Electrokinetic Injection for Stacking Neutral Analytes in Capillary and

Anal. Chem. 2001, 73, 725-731

Accelerated Articles

Electrokinetic Injection for Stacking Neutral Analytes in Capillary and Microchip Electrophoresis James Palmer,† Dean S. Burgi,‡ Nicole J. Munro,| and James P. Landers*,†,§

Department of Chemistry and Department of Pathology, University of Virginia, Charlottesville, Virginia 22901, and Molecular Dynamics, 928 East Arques Avenue, Sunnyvale, California 94086

An on-column mechanism for electrokinetically injecting long sample plugs with simultaneous stacking of neutral analytes in capillary electrokinetic chromatography is presented. On-column stacking methods allow for the direct injection of long sample plugs into the capillary, with narrowing of the analyte peak width to allow for an increase in the detected signal. Low-pressure injections (∼50 mbar) are commonly used to introduce sample plugs containing neutral analytes. We demonstrate that injection can be accomplished by applying an electric field from the sample vial directly into the capillary, with neutral analytes injected by electroosmotic flow at up to 1 order of magnitude faster than the corresponding pressure injections. Since stacking occurs simultaneously with electrokinetic injection, stacking is initiated at the capillary inlet, resulting in an increased length of capillary remaining for separation. Reproducibility obtained for peak height and peak area with electroosmotic flow injection is comparable to that obtained with the pressure injection mode, while reproducibility of analysis time is markedly improved. Electrokinetic stacking of neutral analytes utilizing electroosmotic flow is demonstrated with discontinuous (high conductivity, high mobility) as well as continuous (equal conductivity, equal mobility) sample electrolytes. Injecting neutral analytes by electroosmotic flow affords a 10-fold or greater decrease in analysis times when capillaries of 50-µm i.d. or smaller are used. This stacking method should be exportable to dynamic pH junction stacking and electrokinetic chromatography with capillary arrays. Equations describing this electrokinetic injection mode are introduced and stacking of a neutral 10.1021/ac001046d CCC: $20.00 Published on Web 01/12/2001

© 2001 American Chemical Society

analyte on a microchip by electrokinetic injection using a simple cross-T channel configuration is demonstrated.

Capillary electrophoresis (CE) is a high-resolution technique for separating charged analytes in liquid solutions using electric fields.1 The microscale dimensions of the capillary serve to dissipate Joule heating and control convection efficiently at high separation voltages, allowing for plate heights of micrometers or less. However, because of the small radial dimension of a capillary, it is difficult to detect short injected plug lengths of lowconcentration analytes. Hence, injection of long sample plugs has been necessary to improve the detection of low-concentration analytes. Mechanisms have been developed to narrow, or stack, long analyte zones in sample plugs.2,3 Stacking decreases the length of an analyte zone, thus increasing the analyte concentration and its signal at detection. To separate neutral analytes by CE, it is necessary to provide an electrokinetic vector. Terabe and co-workers were the first to demonstrate this using micellar electrokinetic chromatography which utilized a charged micelle to impart mobility to neutral analytes.4 Electrokinetic chromatography (EKC) is a term that encompasses the use of any electrokinetic vector for separation of neutral analytes. Recent techniques for stacking neutral analytes * Corresponding author: (phone) 804-243-8658; (fax) 412-243-8852 (secure); (e-mail) [email protected]. † Department of Chemistry, University of Virginia. ‡ Molecular Dynamics. | Department of Chemistry, University of Pittsburgh, Pittsburgh, PA 15260. § Department of Pathology, University of Virginia. (1) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 51, 1298. (2) Chien, R.-L.; Burgi, D. S. J. Chromatogr. 1991, 559, 141-152. (3) Foret, F.; Sustacek, V.; Bocek, P. J. Microcolumn Sep. 1990, 2, 229-233. (4) Terabe, S.; Otsuka, U.; Ichihara, K.; Tsuchiya, A.; Ando, T. Anal. Chem. 1984, 56, 111-113.

