Anal. Chem. 2001, 73, 2656-2662
Narrow Sample Channel Injectors for Capillary Electrophoresis on Microchips Chao-Xuan Zhang and Andreas Manz*
Astra Zeneca/SmithKline Beecham Centre for Analytical Sciences, Department of Chemistry, Imperial College of Science, Technology & Medicine, London SW7 2AY, U.K.
In microchip CE, sample injection is generally achieved through cross, double-tee, or tee injector structures. In these reported approaches, channel width and depth are uniform at the injection intersection. Here, we present cross and tee injectors having narrow sample channels. Using a cross injector with reduced sample channel width, resolution, column efficiency, and sensitivity are remarkably improved. Furthermore, no leakage control is required in both injection and separation phases, making the microchip CE system more user-friendly. Good resolution can also be obtained using tee injectors with narrow sample channels, which would otherwise be impossible using conventional tee injectors. Using the narrow sample channel tee injector instead of conventional cross and double-tee injectors, the number of reservoirs in multiplexed systems can be reduced to N + 2 (N, the number of paralleled CE systems), the real theoretical limit. The virtues of the novel injectors were demonstrated with poly(dimethylsiloxane)-glass chips incorporating eight parallel CE channels. Miniaturization of conventional chemical systems has been recognized as a highly rewarding method for increasing sample throughput, improving performance, and reducing the cost, as originally envisioned.1 Among the various conventional separation techniques including chromatographic and electrophoretic, capillary electrophoresis (CE) has proved to be the most successful one in miniaturization as it possesses obvious advantages over high-performance liquid chromatography in terms of separation efficiency, speed, and cost.2 CE on microchips has been demonstrated for analysis of various substances, ranging from small drug molecules,3 amino acids,4 peptides,5 and oligonucleotides6 to large proteins7 and DNA fragments.8 (1) Manz, A.; Harrison, D. J.; Verpoorte, E.; Widmer, H. M. In Advances in Chromatography, Volume 33; Brown, P. R., Crushka, E., Eds.; Marcel Dekker: New York, 1993; p 1. (2) Kutter, J. P. Trends Anal. Chem. 2000, 19, 352. (3) Ramseier, A.; von Heeren, F.; Thormann, W. Electrophoresis 1998, 19, 2967. (4) von Heeren, F.; Verpoorte, E.; Manz, A.; Thormann, W. Anal. Chem. 1996, 68, 2044. (5) Zhang, B.; Foret, F.; Karger, B. L. Anal. Chem. 2000, 72, 1015. (6) Effenhauser, C. S.; Manz, A.; Widmer, H. M. Anal. Chem. 1995, 67, 2284. (7) Liu, Y.; Foote, R. S.; Jacobson, S. C.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 4608. (8) Shi, Y.; Simpson, P. C.; Scherer, J. R.; Wexler, D.; Skibola, C.; Smith, M. T.; Mathies, R. A. Anal. Chem. 1999, 71, 5354.
2656 Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
Most of the strategies developed in conventional CE, in which a fused-silica capillary is normally employed, are readily adaptable to the microchip format in which the micromachined channel serves as the separation column. As the channel length in a microchip is short (a few centimeters) compared with a fusedsilica capillary, fast separation using microchip CE can be achieved in a few seconds; large numbers of parallel CE systems can be incorporated on a single piece of chip. The most radical difference between conventional CE and the chip format is the injection method. In CE on microchips, sample injection is achieved through the channel network, i.e., a sample guiding channel that intersects the separation channel. The first CE chips employed a tee-injector design.9,10 Due to the difficulty in control of the sample plug, the tee injector has rarely been pursued11 and has promptly been replaced with cross12 or double-tee injectors.13 In microchip CE using cross or double-tee injectors, two steps are typically involved: