Anal. Chem. 1997, 69, 1732-1737
Miniaturized Supported Liquid Membrane Device for Selective On-Line Enrichment of Basic Drugs in Plasma Combined with Capillary Zone Electrophoresis S. Pa´lmarsdo´ttir,† E. Thordarson,† L.-E. Edholm,†,‡ J. Å. Jo 1 nsson,† and L. Mathiasson*,†
Analytical Chemistry, University of Lund, P.O. Box 124, S-221 00 Lund, Sweden, and Bioanalytical Chemistry, Preclinical R & D, Astra Draco AB, P.O. Box 34, S-221 00 Lund, Sweden
A hollow fiber miniaturized supported liquid membrane (SLM) device for sample preparation is connected on-line with capillary electrophoresis and used for determination of a basic drug, bambuterol, in human plasma. The analyte is extracted from the outside of the hollow fiber (donor) through the liquid membrane (pores of the fiber impregnated with organic solvent) into the acceptor solution in the fiber lumen. The process is driven by differences in pH between the donor and acceptor solution. The whole volume of the acceptor solution can then be injected into the CZE capillary by using the double-stacking procedure for large volume-injection. Very clean extracts of low ionic strength are obtained from the SLM treatment, making this sample pretreatment method compatible with the CZE double-stacking procedure, which in turn makes it possibile to inject large volumes of sample onto the separation capillary. Good performance of the whole procedure is demonstrated, and detection limits in the low nanomolar range were obtained in spite of the relatively weak UV absorbance of bambuterol. Extractions through the miniaturized SLM unit can be performed for 5-6 h without regenerating the fiber. The regeneration procedure was tested, and no relevant changes in the performance of the extraction could be found after seven regenerations, allowing the same fiber to be used for a week. Miniaturization is a general trend in separation science, offering solutions that are more environmentally and economically attractive than conventional ones. Capillary zone electrophoresis (CZE) is a microtechnique that can be used for determination of substances with widely varying properties and is especially suitable when only small sample volumes are available and high separation power is needed. Major demands in quantitative bioanalysis using CZE are clean samples to ensure good performance and relatively high concentrations in the injected sample plug. One approach to increasing the detectability in miniaturized systems is to increase the analyte concentration before it reaches the detection cell. This can be done by enrichment during sample pretreatment and/or by performing sample focusing during the final analysis step in order to increase the injected sample volume without losses in separation † ‡
University of Lund. Astra Draco AB.
1732 Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
performance. One way to obtain such large-volume injections is to make use of differences in conductivity between the sample plug and background electrolyte buffer, i.e., stacking.1,2 Here the sample is dissolved in a low-conductivity solution, e.g., water or methanol, and injected into a capillary filled with a buffer of higher conductivity. As the electric field depends inversely on the specific conductivity, the ions in the sample plug are accelerated toward the boundary between the solutions. When the analytes enter the background electrolyte, they experience a lower field strength and stack into a narrow zone. The supported liquid membrane technique3 has proved useful for enrichment of ionizable and charged species, giving a high degree of cleanup and enrichment of various analytes in biological samples.4-6 The analyte is extracted from an aqueous (donor) phase into a hydrophobic organic liquid immobilized in a porous membrane, followed by back-extraction into a second aqueous (acceptor) phase. In previous work, we have shown that the supported liquid membrane (SLM) technique for pretreatment of human plasma samples is highly compatible with the CZE double-stacking procedure.7 The double-stacking procedure,8 based on field enhancement, makes it possible to fill the whole capillary with sample without significant loss of separation performance and can improve the detection limit several hundred