Anal. Chem. 2007, 79, 6287-6294
Micelle Stacking in Micellar Electrokinetic Chromatography Braden C. Giordano,†,‡ Carl I. D. Newman,† Philip M. Federowicz,† Greg E. Collins,*,† and Dean S. Burgi§ Naval Research Laboratory, 4555 Overlook Avenue, S.W., Chemistry Division, Code 6112, Washington, D.C. 20375-5342, and dbqp, 767 Mahogany Lane, Sunnyvale, California 94086
In order to understand the role of stacked micelles in sample preconcentration, it is necessary to understand the factors that contribute to the micelle stacking phenomenon. Various MEKC background electrolyte (BGE) solutions were prepared in the presence of Sudan III in order to monitor the micelle stacking phenomenon in the anionic sodium dodecyl sulfate and sodium cholate micelle systems. The data show that micelle stacking is a dynamic process that is strongly dependent upon the relative conductivities of the sample matrix and BGE, the relative column length of the sample plug, and the mobilities of the ions involved in the stacking process regardless of electric field conditions (i.e., field-amplified stacking, sweeping, or high-salt stacking). Conditions under which micelle stacking can be expected to occur are presented, and the extent of micelle stacking is quantified. The micelle stacking phenomenon is correlated to the separation performance of a series of neutral alkaloids. It is shown that neutral analytes migrate rapidly through the evolving stacked micelle region in the initial moments of the separation. As a consequence of this transient interaction, analytes with small retention factors spend less time in the stacked micelle region and experience lower stacked micelle concentrations than analytes with large retention factors that spend more time in the growing stacked micelle region. It is also demonstrated that the extent of analyte enrichment generally increases with injection length, by facilitating greater interaction time with stacked micelles; however, enrichment will eventually plateau with increasing injection length as a function of an analyte’s affinity for the micelle. Finally, it is shown that, in contrast to conventional wisdom, a range of long injection plugs exist where separation efficiency can be dramatically improved due to analyte interaction with an actively growing stacked micelle region. Micellar electrokinetic chromatography (MEKC) is a mode of capillary electrophoresis that allows for the separation of neutral analytes. This is typically achieved by addition of an anionic or * To whom correspondence should be addressed. Phone: 202-404-3337. E-mail:
[email protected]. † Naval Research Laboratory. ‡ Current address: Nova Research Inc., 1900 Elkin Street, Suite 230, Alexandria, Virginia 22308. § dbqp. 10.1021/ac0701987 CCC: $37.00 Published on Web 07/18/2007
© 2007 American Chemical Society
cationic surfactant micelle in the background electrolyte (BGE) or separation buffer to impart a separation vector onto neutral analytes. Neutral analytes are reversibly complexed with micelles based upon their relative affinities. Consequently, the observed migration velocity of an analyte is the weighted average of the velocities experienced traveling as free analyte and as the analytemicelle complex. Although capillary-based separations, such as MEKC, afford analysis of nanoliter volume samples, they are inherently limited by poor sensitivity for absorbance detection as a result of the narrow diameter of the capillary (5-100 µm). In order to address this issue, several groups have developed on-capillary preconcentration techniques, including field-amplified stacking,1-3 fieldenhanced sample injection,4-7 dynamic pH junction stacking,8 sweeping,5,9-13 and high-salt stacking (HSS).14-17 In recent years, there has been discussion with regard to the mechanisms of sweeping and HSS as on-line preconcentration techniques.13,17 More specifically, there remains uncertainty regarding the nature of sweeping as it compares to high-salt stacking. Sweeping was originally presented by Quirino and Terabe as a means to improve sensitivity in MEKC.10 Effective sweeping is made possible by preparing analyte in a matrix absent of an electrokinetic vector and was originally presented under conditions where the conductivity of the sample matrix matched the conductivity of the BGE. This allows for analyte in the sample zone to be “picked up and concentrated” by the electrokinetic (1) Liu, Z. Y.; Sam, P.; Sirimanne, S. R.; Mcclure, P. C.; Grainger, J.; Patterson, D. G. J. Chromatogr.. A 1994, 673, 125-132. (2) Quirino, J. P.; Terabe, S. J. Chromatogr.. A 1997, 781, 119-128. (3) Wang, T. L.; Yuan, L. L.; Li, S. F. Y. J. Chromatogr.. A 2003, 1013, 19-27. (4) Macia, A.; Borrull, F.; Calull, M.; Aguilar, C. J. Chromatogr.. A 2006, 1117, 234-245. (5) Quirino, J. P.; Terabe, S. Anal. Chem. 1998, 70, 1893-1901. (6) Quirino, J. P.; Terabe, S. J. Chromatogr.. A 1998, 798, 251-257. (7) Wang, S. F.; Wu, Y. Q.; Ju, Y.; Chen, X. G.; Zheng, W. J.; Hu, Z. D. J. Chromatogr.. A 2003, 1017, 27-34. (8) Britz-McKibbin, P.; Ichihashi, T.; Tsubota, K.; Chen, D. D. Y.; Terabe, S. J. Chromatogr.. A 2003, 1013, 65-76. (9) Monton, M. R. N.; Otsuka, K.; Terabe, S. J. Chromatogr. A 2003, 985, 435445. (10) Quirino, J. P.; Terabe, S. Science 1998, 282, 465-468. (11) Quirino, J. P.; Terabe, S. Anal. Chem. 1998, 70, 149-157. (12) Quirino, J. P.; Terabe, S. Anal. Chem. 1999, 71, 1638-1644. (13) Quirino, J. P.; Terabe, S.; Bocek, P. Anal. Chem. 2000, 72, 1934-1940. (14) Munro, N. J.; Palmer, J.; Stalcup, A. M.; Landers, J. P. J. Chromatogr., B 1999, 731, 369-381. (15) Palmer, J.; Landers, J. P. Anal. Chem. 2000, 72, 1941-1943. (16) Palmer, J.; Munro, N. J.; Landers, J. P. Anal. Chem. 1999, 71, 1679-1687. (17) Palmer, J. E. J. Chromatogr.. A 2004, 1036, 95-100.