injection and separation (Figure 1). In the injection phase, sample is drawn from the sample reservoir across the injection intersection toward the sample waste reservoir with an electric field along the sample channel. The analyte stream is perpendicular to the separation channel. For separation, an electric field is applied along the separation channel to push a sample plug, which resides in the intersection, away from the intersection downstream toward the detection point. This simplest injection scheme was employed in the first microchip CE systems.12,13 It had soon been realized that voltage control of all device ports in both injection and separation phases was necessary for a welldefined sample plug to be injected into the separation channel.14-16 A short well-defined sample plug is crucial for high separation efficiency. To form a well-defined sample plug at the channel intersection during injection, pinching voltages are applied at the buffer inlet and outlet inducing buffer flows toward the sample waste reservoir to counteract the diffusion of analytes into the (9) Manz, A.; Harrison, D. J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Lu ¨ di, H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253. (10) Harrison, D. J.; Manz, A.; Fan, Z.; Lu ¨ di, H.; Widmer, H. M. Anal. Chem. 1992, 64, 1926. (11) Liang, Z.; Chiem, N.; Ocvirk, G.; Tang, T.; Fluri, K.; Harrison, D. J. Anal. Chem. 1996, 68, 1040. (12) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895. (13) Effenhauser, C. S.; Manz, A.; Widmer, H. M. Anal. Chem. 1993, 65, 2637. (14) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107. (15) Effenhauser, C. S.; Paulus, A.; Manz, A.; Widmer, H. M. Anal. Chem. 1994, 66, 2949. (16) von Heeren, F.; Verpoorte, E.; Manz, A.; Thormann, W. J. Microcolumn Sep. 1996, 8, 373. 10.1021/ac010138f CCC: $20.00
© 2001 American Chemical Society Published on Web 05/03/2001
Figure 1. Schematic diagrams of CE procedures with pinched injection (top panels), floating injection (middle), and (C) simplest injection mode (bottom).
separation channel. In the subsequent separation phase, back voltages are applied at the sample and waste reservoirs to draw the analyte flows back to the sample and waste reservoirs, preventing leakage into the separation channel. This pinched injection method has now become a common practice in CE on microchips.3,8,17-22 In the pinched injection schemes, long injection times, typically 10-150 s, are required for the analytes to migrate from the sample reservoir to the injection intersection. The injection time is comparable to or even longer than the real analytical time. It restricts the analytical speed. Fast injection can be achieved using a gated injection method.7,23-25 Unlike the pinched injection, the analyte stream makes a 90° turn at the injection cross toward the sample waste. During the separation, analytes are continuously moving through the cross while the buffer is electroosmotically pumped toward the sample waste and buffer outlet reservoirs, preventing sample (17) Hutt, L. D.; Glavin, D. P.; Bada, J. L.; Mathies, R. A. Anal. Chem. 1999, 71, 4000. (18) Liu, S.; Shi, Y.; Ja, W. W.; Mathies, R. A. Anal. Chem. 1999, 71, 566. (19) Cheng, J.; Waters, L. C.; Fortina, P.; Hvichia, G.; Jacobson, S. C.; Ramsey, J. M.; Kricka, L. J.; Wilding, P. Anal. Biochem. 1998, 257, 101. (20) Ocvirk, G.; Munroe, M.; Tang, T.; Oleschuk, R.; Westra, K.; Harrison, D. J. Electrophoresis 2000, 21, 107. (21) Wallenborg, S. R.; Bailey, C. G. Anal. Chem. 2000, 72, 1872 (22) Hashimoto, M.; Tsukagoshi, K.; Nakajima, R.; Kondo, K.; Arai, A. J. Chromatogr., A 2000, 867, 271. (23) Jacobson, S. C.; Koutny, L. B.; Hergenro¨der, R.; Moore, A. M., Jr.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472. (24) Kutter, J. P.; Jacobson, S. C.; Matsubara, N.; Ramsey, J. M. Anal. Chem. 1998, 70, 3291. (25) Hadd, A. G.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1999, 71, 5206.