times. In previous work, the supported liquid membrane devices used were based on flat membranes, which allowed construction of devices with acceptor volumes down to 10-15 µL. However, even when using the double-stacking procedures in CZE for largevolume injection, these volumes are 5-10 times too large. An acceptor volume of 1-2 µL is desirable in order to be able to analyze the whole sample and obtain maximum sensitivity. We have recently developed an SLM unit based on a porous fiber with an acceptor volume of this size.9 In this work, we have examined the possibility of connecting such a miniaturized SLM device on-line with CZE. The combina(1) Burgi, D. S.; Chien, R. L. Anal Chem. 1991, 63, 2042-2047. (2) Vinther, A.; Soeberg, H. J. Chromatogr. 1991, 559, 3-26. (3) Jo ¨nsson, J. Å.; Mathiasson, L. Trends Anal. Chem. 1992, 11, 106-114. (4) Audunsson, G. Anal. Chem. 1988, 60, 1340-1347. (5) Lindegård, B.; Bjo ¨rk, H.; Jo ¨nsson, J. Å.; Mathiasson, L.; Olsson, A.-M. Anal. Chem. 1994, 66, 4490-4497. (6) Pa´lmarsdo´ttir, S.; Lindegård, B.; Deininger, P.; Edholm L.-E.; Mathiasson, L.; Jo ¨nsson, J. Å. J. Capillary Electrophor. 1995, 4, 185-189. (7) Pa´lmarsdo´ttir, S.; Mathiasson, L.; Jo¨nsson, J. Å.; Edholm, L.-E. J. Chromatogr. B 1997, 688, 127-134. (8) Pa´lmarsdo´ttir, S.; Edholm, L.-E. J. Chromatogr. A 1995, 693, 131-143. (9) Thordarson, E.; Pa´lmarsdo´ttir, S.; Mathiasson, L.; Jo¨nsson, J. Å. Anal. Chem. 1996, 68, 2559-2563. S0003-2700(96)00668-3 CCC: $14.00
© 1997 American Chemical Society
Figure 1. Structures of the model compound and inhibitor used in this work.
tion of efficient cleanup with SLM and double-stacking procedure for large-volume injection was expected to create good conditions for trace analysis of complicated samples. This approach was evaluated with plasma samples containing the secondary amine bambuterol as a model substance. EXPERIMENTAL SECTION Chemicals and Samples. 6-Undecanone was obtained from Janssen Chimica (Geels, Belgium) and the polypropylene hollow fiber Plasmaphan, with an inner diameter of 330 ( 50 µm, a wall thickness of 170 ( 50 µm, and an average pore size of 0.2 µm, from Akzo Nobel (Wuppertal, Germany). Racemic bambuterol (hydrochloride) was obtained from Astra Draco AB (Lund, Sweden) and the esterase inhibitor physostigmine from Sigma Chemical Co. (St. Louis, MO). Their structures are shown in Figure 1. All other chemicals were of analytical reagent grade from Merck (Darmstadt, Germany). All water used was purified with a Milli-Q system (Millipore, Bedford, MA). Phosphate buffers were prepared from phosphoric acid and its sodium salts, mixing together to the desired pH value. For the phosphate buffers of low concentration (5 mM), daily preparation was necessary to avoid changes in pH values. Standard solutions of bambuterol were prepared in water and were stable for at least 1 month when kept at room temperature and protected from light. A stock solution (0.1 mM) of the esterase inhibitor physostigmine was prepared in 0.9% NaCl solution. The solution was stable for at least 2 weeks when kept in a refrigerator and protected from light. Blank plasma samples were obtained from blood that was collected into Na heparinized Venoject tubes (10 mL). The tubes were turned upside-down seven times and then immediately centrifuged for 10 min at 1400g. All samples were stored frozen (-18 °C). Before the analysis of the plasma samples, these were thawed and centrifuged for 10 min at 2000g. A solution of phosphate buffer, physostigmine, and plasma at pH 10.7 was used as a plasma blank, while phosphate buffer at pH 10.7 is referred to as aqueous blank. The concentration of phosphate buffer as well as physostigmine is the same in all extractions. Instrumentation. The hollow fiber SLM-CZE setup is shown in Figure 2. The inlet to the donor channel of the hollow fiber membrane unit (1) was connected to the valve of a sample processor instrument, Model 233 XL, with a Model 401 syringe pump (2) (Gilson Medical Electronics, Villiers-le-Bel, France) used to process the plasma samples. The inlet of the acceptor channel
Figure 2. Experimental setup: (1) hollow fiber membrane unit, (2) sample processor, (3) pump, (4) flow processor, and (5) miniaturized switching device. For details, see the text.