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007 6287
vector (i.e., micelles) in the BGE.10 At its most basic level, the sweeping mechanism of sample preconcentration requires a continuous electric field and an absence of separation vector (micelle) in the sample matrix. Under these conditions the length of the analyte plug is decreased according to
lsweep ) linj(1/(1 + k))
(1)
where, lsweep is the length of the analyte plug after sweeping, linj is the original length of the injected analyte plug, and k is the retention factor of the analyte by the micelle.10 It is clear from eq 1 that lsweep is dependent upon the analyte’s affinity for the micelle, such that, the more strongly retained analytes are swept more efficiently. Landers and co-workers developed high-salt stacking as an alternative method for analyte preconcentration.16 In this method, the salt concentration in the sample matrix is increased such that the conductivity of the sample matrix is large relative to the conductivity of the BGE. This results in a heterogeneous electric field through the column, wherein a lower apparent field is experienced by the sample matrix relative to the field experienced by the BGE. It was concluded that this field difference results in a stacking of micelles on the detector side of the sample zone.16 This increased micelle concentration causes the analytes to preconcentrate or stack, reducing the effective injection plug length and significantly improving detection limits and resolution. It was also concluded that micelles stack at the detector side of the sample plug when
µsampleEsample < µevEev
(2)
and
µsample > µev
(3)
where µev is the mobility of the electrokinetic vector (i.e., the micelle) in the BGE, Eev is the field in the BGE, Esample is the field in the sample plug, and µsample is the mobility of the sample stacking co-ion (the ion of a charge similar to the micelle present in the sample plug).16 Palmer later noted that micelles will stack at the sample plug/BGE interface when µsample is greater than µev.16 Finally, it has been suggested that optimal micelle stacking and, consequently, optimal preconcentration, occurs when the conductivity of the sample matrix is approximately twice that of the BGE.16 Strictly speaking, high-salt stacking can also be classified as sweeping; however, Palmer et al. demonstrated that an electrokinetic vector could be included in the sample matrix provided its mobility was greater than the mobility of the micelle in the BGE.16 In response to the introduction of HSS by Palmer et al.,16 Terabe and co-workers postulated that the high-conductivity sample matrix will drive analyte into the micelle, thereby artificially increasing the retention factor.13,18 This effect was quantitatively described by the enhancement factor, γ′, through which any expected enrichment in sample concentration would reflect a commensurate increase in micelle concentration. In addition, it (18) Quirino, J. P.; Kim, J. B.; Terabe, S. J. Chromatogr.. A 2002, 965, 357-373.
