End-to-End Differential Contactless Conductivity Sensor for Microchip

Mar 25, 2010 - This measurement technique is applied to a miniaturized CE device fabri- cated in low-temperature cofired ceramics (LTCC) mul- tilayer ...
7 downloads 25 Views 910KB Size
Download PDF
Anal. Chem. 2010, 82, 3270–3275

End-to-End Differential Contactless Conductivity Sensor for Microchip Capillary Electrophoresis Georg Fercher,*,†,‡ Anna Haller,† Walter Smetana,† and Michael J. Vellekoop† Institute of Sensor and Actuator Systems, Vienna University of Technology, Vienna, Austria, and IMA GmbH, Wiener Neustadt, Austria In this contribution, a novel measurement approach for miniaturized capillary electrophoresis (CE) devices is presented: End-to-end differential capacitively coupled contactless conductivity measurement. This measurement technique is applied to a miniaturized CE device fabricated in low-temperature cofired ceramics (LTCC) multilayer technology. The working principle is based on the placement of two distinct detector areas near both ends of the fluid inlet and outlet of the separation channel. Both output signals are subtracted from each other, and the resulting differential signal is amplified and measured. This measurement approach has several advantages over established, single-end detectors: The high baseline level resulting from parasitic stray capacitance and buffer conductivity is reduced, leading to better signal-to-noise ratio and hence higher measurement sensitivity. Furthermore, temperature and, thus, baseline drift effects are diminished owing to the differentiating nature of the system. By comparing the peak widths measured with both detectors, valuable information about zone dispersion effects arising during the separation is obtained. Additionally, the novel measurement scheme allows the determination of dispersion effects that occur at the time of sample injection. Optical means of dispersion evaluation are ineffective because of the opaque LTCC substrate. Electrophoretic separation experiments of inorganic ions show sensitivity enhancements by about a factor of 30-60 compared to the single-end measurement scheme. Capacitively coupled contactless conductivity measurement (C4M) is an universal method for the label-free determination of charged compounds in capillary electrophoresis (CE) systems.1-3 Easy on-chip integration capabilities as well as low power consumption requirements have rendered C4M as the preferred measurement technique for mobile and point-of-use CE applications.4-7 Overall robustness and lifetime of the device are significantly improved because of the avoidance of direct * To whom correspondence should be addressed. E-mail: georg.fercher@ tuwien.ac.at. † Vienna University of Technology. ‡ IMA GmbH. (1) Pumera, M. Talanta 2007, 74, 358–364. (2) Kuban, P.; Hauser, P. C. Electrophoresis 2009, 30, 176–188. (3) Johns, B.; Breadmore, M. C.; Macka, M.; Ryvolova, M.; Haddad, P. R. Electrophoresis 2009, 30, S53–S67. (4) Kuban, P.; Reinhardt, M.; Mu ¨ller, B.; Hauser, P. C. J. Environ. Monit. 2004, 6, 169–174.