Analytical Chemistry, Vol. 73, No. 4, February 15, 2001 725

in EKC require pressure injections of sample matrixes. High-salt stacking5 and sweeping6 have been applied to afford sample plug lengths up to 60% of the effective capillary length (length from the injection end to the detection point). In the “sweeping” approach described by Terabe’s group,6 continuous sample matrix/separation buffer conditions are utilized; i.e., there are no mobility or conductivity differences between the background electrolyte in the sample matrix and the separation buffer. This differs from the high-salt stacking we have described, which requires a discontinuous system induced by adding sodium chloride to the sample matrix. In a previous work, we described the stacking of neutral analytes by augmenting the sample matrix with a high salt concentration.5 That approach exploited discontinuous conditions where the sample matrix co-ion (the ion with the same charge as the electrokinetic vector, e.g., chloride) must have a higher intrinsic electrophoretic mobility than the electrokinetic vector. To afford stacking, the sample co-ion must be present at a higher concentration in the sample matrix than that of the electrokinetic vector in the separation buffer. These conditions coincide with those previously described by Mosher and Thormann7 to induce a pseudo-steady-state field boundary in discontinuous nonmicellar systems. When the sample matrix and separation buffer co-ions have different intrinsic electrophoretic mobilities, the co-ions in the sample matrix and separation buffer sort into zones according to their mobilities and concentrations.3 High-salt5 and sweeping6 methods for stacking neutral analytes utilize low-pressure sample injections. Pressure injections are typically low velocity to diminish mixing at the sample matrix/ separation buffer interface and to maintain reproducibility between analyses. While typical separations take as little as 60 s, the time necessary to introduce long sample plugs (e.g., 50% of the capillary length, ∼50-2000 s) by low pressure can exceed the analysis time by more than 1 order of magnitude. In an effort to reduce injection time and maximize the effective capillary length available for the separation mode, electrokinetic injection of neutral analytes by electroosmotic flow was explored. While electroosmotic flow can be used effectively to inject neutral analytes in EKC without stacking,8 it was reasoned the electrokinetic injection of high-salt sample matrixes by electroosmotic flow was possible, with the stacking of neutral analytes occurring simultaneously with the injection procedure. In this report, the electrokinetic stacking injection of neutral analytes in EKC is described. Electroosmotic flow is exploited for injection of long sample plugs under continuous and discontinuous co-ion conditions where the simultaneous stacking of neutral analytes at a co-ion interface between the sample matrix co-ion and the separation buffer (electrokinetic vector) is key. This allows for decreased analysis times using electrokinetic injection and yields enhanced limits of detection, with better reproducibility in comparison with the pressure injection approach. In smaller inner diameter capillaries, particularly microchip channels, manipulation of zones by electroosmotic flow is strongly indicated for ease of automation and rapid analysis.9 This is illustrated with the first (5) (6) (7) (8) (9)

726

Palmer, J.; Munro, N. J.; Landers, J. P. Anal. Chem. 1999, 71, 1679-1687. Quirino, J. P.; Terabe, S. Science 1998, 282, 465-468. Mosher, R. A.; Thormann, W. Electrophoresis 1985, 6, 477-482. Xue, G.; Pang, H.-m.; Yeung, E. S. Anal. Chem. 1999, 71, 2642-2649. Manz, A.; Graber, N.; Widmer, H. M. J. Chromatogr. 1990, 1, 244-252.