6288
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
was concluded that, while the micelle stacking occurs in HSS, it is effectively mitigated by active destacking via electrodispersion as a result of discontinuities between the sample matrix and BGE conductivities. Recent efforts in our laboratory have focused on the analysis of explosives in seawater for mine detection.19 Ideally, such an analysis should be rapid, with minimal time spent on sample preparation. With this in mind, direct injection of samples prepared in seawater was tested using MEKC-based separation conditions. Given the high ionic strength of seawater, high-salt stacking for sample preconcentration was expected to be an ideal mechanism for improving resolution and sensitivity.16 For borate-containing BGE with cholate micelles, high-salt stacking was achieved19 as outlined by Landers and co-workers.16 However, when employing sodium dodecyl sulfate (SDS) as the surfactant, a systematic increase in sample matrix conductivity relative to the BGE (using NaCl) resulted in a marked reduction in separation efficiency despite the application of conditions where the sample matrix conductivity was approximately twice that of the BGE. Since the underlying mechanism for preconcentration in high-salt stacking is the stacking of the micelle, the importance of a more complete understanding of micelle stacking is evident. The work presented in this report explores the specific phenomenon of micelle stacking as it relates to both the sweeping and high-salt sample stacking methods. The goal of this investigation is to develop a better understanding of micelle stacking and the extent to which it is influenced by the BGE, sample matrix composition, and length of the sample plug. In addition, the effects of micelle stacking on the enrichment of analytes during MEKC separations are explored. EXPERIMENTAL SECTION Reagents and Standards. Disodium tetraborate, 2-amino-2(hydroxymethyl)-1,3-propanediol (Tris), sodium chloride, sodium acetate (NaOAc), SDS, sodium cholate, sodium hydroxide, and Sudan III were purchased from Sigma-Aldrich (St. Louis, MO). Aconitine, benzoin methyl ether, colchicine, colchicoside, emetine dihydrochloride hydrate (emetine), nicotine, strychnine, thiocholchicine, thiocolchicoside, and yohimbine hydrochloride (yohimbine) were purchased from Sigma-Aldrich and prepared as 5 mM stock solutions in methanol. Equipment. All data were collected with a Beckman Coulter P/ACE MDQ capillary electrophoresis (CE) instrument equipped with a UV absorbance detector (Fullerton, CA). Experiments were performed in varying lengths (31-60 cm) of 50 µm i.d. × 360 µm o.d. fused-silica capillaries (PolyMicro, Phoenix, AZ) with applied potentials of 15-25 kV. Data were collected at 254 nm with a sampling rate of 16 Hz. Prior to each series of experiments, the capillary was conditioned with 1 M NaOH (1.5 min at 30 psi), followed by water (1.5 min at 30 psi), and finally with fresh BGE (2 min at 30 psi). Injections were performed hydrodynamically with 0.5-1.0 psi from 5 to 90 s to achieve injection lengths of 5-90 mm. Following sample injection, BGE was injected at 0.5 or 1.0 psi to position the front of the sample plug at the specified distance (lD) from the detector. The capillary temperature was maintained at 25 °C, (19) Giordano, B. C.; Copper, C. L.; Collins, G. E. Electrophoresis 2006, 27, 778786.
and the instrument was utilized at all times per manufacturer recommendations. Background Electrolyte Preparation and Separations. Background electrolytes for MEKC were prepared from stock solutions of Na2B4O7 (100 mM), Tris (500 mM), SDS (200 mM), and sodium cholate (200 mM). Tris and Na2B4O7 were chosen as BGE buffers because they have comparable pH values but dramatically different conductivities. Cholate-containing BGE solutions were prepared at a level of 80 mM cholate and either 10 mM Na2B4O7 (borate-cholate) or 10 mM Tris (Tris-cholate), without pH adjustment (∼pH 9.5). Similarly, SDS-containing BGE solutions were prepared to achieve final concentrations of 80 mM SDS and either 10 mM Na2B4O7 (borate-SDS) or 10 mM Tris (Tris-SDS), without pH adjustment (∼pH 9.5). Sample matrixes were prepared from each of the BGE solutions described above, as well as from 1-100 mM Na2B4O7, 1-300 mM NaOAc, and 1-300 mM NaCl. Samples were prepared by evaporating aliquots of the alkaloid stock solutions to dryness, followed by reconstitution in the appropriate sample matrix. Peak numbers in the figure graphics correspond to (1) nicotine, (2) aconitine, (3) colchicoside, (4) strychnine, (5) thiocolchicoside, (6) colchicine, (7) benzoin methyl ether, (8) yohimbine hydrochloride, (9) thiocholchicine, and (10) emetine dihydrochloride hydrate. Unless otherwise specified, analytes for separations using cholate-containing BGE were prepared at the following concentrations: nicotine and aconitine at 43.3 µM, strychnine and colchicine at 16.7 µM, and colchicoside, thiocolchicoside, benzoin methyl ether, emetine dihydrochloride hydrate, thiocholchicine, and yohimbine hydrochloride at 13.3 µM. Unless otherwise specified, analytes for separations using SDS-containing BGE were prepared at the