3270

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

galvanic contact between the electrodes and the fluid: Chemical reactions, bubble formation, and interactions between the separation field and sensitive measurement circuitry are circumvented. The working principle is based on capacitively coupling the excitation signal through a dielectric substrate wall into a small section of the separation channel which defines the measurement cell. Local changes in conductivity caused by separated analyte zones moving through this cell result in a change of the detector output signal. First, implementations of C4M were shown for conventional CE in 19988,9 and later on adapted for miniaturized devices.10-18 In the beginning, onchip detector electrodes were realized by photolithographic processes on glass substrates.10-13 Detectors constituting externally arranged electrodes were presented thereafter, simplifying the fabrication process.14-19 External electrodes were fabricated by aluminum foil strips,14,15 silver paint,16 adhesive copper tape,19 or screen-printing silver conductive paste.20 These can be placed on movable holders, allowing the measurement of positively and negatively charged species in one separation run and enabling the use of bare CE chips.19,21 On conventional CE, the fused-silica capillary can be fully encircled by the two C4M electrodes, allowing proper capacitive (5) Kuban, P.; Nguyen, H. T. A.; Macka, M.; Haddad, P. R.; Hauser, P. C. Electroanalysis 2007, 19, 2059–2065. (6) Kappes, T.; Galliker, B.; Schwarz, M. A.; Hauser, P. C. Trends Anal. Chem. 2001, 20, 133–139. (7) Xu, Y.; Wang, W.; Li, S. F. Y. Electrophoresis 2007, 28, 1530–1539. (8) Zemann, A. J.; Schnell, E.; Volgger, D.; Bonn, G. K. Anal. Chem. 1998, 70, 563–567. (9) da Silva, J. A. F.; do Lago, C. L. Anal. Chem. 1998, 70, 4339–4343. (10) Lichtenberg, J.; de Rooij, N. F.; Verpoorte, E. Electrophoresis 2002, 23, 3769–3780. (11) Laugere, F.; Lubking, G. W.; Berthold, A.; Bastemeijer, J.; Vellekoop, M. J. Sens. Actuators A 2001, 92, 109–114. (12) Guijt, R. M.; Baltussen, E.; van der Steen, G.; Frank, H.; Billiet, H.; Schalkhammer, T.; Laugere, F.; Vellekoop, M. J.; Berthold, A.; Sarro, L.; van Dedem, G. W. K. Electrophoresis 2001, 22, 2537–2541. (13) Berthold, A.; Laugere, F.; Schellevis, H.; de Boer, C. R.; Laros, M.; Guijt, R. M.; Sarro, P. M.; Vellekoop, M. J. Electrophoresis 2002, 23, 3511–3519. (14) Pumera, M.; Wang, J.; Opekar, F.; Jelinek, I.; Feldman, J.; Lo¨we, H.; Hardt, S. Anal. Chem. 2002, 74, 1968–1971. (15) Wang, J.; Pumera, M.; Collins, G. E.; Mulchandani, A. Anal. Chem. 2002, 74, 6121–6125. (16) Tanyanyiwa, J.; Hauser, P. C. Anal. Chem. 2002, 74, 6378–6382. (17) Wang, J.; Pumera, M. Anal. Chem. 2002, 74, 5919–5923. (18) Wang., J.; Pumera, M. Anal. Chem. 2003, 75, 341–345. (19) Tanyanyiwa, J.; Abad-Villar, E. M.; Fernandez-Abedul, M. T.; Costa-Garcia, A.; Hoffmann, W.; Guber, A. E.; Herrmann, D.; Gerlach, A.; Gottschlich, N.; Hauser, P. C. Analyst 2003, 128, 1019–1022. (20) Wang., J.; Chen, G.; Chatrathi, M. P.; Wang, M.; Rinehart, R.; Muck, A. Electroanalysis 2008, 20, 2416–2421. (21) Wang, J.; Chen, G.; Muck, A., Jr. Anal. Chem. 2003, 75, 4475–4479. 10.1021/ac100041p  2010 American Chemical Society Published on Web 03/25/2010

Figure 1. Cross-sectional view and electrical model of a planar conductivity cell with the microchannel (b) embedded within the chip substrate (a,c).

