8, and regression coefficient, rz, are listed in Table I. Such linear relationship

k’ = anc + 8

Figure 1. Mass phase diagram of the system water-heptaneSDSpentanol with a constant SDSIpentanoi 112 mass ratio. Key: open area, microemulsion systems (L1 and L2 + bicontinuous);dotted area; liquid crystal; hatched area,heterogeneousmixtures. Theckcle focuses on the area containing the compositions used (see Figure 3).

(Darmstadt, Germany) with a 99% purity. The structural formula of SDS is C12HZz-SOa-Na+, the molecular weight is 288.4, the micellar molar volume is 0.246 L/mol. The solutes, toluene, butylbenzene, heptylbenzene, and decylbenzene,were obtained from Merck. Water was deionized and filtered on a Barnstead system. Chromatographic System. A Shimadzu LC6-A pump (Touzard & Matignon, Vitry, France) was used with a pulse dampener. The minimum flow rate was 0.1 mL/min, increasing by 0.1 mL/min steps. A Rheodyne 7125 valve with a 10-pLloop was used for sample injection. A Philips UV detector, Model PU 4025, was connected to a Philips recorder, Model PM 8252 A. A 15-cm- X 4.6-mrn4.d. Spherisorb ODS 2 ((218) column was purchased at SFCC (Eragny,France). The Spherisorb packing has the followingcharacteristics: spherical 10-rmparticles;mean pore diameter, 8 nm; pore volume, 0.4 cm3/g; surface area, 220 m2/g;carbon percentage,12% ;monomer octadecylsilanebonding followed by end-capping. The column efficiencywas in the 4000plate range. The height equivalentto a theoreticalplate (HETP) was about 37 pm or four particle diameters (mobilephase, pure methanol; solute, naphthalene, k’ = 0.93; flow rate, 1mL/min). A 10-pmparticle size stationary phase was chosen to increasethe column permeability, that is to reduce the column back-pressure when working with relatively viscous microemulsion mobile phases. The maximum working pressure was 180 kg/cm2(2600 psi). The detection wavelength was set at 254 nm. Microemulsion System. Heptane was chosen for the oil because of the low toxicity of the odd carbon number alkanes. The heptanewater-SDS-pentanol mass phase diagram was constructed at 22 O C with the constant mass ratio SDS/pentanol = 1/2 (molar ratio, 1 SDS molecule for 6.5 pentanol molecules) as previously described.l8J9 Figure 1 shows that there are two separated microemulsion domains. The large domain of L2 and bicontinuous microemulsions is separated from the narrow L1 domain by a liquid crystaldomain. The circle at the lower aqueous corner focuses on the composition zone that was used in this study.

RESULTS AND DISCUSSION The efficiencies observed with L1 microemulsions were not very high. It was not possible to separate more than three members of the homologous alkylbenzene series (tol(19) Berthod, A. J. Chim. Phys. 1983,80, 407-421.

(1) was first observed by Khaledi.20,z1In reversed-phase LC, log k’ is linearly related to nc. The slope of the log k‘ vs nc line gives the free energy of transfer, AGCHZ, for a methylene group from the mobile phase to the stationary phase.zz From a thermodynamic point of view, Hinze and Weber established why the relationship between log k‘ and nc were not linear in MLCZ3 They demonstrated that k’ is related to nc according to

llk’ = (1- cp) exp(a + bnc) + cp exp(c + dnc) (2) in which cp is the volume fraction of the micellar phase. a-d are four constants related to different free energies of transfer for the methylene incremental group or for the benzene ring, and for micelle-stationary phase exchanges or for aqueous phase-stationary phase exchanges.23 The authors state that there is no theoretical foundation for a linear k‘ v8 nc relationship. However, the llk’ factor should be directly proportional to the organic volume fraction, cp (eq 2). In microemulsion systems, the organic volume fraction includes the contributions of pentanol, heptane, and the surfactant. The organic volume fraction was calculated by assuming an exact volume additivity and using the compound density. With MSDS,MC~OH, Mw, and Mc,H~,,the mass percentage of SDS,pentanol, water, and heptane, respectively, the cp value can be obtained from (Ms~$1.17) + (Mc60~/o-814) + (Mc,~~,/0.684) (MsDs/1.17) (Mc60~/o.814) (M~,,~,/0.684)