Analytical Chemistry, Vol. 73, No. 4, February 15, 2001

stacking injection of neutral analytes in EKC on a microchip. EXPERIMENTAL SECTION Materials. HPLC-grade water (Fisher, St. Louis, MO) was used for all separation buffers and sample matrixes. Sodium tetraborate, sodium hydroxide, sodium chloride, and punctilious ethanol were obtained from Sigma (St. Louis, MO). Cholic acid, sodium salt hydrate, was obtained from Aldrich Chemical Co. (Milwaukee, WI). Corticosteroids were obtained from Steraloids, Inc. (Newport, RI). 4,4-Difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-sindacene-3-propanol (BODIPY) was obtained from Molecular Probes (Eugene, OR), and sodium fluorescein was obtained from Acros. Sodium dodecyl sulfate was obtained from Bio-Rad Laboratories (Hercules, CA). Capillary Electrophoresis. Experiments with corticosteroid analytes were conducted with a Hewlett-Packard HP 3D-CE instrument interfaced with a Hewlett-Packard Kayak XA computer with HP 3D-CE ChemStation Rev. A.06.01[403] software for control of the instrument and data collection. The capillaries were 19-, 27-, or 50-µm i.d., as indicated, bare silica, polyimide surfaced, ∼370-µm o.d., from Polymicro Technologies, Inc. (Phoenix, AZ). The electrokinetic injection and separation polarities were normal (anode at the inlet). Capillary temperature was maintained at 20 °C, and detection was by UV absorption at 242 nm with a 20-nm bandwidth except as noted. Stacking of a fluorescent neutral compound (BODIPY) was performed with a Beckman P/ACE 5510 using a P/ACE system laser module 488 equipped with a 27 µm by 37 cm capillary. Separation Buffer and Sample Matrix Preparation. Separation buffer vial volumes were 0.7 mL. Sample matrix volumes were 100 µL. Separation buffer consisted of 80 mM sodium cholate, 5 mM tetraborate, and 10% punctilious ethanol in HPLC-grade water (pH range 8.9-9.1) or sodium dodecyl sulfate, at concentrations as noted, with 5 mM tetraborate and 20% methanol. Separation buffers were degassed by manual decompression in a syringe and passed through a 0.2-µm pore-diameter filter before use. Sample matrixes were prepared with sodium chloride or sodium tetraborate at stated concentrations in HPLC-grade water. Crystalline corticosteroids were dissolved in punctilious ethanol (316-362 µg/mL). Standards were stored at 4 °C when not in use. Aliquots were dried to remove the ethanol and redissolved in sample matrixes at stated concentrations. Analyte concentrations in the sample matrixes were as stated in each experiment. Capillary Electrophoresis Conditions. Capillary length was 33 or 48.5 cm with an inner diameter of 50 µm except as noted. New capillaries were conditioned by flushing with 1.0 M sodium hydroxide, water, and separation buffer, in order. For separations, the capillary was first flushed by high pressure (∼950 mbar) with fresh separation buffer (1-5 min). Separations were carried out at 30 kV. The separation buffer was replaced after no more than 2 h of total running time, and the capillaries were reconditioned daily by flushing with 1.0 M sodium hydroxide and water. Sample Matrix and Sample Injection Conditions. The velocity of sample plug injection by low-pressure and electrokinetic injection was determined for 33-cm capillaries with inner diameters of 19, 27, and 50 µm using 150 mM sodium chloride sample matrix and 80 mM sodium cholate with 10% ethanol and 40 mM borate separation buffer (pH ∼9). Triplicate injections were made under reverse pressure (-50 mbar) or reverse voltage (-30 kV) from

the outlet (sic) vial, which contained the high-salt matrix. During injection, time of absorbance shift at the detection point under these reversed conditions (distance 8.5 cm) was recorded to determine injection velocity. Injection velocity was similarly obtained with sample matrixes and separation buffer used for sweeping experiments (50 mM borate sample matrix and 40 mM sodium dodecyl sulfate with 20% methanol and 20 mM borate as separation buffer, pH ∼9). Viscosity differences between the sample matrixes and the separation buffer did not affect accurate determination of injection plug length. Relative conductivity of solutions was gauged as previously described.5 With the 50 µm by 33 cm capillary, sample matrixes consisting of 0, 25, 50, 75, 100, 150, 200, and 300 mM sodium chloride were examined for stacking effect by electrokinetic injections at 30 kV for 40 s. With the optimal salt concentration for stacking (150 mM sodium chloride), 5-, 10-, 20-, 40-, and 80-s injections were examined to verify linearity of injection time with peak response of neutral analytes at the detector. Peak area was examined to determine amount of analyte injected with injection duration. Triplicate electrokinetic injections from six identical high-salt samples, with separation mode using a single set of separation buffer vials, were used to determine the robustness of the separation buffer. Eighteen injections from a single high-salt sample matrix vial, using a single set of separation buffer vials, were undertaken to determine the reproducibility of the highsalt electrokinetic injection mode. Peak height, area, and migration times were examined. Sweeping conditions were examined with sample matrixes with conductivity equal to the separation buffer, with electrolyte conductivity in the sample matrix adjusted by addition of the separation buffer background electrolyte (borate), to provide continuous conductivity conditions. Sample matrix pH was previously determined to be unimportant with the selected analytes.5 Stacking of a fluorescent neutral compound (BODIPY) was demonstrated on the capillary format with conditions described above. Buffer and sample preparation were as listed for the corticosteroid separations, with BODIPY at 67 nM in 150 mM sodium chloride. Electrokinetic Stacking Conditions on the Microchip. A simple T-configuration cross-channel injection microchip (Alberta Microelectronics Corp., Edmonton, AB, Canada) was utilized to demonstrate stacking of a neutral analyte on a microchip format. The separation channel was 7.6 cm from the junction of the sample cross channel to the outlet (O), with a hemispherical cross section 50 µm wide by 20 µm deep. The sample T-channel had the same cross-section dimensions as the separation channel, extending perpendicular to the separation channel for 0.6 cm to the sample well (S) and waste well (W). The microchip channels were conditioned by flushing with 1.0 M sodium hydroxide, water, and separation buffer, in order. The apparatus for pressure-flushing the microchip channels was previously described.10 For analyses, the sample well was flushed with 150 mM sodium chloride and then filled with the sample matrix consisting of 150 mM sodium chloride with BODIPY at 67 nM. Wells I, O, and W were filled to equivalent levels with separation buffer. The separation channel was conditioned by inducing a field between I and O (+500/2000) for 100 s while holding S and W at ground to reduce cross(10) Hofgartner, W. T.; Huhmer, A. F. R.; Landers, J. P.; Kant, J. A. Clin. Chem. 1999, 45, 2120-2128.

