Chromatographic Preconcentration Coupled To Capillary

Jun 30, 2004 - With a micro injection valve, sample could be effectively introduced into the CE system in an in-line fashion without seriously affecti...
9 downloads 0 Views 84KB Size
Download PDF
Anal. Chem. 2004, 76, 4432-4436

Chromatographic Preconcentration Coupled To Capillary Electrophoresis via an In-Line Injection Valve F. W. Alexander Tempels, Willy J. M. Underberg, Govert W. Somsen, and Gerhardus J. de Jong*

Department of Biomedical Analysis, Faculty of Pharmaceutical Sciences, Utrecht University, Sorbonnelaan 16, NL-3584 CA Utrecht, The Netherlands

A preconcentration-capillary electrophoresis (CE) system using a small precolumn in combination with an in-line injection valve is presented. The advantage of the present design is the ability to perform the sample preconcentration fully independently from the CE separation and to prevent sample matrix and washing solvents from entering the CE capillary. With a micro injection valve, sample could be effectively introduced into the CE system in an in-line fashion without seriously affecting the CE separation efficiency. Breakthrough volume, desorption efficiency, and elution volume for the C18 microcolumn (5 × 0.5 mm i.d.) were established, yielding values of 750 µL, 70%, and 0.9-1.1 µL, respectively, using enkephalin peptides. The time between the start of the desorption of the analytes from the precolumn and the injection into the CE system was also studied in order to achieve optimal sensitivity and separation efficiency. The performance of the complete system was demonstrated by the preconcentration and separation of an enkephalin mixture. Using a sample volume of 250 µL and a CE injection voltage of -15 kV for 12 s, linearity was observed over 2 orders of magnitude, and detection limits (S/N ) 3) were in the 5-10 ng/mL range. A 1000-fold sensitivity enhancement is obtained using this setup, as compared to a regular CE setup. For 100 ng/mL samples, repeatabilities (RSDs) of migration time and peak area were 1.2 and 11%, respectively.

Focusing on preconcentration techniques in combination with CE, electrophoretically based or chromatographically based techniques can be discerned. Typically, electrophoretic preconcentration takes place in the separation capillary,1-4 whereas the chromatographic preconcentration is carried out either in an offline,5,6 on-line,7,8 or in-line9-11 fashion. In the in-line mode, the preconcentration column is an integrated part of the CE system; in the on-line mode, the precolumn is no part of the CE system, but is coupled to this via an interface. Electrophoretic preconcentration is less suitable for the direct analysis of biological samples and generally has lower preconcentration factors than the chromatographic techniques for which 2-4 orders of magnitude preconcentration have been achieved.12 When using these chromatographic techniques, on-line or in-line methods are regarded as advantageous, as compared to the off-line approach, as a result of their shorter total analysis times, minimum of sample handling, and possibility of automation. For the coupling of chromatographic preconcentration techniques and CE, several aspects are essential. Ideally, the precolumn should provide high loadability and high loading speed. Additionally, possibilities for directing flows to waste during trapping and washing are preferred. Moreover, small elution volumes are desirable to achieve high preconcentration factors and to allow a maximum amount of analyte(s) to be injected in the CE system. Although effective preconcentration has been reported for CE systems with in-line preconcentrators, these systems share the disadvantage that high-speed loading is not

Capillary electrophoresis (CE) has gained increasing importance in pharmaceutical and biomedical analysis because of its high separation efficiency. Most commonly, UV detection is applied in CE because of its simplicity and wide applicability. Using UV detection, the concentration limits of detection (CLODs) of most CE systems are in the low micromolar range, mainly caused by the small injection volumes (1-10 nL) and the short optical path length (often 50 or 75 µm). These CLODs are generally insufficient for (biological) samples with low-abundance components. To overcome these limitations in sensitivity, either more sensitive detection systems, for example, laser-induced fluorescence or mass spectrometry, have to be used or the sample has to be preconcentrated.