9 = Mw

(3) Table I1 lists the slopes, intercepts, and regression coefficients of the llk’ vs cp curves obtained with the results of this study. Regression coefficients in the 0.92 range denote a loose linear correlation. It was not possible to estimate the a-d values of eq 2 because the experimental intercepts listed in Table I1 were negative. The theoretical intercept is exp(a + bnc)which cannot be negative. Table I1lists the theoretical slopes and intercepts computed with the a-d values experimentally obtained by Hinze and Weberz3 with SDS/pentan01 2/1 w/w micellar solutions (1 SDS molecule for 1.6 pentanol molecules). Although the SDS to pentanol ratio we used was 4 times lower (SDS/pentanol 1/2 w/w), the same order of magnitude and the same trend was obtained for the slopes. The very low values of the theoretical intercepts demonstrate how difficult it is to determine them with accuracy (Table 11). (20) Khaledi, M. G.; Peuler, E.; Ngeh-Ngwainbi, J. Anal. Chem. 1987,

59, 2138-2147.

(21) Khaledi, M. G. Anal. Chem. 1988,60,876-887. (22) Tchapla, A,; Colin, H.; Guiochon, G. Anal. Chem. 1984,56,621625. (23) Hinze, W. L.; Weber, S. G. Anal. Chem. 1991,63, 1808-1811.

ANALYTICAL CHEMISTRY, VOL. 64, NO. 19, OCTOBER 1, 1992

2269

Table I. Microemulsion Compositions and Alkylbenzene Capacity Factors.

+B

mass composition k' = a n, caDacitv factors k' cp, % vol SDS + Dentanol, % w/w heptane, % w/w water toluene butylbenzene heptylbenzene decylbenzene a B 2.49 0 97.51 52.1 121.7 164.3 224 2.675 38.1 18.6 2.705 0.021 33.1 21.0 2.49 97.48 50.4 124.1 177.4 243 109.1 2.741 0.046 40.1 15.4 2.49 97.46 49.8 149.5 190 95.0 21.4 5.361 0 33.2 47.1 20.1 5.00 40.6 2.81 5.837 27.0 37.2 4.99 94.67 19.6 30.6 0.34 18.3 1.87 6.215 26.7 37.0 4.93 94.42 19.2 30.3 0.65 17.8 1.9 27.1 36.2 4.95 94.07 20.8 30.6 0.98 6.705 19.6 1.66 7.473 15.5 21.6 0 6.98 93.02 11.3 18.4 10.5 1.12 0 15.8 21.9 7.53 92.47 11.1 18.7 8.059 10.4 1.18 0 12.9 10.0 7.21 9.31 90.0 10.9 10.68 6.65 0.62 1.01 12.07 7.38 10.0 5.99 6.79 89.0 7.86 5.87 0.21 12.52 0 10.1 8.58 11.74 7.32 88.26 5.49 0.45 5.96 1.5 7.72 9.95 6.22 88.55 12.71 8.01 6.09 0.21 6.93 0 13.24 12.42 5.48 6.74 87.58 5.06 0.44 9.34 8.36 2.2 9.88 87.92 6.80 0.22 13.61 8.90 8.40 6.94 7.80 0 13.05 86.95 13.9 8.84 7.32 5.16 6.28 4.69 0.40

r2 0.992 0,996 0.990 0.977 0.984 0.984 0,989 0.993 0.987 0.997 0.986 0.999 0.969 0.992 0.984 0.992

SDS/pentanol mass ratio = 112;5% w/w in the SDS + pentanol column means 1.66% w/w SDS and 3.33% w/w pentanol.

Table 11. Slope, Intercept, and Regression Coefficient of the l / k fvs cp Curves compd toluene

slopeb . interceptb 1.515 -0.030 (0.086) (0.013) butylbenzene 1.367 -0,038 (0.070) (0.011) heptylbenzene 1.205 -0.034 (0.063) (0.010) -0.033 decylbenzene 1.096 (0.092) (0.015)

a

0.935

1.564 1.02 X lW3

0.933 1.355 5.3 X loa 0.891 1.173 2.8 X 1o-B

Equation 2c b = -0.98 c = 0.64

a = -2.97

e= -0.0775

0.920

theoretical valuesn slope intercept 1.788 0.0192

r2

Equations 4-6 g= 0.971 -0.154

+=

f= 1.734

d = -0.048 h= 1.426

r2

0.980

The theoretical slope is calculated wing exp(c + ncd)- exp(a +

+

neb), the theoretical intercept is calculatedusing exp(a ncbLBbThe

numbers in parentheses are the experimental slope and intercept confidence limits. Values taken from ref 23.