channel leakage.11 Electrokinetic injection was performed by applying an electric field between S and O (+250/-1000) for 20, 40, 60, 80, 100 and 120 s, while floating W and I. Identical injections with BODIPY at 134 nM dissolved in separation buffer were made to demonstrate nonstacking injection conditions. Separation mode was initiated by redirecting the electric field from I to O (+500/2000), with S and W at ground. Detection was by laser-induced fluorescence. An argon ion laser (488 nm) was focused by a microscope objective (60×) into the separation channel. Fluorescence (514 nm) was collected collinearly through the microscope objective, focused through a lens (15.6-cm focal length, planoconvex), and passed through a spatial filter (0.3-mm pinhole) and a band-pass filter to a photomultiplier tube detector.12 RESULTS AND DISCUSSION To accurately quantify the amount of an analyte that is injected by electroosmotic flow, it is necessary to characterize electrokinetic injections. Huang et al.13 described two sources of bias for electrokinetic injection: analyte charge and the effect of the sample matrix on electroosmotic flow. Equation 1 describes

Qi ) liACi

(1)

analyte loading by pressure injections, where Qi is the moles of analyte i that are injected, li is the length of the capillary that analyte i occupies at the conclusion of injection, A is the crosssectional area of the capillary, and Ci is the molarity of analyte i in the injection vial. For electrokinetic injections in continuous systems

Qi ) [(µi + µeof)Etinj]ACi

(2)

where the term [(µi + µeof)Etinj] replaces li in eq 1, again under the conditions of continuous systems, with µi the electrophoretic mobility of analyte i, µeof the electrophoretic mobility of the electroosmotic flow, E the applied electric field, and tinj the amount of time the electric field for injection is applied. For neutral analytes, the term µi in eq 2 disappears, and it is indicated that the length of the sample plug is directly proportional to electroosmotic flow velocity. Consequently,

Qi ) µeofEtinjACi

(3)

However, if the sample matrix co-ion is discontinuous from the separation buffer, there is a potential bias from the effect of the sample matrix on electroosmotic flow. Chien and Helmer10 showed the bulk electroosmotic flow velocity is an average of the electroosmotic flow velocities of the sample matrix and the separation buffer, proportional to the length of the capillary each occupy. Hence

Qi ) vBCiAtinj ) [(xvS) + ((1 - x)(vsep))]CiAtinj

(4)

(11) Jacobsen, S. C.; Hergenroder, R.; Moore, A. W., Jr.; Ramsey, J. M. Anal. Chem. 1994, 66, 4127-4132. (12) Roach, M. C.; Gozel, P.; Zare, R. N. J. Chromatogr. 1988, 426, 129-140. (13) Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1988, 60, 377-380.