leakage into the separation channel. For injection, the analyte flow is deflected into the separation channel for a short time, typically 0.1-0.5 s. In consecutive runs, the next sample is transported to the cross while the current separation is underway, ready to be dispensed into the separation channel for the next run. Virtually no time is required for injection phase. However, all those injection schemes involve cumbersome voltage programs for leakage control. The complicated relationships between microchip geometry, operating conditions, and fluidic behaviors add to the complexity of hardware and software in chip operation. It has been reported that gated injection can be achieved using only one voltage output with a chip incorporating a waste channel parallel to the separation channel, but no separation data were shown.26 In recent reports of DNA analysis, floating injection (no pinching voltages but back voltages applied during the separation phase) is shown working well due to the low diffusion coefficients of DNA in polymer sieving media.27,28 The simplest injection (no leakage control at all) has also been used in a few recent approaches of capillary zone electrophoresis.29-31 These reports challenge the necessity of pinched injection. However, in those reports, no separation data were obtained using pinched injection for comparison. Therefore, those floating and simplest injection methods cannot be justified. In the reported approaches using cross, double-tee, and tee injector structures, channel width and depth are uniform at the injection intersection. In this paper, we present double-tee and tee injectors having narrow sample channels. Compared to conventional cross injectors with uniform channel width, the novel injector cross design using reduced sample channel width provides improved resolution and sensitivity. Furthermore, no leakage control is required at all, making the microchip CE system more user-friendly. Good resolution can also be obtained with tee injectors of narrow sample channels, which would otherwise be impossible using conventional tee injectors. Using the narrow sample channel tee injectors instead of conventional cross and double-tee injectors, the number of reservoirs in multiplexed systems can be reduced to N + 2 (N, the number of paralleled CE systems), the real theoretical limit. The virtues of the novel injectors were demonstrated with cross-linked poly(dimethylsiloxane) (PDMS)-glass chips incorporating eight parallel CE channels. EXPERIMENTAL SECTION Chip Design and Fabrication. The layout of the multiplexed CE chips used in this work is schematically shown in Figure 2. It consists of eight parallel CE systems distanced by 4.5 mm, a common buffer inlet reservoir, and a common buffer outlet reservoir. With this design of multiplexed systems, the number of buffer reservoirs is favorably reduced to two, the theoretical minimum. The distances from the injection intersection to the buffer inlet, buffer outlet, sample, and waste reservoirs were (26) Jacobson, S. C.; Ermakov, S. V.; Ramsey, J. M. Anal. Chem. 1999, 71, 3273. (27) Khandurina, J.; McKnight, T. E.; Jacobson, S. C.; Waters, L. C.; Foote, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 2995. (28) Salas-Solano, O.; Schmalzing, D.; Koutny, L.; Buonocore, S.; Adourian, A.; Matsudaira, P.; Ehrlich, D. Anal. Chem. 2000, 72, 3129. (29) Chiem, N.; Harrison, D. J. Anal. Chem. 1997, 69, 373. (30) Chiem, N.; Lockyear-Shultz, L.; Andersson, P.; Skinner, C.; Harrison, D. J. Sens. Actuators, B 2000, 63, 147. (31) Wang, J.; Chatrathi, M. P.; Tian, B. Anal. Chim. Acta 2000, 416, 9.
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
2657
Figure 2. Schematic layouts of a chip composed of eight CE systems with different injection designs. On the left are the expanded views of three typical injectors: conventional cross, NSC, cross and NSC tee injectors.
3, 30, 9, and 9 mm, respectively. The width of all the separation channels was 50 µm while the sample channel width varied. Three of the various injector structures are shown in the expanded inset of Figure 2. The sample channel widths of the conventional cross, narrow sample channel (NSC) cross, and NSC tee injectors were 50, 10, and 5 µm, respectively. The channel depth was uniform for the whole chip, which was either 5 or 10 µm (see the figure legends below). The in-house-made chip was composed of a plain glass substrate of 1.5 mm in thickness and a PDMS layer with the micromachined channels. The fabrication of the PDMS layer of 0.3-0.4 mm in thickness involved three steps: (a) fabrication of a mask, (b) a master, and (c) a device molding. (a) The layout drawn with the AutoCAD package was transferred onto a glass wafer coated with a positive photoresist and chromium (Nanofilm, Westlake Village, CA) using the photolithographic facility in our laboratory. This was followed by chromium etching. (b) A plain glass wafer was spin-coated with a negative photoresist XP SU-8 10 (MicroChem Corp., Newton, MA) at 2000-3000 rpm. The spinning speed determined the thickness of the SU-8 coating and thus the channel depth. The transparent patterns on the chromium mask was transferred onto the SU-8 master by putting them together and exposing to a 10-cm-diameter collimated UV light beam from a 200-W mercury lamp. The unexposed SU-8 was flushed off with a SU-8 developer, leaving the SU-8 structures standing on the surface. (c) PDMS base and curing agent (Sylgard 184, Dow Corning, Wiesbaden, Germany) were mixed in a 10:1 ratio and poured onto the SU-8 master. The device was cured overnight at 40 °C with a hot plate. Two big slots of 35 × 7 mm were cut out at the ends of separation channels on the PDMS device, serving as the buffer reservoirs. Holes, 1 mm in diameter, were punched at the ends of sample channels serving for sample and waste reservoirs. The PDMS layer and a plain glass substrate were brought into contact, forming the finished chip. 2658