Figure 3. Membrane extraction device: (1) compartment for inner O-ring, (2) polypropylene hollow fiber, (3) fused silica capillaries, (4) 1 mm thru-hole male nuts.
was connected to a pump system consisting of a pump (3) (Model 2150, Pharmacia LKB, Bromma, Sweden) connected to an accurate microflow processor (4) (LC Packings, Zurich, Switzerland). A µ-Dumper interface (5) (LC Packings), a miniaturized switching device, was connected to the flow processor and used for switching the flow between the hollow fiber membrane unit and waste. The electrophoresis instrument consisted of a Prince programmable injector for capillary electrophoresis supplied with an outlet buffer replenisher (Butler), a high-voltage power supply (Prince Technologies BV, Emmen, Netherlands), a Spectra 100 UV-visible detector (Spectra-Physics, Mountain View, CA) for on-capillary UV detection at 205 nm, and an untreated fused silica capillary, 75 µm i.d. and 375 µm o.d. (Polymicro Technologies, Phoenix, AZ) with a length of 50 cm from the inlet to the detector (2.2 µL volume) and 61 cm from the inlet to the outlet. The temperature in the autosampler and the capillary housing was maintained at 22 °C. Data analysis and collection were accomplished using System Gold software (Beckman, Palo Alto, CA), version 712. Figure 3 is a cross-sectional depiction of the membrane device. It was manufactured from Kel-F in one piece with a cylindrical channel at the center, 15 mm long and 0.9 mm in diameter. Two identical holes, 2.7 mm in diameter, were drilled at both ends of the channel (not shown in the picture) where two O-rings (2.8 mm o.d.) could be fixed in position (1). The polypropylene fiber (2) was inserted into the channel with a 350 µm o.d. fused silica capillary and an extra O-ring in each end (3). The position of the connecting capillaries was adjusted, leaving a space of about 1 cm between their ends before inserting the O-rings, thus stabilizing the arrangement. This gave an approximate volume of 1.2 Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
1733
µL in the fiber lumen. Now, an additional O-ring was inserted in each end, making contact with the inner O-ring. Two screws were used to tighten the fiber against the capillary tubes used to connect the hollow fiber to the flow stream (4). These two screws also provided a seal toward the surroundings when the O-rings were depressed. The large frictional force established between the O-rings when turning the screws prevented the O-rings from rotating, leaving the arrangement stable and reducing the risk of the fiber being crumpled. The fiber was impregnated in situ by introducing 6-undecanone into the donor and acceptor channels respectively, keeping it stagnant for 0.5 h. Before commencement of the first enrichment, the fiber was washed for some minutes with water to remove excess solvent. Analysis Procedure. Sample Preparation. The plasma samples are automatically prepared. To 350 µL of blank plasma is added 3.5 µL of a 0.1 mM solution of esterase inhibitor physostigmine. This prevents hydrolysis of bambuterol, which is next added as 70 µL of standard solution. Finally, 276.5 µL of 0.3 M phosphate buffer of pH 11.1 is added, to give a total sample volume of 700 µL. The pH in the resulting sample solution is ∼10.7, which is chosen so that bambuterol is predominantly uncharged (as pKa is 9.8)10 and sufficiently stable. Bambuterol is degraded in alkaline solutions, but at pH 10.7 the degradation is negligible. After mixing, 500 µL is extracted, corresponding to 250 µL of the original plasma. SLM Enrichment. The prepared sample solution is pumped with pump 2 in Figure 2 at a flow rate of 25 µL/min through the donor channel of the hollow fiber membrane unit. The uncharged bambuterol molecules diffuse through the membrane liquid into the stagnant acidic acceptor phase, where they are protonated and trapped in a solution of phosphoric acid (0.5 mM) containing 20% methanol at a pH of 3.4. After each enrichment, the needle and the sampling loop of the sample processor are washed with 10 mL of water. The donor channel is washed with 20 channel volumes of water and the acceptor channel with 20 channel volumes of acceptor phase. The washing procedure is conducted simultaneously with the final CZE analysis. After 6 h, or before exhaustion of the fiber, it is regenerated in three steps. First, the fiber is washed with 5 mL of water and then with 5 mL of ethanol on both donor and acceptor to get rid of remaining plasma solution, buffer solutions, and 6-undecanone left in the fiber pores. Second, the fiber is dried on both sides for 10 min in an air flow (2 bar pressure). Finally, the fiber is impregnated again with the organic solvent as described above. CZE Analysis. Before each run, the capillary is washed with four column volumes each of 0.1 M NaOH, water, methanol, and electrolyte solution. The capillary is stored filled with methanol between working days but dried with air if the system was not used for longer periods. The Prince system is equipped with a facility allowing the application of back-pressure. This is used in the different steps of the double-stacking procedure to either balance the electroosmotic flow or pump the stacked band from the outlet to the inlet part of the capillary. The outlet of the SLM acceptor channel was manually connected to the outlet of the CE capillary through a piece of Teflon tubing and disconnected after transfer of the extract, i.e., prior to (10) Rosberg, B.; Schro¨der, C.; Nyberg, L.; Rosenborg, J.; Wire´n, J. E. Eur. J. Clin. Pharmacol. 1993, 45, 147-150.