727

When the conductivity of the sample matrix is made to exceed that of the separation buffer by increasing the sodium chloride concentration to 75 mM or higher, analyte stacking is observed. It is apparent the stacking efficiency in this mode is dependent on the concentration of salt in the sample matrix (Figure 1). The co-ion in the sample matrix (chloride in this case) must be present at a concentration sufficient to reduce its velocity to less than that of the electrokinetic vector in the separation buffer. This causes the formation of a pseudo-steady-state co-ion boundary that forces an increased concentration of the electrokinetic vector in the separation buffer at the cathode side of the co-ion boundary. For the unique case here in which the leading electrolyte is electrokinetically injected adjacent to the separation buffer containing the trailing electrolyte (the electrokinetic vector), a pseudo-steadystate co-ion boundary will be formed under the following conditions:14

Figure 1. Stacking effect at different sample matrix salt concentrations. Sample matrix sodium chloride concentrations as listed in the figure; 30-kV injection for 40 s. Separation conditions: 80 mM sodium cholate, 10% ethanol, 5 mM tetraborate, pH ∼9 separation buffer; 50 µm i.d. by 48.5 cm capillary. Separation at 30 kV. Separation buffer was approximately the same conductivity as the 50 mM sodium chloride sample matrix. Peak order, 1-5: cortisone, cortisol, 11deoxycortisol, 17-hydroxyprogesterone, and progesterone. CB: coion boundary.

where vB is the bulk electroosmotic flow velocity, x is the length of the sample plug injected expressed as a fraction of the total capillary length, vS is the velocity of the electroosmotic flow solely in the presence of sample matrix, and vsep is the velocity of the electroosmotic flow solely in the presence of separation buffer. Including the electric field effect on velocity leads to

Qi ) [(xµeof,S) + ((1 - x)µeof,sep)]ECiAtinj

(5)

However, because the high-salt sample matrix has a resistance that differs from the separation buffer, the different electric fields in the two zones require eq 5 to be rewritten as

Qi ) [(xµeof,S)ES] + [((1 - x)µeof,sep)Esep]CiAtinj

(6)

Examination of electrokinetic injections under a variety of conditions for stacking neutral analytes was undertaken. Effect of Varied Sample Matrix Salt Concentration on Stacking Efficiency. Using different concentrations of sodium chloride (a high-mobility co-ion) in the sample matrix, electrokinetic injections of corticosteroids (neutral analytes) in sample matrixes ranging from 0 to 300 mM sodium chloride were examined (Figure 1). These experiments examined the effect of the leading co-ion (chloride) on inducing stacking conditions with the electrokinetic vector, cholate. For sample matrixes with conductivity below that of the separation buffer (0 and 25 mM sodium chloride), no evidence of sample stacking is apparent. Increasing the sodium chloride concentration to 50 mM yields a sample matrix that has roughly the same conductivity as the separation buffer, yet negligible analyte stacking is observed. 728


µSES < µevEev

(7)

µS > µev

(8)

with the restriction that

where µS and µev are the electrophoretic mobilities of the co-ion in the sample matrix and separation buffer, respectively, and ES and Ev are the electric field strengths in the sample co-ion zone and the electrokinetic vector zone. This is consistent with chloride having a postulated intrinsic mobility of roughly 1.5-3 times that of the cholate micelle under these conditions. This is based on the necessity of supplying chloride ion in the sample matrix at a concentration of 1.5-3 times that of the electrokinetic vector in the separation buffer for stacking to occur. Absolute mobility is difficult to interpret due to variations in ionic constituents that induce isotachophoretic effects as well as field differences induced by zonal conductivity differences. Effect of Electrokinetic Injection Length on Detection. The velocities of electrokinetic and low-pressure injected sample matrix plugs over a range of capillary inner diameters and lengths were compared. The time at which neutral components of the sample matrix (e.g., water or organic modifier) reached the detection point with 50-mbar pressure or 30-kV injection was used to determine velocity of respective injections. The sample plug length for neutral analytes can be determined by multiplying the pressure-induced flow velocity by duration of pressure injection and by multiplying the electroosmotic flow velocity by duration of electrokinetic injection. With the sample and separation buffer conditions described above, a 33 cm by 50 µm i.d. capillary has a low-pressure injection velocity of 0.081 cm/s or an electroosmotic flow injection velocity of 0.372 cm/s (with an applied voltage of 30 kV). Electrokinetic injection is 4.6 times faster at this capillary length and internal diameter. However, a 33-cm capillary with a 19-µm i.d. has a low-pressure injection velocity of 0.012 cm/s, or an electroosmotic flow injection velocity of 0.295 cm/s, representing an injection rate 25 times faster when electroosmotic flow is utilized. With longer capillaries, low-pressure injection is slowed by an increase in flow resistance proportional to the capillary (14) Palmer, J.; Landers, J. P. Anal. Chem. 2000, 72, 1941-1943.