The linear relationship between k' and nc was obtained for all microemulsion compositions. We tried to analyze these retention results from an empirical point of view by considering the whole alkylbenzeneseries and not asingle compound the k' vs nc slopes and intercepts (Table I) were studied as a function of 9. It was found that the slopes a and the intercepts /3 of the k' vs nc plots (eq 1)could be related to cp according to

log a = e q + f

(4)

log0 =gcp+ h

(5)

The e-h parameter values are listed in Table I1 along with the correlation coefficients in the 0.96 range. Combining eqs 1, 4, and 5 yields

kf = loe+fnc+ I@+''

(6) Empirical eq 6 relates the alkylbenzenecapacity factor to the parameters 9 and nc as the theoretical eq 2 does. Both equations use the same number of adjustable parameters. Figure 2 is the three-dimensional plot of k' vs 9 and nc. The experimental points of Table I were posted on the calculated surface. The average error between experimental and calculated k' values was 11.2% (Figure 2). The surface k' = f ( ~ ,

nc),generatedby eq 2,resembles the one of Figure 2. However it was not possible to adjust the a-d parameters to obtain a comparably low 11.2% error. From a practical point of view, it is essential to realize that a small addition of heptane to a SDS-pentanol micellar solution produces a dramatic decrease of the capacity factors of the most retained solutes. Figure 3 shows the k' contour plot of toluene (top) and decylbenzene (bottom) in the L1 microemulsion domain investigated (Figure 1). The capacity factor decreasewhen the organic-phasepercentage increases is 10 times faster for decylbenzene than for toluene (Figures 2 and 3). Figure 4 shows the chromatogram evolution when 1% w/w heptane was added to a 91 % water-9 5% SDS-pentan01 blend (9% w/w) micellar solution. The dramatic loss of resolution is due to both a decrease of selectivityand efficiency. Efficiency. The chromatographic efficiencywas measured using the plate number, N, N = 4[tr/W0.6H12 (7) in which wO.6H is the peak width at 60% of the peak height (expressed in time units) and t, is the retention time. The low efficiency observed in MLC was extensively studied.943J8@ It was shown that the adsorbed surfactant layer was responsible for the slow mass-transfer process." The addition of a short-chain alcohol to the micellar phase decreased the amount of adsorbed surfactant and increased the effi~iency.l'-'~Pentanol competes with SDS to adsorb onto the cl8 bonded surface, enhancing the efficiency. However, the ratio SDS/pentanol was kept constant for all microemulsion compositions. The amount of adsorbed SDS was determined using the procedure already described.718,",'3,la For all compositions without heptane, the adsorbed SDS amount was in the 40-mg range (-0.4 pmol/m2). For comparison, the adsorbed SDS amounts on CISphases were in the 3-5 pmol/m2 range with micellar solutions without a l ~ 0 h 0 1 ? JHeptane ~ ~ ~ ~ is ~ another ~~ competitorfor stationaryphase adsorption.26 It displaces further SDS from the stationary phase. The adsorbed amount of SDS was 20 mg (-0.2 pmol/m2)with a microemulsion containing only 0.4% w/w heptane. This amount is close to the limit of detection of the method used. It was not possible to determine accurately the amount of adsorbed SDS with microemulsions containing more than 0.5 % w/w heptane. Also, it was (24)Borgerding, M. F.;Quina, F. H.; Hinze, W. L.; Bowmaster, J.; MacNair, H. M. Anal. Chem. 1988,60, 2520-2527. (25)Borgerding, M. F.;Williams, R. L.; Hinze, W. L.; Quina, F. H. J. Liq. Chromutogr. 1989, 12, 1367-1406.

2270

ANALYTICAL CHEMISTRY, VOL. 64, NO. 19, OCTOBER 1, 1992

Figure 2. 3D plot of the capacity factor vs the carbon side chain number and the organic voiume percentageof the microemuislon mobile phase. The bottom of the lines leadlng from the posted numbers corresponds to the capacity factor data (Table I). Due to overlapping, some numbers are illegible, they can be read in Table I.

