Anal. Chem. 1908, 60,617-621
617
Gradient Elution for Micellar Electrokinetic Capillary Chromatography Anthony T.Balchunas’ a n d Michael J . Sepaniak*
Department of Chemistry, The University of Tennessee, Knoxville, Tennessee 37996-1600
A method for performlng gradient elution in micellar electrokinetic capillary chromatography Is developed. The lnfluence of temperature and mobile phase organic solvent concentratlon on capaclty factors Is presented and attributed to changes in phase ratlo and solute distribution coefficient. Solvent programming Is more effective than temperature programming at adjusting retention due to its greater Influence on dlstributlon coefflclent. As anticipated, phase ratio changes could not be exploited due to concomitant losses in column efflciency. A stepwlse solvent gradlent lnvolvlng Increasing concentrations of 2-propanol and Triton-X-100 is used for the Separation of a test mixture of derivatized amlnes.
I , applied voltage, V, and electroosmotic, peo, and micelle electrophoretic, pme,mobilities is shown in eq 1.
While extending the elution range in MECC does result in a larger theoretical peak capacity, in practice it does not necessarily provide the best chromatographic resolution in the shortest time. This is particularly true for compounds that are well-solubilized by the micelles and, hence, have large capacity factors. This can be better understood by considering the resolution, R,, equation for MECC (eq 2), where N is the
Rs =
(?)( ?)( L)( K2‘ + +
1- t o / t m
1
Micellar electrokinetic capillary chromatography (MECC) is a highly efficient separation technique that has been used to separate neutral species, based on their differential distribution between an electroosmotically pumped, aqueous mobile phase, and a slower moving, electrophoretically retarded, micellar phase. Since first being reported ( I ) , fundamental studies that characterized the effects of experimental parameters on retention behavior (2) and column efficiency ( 3 , 4 )have appeared in the literature. The technique has been used for the separation of many water-soluble classes of compounds (5-10). More recently, changes in selectivity have been demonstrated, either by doping the mobile phase with transition metal ions (7) or, more straightforwardly, by simply using surfactants other than the routinely used sodium dodecyl sulfate (SDS)(11). In conventional elution chromatography, a totally retained compound (i.e., one with an infiiite capacity factor, k’) is never eluted. Conversely, a compound that is totally solubilized by the micellar phase in MECC is eluted in a time that is equivalent to the effective retention time, t,, of the retarded micelles. Hence, MECC is characterized by a limited elution range. The elution range can be extended by silanating the capillary column walls (12),coating the walls with polymers (13),adding certain metal ions to the mobile phase (7) or, as demonstrated in this work, adding certain organic solvents to the mobile phase. The increase in elution range can be desirable in the analysis of complex samples in that it is accompanied by an increase in peak capacity (2). Under most conditions employed in MECC the electroosmotic flow velocity, ueo,opposes the electrophoretic flow velocity of the micelles, ume,and is greater in magnitude. An effect of silanating the column walls or adding certain organic solvents (e.g., 2-propanol) to the mobile phase is to reduce the electric double layer potential (zeta potential) at the column wall-mobile phase interface and this reduces u, (14). The elution range can be increased greatly as u,, and urn,approach the same absolute value. The relationship between t , and total column length, L, effective column length (inlet to detector length), ‘Present address: The Procter and Gamble Co., Sharon Woods Technical Center, Cincinnati, OH 45241. 0003-2700/88/0360-0617$01.50/0
1
)
(2)
(t,/t,)R,’
average efficiency of the two compo_nentsof interest, a is the selectivity coefficient, given by k2)/k{, and t o / t mis an elution range parameter with to equal to the elution time of a nonretained solute. Capacity factors in MECC are calculated by using eq 3, where t , is the retention time of the solute of interest.