Figure 2. Comparison of electrokinetic and low-pressure injections. Sodium chloride sample matrix, 150 mM. Duration of 30-kV injections as listed in the figure. Separation conditions: 80 mM sodium cholate, 10% ethanol, 5 mM tetraborate, pH ∼9 run buffer; 19 µm i.d. by 33 cm capillary. Separation at 30 kV. Total analysis time with an 80% effective-length plug injection by pressure was 1928 s. With electrokinetic injection, the total analysis time was 250 s. Peak order, 1-5: cortisone, cortisol, 11-deoxycortisol, 17-hydroxyprogesterone, and progesterone.

length. Likewise with electrokinetic injection, the field is reduced proportional to capillary length. Linearity of peak area to injection length was examined to determine whether eq 3 could be corroborated by experimental data. With the 50 µm i.d. by 33 cm capillary, peak area for progesterone is linear with electrokinetic injection lengths (R2 ) 0.9873), from 5 to 80 s (125% of the effective capillary length, i.e., 80-s electrokinetic injection, capillary length to detector, 24.5 cm, injected plug length, 30 cm). Similar analysis was undertaken with 50-µm-i.d. capillaries of 48.5- and 64.5-cm length, with peak areas corresponding to injection times by R2 values of 0.9979 and 0.9799. The linearity of peak area with injection duration indicates that the amount of neutral analyte injected is linear with injection duration and that eq 3 holds for predicting the amount of neutral analyte injected via electroosmotic flow with high-salt sample matrixes. There appears to be no bias due to the bulk electroosmotic flow changing as a function of sample plug length. Further study is ongoing to investigate this effect. The total analysis times for electrokinetic versus pressure injection of long sample plugs were compared by summing the injection and analysis times. For a 50% capillary-length injection using high-salt sample matrixes with a 33 cm × 50 µm capillary, electrokinetic injection is ∼50% faster (∼2.5 min total analysis versus ∼3.5 min with pressure injection). However, with smaller inner diameter capillaries, total analysis times can be decreased by roughly 1 order of magnitude with the use of electrokinetic injection. For example, an 80% effective capillary length injection takes 1808 s by low-pressure injection with a 33 cm × 19 µm capillary. The same injection can be made by a 73-s electrokinetic injection at 30 kV. With separation times of ∼150 s in either case, the total analysis time with electrokinetic injection is ∼9 times faster (Figure 2). For electrokinetic injection with a stacking boundary that is moving against the electroosmotic flow, capillary injections of more than 100% of the effective capillary length can be made (Figure 3), with capillary length still remaining for separation of the stacked analytes. Likewise, with continuous conditions, an electrokinetic injection 170% of the effective capillary length (∼42 cm sample plug length with an effective capillary length of 24.5 cm) can be made. The resulting resolution is similar to that observed when using a pressure injection of ∼64 cm with

Figure 3. Depiction of co-ion boundary movement into a capillary. A 1-m sample plug is injected by EOF into a 1-m capillary. The velocity of the co-ion boundary is one-fourth that of the EOF (moving left to right). The co-ion boundary is designated by the arrows. The sample plug length is unchangeable, but stacking of neutral analytes at the co-ion boundary can decrease the zonal width of the neutral analytes. Analytes are shown in the sample plug by dark gray background. At time t0, the sample plug with analytes is shown to the left of the capillary. At t1, the sample plug has moved 0.5 m into the capillary. The stacking boundary has advanced only 0.125 m. The leading portion of the sample plug has no analytes (light gray), as they are retained at the co-ion boundary (dark line). At time t2, the sample plug has completely filled the capillary, and the co-ion boundary has moved 0.250 m into the capillary. To finish injection, the capillary inlet is switched from the sample matrix to the separation buffer at this point. At time t3, the sample plug has completely transited the co-ion boundary, and analytes are undergoing separation with more than half of the capillary length remaining.