coupling of the excitation signal into the fluid. Placing additional grounded shield electrodes greatly reduces cross-coupling between the excitation- and the measurement-electrode.9,22 An adoption of such electrode configurations for on-chip devices is technologically harder to achieve since microchannels are embedded in a substrate material. Several alternative electrode designs were proposed for on-chip C4M and were recently reviewed.1 A common detector arrangement is a planar electrode allocation at one side of the chip, as schematically depicted in Figure 1.11-21 It also shows the electric equivalent circuit which is used for modeling this type of contactless conductivity cell.9,23 In this arrangement, the measurement signal is fed into and retrieved from the microchannel via a coupling capacitance Ccpl and a double layer capacitance Cdl. For proper coupling and, thus, high measurement sensitivity, these capacitances should be as large as possible.22,24 Whereas the value of Cdl is defined by the fluid/surface chemistry, the value of Ccpl can be increased by reducing substrate thickness and/or increasing the material permittivity.25 A significant fraction of the measurement signal is not coupled into the microchannel at all. It is rather short-circuited by substrate (Csub) and ambient air (Cair) capacitance. These parasitic effects have to be kept as low as possible for adequate coupling. Although the influence of Cair can be reduced using grounded shields, cross-coupling through the substrate persists particularly for planar electrode arrangements.2,24 The influence of stray capacitance induces high baseline levels with respect to the output signals of interest. This reduces signal-to-noise ratio and, thus, affects limits-of-detection (LOD) on miniaturized devices. Compared to conventional capillaries, LODs for onchip C4M are around 1 order of magnitude higher.19,26 Apart from parasitics, the electric equivalent conductivity of the buffer solution has to be kept low to minimize the baseline level of the output signal. Amphoteric buffer electrolytes with relatively low equivalent conductivity fulfill this criterion for CE(22) Brito-Neto, J. G. A.; da Silva, J. A. F.; Blanes, L.; do Lago, C. L. Electroanalysis 2005, 17, 1198–1214. (23) Kuban, P.; Hauser, P. C. Electrophoresis 2004, 25, 3387–3397. (24) Kuban, P.; Hauser, P. C. Lab Chip 2005, 5, 407–415. (25) Fercher, G.; Smetana, W.; Vellekoop, M. J. Electrophoresis 2009, 30, 2516– 2522. (26) Tanyanyiwa, J.; Galliker, B.; Schwarz, M. A.; Hauser, P. C. Analyst 2002, 127, 214–218.

separations at pH values ranging from 5 to 9.27 Thermal noise caused by Joule heating effects of the buffer, however, nevertheless compromises sensitivity. Several approaches were proposed in combination with C4M to reduce the baseline level and, thus, achieve high measurement sensitivity: A common implementation is electronic baseline suppression. An adjustable bias is subtracted from the output signal before analog-to-digital conversion.9,14-24 Recently, a highresolution version of an analog-to-digital converter was introduced, eliminating the necessity of electronic baseline suppression.28 Another way of electronic baseline suppression is by means of a four-electrode arrangement in combination with the differential input of a lock-in amplifier.29 A sensitivity increase by a factor of 10 is achieved. The design, however, involves a widening of the channel at the detector area and cumbersome photolithographic steps for electrode fabrication. Furthermore, the measurement is carried out at one single point near the end of the channel only. The baseline level can also be reduced by keeping parasitics low which is the case when detector electrodes are arranged in an opposite way at two sides of the channel.10,25,30 The measurement signal is applied perpendicular to the direction of the flow; cross-coupling through the substrate wall is minimized. This method is advantageous in combination with high-permittivity tapes and also results in improved coupling of the measurement signal into the microchannel. Cell volumes, however, are smaller and affect LODs.1 A different approach of baseline suppression is based on the operation of the conductivity cell as part of a resonator.31 When the measurement is carried out at the resonance frequency, capacitive influences are compensated. A new way for baseline compensation is end-to-end differential C4M as presented herein. Instead of the use of one pair of electrodes which usually form one detector near the end of the separation channel (downstream detector), an identical electrode pair is placed at the beginning of the channel near the sample injector (upstream detector). The baseline offset is canceled by subtracting the two detector output signals from each other. A major advantage is that global fluctuations in conductivity which are, e.g., caused by Joule heating effects of the buffer are removed from the final output signal. Only local changes in conductivity are measured owing to the differentiating nature of the circuit. A conductive zone (“peak”) migrating down the separation channel, therefore, passes two detectors and causes two output signals with opposing amplitudes. Comparison of these two signals gives valuable information about dispersion effects that occur during the separation and at the time of injection. This information cannot be obtained by optical means because of the opaque device material. A further merit of of the differential approach is the extended utilizable dynamic range of the subsequent analog/digital conver(27) Good, N. E.; Winget, G. D.; Winter, W.; Connolly, T. N.; Izawa, S.; Singh, R. M. M. Biochemistry 1966, 5, 467–477. (28) Francisco, K. J. M.; do Lago, C. L. Electrophoresis 2009, 30, 3458–3464. (29) Laugere, F.; Bastemeijer, J.; van der Steen, G.; Vellekoop, M. J.; Sarro, P. M.; Bossche, A. Proc. IEEE Sensors 2002, 450–453. (30) Lee, C. Y.; Chen, C. M.; Chang, G. L.; Lin, C. H.; Fu, L. M. Electrophoresis 2006, 27, 5043–5050. (31) Kang, Q.; Shen, D.; Li, Q.; Hu, Q.; Dong, J.; Du, J.; Tang, B. Anal. Chem. 2008, 80, 7826–7832.