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
Apparatus. An inverted confocal microscope equipped with a 50-W mercury lamp (Leica, Milton Keynes, U.K.) was used for fluorescence detection. A filter cube consisted of two filters (BP 450-490, 515-560) and a dichroic mirror (RKP 510) allowed the passage of exciting and emission beams. The fluorescent emission was registered by a photomultiplier tube (MEA 153, Seefelder Messtechnik, Seefelder, Germany) or by a CCD color camera (Sony). Eight power supplies (F.u.G. Elektronik GmbH, Roseheim, Germany), which could only be controlled manually, were used in this work. To perform microchip CE, four electrodes are generally needed to deliver voltages to buffer inlet, buffer outlet, sample, and waste reservoirs. Each electrode was connected to two power supplies and one ground line through three relays, which could be closed and opened independently by computer with the Labview program (version 5.0, National Instruments, Austin, TX). In this configuration, each electrode can apply two different voltages and ground potential in a run, providing a versatile power supply module for optimization of voltage programs. Each electrode was also coupled to a multimeter so that the current through each electrode can be monitored. Monitoring current is very helpful in troubleshooting. Reagents, Buffers and Sample Solutions. Tris(hydroxymethyl)aminomethane (Tris) was purchased from Lancaster Synthesis (Morecambe, U.K.), N-tris(hydroxymethyl)methyl-3aminopropanesulfonic acid (TAPS) from Merck ((Leics, U.K.), fluorescein from Fluka (Buchs, Germany), and fluorescein-5isothiocyanate (FITC) from Merck (Darmstadt, Germany). The running buffer was composed of 50 mM each of Tris and TAPS (pH 8.8). A sample solution composed of 1 µg/mL fluorescein and 5 µg/mL FITC was used for all electropherograms and prepared with the running buffer daily. It should be mentioned that FITC was not stable in the running buffer and decomposed significantly from hour to hour. Continuous decrease in the peak height of FITC while at the same time increases in peak heights
of FITC degradation products were observed for the same sample solution within 1 day. For recording fluorescent images, the sample solution contained 5 µg/mL fluorescein and was prepared with the buffer as well. Chip Operation. A long platinum wire was placed in each of the buffer reservoirs, as shown in Figure 2, to minimize the problems of air bubbles generated by electrolysis on the surface of the electrodes.32 The electrolysis problem is especially evident in multiplexed systems as the current is many times higher than in single-channel systems. The long electrode also guarantees that the electrical potential is the same at same length points among the different separation channels. The empty chip was filled first with ethanol by filling the buffer inlet reservoir leaving all other reservoirs empty. All the channels were automatically filled in seconds by capillary action; neither vacuum nor pressure was required. The ethanol filling was crucial to the subsequent chip operation without introducing air bubbles into the somewhat hydrophobic PDMS channels. Then the buffer inlet and outlet reservoirs were flushed with water and loaded with 1.5 mL of buffer each. An electric field was applied to the two buffer reservoirs to fill the channels with buffer by electroosmotic flow. For CE separation, 3-µL aliquots of buffer and sample were pipetted into the waste and sample wells, respectively. To record electropherograms, fluorescent emissions from a detection window of 50 × 50 µm, which was positioned 28 mm downstream from the injection intersection, were collected with the PMT. Data were acquired with the Labview program at 50 points/s. No flushing was executed between runs. The sample injection length data were obtained by eye-checking the fluorescent images around the injection intersection, which were recorded on videotapes through the CCD camera. Although the boundary of sample plugs was arbitrary, the data obtained under the same standards are eligible to be compared to each other. RESULTS AND DISCUSSION Conventional Cross Injectors. In conventional cross injectors, the channel width and depth of sample channels are equal to the width and depth of the separation channel. Figure 3 shows the results obtained with such a cross injector. Fluorescein, FITC, and two degradation products of FITC can be separated well from each other. It has been reported that native PDMS walls possess cathodic electroosmotic flow (EOF), which is strong enough to carry the anionic fluorescein to the cathode.20 Despite the fact that the magnitude of EOF in PDMS channels is ∼20% lower than that in glass channels,20 moderate column efficiency has been obtained with chips made of the two materials. The data presented in Figure 3A were obtained using the popular pinched injection method. Theoretically, the sample injection plug is defined by the injector geometry and is independent of injection time. Our results show little change with injection time (Table 1) when the injection time is above the threshold of 10 s, the time for the analytes to move from the sample reservoir to the cross intersection. It should be noted that the sample length and column efficiency were heavily dependent on the pinching voltages. Longer sample length was accompanied by lower column efficiency. When the pinching voltages were not proper so as not to induce sufficient buffer flows (32) Zhu, T.; Sun, Y.-L.; Zhang, C.-X.; Sun, Z.-P.; Ling, D.-K. J. High Resolut. Chromatogr. 1994, 17, 563.