a capillary length of 80.5 cm15 with capillaries of 50-µm i.d. It is possible to use a longer capillary to increase the injected plug length, which leads to an increase in the effective capillary length remaining for separation. However, reduction in the maximum electric field is proportional to the length of the capillary, and analysis and injection times are proportionally longer, as well. In addition, a better signal-to-noise ratio is observed with electrokinetic injection, and analysis time with a 50-µm-i.d. capillary is reduced from ∼2000 s with a 42-cm sample plug in an 80.5-cmlong capillary to ∼200 s with a 33-cm capillary (Figure 4A). Limits of Detection with Electrokinetic Stacking Injections. With electrokinetic injection, extremely long, fast sample injections can be made. Since the stacking process occurs at a co-ion boundary that moves into the capillary at a lower velocity than the electroosmotic flow, it is possible to inject sample plugs longer than the effective length of the capillary. As can be seen (Figure 4B), the UV absorption signal dip of the electroosmotic flow at the detector immediately after the initiation of separation shows that the injection was ∼100% of the effective capillary length (i.e., the electroosmotic flow-injected plug was electrokinetically injected nearly to the detector window). High-k analyte/micelle systems are particularly well suited to this type of extended injection. In this case, sodium dodecyl sulfate was utilized as the electrokinetic vector for corticosteroid analytes. The sample matrix was 10 mM tetraborate with a leading co-ion hydrogen phosphate at 5 mM. Electrokinetic injection at 30 kV was made for 60 s directly from the sample vial. Separation was carried out at 30 kV. Three corticosteroids are resolved in less than 2 min at 50 ng/mL. While the signal-to-noise ratio with electrokinetic injection under sweeping conditions (Figure 4A) is similar to that already (15) Quirino, J. P.; Terabe, S. Anal. Chem. 1999, 71, 1638-1644.


729

Figure 4. (A) Electrokinetic stacking injection with continuous buffer conditions. Sample matrix was 50 mM borate (approximately the same conductivity of the separation buffer). Cortisone, 11-deoxycortisol, and progesterone at 10 ng/mL. Separation buffer: 50 mM SDS, 20% methanol, 5 mM tetraborate, pH ∼9; 50 µm i.d. by 33 cm capillary. Injection at 30 kV for 110 s. Separation at 30 kV. (B) Electrokinetic stacking injection instigated by a leading co-ion. Sample matrix was 40 mM borate with 5 mM hydrogen phosphate, pH ∼9. Cortisone, 11-deoxycortisol, and progesterone analytes at 50 ng/mL; 30 kV injection for 60 s. Separation conditions and peak order as in (A).

Figure 5. Electrokinetic stacking injection on a microchip. Sample matrix was 150 mM sodium chloride. BODIPY analyte was 67 nM. Injection at 183 V/cm for number of seconds as shown in the figure. Separation conditions: 80 mM sodium cholate, 10% ethanol, 5 mM tetraborate, pH ∼9 run buffer. Separation at 366 V/cm; 80-s injection of 134 nM BODIPY in separation buffer included to show nonstacking injection. Inset: microchip layout, with reservoirs S, sample, I, inlet, W, waste, and O outlet.

reported,16 the results (Figure 4B) show ∼1 order of magnitude better detection for the progesterone peak, with a theoretical plate number of 2 000 000/m. 11-Deoxycorticosterone exhibited a plate number of 1 100 000/m, while the cortisone peak is not as well sharpened. This high-salt stacking mechanism is thus ∼1 order of magnitude more sensitive, as well as 1 order of magnitude faster than that previously reported in the presence of electroosmotic flow.16 Reproducibility for Electrokinetic Stacking Injections. To determine the reproducibility of the electrokinetic injection method, the robustness of both the sample matrix and separation buffer was examined by repeated injection from a single sample matrix vial (n ) 18) and triplicate separation cycles with injections from six identical high-salt sample matrix vials (n ) 18). Peak height, peak area, and migration time for three analytes were examined. With injections from three separate identical high-salt sample matrixes, using a single set of separation buffer vials, analyte migration times have a coefficient of variance (CV) of 0.2%, peak heights between 3.2 and 3.9%, and peak areas between 6.4 and 8.2%. It is notable that the separation buffer is not affected by repeated high-salt sample matrix injections. Repeated injections from a single high-salt sample vial show migration time CV values of 0.2-0.3% for up to 18 injections (all that were attempted), with CV values for peak height and area comparable for up to 10 injections. For more than 10 injections, a trend of decreasing peak height and area is observed. Hydrolysis of the sample matrix can be severe with low-ionic strength sample matrixes,16 limiting sampling from a single vial to one electrokinetic injection. However, it has been shown that multiple injections can be made from high-salt sample matrixes with extremely reproducible analyte responses, thus expanding the utility of this mode. Electrokinetic Stacking Injections on a Microchip. Fieldamplified stacking of charged analytes on a microchip has been