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3271

Figure 2. Exploded view (A), enlarged detector areas (B,C) of the LTCC device (not to scale) and differential measurement setup with postamplification stage.

sion stage which can now be set to more sensitive values. Furthermore, correlated noise components affecting both detector areas simultaneously are canceled, and therefore, better signalto-noise ratio is achieved. Differential contactless conductivity detection was previously reported for conventional CE, however, in a substantially different way. The differential detector is placed at one end of the channel only and is composed of one single excitation electrode and two adjacent pick-up electrodes.32 We show first successful CE separations using the end-to-end differential C4M approach on a miniaturized device fabricated in low-temperature cofired ceramics (LTCC) multilayer technology to highlight the advantages of the novel measurement scheme. The cost-efficient LTCC is an attractive substrate material for applications where robustness and longevity of analysis devices are needed. Chemical inertness and high resistivity against the penetration of foreign gases and liquids as well as easy threedimensional integration capabilities of fluidic, electrical, and mechanical components are other key benefits of the ceramics.25,33-36 EXPERIMENTAL SECTION Device Description. The LTCC CE device as used for all consecutive experiments was fabricated in-house. Figure 2A shows an exploded view of the multilayered device consisting of five ceramic tapes with a thickness of 100 µm each after sintering and (32) Shaw, G. A.; Ross, D.; Fick, S. E.; Vreeland, W. N. Proc. MicroTas 2008, 1630–1632. (33) Smetana, W.; Balluch, B.; Stangl, G.; Lu ¨ ftl, S.; Seidler, S. Microelectron. Reliab. 2009, 49, 592–599. (34) Fercher, G.; Haller A.; Smetana, W.; Vellekoop, M. J. Analyst 2010, DOI: 10.1039/b922501c. (35) Ciosek, P.; Zawadzki, K.; Lopacinska, J.; Skolimowski, M.; Bembnowicz, P.; Golonka, L. J.; Brzozka, Z.; Wroblewski, W. Anal. Bioanal. Chem. 2009, 393, 2029–2038. (36) Martinez-Cisneros, C. S.; da Rocha, Z.; Ferreira, M.; Valdes, F.; Seabra, A.; Gongora-Rubio, M.; Alonso-Chamarro, J. Anal. Chem. 2009, 81, 7448–7453.

3272

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

a dielectric constant of 7.9. A top layer (a) encompasses the sensor and fluidic inlet/outlet ports. The tape thickness defines the height of the microchannel which was formed into one single ceramic layer (b). Three support layers (c) form the bottom of the channel and ensure sufficient mechanical stability of the CE device. A detailed description of the fabrication process was recently published.34 The microchannel has a width of 100 µm. A double-T injector with an offset of 450 µm is placed 5 mm away from buffer- and sample-inlet ports. Two identical planar detector areas, each consisting of two screen-printed silver electrodes, are located at opposing ends of the separation channel. Close-up drawings are depicted in Figure 2B,C. The upstream detector near the beginning of the channel is placed 7 mm away from the midpoint of the injector, and the downstream detector is placed in a distance of 46 mm from this midpoint. The electrodes have a width of 100 µm and fully cover the width of the microchannel. Further electrode dimensions are given in Figure 2B. Measurement Setup. Both detectors are excited with a sinusoidal voltage from a signal generator (Model 33120A, Agilent) at a frequency of 2.5 MHz. At this frequency, the maximum peak response is observed. It was determined by measuring the output signals of separated inorganic ions at frequencies up to 6 MHz. Figure 2 shows the differential measurement setup. The output signals of both detectors are separately preamplified by two current-to-voltage converters with a feedback resistance of 40 kΩ. A balancing circuit is incorporated on one signal branch to equalize the signal amplitudes of both preamplifiers by an adjustable offset. Both preamplified signals are then fed into a single differential amplifier stage which was constructed in-house in surface mount technology based on the OPA656 (Texas Instruments). After subtraction, the obtained differential signal is postamplified by a selectable factor of 20 or 100. Preamplifiers and the differential amplifier are placed symmetrically in close