(3) As theoretically shown by eq 2, and as verified experimentally (12), extending elution range does yield improvements in resolution by reducing the ratio to/tm,provided that efficiency remains constant. However, these improvements might be insignificant for highly micelle solubilized compounds. Examinatio? of eq 3 reveals that solutes eluting very close to t , will have k’values approaching infinity as the ratio t,/t, approaches unity. Since k’also appears in the denominator of the last term in eq 2, one can readily see that optimization of resolution in MECC will require techniques that not only extend the elution range bilt also_ provide for “programming”, or specifically, adjustment of k ’to optimum values. In this paper, we report the results of our studies concerning techniques for adjusting k’ and effecting gradient elution in MECC. EXPERIMENTAL SECTION Reagents. Sodium dodecyl sulfate (SDS), 99% purity, was supplied by Sigma Chemical Co. Triton-X-100 (ScintillAR) was obtained from Malinckrodt. The highly fluorescent laser dyes Coumarin 153 and Coumarin 343 were supplied by Exciton Corp. All other chemicals were reagent grade (Fischer Scientific). Amines were derivatized with the fluorescent label 7-chloro-4nitrobenz-2-oxa-1,3-diazole (NBD chloride) obtained from Sigma Chemical Co., using a procedure reported previously (15). All mobile phases were prepared in triply distilled, deionized water. Apparatus. Fused silica capillaries were obtained from Scientific Glass Engineering (Austin, TX) and were used without further chemical modification. All columns used in these studies were placed in thermostated, circulating water jackets for ambient temperature control to within f0.05 O C . Columns were electroosmoticallypumped with a regulated Hipotronics Model MOA, high-voltage dc power supply. The electroinjection procedure used 0 1988 American Chemical Society
818
ANALYTICAL CHEMISTRY, VOL. 60. NO. 7, APRIL 1, 1988
Table I. Changes in A’ and Efficiency as a Function of Temperature for Two Surfactant Concentrations 0.015 M SDS
NBD n-butylamine 28 39 47 57
0.10 M SDS
NBD rz-hexylamine
NBD rz-butylamine
NBD n-hexylamine
A,
N
A/
N
6l
N
kt
N
1.20 0.93 0.86 0.77
12 160 58 600 24 100 6 800
12.20 9.68 8.47 7.10
6 500 43 700 14 400 4 500
3.26 3.03 2.87 2.60
530 000 404 800 123 200 102 600
34.10 31.60 30.42 25.10
47 1000 451 900 98 700 77 700
has been previously described (4). On-column laser-based fluorescence detection was performed by using a Cyonics Model 2001-20BL argon ion laser for excitation (488 nm and 20 mW). Fluorescence emission signals were isolated at 525 nm (6 nm band-pass) by using an Instruments SA Model H-10 monochromator, and detected with an RCA Model 1P28 photomultiplier tube. Photocurrents were processed with a Keithley Model 485 picoammeter, and chromatograms were recorded on a Kipp and Zonen Model BD40 strip-chart recorder. Gradient Elution. The column was first placed in the thermostat& water jacket and then fiied with the starting mobile phase. All gradient elutions were performed at 38 “C. The injection end of the column was placed into a 5-mL beaker, which was positioned on a magnetic stir plate. This inlet reservoir was filled with exactly 2.5 mL of the starting mobile phase. Stepwise gradients were produced by pipetting aliquots of a gradient solvent containing 2-propanol into the 5-mL beaker. A small magnetic stirring bar was used to ensure thorough mixing of the added gradient solvent with the starting mobile phase. Upon injection of sample, the chromatogram was initiated via application of the desired starting voltage (typically 15-20 kV). Gradient elution solvent was then manually added in four 0.5-mL increments, spaced 5 min apart, the first increment being added 5 min after injection. Prior to solvent addition, the power supply was shut down to avoid injury to the operator. Voltage was then restored to its previous value. Solvent additions generally took less than 5 s. All gradient elutions were run at approximately constant current, by monitoring the power supply’s current meter during the chromatogram and appropriately adjusting the applied voltage when current decreased due to changes in mobile phase composition. Temperature and Solvent Studies. Effects of temperature and solvent composition on k’were determined by using a 0.05 mm i.d. X 1.5 m