not possible to determine the exact amount of pentanol and heptane adsorbed in (or on) the ODS stationary phase. The sharp decrease in capacity factors with increased heptane concentration clearly place the heptane in the mobile-phase microemulsion. For any microemulsion composition, a dramatic decrease of efficiency was observed with the alkyl chain length increase. As a typical example, the efficiencies corresponding to the Figure 4 chromatogram are listed in the inset tables. In the same chromatogram,the efficiencywas 7000plates for toluene and only 1100 plates for decylbenzene. The addition of a mere 1 % wlw heptane to the micellar mobile phase produced a 50% decrease of the toluene efficiency (3300 plates) and an 8-fold decrease of the decylbenzene efficiency (140 plates). The efficiencyobserved with the long-chain alkylbenzenewas often so low that it was necessary to inject each solute separatelyto measure it (Figure 4). Very different efficiencies were observed in MLC chromatograms for solutes with different fun~ti0nalities.l~ The efficiency decreasewith alkyl chain length increase can be noticed in the MLC literature (for example, Figure 1, ref 24). The HETP, H, can be expressed as the s u m of three terms: (i) a term representing the flow contribution, (ii) a term depending on the molecular longitudinal diffusion, (iii) and a term representing the mass-transfer processes.26 Knox expressed the reduced plate height, h = Hld,,with the particle diameter d,, as h = Au1I3+ Blu -+ Cu

in which u is the linear mobile-phase velocity and the A-C constants are related to the flow anisotropy, longitudinal diffusion, and mass transfer, re~pectively.~'Due to experimental constraint (minimum flow rate 0.1 mllmin), it was not possible to obtain the B constant with accuracy. In the 0.2-1.2 mL/min flow rate range, the h vs u curves were linear. The slopes are a combination of the A and C parameters that is expressed by dhldu = = A/(3u2I3)- B/u2+ C (9) The flow anisotropy is dependent on the column-packing quality. It is essentially the same for all compounds of the same chromatogram.11126The B/u2term is small in the 0.21.2 mL/min flow rate range studied. So, changes in the experimental slopes of the h vs u curves represent mainly changes in the solute mass-transfer process (C term). For this reason, these h vs u slopes were designated by c. Table I11 lists the c values for the alkylbenzenes studied and for different microemulsion compositions. For most microemulsions, an exponential increase of the term was observed with the alkyl chain length increase. The plots of log c v s nc were relatively linear (lowestregression coefficient: 0.946, Table 111). The C term is mainly responsible for the changes observed on the c parameter (eq 9). It can be written as26

e

(8)

(26) Giddings,J.C. Dynamics of Chromatography;Dekker: New York, 1965.

(27) Knox, J. H.J. Chromatogr. Sci. 1977, 15, 352-364.

ANALYTICAL CHEMISTRY, VOL. 64, NO. 19, OCTOBER 1, 1992

2271

Y

Y

1

98

96

34

92

I

90

08

WATER Figure 3. k'contour plot of toluene (top) and decylbenrene (bottom) for the L1 microemulsion domaln (circle, Flgure 1). The triangles carrespond to the two compositions of the Flgure 4 mobile phases.

in which 4 is the stagnant mobile-phase fraction, q is a geometrical fador dependent on porosity, d, is the particle diameter, the 7 terms are obstruction factors,and the D terms are solute diffusion coefficients. The subscripts 8, m, and sm stand for stationary, mobile, and pore stagnant mobile phase, respectively. Only q and d, are constanta. All other eq 10 parameters depend on the microemulsion composition and/ or the solute. The stagnant mobile-phase volume fraction, 4, is in the 0.3-0.4range, and surfactant adsorption tends to decrease it.7Jsv2435 For most microemulsion compositions studied, the k'values were higher than 6 (Table I). Then, the fiit term of eq 10 is close to unity. The exponential changes of the term (Table 111)seem to be due to the second term of eq 10. In the selectivity study, it was shown that k' increased linearlywith the alkyl chain length. Such an increase should decrease the c parameter (eq 10). It does not, which means that the 79, term decreases dramatically. The solute mobility in the stationary phase becomes very limited when the alkyl chain length increases. It was shown that very hydrophobicsoluteswere directly transferred from the organic phase to the stationary ~hase.~~,25 They did not and could not go into the aqueous phase in which their solubility was too far 10w.11v24925It was demonstrated that the D, diffusion coefficient depends on the solute micellar partition coefficient?~'~The lowest possible D, value corresponds to the micelle, and in our case the microemulsion droplet, diffusion

Flgurr 4. Effect of a small addltlon of heptane to an L1 microemuldon. Condltlons: d u m n Cia, 5 pm, 15 cm X 4.6 mm 1.d.; flow rate 0.6 mL/mln; UV detection 254 nm, 0.32 aufs. Peaks: (1) toluene, (2) bvtylbenzene,(3) heptylbenrene, (4)decylbenrene(- 7 & InJecUons). ~ Top: mlcroemulslon 91 % water, 9% SDS pentanol acthre bknd (% w/w). Bottom: mlcroemulslon90.1 % water, 8.9% active blend, 1% heptane. Dotted lines: single solute Injections. Inset t a b k . peak efficlencles In plate number.