following concentration: nicotine, aconitine, colchicoside, strychnine, thiocolchicoside, colchicine, and benzoin methyl ether at 28.5 µM. Conductivity Measurements. Conductivities of the sample matrixes and BGEs were determined by filling the capillary with the solution of interest and applying a potential of 5 kV while measuring the current at a fixed temperature (25 °C). The conductivity (σ) was calculated from the relationship, σ ) (ILT)/ (VA), where I is the measured current, LT is the length of the capillary, V is the applied voltage, and A is the cross-sectional area.20 RESULTS AND DISCUSSION The hydrophobic dye, Sudan III (50 µM), was included in each BGE to enable direct monitoring of the micelles. This dye is poorly soluble in water and has been used previously by Terabe and coworkers to determine micelle electrophoretic mobility.9 Sample matrixes of varying composition and concentration were utilized to obtain a more complete picture of micelle stacking as it relates to both the high-salt stacking and sweeping mechanisms of preconcentration in MEKC. Micelle Stacking in Borate-Cholate BGE. Effects of Sample Matrix Concentration. Figure 1A shows the effects of sample matrix concentration on micelle stacking. It is evident that the (20) Giordano, B. C.; Horsman, K. M.; Burgi, D. S.; Ferrance, J. P.; Landers, J. P. Electrophoresis 2006, 27, 1355-1362. (21) ABBREVIATIONS: micellar electrokinetic chromatography - MEKC, background electrolyte - BGE, sample matrix - SM, capillary electrophoresis - CE, field-amplified stacking - FAS, high-salt stacking - HSS
Figure 1. (A) Effect of sample matrix concentration on micelle stacking. SM: 5-100 mM Na2B4O7, linj ) 0.5 cm. (B) Effect of sample volume on micelle stacking. SM: 25 mM Na2B4O7, linj ) 0.5, 2.0, and 4.5 cm. (C) Effect of % RCL and sample matrix on the % IMC. SM: 5-100 mM Na2B4O7 (X), 5-300 mM NaOAc (0), and 5 - 300 mM NaCl (4), linj ) 0.5, 2.0, and 4.5 cm. Experimental conditions: lT ) 31 cm (lD ) 15 cm), V ) 15 kV (56 µA). Background electrolyte: 10 mM Na2B4O7, 80 mM cholate, 50 µM Sudan III.
peak height of the stacked cholate micelle increases with sample matrix concentration. It is also apparent that the vacancy following the stacked micelle band increases with sample matrix concentration, apparently due to electrodispersion effects. Effects of Injection Volume. While sample preconcentration in the form of sweeping and high-salt stacking can vastly improve detection limits, the realized improvement is dependent upon the amount of analyte injected. Arguably the most common mechanism for improving sensitivity is to simply inject longer plugs (i.e., larger volumes) of sample. When coupled to some sort of preconcentration mechanism, including field-amplified stacking, sweeping, or high-salt stacking, this approach not only lowers detection limits but also maintains or even improves resolution. Figure 1B illustrates the effects of a 25 mM Na2B4O7 sample matrix on micelle stacking with 0.5-, 2.0-, and 4.5-cm injection lengths with an lD of 15 cm. This concentration of Na2B4O7 is such that the conductivity of the sample matrix and BGE closely approximate one another (i.e., sweeping). It is apparent that the magnitude of the stacked micelle band increases almost 2-fold from 0.5- to 2.0-cm injection lengths, but decreases slightly from 2.0- to 4.5-cm lengths. Effects of Sample Matrix Conductivity and Injection Volume on Micelle Stacking. In order to compare the extent of micelle stacking between the different sample matrixes, the percent increase in micelle concentration (% IMC) is evaluated as a function of dimensionless percent relative conductivity length (% RCL). Values of RCL are calculated as
% RCL ) 100
( )(
)
σSM linj σBGE lT - linj
(4)
where, σSM is the conductivity of the sample matrix, σBGE is the conductivity of the background electrolyte, and lT is the entire length of the capillary. For these data, the % IMC was calculated by dividing the peak height of the stacked micelle by the signal associated with the BGE containing Sudan III and multiplying by 100. Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6289
Table 1. BGE, Sample Matrix, and % RCL under Which Maximum Micelle Stacking Was Observed for a 4.5-cm Sample Plug and lD ) 15.0 cm (lT ) 31.2 cm) BGE
sample matrix
% RCL
40 mM Na2B4O7 21.9 (HSS) 10 mM Na2B4O7 80 mM sodium cholate 100 mM NaOAc 27.0 (HSS) 300 mM NaCl 112.9 (HSS) 10 mM Tris 20 mM Na2B4O7 13.5 (sweeping) 80 mM sodium cholate 80 mM NaOAc 23.6 (HSS) 80 mM NaCl 38.8 (HSS) 10 mM Na2B4O7 11 mM Na2B4O7 6.7 (FAS) 80 mM SDS 80 mM NaOAc 18.5 (sweeping) 10 mM NaCl 5.1 (FAS) 10 mM Tris 10 mM Na2B4O7 20.2 (sweeping) 80 mM SDS 10 mM NaOAc 8.4 (FAS) 15 mM NaCl 18.5 (sweeping)
% IMC 580 1040 620 320 3360 920 200 320 30 80 250 40