long, fused-silica column, thermostated to the appropriate temperature. A test sample consisting of the NBD chloride derivatives of ethylamine, n-propylamine, n-butylamine, cyclohexylamine,and n-hexylamine, and a fluorescent laser dye, Coumarin 343, listed in order of elution, was used to perform these studies. Chromatogramsobtained for each temprature or solvent composition were used with eq 3 to determine k’values. For these calculations, t , was assigned to the retention time of the highly lipophilic Coumarin 343 compound. We were not able to locate a highly hydrophilic, nonionic, fluorescent (when excited at 488 nm) compound to accurately mark t,,. Similarly, solvent-related base-line disturbances that can be used to mark to are often not discernible with this mode of detection. Since we were investigating procedures for adjusting retention, which does not require the exact determination of thermodynamic parameters for our test solutes, we used the first component in our test mixture to elute (NBD ethylamine) to mark to. In acutal fact, NBD ethylamine has a k’ of approximately 2-3 when room temperature separations are performed with the aqueous mobile phases described herein. We-have denoted capacity factors determined in this work with k’ to indicate- that actual values would be somewhat higher. The reported k ’values represent the average of duplicate chromatograms. Column efficiencies were obtained directly from the chromatograms, using eq 4, where W,,, is the peak width of a component at half maximum. N = 5.5(t,/WI,J2
(4)
RESULTS AND DISCUSSION As shown in eq 5, k’ depends on K , the thermodynamic
distribution coefficient for a solute between the stationary (or micellar) and mobile phases, and on @,the so-called phase ratio
R f = Kp
(5)
In conventional chromatography, fi is defined as the ratio of column volume occupied by stationary phase to that occupied by the mobile phase (V,/ Vm). Likewise, in MECC, @ is defined as Vm,/Vm, where V,, is the column volume occupied by micelles. Established liquid chromatographic gradient elution methods usually involve variation of mobile phase composition or, to a lesser extent, temperature. Since the phase ratio generally remains constant for these techniques, variation of these parameters causes changes in k’ by affecting K , the distribution coefficient. In contrast, the phase ratio in MECC is not always constant, _since it depends on micelle concentration (2). Changes in k’in MECC are therefore more complicated, since both distribution coefficient and phase ratio can be involved. We have studied the influence of both temperature and solvent composition on k’ in an effort to develop effective procedures for gradient elution in MECC. Since one of the most attractive features of MECC is its high efficiency, we were also concerned about how temperature and solvent composition affected this parameter. We had previously established that efficiency degrades as micelle concentration is reduced (3). It was anticipated that the gradient technique (temperature versus solvent programming) that resulted in large K changes, relative to fi changes, would be more desirable. Temperature Studies. The variation of K with temperature was used by Terabe and co-workers to determine thermodynamic parameters associated with micelle solubilization for several substituted benzene compounds (2). Average values for the enthalpy of solubilization, AH”, were approximately -15 kJ/mol. Using this value in the van’t Hoff equation, one would predict roughly a factor of 2 decrease in K when the temperature is increased from 300 to 340 K (approximately the available range in MECC). While this relatively small change indicates that temperature programming would be only marginally useful for gradient elution in MECC, the reported AH” values were obtained for a small set of solutes, over a very small temperature range (14 K), and no corrections for temperature related changes in critical micelle concentration (cmc) were made in the determinations. Thus we decided to investigate the influence of temperature on k ’ and efficiency. Table I shows the results of these studies for NBD n-butylamine and NBD n-hexylamine a t two surfactant concentrations. As expecte$ the direct proportionality-between micelle volume and k’ is not observed for these k’ values; however, important-trends are evident. The temperaturerelated changes in k’ were greate; for the lower SDS concentration. For the test solutes, k’values were reduced by approximately a factor of 1.7 as the temperature was increased from 28 to 57 “C. The high SD_Sconcentration resulted in only a factor of 1.3 reduction in k’values over the same temperature range. According to Mukerjee et al. (16),cmc is a function of temperature. For a given concentration of S D S ,