+

coefficient? The microemulsion droplet size depends on the composition. It can be seen in Table I11that slopes, S,of the log vs nc lines change with the microemulsion composition. Two opposite trends can be noted: (i) at a constant heptane content, the slopesSdecrease when the SDS/pentanolcontent increases; (ii) at a constant active blend content, the slopes S increase with the heptane percentage. More pentanol goes into the aqueous phase when the active blend content is increased. This decreases the polarity difference between the aqueous and the organic phase and renders the solute mass transfer faster. When heptane is added to the micellar solution, it goes exclusively inside the micelle core. The polarity difference between the aqueous and organic phases increases. Long alkylbenzenes are located inside the microdroplet withaslow anddirectmaw transferwiththestationary phaee. Also, heptane addition may increase the microdroplet size which would decrease the D, diffusion coefficients. Similar results were observed in SDS and nonionic Brij micellar solutions.25 However, we do not have any data concerning a relationshipbetween microemulsion droplet size and composition because the system studied is original. Heptane and pentanol have the same effect on the amount of adsorbed surfactant, both decreasing it. They have opposite effects on the solute mass transfer. Pentanol increases the efficiency. Heptane decreases it.

2272

ANALYTICAL CHEMISTRY, VOL. 64, NO. 19, OCTOBER 1, 1992

Table 111.

?i

Parameter in the Efficiency Study

c

mass composition

v, % vol heptane, % w/w SDS + pentanol, % w/w toluene butylbenzene 2.28 5.36 5.77 6.18 6.55 7.47 10.68 12.08 13.19 13.63

0 0 0.336 0.649 0.980 0 0 1.0 1.2 2.21

2.49 5.0 4.93 4.90 4.80 6.98 10.0 10.0 10.8 9.88

0.5 9.0 2.52 2.52 3.24 14.5 18.2 27.8 20.8 23.4

To conclude this work, the unique solubility power of the microemulsion system studied should be pointed out. Decylbenzene is barely soluble in a 75% v/v methanol-rich solution. However it is possible to dissolve 4 g/L of decylbenzene in a heptane-SDS-pentanol microemulsion containing 90% w/w water. The solubility power of microemulsions is observed in the selectivity study. The decylbenzene k’ value of a 90% w/w water microemulsion is around 10 (Table I). This k’ value could not be experimentally measured on the same CISphase with a 90/10 watedmethanol v/v mobile phase. It can be estimated theoretically to be in the 2 OOO OOO range (log k’ = 6.4).2S Decylbenzene is located in the microemulsion droplet core. It cannot go in the aqueous phase due to its extremely low water solubility. This renders the mass transfer very slow, and the corresponding chromatographic peak is broad. The low-efficiency problem (28) Colin, H.; Krstulovic, A. M.; Gonnord, M. F.; Guiochon, G.; Yun, Z.; Jandera, P. Chrornatographia 1983, 17, 9-15.

16.8 20.0 31.7 39.2 66.2 25.5 37.1 123 81.8 239

heptbenzene

decylbenzene

dope

log c = f ( n J intercept

227 565 257 545 1288 225 84.2 295 210 2300

9500 1970 3530 8260 21900 264 110 983 483 21600

0.46 0.28 0.345 0.39 0.43 0.16 0.09 0.17 0.15 0.33

4-69 0.59 0.07 0.02 0.10 0.97 1.21 1.33 1.23 1.05

r* 0.990 0.982 0.998 0,999 0.999 0.946 0.963 0.991 0.986 0.999

hinders the use of the unique selectivity of microemulsion mobile phases. However,we have shown that normal LC can give interesting information on the physicochemicalstructure of L2 microemulsions (water in oil).lS The present results show that low-viscosity L1 microemulsions could be studied by reversed-phase LC as well. The dramatic decrease of the decylbenzene efficiency observed on the addition of a small amount of heptane in a SDS-pentanol micellar solution (Figure 4) can be related to physicochemical changes. The heptaneless SDS-pentanol solution has a loose structure. Upon addition of heptane, the L1 oil-in-water structure becomes well-organized. Efficiency measurements could be used to follow microemulsion structure evolutions.

RECEIVED for review March 30, 1992. Accepted July 7, 1992.