Figure 1C illustrates the effect of RCL and sample matrix on the %IMC at an lD of 15 cm. For the Na2B4O7 and NaOAc sample matrixes, the extent of micelle stacking increases with % RCL until a maximum value is achieved (∼6% RCL for Na2B4O7 and ∼27% RCL for NaOAc). The data beyond these maximums are attributed to growth of the stacked micelle bandsat the detection point the micelle has not yet completely stacked, whereas the data prior to the maximums are attributed to bands that have achieved a maximum extent of stacking and are diffusion limited. It is apparent that the extent of micelle stacking increases over the entire range of % RCL for the NaCl sample matrix. This feature is attributed to the detection of diffusion-limited stacked micelles. The shaded symbols along the % RCL axis correspond to 4.5-cm injections of sample matrixes where σSM is significantly less than σBGEsconditions typical of field-amplified stacking (FAS). This may be a consequence of the significant electric field discontinuity present under these conditions preventing/disrupting the formation of the stacked micelle. The conditions under which maximum micelle stacking occur for each sample matrix-BGE system are presented in Table 1. Included in the table are the sample matrix concentrations, the corresponding % RCL with electronic condition (FAS, sweeping, or HSS), and % IMC for constant values of linj ) 4.5 cm and lT ) 31 cm (lD ) 15 cm). It is clear from the data presented in Table 1 that the maximum extent of micelle stacking occurs under a range of electric field conditions corresponding to FAS (σsample < σBGE), traditional sweeping (σsample ≈ σBGE), and HSS (σsample > σBGE). In addition, the extent of micelle stacking is significantly greater in cholate-containing BGEs than in SDS-containing BGEs. The corresponding separations of alkaloids under conditions identified in Table 1 are presented in Supporting Information. In all cases, no effective separation, resolving all analytes, is achieved in the 15-cm distance allotted for separations. Micelle Stacking and the Separation of Alkaloids. In order to further understand the role micelle stacking has on MEKCbased separations, it is important to investigate the apparent role of electric field homogeneity or lack thereof on analyte enrichment in the absence of micelle stacking. Figure 2 shows the separation of the alkaloids under conditions where the sample matrix contains the borate and cholate at a molarity ratio of 1:8 (as with the borate-cholate BGE) but at concentrations sufficient to achieve σsample/σBGE ) 0.1, 0.5, 1, 1.5, and 2.0. Electropherograms are presented pairwise as separation of the alkaloids (dark trace) and 6290 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
Figure 2. Effect of relative conductivity and separation distance on the separation of alkaloids in borate-cholate sample matrix. BGE, 10 mM Na2B4O7, 80 mM cholate; sample matrix (bottom to top), BGE at 0.1, 0.5, 1.0, 1.5, and 2.0 σsample/σBGE. All samples prepared as indicated in the Experimental Section. Separation of the alkaloids (dark trace); extent of micelle stacking (light trace). Experimental conditions: lT ) 60 cm, lD ) 1 cm (A) and 45 cm (B), linj ) 2.0 cm, V ) 25 kV (56 µA).
extent of micelle stacking (light trace) as determined by the presence of Sudan III in the BGE. Cumulatively, each pair illustrates electropherograms obtained under conditions from FAS (σsample/σBGE ) 0.1) to sweeping (σsample/σBGE ) 1) to HSS (σsample/ σBGE ) 2.0). Panel A shows the progress of each separation and any associated micelle stacking after 1 cm of separation distance; whereas, panel B shows the progress of each separation and any associated micelle stacking after 45 cm of separation distance. These data indicate that the most effective analyte enrichment is observed under conditions that facilitate FAS and that the extent of enrichment decreases as σsample/σBGE increases. It is also apparent for the traces corresponding to σsample/σBGE ) 0.1 that a small region of stacked micelles is present in panels A and B, as indicated by *. Although surprising, this is not entirely unexpected given that, in this pair of electropherograms, the concentration of cholate in the sample matrix is 8 mM, whereas the critical micelle concentration for cholate is ∼10 mM.16 Thus, while this sample matrix contains surfactant monomer, it is devoid of micelles, and the resulting mobility difference between the surfactant monomer and micelle is sufficient to support a nominal degree of micelle stacking. As σsample/σBGE increases, it is apparent that, not only does the extent of analyte enrichment decrease but separation efficiency decreases as well. Under conditions of a homogeneous electric field (σsample/σBGE ) 1.0), flat-top peaks are present, indicating no analyte stacking. Further increases in σsample/σBGE result in significant peak broadening, attributed to active electrodispersion of micelles from the relatively high
Table 2. Effect of Sample Matrix and Relative Conductivity on % IMC, Resolution, and Peak Height for Borate-Cholate BGE sample matrix Na2B4O7
NaOAc
NaCl
Figure 3. Effect of relative conductivity and separation distance on the separation of alkaloids in Na2B4O7 sample matrix. BGE, 10 mM Na2B4O7, 80 mM cholate; sample matrix (bottom to top), Na2B4O7 at 0.1, 0.5, 1, 1.5, and 2.0 σsample/σBGE. All samples prepared as indicated in the Experimental Section. Separation of the alkaloids (dark trace); extent of micelle stacking (light trace). Experimental conditions: lT ) 60 cm, lD ) 1 cm (A) and 45 cm (B), linj ) 2.0 cm, V ) 25 kV (56 µA).