(1) Waterval, J. C. M.; Hommels, G.; Teeuwsen, J.; Bult, A.; Lingeman, H.; Underberg, W. J. M. Electrophoresis 2000, 21, 2851-2858. (2) Thompson, T. J.; Foret, F.; Vouros, P.; Karger, B. L. Anal. Chem. 1993, 65, 900-906. (3) Shihabi, Z. K.; Friedberg, M. J. Chromatogr., A 1998, 807, 129-133. (4) Stegehuis, D. S.; Irth, H.; Tjaden, U. R.; van der Greef, J. J. Chromatogr. 1991, 538, 393-402. (5) Loos, R.; Niessner, R. J. Chromatogr., A 1998, 822, 291-303. (6) Carabias-Martı´nez, R.; Rodrı´guez-Gonzalo, E.; Domı´nguez-AÄ lvarez, J.; Herna´ndez-Me´ndez, J. J. Chromatogr., A 2000, 869, 451-461. (7) Veraart, J. R.; Gooijer, C.; Lingeman, H.; Velthorst, N. H.; Brinkman, U. A. Th. J. Pharm. Biomed. Anal. 1998, 17, 1161-1166. (8) Stroink, T.; Schravendijk, P.; Wiese, G.; Teeuwsen, J.; Lingeman, H.; Waterval, J. C. M.; Bult, A.; de Jong, G. J.; Underberg, W. J. M. Electrophoresis 2003, 24, 1126-1134. (9) Tomlinson, A. J.; Naylor, S. J. Capillary Electrophor. 1997, 2, 225-233. (10) Guzman, N. A. J. Liq. Chromatogr. 1995, 18, 3751-3768. (11) Waterval, J. C. M.; Bestebreurtje, P.; Lingeman, H.; Versluis, C.; Heck, A. J. R.; Bult, A.; Underberg, W. J. M. Electrophoresis 2001, 22, 2701-2708. (12) Sentellas, S.; Puignou, L.; Galceran, M. T. J. Sep. Sci. 2002, 25, 975-987.

* To whom correspondence should be addressed. Phone +31 30 253 6591. Fax +31 30 253 5180. E-mail [email protected].

4432 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

10.1021/ac0499221 CCC: $27.50

© 2004 American Chemical Society Published on Web 06/30/2004

possible because of the small capillary diameter and that the sample matrix is introduced into the CE separation capillary. Moreover, elution volumes often exceed levels that are acceptable in CE and, thus, demand additional stacking procedures to achieve good separation efficiencies. On-line systems that allow switching of the precolumn could circumvent these problems. The main problem for the on-line introduction of eluates into a CE system is the difference in volume dimensions and requirements of the CE and the preceding (chromatographic) system. Typically, only a small part of the eluate is introduced in the CE system, which can be performed using different approaches. The eluate can be split, thereby flowing the major part of the eluate to waste and a minor part into the CE system, and thus, a continuous sampling out of the complete elution plug occurs. Examples of this approach can be found in the literature, with a vial interface,7,8 a valve with a loose nut,13 or a combination of a valve and a tee.14 In another procedure, a specific zone of the elution plug is introduced into the CE separation capillary, that is, a zone-cutting method. This procedure has been followed by several researchers, who used a cross15,16 or a valve.17,18 For this purpose, (homemade) injectors to introduce specific volumes into the CE system19-21 may be useful. In this paper, a micro trapping column is coupled on-line to a micro injection valve that is positioned in a CE system. The preconcentration-CE procedure includes analyte trapping and desorption, capturing of a part of the elution plug in the loop of the in-line injection valve, and (partial) injection of concentrated sample out of the loop into the separation capillary with subsequent separation. For preconcentration, a small C18 column is chosen, providing high loading capacity and small elution volumes. The precolumn is placed not in the loop position of a valve in the preconcentration system, but in the permanent flow path to avoid additional broadening of the elution plug. Moreover, the precolumn is no part of the CE system, thereby avoiding adverse effects of the column and the sample matrix on the CE performance. An in-line injection valve is used to partially inject an eluted sample plug from the trapping column into the CE. The various parts of the system are optimized, and the analytical performance of the total system is demonstrated. EXPERIMENTAL SECTION Materials. Boric acid was purchased from OPG Pharma (Utrecht, The Netherlands), acetonitrile from Biosolve BV (Valkenswaard, The Netherlands), and sodium hydroxide (NaOH) from Bufa BV (Uitgeest, The Netherlands). Acetates of Leu-enkephalin (Tyr-Gly-Gly-Phe-Leu-OH, pI ) 5.93), des-Tyr1-[D-Ala2-D-Leu5]enkephalin (D-Ala-Gly-Phe-D-Leu-OH, pI ) 6.02), and [Met5]enkephalin (Tyr-Gly-Gly-Phe-Met-OH, pI ) 5.93) were obtained (13) Bushey, M. M.; Jorgenson, J. W. Anal. Chem. 1990, 62, 978-984. (14) Lemmo, A. V.; Jorgenson, J. W. J. Chromatogr. 1993, 633, 213-220. (15) Cabaleiro, O.; Lores, M.; Cela, R. Analusis 1999, 27, 468-471. (16) Lewis, K. C.; Opiteck, G. J.; Jorgenson, J. W.; Sheeley, D. M. J. Am. Soc. Mass Spectrom. 1997, 8, 495-500. (17) Debets, A. J. J.; Mazereeuw, M.; Voogt, W. H.; Iperen, D. J. v.; Lingeman, H.; Hupe, K.-P.; Brinkman, U. A. Th. J. Chromatogr. 1992, 608, 151-158. (18) Cooper, J. W.; Chen, J.; Li, Y.; Lee, C. S. Anal. Chem. 2003, 75, 10671074. (19) Iizuka, E.; Tsuda, T.; Munesue, M.; Samizo, S. Anal. Chem. 2003, 75, 39293933. (20) Tsuda, T.; Mizuno, T.; Akiyama, J. Anal. Chem. 1987, 59, 799-800. (21) Ponton, L. M.; Evans, C. E. Anal. Chem. 2001, 73, 1974-1978.