1734 Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
Figure 4. Double-stacking procedure in CZE. From ref 8 with permission from Elsevier, Amsterdam.
analysis. After enrichment, the extract was transferred to the CZE capillary simply by pumping it (flow 2.2 µL/min). A total of 4 µL is pumped, i.e., the nominal volume of the fiber lumen (1.2 µL) plus an additional aliquot to account for the volume of the transfer tubing and to ensure complete transfer of the extract to the capillary. Before starting the first step in the double-stacking procedure, the part of the capillary between the outlet and the detector (back-pressure 100 mbar for 0.7 min) is filled with a 5 mM phosphate buffer (pH 7.5). The following passage will continuously relate to Figure 4,8 step by step, the current situation being as in picture 1 in the figure. Now, the analyte is concentrated on the CZE capillary using double-stacking, which is fully automated. The first step in the stacking procedure is carried out by applying a voltage of 30 kV for 4 min, at the same time as a back-pressure of 100 mbar is applied. The back-pressure is used to oppose the electroosmotic flow and thus to prevent the stacking analyte from moving too fast toward the outlet of the capillary and allowing some of the first bulk electrolyte solution to remain in the capillary end. At the end of this step, all cationic sample components have been collected in a short band near the detection window (2). The stacked peak is now forced back to the injection end of the capillary by introducing 0.1 M phosphate buffer, pH 2.5, into the outlet end of the capillary. This is done by applying a backpressure of 180 mbar for 0.8 min to the system after turning off the high voltage. This results in more than half of the original sample volume being pushed out of the capillary (3). The second stacking step is now carried out, with the 0.1 M phosphate buffer, pH 2.5, present at both outlet and inlet, by applying 30 kV over the capillary. The positive species will now stack up at the concentration boundary between the 0.1 M buffer and the 5 mM buffer (mixed with methanol solution), and the band broadening
Figure 5. Electropherograms showing blank aqueous sample (a) and blank plasma sample (b) after enrichment using the miniaturized SLM device. Membrane extraction conditions were as follows: donor flow rate, 25 µL/min; donor pH, 10.7; acceptor pH, 3.4.
obtained after transportation of the previously stacked sample will be compensated for (4). At the same time as the stacking is in progress, a back-pressure of 105 mbar is used to push the methanol-containing band out of the capillary. After 1.5 min, the separation voltage is lowered to 18 kV to prevent too much Joule heating in the system. When the current indicates that most of the highly resistant solution has been pushed out of the capillary (60 µA), the back-pressure is switched off automatically (5). The final separation of the stacked compounds is then carried out, which brings us to the bottom picture in Figure 4. RESULTS AND DISCUSSION Performance of the Miniaturized SLM-CZE System. Good performance of the double-stacking procedure demands samples of low ionic strength. Such samples can be obtained after hollow fiber SLM treatment of plasma samples by choosing an aqueous acceptor solution containing methanol. Higher methanol concentration in the sample extract will result in higher separation efficiency when applying the double-stacking procedure.8 On the other hand, the stability of the hollow fiber liquid membrane will decrease when an organic solvent such as methanol is present in the acceptor phase. A good compromise was reached with 20% in the acceptor phase. This composition only slightly decreased the stability of the hollow fiber liquid membrane and gave acceptable separation efficiency in the double-stacking procedure. Electropherograms resulting from samples being injected into the miniaturized SLM-CZE system are shown in Figures 5 and 6. In Figure 5, almost no difference is visible between a blank aqueous sample (a) and a blank plasma sample (b), signifying a high degree of cleanup from the plasma matrix. The electropherograms in Figure 6 show an aqueous sample (a) and a plasma sample (b), both containing 5 µM physostigmine and 4 nM bambuterol. The extraction is obviously highly selective for bambuterol, clearly seen as peak B. The concentration level of 4
Figure 6. Electropherograms showing aqueous sample (a) and plasma sample (b) containing 1 µM physostigmine and 4 nM bambuterol. Peak A is the physostigmine inhibitor, and peak B is bambuterol. Conditions as in Figure 5.