concentration sample matrix into the relatively low concentration BGE. Figure 3 shows the separation of 10 alkaloids under conditions that promote micelle stacking with a Na2B4O7 sample matrix at concentrations sufficient to achieve σsample/σBGE ) 0.1, 0.5, 1, 1.5, and 2.0 relative to the borate-cholate BGE. Electropherograms are presented as in Figure 2. It is clearly demonstrated in panel A that analyte enrichment and micelle stacking are achieved within each sample matrix and that greater micelle stacking is achieved when σsample/σBGE ) 0.5, 1.0, and 1.5 than when σsample/σBGE ) 0.1 or 2.0. Panel A also illustrates that analyte has separated from the stacked micelle region after less than 1 cm of separation distance, intimating rapid analyte migration through the evolving stacked micelle region. It is apparent from panel B that, after 45 cm of separation distance, analytes are effectively separated, the stacked micelle region is traveling well behind the analytes, and the extent of micelle stacking is maintained (σsample/σBGE ) 0.1 or 0.5) or substantially greater than after 1 cm (σsample/σBGE ) 1.0, 1.5, and 2.0). The best separation, as determined by peak symmetry and resolution of the limiting pair (yohimbine hydrochloride and thiocolchicinespeaks 8 and 9, respectively), occurs under conditions consistent with sweeping (σsample/σBGE ) 1.0). As expected, preconcentration efficiency is a function of retention factor with more retained analytes experiencing greater enrichment than less retained analytes. Similar experiments were performed using the NaOAc and NaCl sample matrixes at concentrations sufficient to achieve
σSM/ % IMC at % IMC at R (limiting % increase σBGE lD ) 1 cm lD ) 45 cm pair) in peak height 0.1 0.5 1.0 1.5 2.0 0.1 0.5 1.0 1.5 2.0 0.1 0.5 1.0 1.5 2.0
140 630 420 330 490 30 450 620 440 650 0 170 460 680 620
200 590 900 1000 1000 90 200 260 320 400 80 120 160 180 190
0.38 0.86 0.54 0.64 0.37 0.60 0.71 0.39 0.43 0.68 0.83
130 420 800 470 420 270 600 1140 1350 890 340 390 750 1060 1250
σsample/σBGE ) 0.1, 0.5, 1, 1.5, and 2.0 relative to the borate-cholate BGE (data not shownsfigures presented in Supporting Information). The % IMC at 1 and 45 cm from the detector, the resolution of the limiting pair (yohimbine and thiocholchicine), the percent increase in peak height of the most retained peak (emetine)s calculated from relative peak heights achieved under stacking and conventional MEKC conditionssfor each sample matrix are summarized in Table 2. Analyte enrichment and micelle stacking are observed for both the NaOAc and NaCl sample matrixes, and the effectiveness of analyte stacking (determined by resolution and peak height) generally improves with increasing σsample/σBGE. In a fashion similar to that of the Na2B4O7 sample matrix, analytes injected in NaOAc and NaCl separate from the stacked micelle region after less than 1 cm of separation distance. For the NaOAc sample matrix, after 45 cm of separation distance, the % IMC decreased for σsample/σBGE ) 0.5, 1.0, 1.5 and 2.0. The best separation using the NaOAc sample matrix, as determined by peak symmetry and resolution of the limiting pair, occurs under conditions consistent with HSS (σsample/σBGE ) 1.5). For the NaCl sample matrix, after 45 cm of separation distance, a similar decrease in % IMC is observed for σsample/σBGE ) 0.5, 1.0, 1.5 and 2.0. The best separation using the NaCl sample matrix, as determined by peak symmetry and resolution of the limiting pair, occurs under conditions consistent with HSS (σsample/σBGE ) 2.0). Figure 4 shows the corresponding separations of alkaloids where sample is prepared under conditions that afford the best analyte enrichment as determined by peak shape and the ability to resolve the alkaloids over the entire separation window. Panels A and B show the separation of 10 alkaloids in the borate-cholate and Tris-cholate BGE systems, respectively. Similarly, panels C and D show the separation of seven alkaloids in the borateSDS and Tris-SDS BGE systems, respectively. The bottom trace in each panel shows the separation corresponding to sample prepared in the respective BGE. The remaining traces in each panel show separations corresponding to sample prepared in Na2B4O7, NaOAc, and NaCl sample matrixes, from bottom to top. The % RCLs, with electronic conditions under which separations occur, are summarized in Table 3. Efficient analyte preconcentration typically occurs under the entire range of electronic conditions (FAS, sweeping, and HSS). The common trend Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6291
Figure 4. Effect of optimized sample enrichment on the separation of alkaloids. BGE, 10 mM Na2B4O7, 80 mM cholate (A), 10 mM Tris, 80 mM cholate (B), 10 mM Na2B4O7, 80 mM SDS (C), and 10 mM Tris, 80 mM SDS (D). Sample matrixes are BGE, Na2B4O7, NaOAc, and NaCl, from bottom to top at concentrations presented in Table 3. All samples prepared as indicated in the Experimental Section. Experimental conditions: lT ) 60 cm, lD ) 1 cm (A) and 45 cm (B), linj ) 2.0 cm, V ) 25 kV (56 µA).