ANALYTICAL CHEMISTRY, VOL. 60,
Table 11. Variation of &’and Efficiency with 2-Propanol Concentration NBD n-butylamine 70
2-propanol
0 5 10
20 30 35
NBD n-hexylamine
E’
N
if
N
1.02 0.67 0.41 0.28 0.18 0.17
470 000 471 500 379 800 279 800 143700 22 000
44.70 27.80 16.70 14.02
444 300 465 500 245 600 187 300 111000 14 600
12.00 5.14
temperature increases cause a reduction in micelle concentration due to an increase in cmc. Since the room temperature cmc of SDS is approximately 8 x lo9 M, we are already quite close to this value for the low SDS concentration case shown in Table I. Consequently, as temperature is increased the relative changes in (Iare more significant than for the high SDS concentration cas_e. For the higher SDS concentration, observed changes in k’ are more likely to be primarily attributable Lo changes in the distribution coefficient. The fact that these k’values are less than the true k’values causes an underestimation of the reduction in capacity factor with increasing temperature. Thus our observations are reasonably consistent with the aforementioned results of Terabe and co-workers. The efficiency data in Table I are consistent with the observed trends in k’ and illustrate the synergistic influences of micelle concentration, column temperature, and column current on efficiency in MECC. The efficiency data in the table can be explained in terms of resistance to mass transfer in the mobile phase due to poor solute diffusivity, dispersive temperature gradients within the column, and the polydispersity of the micelles (17). The improvements in efficiency with increases in temperature can be partly attributed to increases in solute diffusivity. However, as temperature is increased, column current is observed to increase. At some point, increases in temperature result in excessive Joule heating and a transverse temperature gradient is created within the column (temperature within the column is lower near the walls where the heat is dissipated to the environment). This can lead to dispersion due to the dependence of electrophoretic mobility on temperature (3). As observed in this work, the problem is more severe with high surfactant concentrations since the current is greater. The efficiency trends in Table I can also be explained in terms of the polydispersity of the micelles. This can cause band dispersion as solutes are solubilized by micelles that have a distribution of electrophoretic mobilities. This dispersion should be reduced with higher temperatures and surfactant concentrations due to an increase in the kinetics of micelle-surfactant monomer interactions (17). The dramatic improvement in efficiency when the SDS concentration is increased from 0.015 to 0.10 M is probably mostly due to this effec!. Solvent Studies. Table I1 illustrates how k’varies with 2-propanol concentration a t 28 “C. For each of the mobile phases used in this study, SDS and buffer salt concentrations were kept constant at 0.075 M SDS and 0.01 M Na2HP04/ 0.005 M NazB4O7(pH 7 ) by diluting concentrated stock solutions of these reagents with the appropriate volumes of 2-propanol and distilled water. Comparing the results given in Tables I and 11, it is immediately ob$ous that the addition of 2-propanol has a greater influence on k’ than does temperature. When changed from zero to 10% 2-propanol, k’ was reduced by approximately a factor of 2.5 for both test solutes, without a substantial reduction in efficiency. However, as the concentration of the 2-propanol is further increased, efficiency begins to degrade.
NO. 7, APRIL 1, 1988
619
Table 111. Flow Rate and Elution Range Variation with 2-Propanol Concentration % 2-propanol
uBo,m m d
u,, m m d
toltm
0 5 10 15 15” 15b 20 30
0.55 0.44 0.35 0.31 0.31 0.31 0.26 0.13
0.36 0.18 0.06 0.03 0.09 0.15 0.02 0.07
0.65 0.39 0.14 0.10 0.29 0.48 0.06 0.49
a 15% 2-propanol and 0.5% Triton-X-100. 15% 2-propanol and 1.5% Triton-X-100.