previously demonstrated.17 Detection of neutral analytes on the microchip apparatus used was constrained to laser-induced fluorescence excitation by an argon ion laser at 488 nm. A further constraint was the use of a fluorescent analyte that was uncharged; this severely limited the number of analytes available. Consequently, BODIPY, a neutral and fluorescent analyte, was used to demonstrate stacking on a microchip. Using a simple T- configuration cross-channel injection microchip (Figure 5, inset) it was possible to emulate the electrokinetic injection method used on the capillary format by placing the separation buffer in the inlet (I), outlet (O), and waste (W) wells and the high-salt sample matrix in the sample well (S). By applying an electric field from the sample well to the outlet, electrokinetic injection of the highsalt sample matrix was initiated. Electrokinetic injection sample stacking was terminated and the subsequent separation initiated by applying voltage between the inlet and the outlet. Electrokinetic injections from 20 to 120 s were carried out to determine whether stacking could be observed. For comparison to nonstacking injections, the analyte was dissolved in the separation buffer and injected for 20-120 s, with the 80-s injection shown (note the control had twice the initial concentration of analyte as used in the stacking experiments) (Figure 5). The injection field was 50% that of the separation field. The linear increase in analyte peak height with injection time is evidence that the stacking of the neutral analyte is occurring. Note the 100-s injection has a fluorescence response ∼10 times greater than the analyte injected in separation buffer (nonstacking injection). Similarly, the nonstacking conditions had twice the concentration of BODIPY;

(16) Quirino, J. P.; Terabe, S. Anal. Chem. 2000, 72, 1023-1030.

(17) Jacobsen, S. C.; Ramsey, J. M. Electrophoresis 1995, 16, 481-486.

730 Analytical Chemistry, Vol. 73, No. 4, February 15, 2001

hence, a 20-fold peak height improvement is evidenced by utilizing a high-salt stacking injection. CONCLUSIONS The electrokinetic stacking of neutral analytes driven by electroosmotic flow for subsequent EKC separation has been demonstrated. With discontinuous sample matrix/separation buffer co-ions or continuous sample matrix/separation buffer coions, neutral analytes can be injected at the velocity of the electroosmotic flow. The stacking boundary in each mode allows electrokinetic sample plug injections greater than the actual capillary length. Not only was a marked increase in the subsequent resolution of analytes at injection lengths required for low detection limits of neutral analytes possible, injection times were reduced by 4-50-fold versus pressure injections. Injection of neutral analytes by electroosmotic flow is a logical and expedient stacking method for EKC. Electrokinetic injection has the potential to provide a convenient mode for stacking injections with capillary arrays8 and might also be exportable to other methods that utilize electrophoretically active stacking junctions, such as pH-mediated stacking.18 In addition, the use of electrokinetic injection allows the translation of neutral analyte stacking in EKC from the capillary to the microchip format. Many methods for gating injections in the microchip format have been devised. In this case, (18) Britz-McKibbin, P.; Chen, D. Y. Anal. Chem. 2000, 72, 1242-1252.

a co-ion boundary is used to maintain an interface between the sample matrix and separation buffer during injection. This has been demonstrated with a single fluorescent neutral analyte, but the method could find utility for stacking other neutral and charged species on the microchip format. ACKNOWLEDGMENT The authors thank Robert Oda (Mayo Clinic, Rochester, MN), Apryll M. Stalcup (University of Cincinnati, Cincinnati, OH), and Steve Weber (University of Pittsburgh, Pittsburgh, PA) for discussions on this topic, and Nicole Munro (University of Pittsburgh) and Andreas Hu¨hmer (Thermoquest, Mountain View, CA) for assistance with the microchip apparatus. They also acknowledge Beckman Instruments (Palo Alto, CA) and HewlettPackard (Waldbronn, Germany) for providing instrumentation, and Trevor Arends at Steraloids, Inc. (Newport, RI) for providing the corticosteroids for this study. Note Added after ASAP. The version posted ASAP on 1/12/01 inadvertently omitted N.J.M. as one of the authors. This was corrected in the final printed version posted on 2/14/01.

Received for review December 12, 2000.

September

1,

2000.

Accepted

AC001046D


731

Electrokinetic Injection for Stacking Neutral Analytes in Capillary and

Recommend Documents