from Sigma (St. Louis, MO). Throughout the experiments, deionized water (Milli-Q system, Millipore, Bedford, MA) was used. The background electrolyte (BGE) in all CE experiments was 50 mM borate adjusted to pH 8.0 with NaOH. Peptides were dissolved in water, either at a concentration of 100 µg/mL for experiments without preconcentration or at 0.1 µg/mL for experiments with preconcentration, unless otherwise stated. Apparatus and General Procedures. For all CE experiments, a CE apparatus (PrinCE Technologies, Emmen, The Netherlands), a UV detector (model K-2501, Knauer Wellchrom, Berlin, Germany) set at a wavelength of 214 nm and fused-silica capillaries with 75-µm i.d. and 375-µm o.d. (BGB Analytik AG, Anwil, Switzerland) were used. Two microvalves (model M-435, Upchurch Scientific, Oak Harbor, WA) for introduction of desorption solvent and in-line CE injection and a regular valve (model 7010, Rheodyne LLC, Rohnert Park, CA) for sample introduction were used. The microvalves have port-to-port volumes of 51 nL, and all their fluid-contacting materials (polyetheretherketone (PEEK) and ceramics) have insulating properties. All experiments were performed at room temperature. Daily startups of the CE systems consisted of successively rinsing the capillaries with 0.1 M NaOH, water, and BGE (each for 5 min at 1500 hPa). Samples were injected at -15 kV for 12 s unless otherwise stated, and analyses were carried out at -15 kV for 20 min. Each experiment was followed by a rinse of water and BGE (each for 5 min at 1500 hPa). In the preconcentration system, the trapping column was rinsed for 60 min with acetonitrile and water prior to first use. Water was used to transfer the peptide sample (250 µL) and to rinse the trapping column. Trapping and rinsing was done for 8 min at a flow rate of 50 µL/min. Desorptions of trapped peptides were carried out at 3 µL/min with 426 nL acetonitrile. After each run, the trapping column was rinsed three times with 426 nL of acetonitrile. CE System with an In-Line Injection Valve. The regular CE setup comprised the CE apparatus, fused-silica capillary, and the UV detector. The total length of the capillary was 120 cm, and detection was performed at 82 cm from the inlet. For in-line injection, a microvalve (valve 1) was incorporated into the CE setup as depicted in the framed part of Figure 1. Using the in-line microvalve, partial loop injections into the CE were performed by manually switching the valve to the inject position for a set amount of time, then returning to the load position while the electric field was on. For safety, the handle of this valve was extended with an 8-cm plastic tip while the valve body was wrapped in a layer of 1.5-cm insulating material. On-Line Preconcentration-CE System Using In-Line Injection. A micro trapping setup for sample preconcentration was connected to the CE system via the in-line injection valve (Figure 1). The preconcentration part consisted of an LC pump (LC10ADVP; Shimadzu, Kyoto, Japan) or a nano LC pump (1100 series; Agilent Technologies, Waldbronn, Germany), a microvalve (valve 2) with a 375-nL fused-silica loop (75 µm i.d.) for introduction of acetonitrile, a regular valve (valve 3) with a 250-µL PEEK loop (1235 × 0.254 mm i.d.; Upchurch Scientific, Oak Harbor, WA) for sample loading, and a PepMap C18 reversed-phase trapping column (5 × 0.5 mm i.d.; LC Packings, Amsterdam, The Netherlands). Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4433