Nevertheless, the data do indicate that 2-propa_nolconcentrations as high as 30% can be used to reduce k’values by approximately a factor of 4. These results indicate that of the two approaches studied, the more effective procedure for gradient elution in MECC is solvent programming. Unfortunately, changes in mobile phase composition also affect electroosmotic flow rates with accompanying changes in the elution range. As shown in Table 111, ueo,as determined using the retention time of NBD-ethylamine, and the net micellar velocity, u,, decrease as the 2-propanol concentration is increased. Unless otherwise stated in the table, separation conditions used to obtain these data were the same as employed to obtain the data for Table 11. Although some extension of the elution range is desirable, the addition of as little as 20% 2-propanol can result in excessiyely long analysis times. For example, without 2-propanol the k’for NBD n-butylamine is 1.02 and it is eluted in 54.5 min with t , equalto 70 min. With a mobile phase that is 20% 2-propanol, the k’of NBD n-butylamine is reduced to 0.28, but it requires 121 min to elute due to the fact that t , has been extended to 26 h. The 2-propanol percentage can be increased to the point where the reduction in k‘more than compensates for the extension in elution range; however efficiency is seriously degraded. In an attempt to circumvent this analysis time problem, different mobile phase additives are being tested with the aim of controlling capacity factors without excessive reductions in u,, (18). It is possible to partially compensate for the reduction in flow rates by increasing applied voltage during the course of the gradient. As the 2-propanol concentration increases, the column current decreases, presumably due to a decrease in the dielectric constant of the mobile phase, and higher voltages can be applied without increasing the thermal load. Thermal loads of 1.0 W/m can be tolerated without noticeable losses in efficiency ( 3 ) . The inverse relationship between applied voltage and analysis time is shown in eq 1. Although not as desirable as controlling u,, itself, we were able to compensate for solvent-related reductions in u,, by effecting a similar reduction in the opposing electrophoretic velocity of the micelles. This was accomplished by adding the large, nonionic surfactant, poly(ethy1ene glycol) p-isooctylphenyl ether (Triton-X-loo), to the mobile phase. As illustrated in Table 111,the addition of Triton-X-100 increased the net velocity of the micelles. The most likely explanation for this observation is that mixed SDS/Triton-X-100 micelles are formed, and possess smaller electrophoretic mobilities than pure SDS micelles. These studies suggest the possibility of performing gradient elution in MECC by adding a gradient forming solvent, composed of Triton-X-100 and a relatively large concentration of 2-propanol, to the starting mobile phase. Simultaneous with the addition of the gradient forming solvent, the voltage can be increased to partially offset reductions in e_lectroosmotic flow. Thus it should be possible to program k’while main-
620
ANALYTICAL CHEMISTRY, VOL. 60, NO. 7, APRIL 1, 1988
Table IV. Reproducibility of 2-Propanol Step-Gradient
tR,min % RSD
1
2
3
4
5
6
7
8
9
10
28.0 1.9
29.4 1.2
30.8 1.6
35.5 2.1
40.0 1.4
43.2 1.6
47.3 1.1
50.9 1.3
55.7 1.4
59.8 2.1
A
I
,-------c--l>JL
"minutes
0
20
B
c d
TIME(rninutes)
Flgure 2. Separation of the test mixture using a stepwise solvent program. See text for conditions.
Flgure 1. (A) Separation of test mixture by using a 0.05 mm i.d. X 850 mm long column with 0.05 M SDS, 0.01 M Na,HP04/0.005 M Na,B40, (pH 7) mobile phase. (B) Separation of test mixture with 22.5% 2-propanol in the mobile phase.