nM is close to the detection limit both in plasma and in water solutions. Peak A is the esterase inhibitor physostigmine, which is unstable and can give rise to extra peaks in the electropherogram. The origin of the additional peaks in the electropherograms is somewhat obscure (especially for the aqueous blank), but they seem to hail from the hollow fiber. Sample matrix removal was the principal achievement of the hollow fiber SLM treatment. The concentration enrichment per minute through the fiber was about 0.7, which resulted in 14 times higher concentration inside the fiber after carrying out the extraction for 20 min. When taking into account that a volume of 2.2 µL (which filled the capillary) was used to transfer an acceptor volume of 1.2 µL to the capillary to ensure complete transfer of the analyte, the gain in concentration was about 7 times compared to hypothetically filling the CZE capillary with the plasma sample. The enrichment factor can, if necessary, be further increased by decreasing the flow rate, which gives higher enrichment over the membrane but also markedly increases the analysis time. Another possibility is to decrease the diameter of the donor channel. Decreasing the inner diameter of the hole at the center of the Kel-F piece from 0.9 to 0.7 mm, and thereby decreasing the donor channel thickness, almost doubled the concentration enrichment, as expected. However, the regeneration of the fiber turned out to be a problem, probably due to the high back-pressure in the narrow donor channel. In CZE, an injected sample plug occupying about 0.5-1% of the capillary volume is often the limit, whereas larger injections lead to severe band broadening. Here, we load a sample plug 45-90 times that volume onto the capillary; because of the stacking effect, this can be done without a large decrease in efficiency. Linearity and Reproducibility of the Miniaturized SLMCZE System. Calibration curves in plasma based on peak area Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
1735
and double injections at six different concentrations were made in the concentration range 4-128 nM bambuterol. Good linearity was obtained, and the intercept did not deviate significantly from zero. The correlation coefficient (r) was 0.999, and the confidence intervals at the 95% level were (arbitrary units) -5 ( 47 for the intercept and 15 ( 1 for the slope. The repeatability relative standard deviation based on peak areas for five samples was 5% at a concentration of 32 nM and 7% at a concentration of 128 nM. Memory effects in SLM may sometimes be a problem. Also, when using hollow fibers such memory effects appeared and could not be totally eliminated with the washing procedure applied between the extractions. Further work needs to be done aiming at the elimination of this problem. Stability of the Miniaturized SLM-CZE System. Extractions through the miniaturized SLM unit could be performed for 5-6 h without regenerating the fiber. After exhaustion, the fiber could be regenerated as described in the Experimental Section. The same fiber was used throughout the experiments. Regeneration could be conducted seven times without noticeable changes in the extraction efficiency. On further regeneration, it was found that the back-pressure in the donor channel increased, and so did the extraction efficiency. The problem is probably caused by plasma proteins sticking to the walls of the donor channel, decreasing the channel thickness. More thorough washing, for example with detergent, in the regeneration process could possibly solve this problem. Replacing the fiber with a new one resulted in slightly changed extraction efficiency. One reason for this could be a slight difference in positioning of the connecting capillaries, resulting in different acceptor volume of the hollow fiber. Another reason might be the variation in the wall thickness of the fiber, which could be (50 µm according to the manufacturer. The high degree of cleanup from the plasma matrix by the hollow fiber SLM extraction resulted in samples that did not cause any adsorption problems in the CZE capillary. The same capillary could be used for weeks without problems by using the washing steps described above between runs. The main complication in the CZE step was a variation in the electroosmotic flow velocity, especially if the capillary had been unused for a period of time. Then it took some effort to equilibrate the capillary surface. This was usually done by running the system a couple of times with the washing procedure in-between. The most stable conditions were definitely obtained when running the system constantly over a long period of time. Analysis Time Considerations. Using the conditions described above with a flow rate of 25 µL/min through the donor side of the SLM device, a 500 µL sample (250 µL of plasma) was processed within 20 min. Hence, the throughput is ∼25 samples/ day. This is about the same time it took to carry out the CZE double-stacking analysis. By using a hollow fiber with an even smaller inner diameter, the acceptor volume can be decreased further; consequently, a shorter or a smaller inner diameter CZE capillary could be used, resulting in shorter CZE analysis time. Then the SLM extraction will be the time-limiting step, but this can be solved by having two or more units in parallel, coupled to the CZE capillary, increasing the sample throughput. Such a system has been developed for trace analysis of metals with four (11) Malcus, F.; Djane, N.-K.; Mathiasson, L.; Johansson, G. Anal. Chim. Acta 1996, 327, 295-300.