Figure 2. (a) Electropherogram of a mixture (100 µg/mL each) of des-Tyr1-[D-Ala2-D-Leu5]-enkephalin (1) and [Met5]-enkephalin (2) using the regular CE setup. Injection: -15 kV for 12 s; analysis voltage -15 kV. (b) Electropherogram of a mixture (100 ng/mL each) of desTyr1-[D-Ala2-D-Leu5]-enkephalin (1) and [Met5]-enkephalin (2) using the complete preconcentration-CE setup. The negative signal at ∼13 min is due to the presence of acetonitrile in the injected sample plug and migrates with the electroosmotic flow. Sample volume, 250 µL; CE injection, -15 kV for 12 s; analysis voltage, -15 kV. For further conditions, see the Experimental Section.

Figure 1. Schematic diagram of the preconcentration-CE system with in-line injection valve. The framed part represents the in-line injection valve-CE system. Lengths of fused-silica capillaries (75 µm i.d.) are shown in italics (cm). Valves 1 and 2 are microvalves; valve 3 is a regular valve. Valve arrangements are shown during (a) loading, (b) desorption, and (c) sample injection into the CE. After sample injection, valve 1 is switched back to its original position, as in (b). See Experimental Section for further details.

RESULTS AND DISCUSSION CE System with an In-Line Injection Valve. A preliminary study was performed to investigate the separation of a mixture of two closely related peptides, des-Tyr1-[D-Ala2-D-Leu5]-enkephalin and [Met5]-enkephalin, using the regular CE setup. With a BGE of 50 mM borate (pH 8.0), the peptides had a net negative charge, 4434 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

and reproducible baseline separations were obtained (Figure 2a). For injections of 12 s at -15 kV, plate numbers for des-Tyr1-[DAla2-D-Leu5]-enkephalin and [Met5]-enkephalin are 270 000 and 190 000, respectively. To allow in-line injection, a microvalve was incorporated into the regular CE system (Figure 1, framed part). Sample injection was now performed by filling the loop with sample, turning on the high voltage (-15 kV), and switching the microvalve to the inject position. After 12 s of electrokinetic injection, the valve was switched back to the load position, and analysis was started. The resulting separation of the mixture shows reproducible baseline resolution with plate numbers of 150 000 and 105 000 for desTyr1-[D-Ala2-D-Leu5]-enkephalin and [Met5]-enkephalin, respectively. Comparison with the normal CE system (see above) indicates that under the given conditions, the microvalve causes some band broadening, but the obtained plate numbers are still acceptable. A similar effect was observed by Ponton and Evans,21 who also used an in-line injection valve for CE.

Figure 4. Relative peak area versus delay time. Data obtained from a sample of des-Tyr1-[D-Ala2-D-Leu5]-enkephalin (0.1 µg/mL) using the preconcentration-CE system. Sample volume, 250 µL; CE injection, -15 kV for 12 s; analysis voltage, -15 kV. See Experimental Section for further details.

Figure 3. Relative peak area (a) and plate number (b) of des-Tyr1[D-Ala2-D-Leu5]-enkephalin (100 µg/mL) versus injection time obtained with the in-line injection valve-CE system. Electrokinetic injections and analyses are performed at -15 kV. Shown data points are mean values of two measurements. See Experimental Section for further details.