taining reasonable analysis times. Development of Gradient Elution. To illustrate the principles derived from these studies, the chromatograms shown in Figures 1and 2 were obtained by using isocratic and gradient elution. Figure 1A is the isocratic separation of an 11-component test mixture, without 2-propanol in the mobile phase, performed at 15 kV. Separation conditions are given in the figure caption. The components, in order of elution, are (a) NBD methylamine, (b) NBD ethylamine, (c) NBD dimethylamine, (d) NBD n-propylamine, (e) NBD diethylamine, (0 NBD n-butylamine, (9) NBD cyclohexylamine, (h) NBD di-n-propylamine, (i) NBD n-hexylamine, and (j, k) Coumarin 153 and Coumarin 343. Peak identification for this chromatogram was verified by standard addition. Since 2propanol was not used, the elution range was too narrow and capacity factors were too large to adequately resolve all of the components near the end of the chromatogram. This was particularly true for the last two solutes, Coumarin 153 and Coumarin 343, which coelute as the last peak in Figure 1A. Figure 1B is the isocratic separation of the test mixture at the same voltage, under conditions where the 2-propanol concentration is considered too high (22.5% (v/v)), as many of the early eluting compounds are beginning to coelute. Additionally, solute elution order for this chromatogram was changed. On comparison of parts A and B of Figure 1, the most distinctive difference is the extended elution range. In fact, the actual elution range that would have resulted from
using 2-propanol would have been much greater than that shown in Figure 1B; however, due to the aformentioned analysis time problems, the mobile phase used for this chromatogram contained 0.5% Triton-X-100. The significance of this is that the elution range can be adjusted to virtually any size dictated by sample complexity by simply optimizing both 2-propanol and Triton-X-100 concentrations. Note that in Figure 1A the two coumarin compounds coeluted, whereas in Figure 1B they were widely separated so that many more components could potentially be resolved between them under these conditions. By use of the injection-related solvent base line disturbances that are discernible in Figure 1 (just barely with 2-propanol in the mobile phase) to mark to and the Coumarin 343 to mark t,, the k'of Courmarin 153 is reduced from infinity to 4.1 with the addition of the 2-propanol. A rather surprising result of using a mobile phase containing 2-propanol and Triton-X-100 was a change in selectivity. The secondary amine derivatives experienced much larger reductions in k ' than the primary amine derivatives, thereby changing elution order. This emphasizes that potentially useful selectivity changes can be effected by using mobile phase additives in MECC, justifying research in this area. For illustrative purposes, gradient elution for the test mixture was performed with a stepwise solvent program. This separation was carried out on the same column used for Figure 1. The use of gradient elution revealed two previously unresolved impurities eluting about the eighth component (h) of Figure 2. Previous experience with NBD-derivatized amines has shown us that such impurities, presumably due to further reactions of excess derivatizing agent, are quite common. The starting mobile phase consisted of exactly 1.5 mL of 0.05 M SDS, 0.01 M Na2HP04,0.005 M Na2B407,and 10% (v/v) 2-propanol. A gradient solution, consisting of identical SDS and buffer salt concentrations, and 50% (v/v) 2-propanol and
ANALYTICAL CHEMISTRY, VOL. 60, NO. 7, APRIL 1, 1988
2.5% (v/v) Triton-X-100, was added in four 0.5-mL increments resulting in the 2-propanol concentrations shown in Figure 2. Starting voltage for this chromatogram was 15 kV (32 PA) and was adjusted as appropriate to maintain constant current, up to 27 kV. It is emphasized that the gradient profile shown in the figure represents that of the column inlet reservoir and not the actual profile within the column. This gradient elution procedure involved many manual operations and, clearly, automated instrumentation must be developed for the technique to be generally useful. Nevertheless, retention times were quite reproducible as the data shown in Table IV for five replicate gradient elution chromatograms illustrates. Column reequilibration required only that the original starting mobile phase be restored at the column inlet reservoir and that approximately one or two column volumes be allowed to flow through the column. This equilibration generally took 20-25 min. Equilibration and analysis time can be reduced by employing a shorter column. Because of the high efficiency of this technique, columns as short as 50 cm often provide adequate plate numbers. The purpose of developing this gradient program was to illustrate the implementation of the principles derived from this work and not to imply that this particular test mixture required such a method. We have separated this mixture by using isocratic elution under conditions intermediate to those of Figure 1. For “real” samples that are likely to be more complex than our test sample, however, the need to exploit the principles described herein would become more apparent. This would be particularly true for samples containing constituents of varying polarities. In summary, the effects of temperature and solvent composition on both retention and efficiency in MECC have been described for a particular test mixture. Solvent programming appears to be the most effective way to induce desired changes in k ’, Temperature programming at SDS concentrations required for maximum efficiency is not as effective as solvent programming, due to an apparent greater dependence of distribution coefficient on solvent composition. Control of column temperature is, however, very important for maximum efficiency. For SDS concentrations that produce only moderate current (Le., 0.014.08 M), 40 O C appears to be optimum for the column dimensions and buffer concentrations used in
621
this work. At higher SDS concentrations, larger currents prohibit the use of elevated temperatures, and maximum efficiencies are obtained at room temperature. While lower buffer concentrations would reduce the high currents associated with these SDS concentrations, our experience has been that there exists a minimum allowable buffer concentration, below which reproducibility in retention times is poor. Registry No. 2-Propanol, 67-63-0;Triton-X-100, 9002-93-1.