Figure 3. Electropherograms obtained with a conventional cross injector using different injection schemes. (A) Pinched injection. In injection phase, a voltage of 1.08 kV was applied at the sample reservoir and the waste reservoir was grounded. The pinching voltages applied at buffer inlet and outlet reservoirs were 0.94 kV. Injection time was 20 s. To switch for separation, a voltage of 5.40 kV was applied to the buffer inlet reservoir and the buffer outlet reservoir was grounded. A back voltage of 3.82 kV was applied to the sample and waste reservoirs. (B) Floating injection. The conditions were the same as for the pinched injection except that in injection phase both the buffer reservoirs were left floating. (C) Simplest injection. The conditions were the same as for the floating injection except that the sample and waste reservoirs were left floating in the separation phase. The channel depth was 5 µm. Peak identification: (F) fluorescein, (Fc) fluorescein-5-isothiocyanate, and (Fcd1, Fcd2, Fcd3) degradation products of FITC. Other conditions were given in the Experimental Section.
to prevent leakage, sample length and peak height increased with injection time (data not shown). The dependence of peak height on injection time in pinched injection mode has also been reported elsewhere.18 Nevertheless, the pinched injection scheme has now become a common practice in microchip CE. However, it suffers from low sensitivity because of the short sample plug injected. In evidence, using a similar fluorescent microscope, a few tens of micrograms per milliliter fluorescein is generally employed in microchip CE with pinched injection.33 With floating injection, i.e., no pinching voltages applied at the buffer reservoirs during injection while keeping the back voltages in the separation phase, much higher peaks were obtained but at (33) Sirichai, S.; de Mello, A. J. Analyst 2000, 125, 133.
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
2659
Table 1. Comparison of the Performances of Different Injectors injectora
injection method
tinj (s)
S/Nb
conventional cross
pinched
20 40 15 20 20 40 15 20 0.5 2 3 5 6 7
10 12 56 80 8 9 76 104 108 140 164 68 88 116
floating NSC cross
pinched floating simplest
NSC tee
simplest
Linjc (µm)
W1/2d (µm)
Nd (k)
Re
Hplated (µm)
110 110 190 310 60 60 150 220
757 746 875 1018 758 759 680 755 522 574 687 931 1030 1230
7.59 7.81 5.68 4.19 7.56 7.55 9.40 7.62 15.96 13.16 9.20 5.01 4.09 2.87
1.68 1.75 1.55 1.36 1.80 1.73 2.03 1.84 2.46 2.29 1.87 1.53 1.47 1.20
3.69 3.59 4.93 6.69 3.70 3.71 2.98 3.68 1.75 2.13 3.04 5.59 6.84 9.75
a Unless stated otherwise, the data for conventional cross, NSC cross, and NSC tee injectors were obtained under the conditions for Figures 35, respectively. b The signal-to-noise ratio was calculated based on the peak height of 1 µg/mL fluorescein. c The sample plug length injected (Linj) was measured by eye-checking the fluorescent images around the intersection with a sample solution containing 5 µg/mL fluorescein. d The peak width at half peak height (W1/2 ), theoretical plate number (N), and plate height (H plate) were for FITC. e R was the resolution between FITC and fluorescein.