To further study the performance of the in-line valve in the CE system, a sample of des-Tyr1-[D-Ala2-D-Leu5]-enkephalin (100 µg/mL in H2O) was filled into the loop and electrokinetically injected using increasing injection times. Between 12 and 90 s, the obtained peak area increased proportionally with injection time (Figure 3a), indicating that in-line injection functioned properly. At injections above 100 s, the resulting peak area reached a plateau, because the full loop volume was injected at these injection times. Extrapolation of both lines in Figure 3a (dotted lines) reveals that the minimum time needed to inject the entire loop is 102 s. This implies that in 12 s, a volume of ∼50 nL is injected. Using the same analyses, the injection time versus plate number is examined (Figure 3b). As can be expected, when larger sample volumes are introduced, plate numbers decrease from ∼150 000 (for a 12-s injection) to around 60 000 (for 120-s and longer injections). Clearly, the choice of the injection time will depend on the aim of the system performance: if high resolution is needed, a relatively low injection volume has to be selected; for high sensitivity, a relatively large volume can be introduced at the expense of separation efficiency. On-Line Preconcentration-CE System Using In-Line Injection. A reversed-phase C18 micro trapping column was selected for concentration of peptide samples prior to in-line valve injection into the CE. Using a setup in which the microcolumn is directly connected to the UV detector, preliminary studies were performed to investigate the volume loadability, analyte recovery, and elution volume after desorption. When loading a solution of Leu-enkephalin (100 µg/mL in H2O) at 50 µL/min, breakthrough was observed

after 15 min, indicating a maximum volume loadability of 750 µL. The relatively high trapping/washing speed of 50 µL/min was possible because the microcolumn was not in-line with the CE capillary. Desorption of the trapped peptide was performed by flushing a 426-nL plug of acetonitrile over the microcolumn at a flow rate of 3 µL/min. Under these conditions elution volumes (measured at 4σ) were found to be in the range of 0.9-1.1 µL. At higher flow rates, the elution volume increases. For example, using a flow rate of 10 µL/min, an elution volume of 2.5 µL was obtained. Determination of the desorption recovery with this setup is difficult, because acetonitrile and the peptide are simultaneously detected because no separation is carried out after analyte desorption from the trapping column. Therefore, the complete preconcentration-CE system (Figure 1) is used to determine the relative recovery (see below). The preconcentration system (sample loading valve, desorption valve and micro trapping column) was connected to the CE system via the in-line injection valve (Figure 1). The time between the start of the desorption and the actual injection into the CE, hereafter referred to as the delay time, is obviously an important parameter. During the delay time, the acetonitrile plug goes from valve 2 through the trapping column, where analytes will be desorbed, into the loop of the in-line injection valve 1 (Figure 1). The delay time is regarded as optimal when maximum analyte peak areas are obtained in CE analysis. To find the optimum delay time, 250 µL of a solution of des-Tyr1-[D-Ala2-D-Leu5]-enkephalin was trapped on the microcolumn, desorbed, and injected after various time intervals (33-45 s in 3-s steps). This specific delay time interval was chosen on the basis of the desorption flow rate (3 µL/min) and estimated volume between valve 2 and valve 1, that is, ∼2 µL. Figure 4 shows the plot of delay time versus peak area, indicating that a delay time of 39 s is the optimum. Moreover, Figure 4 reflects the distribution of analyte in a part of the eluted plug. Because of this distribution, plugs injected into the CE will have higher concentrations than the mean concentration of the eluted plug when using an optimum delay time, which obviously favors sensitivity (see below for a calculation based on experimental data). It should be noted that a constant flow during all desorptions, that is, a well-performing pump at 3 µL/min, is critical to obtain reproducible results using the various delay times. To determine the relative recovery of the first analyte desorption with 426 nL of acetonitrile, the trapping column was loaded Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