LITERATURE CITED (1) Terabe, S.; Otsuka, K.; Ichikawa, K.; Tsuchiya, A,; Ando, T. Anal. Chem. 1984, 56, 113-116. (2) Terabe, S.; Otsuka, K.; Ando, T. Anal. Chem. 1985, 5 7 , 834-841. (3) Sepaniak, M. J.; Cole, R. Anal. Chem. 1987, 59,472-476. (4) Burton, D. E.; Sepaniak, M. J.; Maskarinec. M. P. Chromatographia 1986, 27(10), 563-586. (5) Otsuka, K.; Terabe, S.; Ano, T. J.Chromatogr. 1985, 348, 39-47. (6) Burton, D. E.; Sepaniak, M. J.; Maskarinec, M. P. J . Chromatogr. Sci, 1986, 24, 347-351. (7) Cohen, A. S.;Terabe, S. Smith, J. A.; Karger. B. L. Anal. Chem. 1987, 59, 1021-1027. (8) Terabe, S.; Ozaki, H.; Otsuka, K.: Ando, T. J. Chromatogr. 1985, 332,211-217. (9) Otsuka, K.; Terabe, S.;Ando, T. J . Chromatogr. 1985, 332,219-226. (IO) Gozel, P.; Gassman, E.; Michelsen, H.: Zare, R. N. Anal. Chem. 1987, 59,44-49. (11) Burton, D. E.; Sepaniak, M. J.; Maskarinec, M. P. J . Chromatogr. Sci., in press. (12) Balchunas, A. T.; Sepaniak, M. J. Anal. Chem. 1987, 59, 1466-1470. (13) Terabe, S.;Utsumi, H.; Otsuka, K.; Ando, T.; Inomata, T.; Kuze, S.; Hanaoka, Y. HRC CC , J . High Resolut . Chromatogr . Chromatogr . Commun. 1986, 9 , 686-670. (14) Pretorius, V.; Hopkins. B. J.; Schieke, J. D. J. Chromatogr. 1974, 99, 23-30. (15) Murray, G. M.; Sepaniak, M. J. J . Li9. Chromatogr. 1983, 6(5), 93 1-939. (16) Mukerjee, P.; Mysels, K. Critical Micelle Concentrations of Aqueous Surfactant Systems ; National Standards Reference Data Series; National Bureau of Standards: Washington, DC, 1971; Vol. 36. (17) Aniansson, E. A. G.; Wall, S. N.; Almgren, M.; Hoffmann, H.; Kielmann, I.; Ulbricht, W.; Zana. R.; Lang, J.; Tondre, C. J . Phys. Chem. 1976, 8 0 , 905-922. (18) Balchunas, A. T.; Swaile. D. F.; Powell, A. C.; Sepaniak. M. J. Proceedings from 5th Symposium on Separation Science and Technology for Energy Applications, submitted for publication.
RECEIVED for review August 31, 1987. Accepted November 24, 1987. This research was sponsored by the Division of Chemical Sciences, Office of Basic Energy Sciences, U.S. Department of Energy, under Contract DE-FG05-86ER13613 with the University of Tennessee (Knoxville).