Table 3. BGE, Sample Matrix, and % RCL under Which Optimal Analyte Enrichment Was Observed for a 2.0-cm Sample Plug and 45.0-cm Effective Length (60.0-cm Total Length) BGEHR
sample matrix
%RCL
10 mM Na2B4O7 80 mM sodium cholate
25 mM Na2B4O7 100 mM NaOAc 90 mM NaCl 20 mM Na2B4O7 80 mM NaOAc 70 mM NaCl 10 mM Na2B4O7 80 mM NaOAc 40 mM NaCl 10 mM Na2B4O7 10 mM NaOAc 15 mM NaCl
2.8 (sweeping) 5.5 (HSS) 7.2 (HSS) 2.8 (sweeping) 4.8 (HSS) 14.1 (HSS) 2.8 (sweeping) 3.8 (sweeping) 6.2 (HSS) 4.1 (sweeping) 1.7 (FAS) 3.8 (sweeping)
10 mM Tris 80 mM sodium cholate 10 mM Na2B4O7 80 mM SDS 10 mM Tris 80 mM SDS
observed for the borate-cholate, borate-SDS, and Tris-cholate systems is that the % RCL affording the best separation increases as a function of the increasing negative mobility of the sample stacking co-ion (borate ) -36.2 × 10-5 cm2/V‚s; acetate ) -42.4 × 10-5 cm2/V‚s; chloride ) -79.1 × 10-5 cm2/V‚s) relative to the micelle mobility (cholate ) -23.1 × 10-5 cm2/V‚s in borate; cholate ) -25.3 × 10-5 cm2/V‚s in Tris; SDS ) -35.7 × 10-5 cm2/V‚s in borate; SDS ) -27.4 × 10-5 cm2/V‚s in Tris). It is clear at this point that micelle stacking is a phenomenon that is common to the range of conditions that correspond to FAS, 6292 Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
sweeping, and HSS preconcentration techniques, regardless of electronic conditions. Moreover, the interaction analyte has with stacked micelles occurs within the first few moments of a separation; during which time the stacked micelle band is actively developing. These conclusions notwithstanding, the extent to which stacked micelles affect sample enrichment is yet to be addressed. Terabe and co-workers have indicated that analyte preconcentration efficiency should follow eq 1 regardless of micelle stacking due to destacking via electrodispersion.13, 18 Upon closer inspection of eq 1, it is obvious that k is assumed constant regardless of whether micelle stacking or destacking occurs and that a plot of lsweep versus linj should be linear with a slope of 1/(1 + k). Therefore, analytes that are weakly retained by the micelle (smaller k) elute earlier and should demonstrate more positive slopes than analytes that are strongly retained by the micelle (larger k). Moreover, because eq 1 assumes that k is constant, the slope of lsweep versus linj should also be positive and constant. Figure 5A-C shows the effect of injection length (linj) on peak width (lsweep) for colchicoside, thiocolchicoside, and strychnine (peaks 3, 5, and 4 in Figure 4C and D, respectively). Panels A and B illustrate the trends for the borate-SDS BGE with optimized Na2B4O7 and NaCl sample matrixes (Table 3), whereas panel C illustrates these trends with a Na2B4O7 concentration such that σsample/σBGE ) 1.0 (sweeping). These data demonstrate that the value of k experienced by individual analytes generally increases with short linj and tends to plateau after ∼2 cm,
Figure 5. Effect of injection length (linj) and sample matrix on peak width (lsweep) for colchicoside (X), thiocolchicoside (0), and strychnine (4) in borate-SDS and Tris-SDS BGE systems. (A) BGE, 10 mM Na2B4O7, 80 mM SDS; SM, 10 mM Na2B4O7 (σsample/σBGE ) 0.8). (B) BGE, 10 mM Na2B4O7, 80 mM SDS; SM, 40 mM NaCl (σsample/σBGE ) 1.8). (C) BGE, 10 mM Na2B4O7, 80 mM SDS; SM, 15 mM Na2B4O7 (σsample/σBGE ) 1.0). (D) Effect of injection length (linj) on the separation, peak width, and resolution of 5 alkaloids: thiocolochicoside (5), benzion methyl ether (7), colchicine (6), strychnine (4), and aconitine (2). BGE, 10 mM Na2B4O7, 80 mM SDS; SM, 10 mM Na2B4O7 (σsample/σBGE ) 0.8). All samples prepared as indicated in the Experimental Section. Experimental conditions: lT ) 60 cm, lD ) 45 cm, linj ) 0.5-4.5 cm, V ) 25 kV (56 µA).