4435

with 250 µL of a solution of des-Tyr1-[D-Ala2-D-Leu5]-enkephalin, followed by five successive desorptive injections of 426 nL of acetonitrile. Repeatedly, 39 s after each desorption, the eluting plugs were captured in the loop of the in-line injection valve and partially injected into the CE (-15 kV for 12 s) and analyzed. The first 426-nL desorption appeared not to be complete, since some peptide was still detected during analysis of the second and third desorption. The relative recovery of the first desorption, as calculated by the ratio of the peptide peak area of the first desorption to the total peptide peak areas of all successive desorptions, was 70%. Selecting a larger desorption volume would improve the absolute recovery, but it would also induce larger elution volumes. This may lead to lower analyte concentrations and, thus, an overall decrease of analyte mass injected into the CE. The separation of two enkephalins using the complete system is shown in Figure 2b. Des-Tyr1-[D-Ala2-D-Leu5]-enkephalin and [Met5]-enkephalin are baseline-separated with 220 000 and 170 000 plates, respectively. The increase in plate numbers, as compared to those obtained with the CE system with the in-line injection valve, is most likely caused by focusing effects of the acetonitrile in the sample plug. As expected, peak areas do not correspond to the mean concentration of the eluted plug, but are higher. Taking into account the amount of trapped analyte (250 µL of a solution of 0.1 µg/mL peptide), the desorption efficiency (70%) and the elution volume of ∼1 µL, the amount injected (assuming a uniform plug concentration) would be 0.88 ng. In comparison, ∼5 ng (50 nL of a solution of 100 µg/mL peptide) was injected using the in-line injection valve-CE system. Therefore, when using the complete system, peak areas are expected to be ∼5 times lower than those using the CE system with the in-line valve. However, peak areas obtained with the complete system are only ∼2 times lower. This relatively high peak area obtained with the complete system is due to the injection of the optimum part of the elution plug, in which the analyte is nonhomogeneously distributed. Using enkephalin concentrations in the range of 10-1000 ng/ mL, linearity (area vs concentration) was established with good correlation (n ) 7). The equations are y ) 0.37x - 0.91 (R2 ) 0.9999) and y ) 0.45x - 2.04 (R2 ) 0.9994) for des-Tyr1-[D-Ala2D-Leu5]-enkephalin and [Met5]-enkephalin, respectively. With a 250-µL sample volume and a 12-s electrokinetic injection into the CE at -15 kV, CLODs (S/N ) 3) can be estimated to be around 5-10 ng/mL for the complete preconcentration-CE system. These CLODs can be further improved by preconcentration of larger sample volumes or increased injection times (see above). Determination of exogenous enkephalins with low microgramper-milliliter concentrations in biological samples seems relatively easy, but for endogenous enkephalins with concentrations of ∼1 ng/mL, some improvement is still needed. Reproducibility of all systems is examined (n ) 5), using enkephalins at concentrations of 100 µg/mL for the regular CE setup and the in-line injection valve-CE system, and enkephalins at concentrations of 0.1 µg/

4436

Analytical Chemistry, Vol. 76, No. 15, August 1, 2004

mL for the complete preconcentration-CE system. Migration times varied only slightly for the regular CE setup, having a relative standard deviation (RSD) of 0.2%, which increased to 0.9 and 1.2% for the in-line injection valve-CE system and the total preconcentration-CE system, respectively. Regarding peak areas, RSDs of 2.7, 3.4, and 11% have been observed for the regular CE system, the in-line injection valve-CE system, and the complete preconcentration-CE system, respectively. The rather high RSD for the complete system is probably caused by variation of the delay time and the recovery of the trapping column during desorption. Use of a pump with a higher precision and automated control of valve switching may improve the reproducibility. CONCLUSIONS An in-line injection valve appears to be an effective tool for the introduction of precolumn eluates into a CE system in an online mode. The system allows effective and fast preconcentration of samples without introducing sample matrix into the separation capillary. In fact, the in-line valve as used here provides a high independency of the preconcentration and the CE separation step. The feasibility of the complete system for the concentration and efficient separation of enkephalin peptides has been demonstrated. Under the applied conditions, CLODs were 5-10 ng/mL, but further improvement is still possible by loading larger sample volumes. In practice, the sample volume can be increased up to near the breakthrough volume of the trapping column (i.e. 750 µL for the studied enkephalins). In principle, a higher sensitivity while maintaining high separation efficiency may be reached by further minimizing the microcolumn elution volume so that a larger part of the elution volume can be injected into the CE system. This requires smaller precolumns which, on the other hand, imply smaller breakthrough volumes and, thus, lower loadabilities. Another critical point is that void volumes of valves and connections might be the limiting factor when reducing elution volumes. Future developments include automation of the system and the analysis of biological samples that will be directly injected into the preconcentration-CE system. As for any SPE-based preconcentration, elution volumes might differ for various sample matrixes and should always be verified. Delay times are also dependent on the elution volume and, thus, have to be optimized again. Since coupling with a mass spectrometer requires a grounded outlet, we will investigate whether our present system allows high voltages at the inlet (i.e. near to the in-line injection valve) side. Another interesting option is to use the developed setup for coupling LC with CE. The LC eluate could be trapped on a microcolumn, which is then desorbed toward the in-line injection valve and analyzed by CE.

Received for review January 13, 2004. Accepted May 20, 2004. AC0499221