1736
Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
SLM units connected to a graphite furnace.11 The parallelization also makes it possible to increase the enrichment time, resulting in higher concentration enrichment without any loss in sample throughput. When the volume of the plasma sample available is very limited, it becomes crucial to extract as much analyte as possible out of the sample. A very slow extraction flow rate is then desirable to maximize the fraction of analyte extracted.9 The detection limit (∼4 nM) obtained in this work equals what has been obtained in previous work,7 where we used supported liquid membrane devices based on flat membranes in an off-line combination with CZE. However, larger sample volumes were enriched in the off-line combination, and a longer CZE capillary was used, resulting in more than 2 times longer analysis times. Minimal risk of sample loss and contamination is another advantage of the on-line combination. The hollow fiber SLM device turned out to be even more robust, when considering the risk of leakage between the donor and acceptor solution, than the flat membrane design. Concluding Discussion. During the development of a suitable cleanup procedure for CZE using SLM methodology, three different technical solutions have been presented up to now. The simplest one is the off-line approach using flat membranes in the SLM procedure, resulting in only a part of the sample being used for final analysis by CZE. This leads to a higher detection limit with the same total analysis time, necessitates more manual input, and increases the risk of sample losses and contamination in comparison to on-line procedures. However, since a number of samples can easily be run in parallel during SLM enrichment by using, for example, multichannel peristaltic pumps, the time for the SLM procedure can generally be kept low in this approach, and the total analysis time will then be determined by the CZE procedure. Even this procedure can be shortened, since long enrichment times in SLM give higher concentration in the acceptor and decrease the need for enrichment through stacking procedures, thus making it possible to reduce the length of the capillary column in CZE. The procedure envisioned in this paper with on-line connection of SLM to CZE seems to be relatively simple to automate, with the obvious advantages of less labor demand and lower risk for contamination and losses. The memory effects, however, may be somewhat more difficult to handle. The detection limits will be comparable to those of a parallelized off-line system. However, the on-line system can also be parallelized, although other pumps capable of delivering low flow rates with good precision will be needed, which will be considerably more expensive. With such an approach, the on-line system will still give a shorter total analysis time. In the third approach presented recently,12 SLM was connected via a CLC microcolumn system on-line to the CZE system. This system, being the most complicated one, gives very high enrichment factors and high selectivity toward matrix components. The system utilizes two selective, sequential enrichment steps before the third analyte focusing and separation step with double-stacking CZE. With the combination of a concentration enrichment on the order of 40 000 and high selectivity, determinations in the subnanomolar-range are achievable. This approach is valuable if samples need to be analyzed at very low concentrations. Otherwise, the on-column system with a miniaturized SLM unit (12) Pa´lmarsdo´ttir, S.; Mathiasson L.; Jo ¨nsson, J. Å.; Edholm, L.-E. Submitted.
connected to the CZE equipment seems to be the best approach for future routine analysis. CONCLUSION The results of this work demonstrate the potential of trace analysis of plasma samples by CZE after efficient cleanup with the SLM technique. The small volume of sample and a tailormade matrix obtained from the SLM cleanup makes the doublestacking procedure for sample injection in CZE possible and result in nanomolar detection limits, with a total analysis time of only ∼20 min. These figures are obtainable using UV detection even for molecules with relatively low absorption coefficients. Further
work will focus on development of procedures which minimize memory effects, systems which improve the sample throughput, and systems which minimize manual operations. Fully automated systems can be envisaged using a suitable computer-controlled microvalve, allowing for transfer into the CZE capillary with minimal band broadening. Received for review July 9, 1996. Accepted February 10, 1997. AC960668P X
Abstract published in Advance ACS Abstracts, March 15, 1997.
Analytical Chemistry, Vol. 69, No. 9, May 1, 1997
1737