regardless of electronic conditions (FAS, sweeping, or HSS). Data and discussion pertaining to all BGEs are presented in Supporting Information. It is interesting to note that, in Figure 5A, it is apparent that initially a negative slope is observed for each analyte, that an increasing slope is then observed for colchicoside after linj ) 1 cm, a zero slope is observed for thiocolchicoside after linj ) 2 cm, and a less negative slope is observed for strychnine after linj ) 3 cm. These data indicate that colchicoside experiences a relatively stable k value for linj > 1 cm, that thiocolchicoside experiences a monotonically increasing k value for linj > 2 cm, and that the k value experienced by strychnine is increasing less rapidly after linj > 3 cm than at linj < 3 cm. Moreover, these trends illustrate the effect of retention order on the extent of analyte enrichment.
In each of these figures, analytes that are weakly retained by the micelle will rapidly enter and exit the growing stacked micelle region, experiencing marginal benefit from the gradually increasing localized micelle concentration. However, analytes that are strongly retained by the micelle will experience greater enrichment from the growing stacked micelle region. Thus, the extent of analyte enrichment from micelle stacking is intrinsically linked to the extent of interaction with the micelle. Finally, it is common knowledge that resolution in CE is a diffusion-limited process and the consequence of increasing injection plug length is a concomitant loss in resolution. Figure 5D illustrates the effect of increasing linj on the separation and resolution of thiocolchicoside, benzoin methyl ether, colchicine, strychnine, and aconitine in the borate-SDS BGE system. Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
6293
Included in this figure is the peak width at baseline for benzoin methyl ether and the resolution of the limiting pair, colchicine and strychnine. Under typical MEKC conditions and under sweeping conditions with the assumption of a constant k value (eq 1), one would expect that peak width should increase with linj. However, as demonstrated in Figure 5A-C, under conditions where micelle stacking occurs, the ratio of linj to lsweep is not constant, presumably due to increasing k afforded by the micelle stacking phenomenon. This is realized for the benzoin methyl ether peak, as a consistent increase in peak width (lsweep) as a function of linj is not observed. Similarly, the resolution between colchicine and strychnine does not behave in a conventional manner. In fact, the best resolution between the peaks is observed for the longest injection time. It is important to note that this improved resolution is strongly dependent upon the analytes’ affinity for the micelle. Landers and co-workers observed that analytes with low affinity for the micelle would broaden as the injection plug length lengthens; consequently a maximum peak height is observed as a function of affinity for the micelle and a simultaneous loss of resolution is realized.16 Nevertheless, this improved resolution as a function of injection time is another example of the beneficial contribution micelle stacking has on analyte preconcentration in MEKC. CONCLUSIONS We have demonstrated the first comprehensive picture of the micelle stacking phenomenon as it affects analyte enrichment in MEKC using electronic conditions conducive for field-amplified stacking, sweeping, and high-salt stacking. The data show the following: (a) Micelle stacking occurs under conditions traditionally attributed to field-amplified stacking, sweeping, and high-salt stacking as determined by the relative conductivities of the sample matrix and BGE; (b) Micelle stacking is a dynamic process that is dictated by the nature and concentration of the sample matrix and BGE, as well as by the nature of the surfactant;
6294
Analytical Chemistry, Vol. 79, No. 16, August 15, 2007
(c) A greater extent of micelle stacking is achieved with cholate than with SDS. Moreover, the greatest extent of micelle stacking is observed with a NaOAc sample matrix in Tris-cholate BGE, whereas, the borate-cholate BGE exhibits the best dynamic range for facilitating micelle stacking. (d) Analyte interaction with the evolving stacked micelle region is transient and occurs early in the separation (∼1 cm). (e) Increasing injection plug length allows for a greater extent of analyte enrichment due to prolonged interaction with an increasing retention factor (k) that results from the localized increased micelle concentration. This enhancement is a function of analyte affinity for the micelle and eventually plateaus. (f) Even under conditions where the extent of micelle stacking is small (i.e., conditions where SDS is the surfactant), a tangible improvement in resolution is observed with increasing injection length. Cumulatively, this work illustrates that micelle stacking plays a significant and beneficial role in the overall effectiveness of analyte preconcentration in MEKC. Future work will focus on understanding the effects of micelle stacking under low EOF conditions and the development of methodologies by which a larger degree of interaction can be engendered between analytes and stacked micelles under high EOF conditions. ACKNOWLEDGMENT The authors acknowledge the efforts of Mark Hammond, U.S. Naval Research Laboratory-Code 6181. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review January 31, 2007. Accepted June 12, 2007. AC0701987