the cost of resolution (see Figure 3B and Table 1). The sample length and thus peak height increased with injection time in floating mode, while the resolution and column efficiency decreased (Table 1). There is a balance between resolution versus sensitivity. Stretched sample plugs in floating injection have been well documented elsewhere.14,16 The simplest injection method gave only distorted small peaks with an elevated baseline due to the severe leakage (Figure 3C). It should be noted that migration times could change by 1-3 s within a day of continuous running. It is ascribed to the elevated temperature inside the channels caused by Joule heating or the change in PDMS walls. Although the simplest injection method has been demonstrated in a few reports,29-31 no data from other injection methods obtained under the same conditions are available for comparison. Higher and sharper peaks could have been achieved with floating injection according to our findings. The results presented here are consistent with the previous conclusions but provide direct comparison of electropherograms obtained with the different injection schemes. Narrow Sample Channel Cross Injectors. Not unlike the conventional cross injectors, using the NSC cross injector with pinched injection, sample length and thus sensitivity and column efficiency were independent of injection time (Table 1). Compared to the conventional cross injector, little advantage is visible when the sample channel width decreases using the pinched injection method (comparing Figure 4A with 3A). In contrast, using floating injection, much sharper (higher separation efficiency) and higher peaks (higher sensitivity) were achieved with the NSC cross injector (comparing Figure 4B to Figure 3B). From the fluorescent images taken of the injection intersections, shorter sample plugs were observed with the NSC than the conventional cross injector in both pinched and floating injection modes (Table 1). The results show that the higher column efficiency and higher sensitivity obtained with the NSC cross injector result from the shorter and more concentrated sample plug formed at the cross intersection. Furthermore, using the same NSC cross injector, floating injection (simpler procedure) provided higher sensitivity without compromising resolution as compared to the pinched injection (compli2660 Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
cated procedure) (see Table 1). Despite the fact that the sample plug length for Figure 4B (floating injection mode) was more than 3 times the sample plug length for Figure 4A (pinched injection mode), the resolution was even a little bit better in the former. In theory, when the sample plug length is shorter than the threshold, its contribution to the plate height is negligible.14 Our results show that sufficiently well shaped sample plugs can be obtained using the NSC cross injector without pinching voltages. The spreading of the sample plug is contained sufficiently with the geometry of NSC injector. Thus, in contrast to the conventional cross injector, floating injection with the NSC cross injector is obviously superior to the pinched one. More importantly, higher separation efficiency was obtained with the NSC cross injector using the simplest injection method (Figure 4C) instead of floating injection. It is due to the shorter injection time required in the simplest injection mode, suppressing the diffusion of sample during injection. Of course, the sensitivity increases with the injection time while the resolution decreases in both simplest and floating injection modes (Table 1). For a given peak height (sensitivity), the simplest injection mode gives the best separation. The baseline elevation, which is commonly seen with conventional injectors due to severe leakage (see Figure 3C and also ref 20), was not observed with the NSC injector (Figure 4C). From the fluorescent image of the injection intersection, in the separation phase, the analyte flows in the fine sample channels were seen to withdraw automatically toward the sample and sample waste reservoirs, which were left floating. This phenomenon was not seen with the conventional injector with big sample channels. The reasons for the withdrawing of analyte flows in the fine sample channels of NSC injectors are not clear. However, the fact is clear that the geometry of NSC itself can prevent the leakage effectively. Even if there were leakage, the leakage of sample from the small sample channels into the big separation channel must be much less severe than it would otherwise be with the conventional cross injector and would be small enough to be imperceptible. Using the simplest injection scheme with the NSC cross injector, the sensitivity is increased by a factor of 10 and the
Figure 4. Electropherograms obtained with a NSC cross injector using (A) pinched, (B) floating, and (C) simplest injection. The width of sample channels was 10 µm, one-fifth of the separation channel width 50 µm. The injection times were 40, 20, and 0.5 s for panels A-C, respectively. The back voltage was 4.32 kV for panels A and B. Other conditions for panels A-C were the same as for Figure 3AC, respectively.
column efficiency more than doubled compared to the popular pinched injection with conventional cross injectors (Table 1). And these are achieved without added complexity to the microchannel structure. In fact, the operational procedure with the NSC injector is much simpler: no leakage control is required at all. To obtain a sample plug narrower than the separation channel width, a sixport injection structure has been proposed, adding complexity amid the complexity of chip operation.34 If such a narrow sample plug is really wanted, the NSC cross injector with pinched injection would provide a much better solution. The principle shown with the PDMS-glass chips is applicable to chips made of other materials because the benefits come from the geometry. Narrow Sample Channel Tee Injectors. With conventional tee injectors, in which the widths of sample and separation channels are equal, it is very difficult, if not impossible, to achieve a decent separation. Although the conventional tee injector has been demonstrated with glass chips,9-11 our attempts to realize it with the glass-PDMS chips failed. However, successful separa(34) Deshpande, M.; Greiner, K. B.; West, J.; Gilbert, J. R.; Bousse, L.; Minalla, A. In Proceedings of the Micro Total Analysis System 2000 Symposium, van den Berg, A., Olthuis, W., Bergveld, P., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2000; p 339.
Figure 5. Electropherograms obtained with a NSC tee injector. The injection times were 5, 6, and 7 s for panels A-C, respectively. In injection phase, a voltage of 1.08 kV was applied at the sample reservoir while the buffer inlet was grounded and buffer outlet left floating. For separation, a voltage of 5.40 kV was applied at the buffer inlet while the buffer outlet was grounded and the sample reservoir was left floating. The width of sample channels was 5 µm, one-tenth of the separation channel width. The channel depth was 10 µm. Other conditions were given in the Experimental Section.
tions were obtained using a NSC tee injector when the sample channel width decreased to one-tenth of the separation channel width (Figure 5). Like the NSC cross injector, no voltage control is needed with the NSC tee injector. The sample plug length, peak height, and resolution were dependent on injection time (Table 1). Compared to the NSC cross injector, the NSC tee injector provided somewhat lower resolution. However, better resolution is anticipated with an even narrower sample channel. The significance is that use of NSC tee injectors drastically reduces the number of reservoirs in multiplexed microchip CE systems. A theoretical minimum of N + 3, where N is the number of paralleled systems, has been predicted.35 Practically a multiplex system consisting of 2N + 1 reservoirs has been realized.8 In the layout presented here combined with NSC tee injectors, only N + 2 reservoirs are required, less than the predicted theoretical minimum. We believe that N + 2 is the real theoretical minimum. Our results show that a multiplexed CE system with N + 2 (35) Woolley, A. T.; Sensabaugh, G. F.; Mathies, R. A. Anal. Chem. 1997, 69, 2181.
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
2661
reservoirs is feasible by using NSC injectors. It should be noted that all the data presented in this paper were obtained using the multiplexed chips composed of eight parallel CE systems. Due to the single-point detection scheme used, only the signal from one CE system was recorded at one time. CONCLUSIONS The NSC cross and tee injectors possess obvious advantages over their conventional counterparts. Without optimization on injector dimensions, the prototype NSC cross injector, in which the sample channel width is one-fifth of the separation channel width, leads to 10 times higher sensitivity and at the same time the column efficiency is doubled. Furthermore, the operation procedure is greatly simplified: no leakage control is needed at all. This is imperative as one precondition for the microchip system to be widely adopted is that the interface between the system and operators must be simple. Currently available techniques allow easy fabrication of the NSC injectors of different channel widths
2662
Analytical Chemistry, Vol. 73, No. 11, June 1, 2001
but with the same depth. There is no reason for the NSC cross injector not to replace the conventional one. Although the NSC tee injector provides separation efficiency that is not as good as the NSC cross injector, the NSC tee injector is very useful in simplifying multiplexed systems when the resolution is not a critical matter. ACKNOWLEDGMENT This research was sponsored by the European Community under research grant BIO4980382. The clean room used for fabrication of microchips was financially supported by Smithkline Beecham and AstraZeneca. The authors also thank Gareth Jenkins for assistance in chip fabrication and much appreciated comments on the manuscript. Received for review January 30, 2001. Accepted March 21, 2001. AC010138F