proximity and on the same printed circuit board to avoid signal drifts due to nonuniform heating effects. Yet, the conducting paths are designed in a way that minimizes crosstalk between both signal branches. External interferences induced by electromagnetic fields are reduced by putting the circuit board together with the CE device in a grounded shielding box. The amplified differential output is fed into a benchtop lock-in amplifier (Model SR844, Stanford Research Systems) set to an input bandwidth of 0.5 Hz for low-noise analog/digital conversion. Data is acquired by a laptop via the IEEE-488 bus. The separation voltage is generated by a dedicated power supply (Model CZE2000, Spellman). Platinum wires serve as highvoltage electrodes and are mounted into the buffer inlet/outlet ports of the microchannel. The LTCC device is placed in a poly(methyl methacrylate) (PMMA) chip holder fabricated inhouse for fluid-tight filling. To make LODs comparable to results we reported earlier,25,34 the same procedure for its determination is used. LODs are calculated according to three times the signal-to-noise ratio, which is defined as maximum peak height divided by the background noise. The noise level is obtained by integration over a time span of 10 s in an area of the electropherogram where no peaks occur. All data is fitted using the integrated second order Gaussian algorithm supplied by the PeakFit software bundle (Hearne Scientific Software). Numerical measurement results presented herein are mean values obtained from five consecutive separation runs. Samples, Reagents, Injection Procedure. Chemicals were purchased from Sigma-Aldrich at reagent grade. Deionized water supplied by a Millipore SA system (Molsheim, France) was used throughout the experiments. Stock solutions of potassium (K+), sodium (Na+), and lithium (Li+) were prepared from their chloride salts. Sample and MES/histidine (His) buffer solutions were prepared daily and freed of gas bubbles and particles by 5 min degassing and 0.2 µm syringe filtration. The channel was flushed with buffer (about 30 channel volumes) at the beginning of the experiments. At the end of the experiments, the channel was rinsed thoroughly with deionized water to prevent clogging and then dried by applying clean, pressurized air. The CE separations are carried out after a vacuum-driven double-T sample injection: The buffer is first pumped into the channel. Then, the sample inlet is filled with sample, and vacuum is applied to the sample waste reservoir using a plastic syringe attached to the chip holder. In this way, reproducible sample plugs are introduced into the separation channel. The injection procedure also causes flow from the buffer inlet and waste reservoirs, effectively leading to a pinched injection.37 Immediately after the sample injection, the separation voltage is switched on. All experiments were conducted at a laboratory temperature of 22 ± 1 °C. RESULTS AND DISCUSSION An initial characterization of the differential measurement scheme was carried out by injecting diluted background buffer as a neutral marker. Figure 3 presents the electropherogram of the differential signal. The plug is driven by EOF forces alone (37) Lichtenberg, J.; Verpoorte, E.; de Rooij, N. F. Electrophoresis 2001, 22, 258–271.

Figure 3. Electropherogram obtained for the separation of a 5 mM MES/His sample. Separation parameters are as follows: 10 mM MES/ His buffer at pH 6; separation field strength is 100 V/cm. Excitation voltage is 3 Vp-p. The factor of postamplification is set to 20.

and reaches the upstream detector after about 17 s. This detector is connected to the noninverting input of the differential amplifier. The local drop in conductivity results in a negative peak. As the migration proceeds, the plug eventually moves past the downstream detector. Because this measurement branch is connected to the inverting input of the amplifier, the local conductivity drop results in a positive peak in the differential signal. During a CE experiment, time-dependent and time-independent processes give rise to a dilution of the injected sample plug into the surrounding buffer which is known as peak dispersion.38-40 By comparing the peak widths obtained by both detectors, information about time-dependent dispersion contributions that occur during the separation run is obtained. The main factors for this type of dispersion are molecular diffusion, Joule heating, and analyte/wall interactions. The average temporal peak broadening39 (with units of time) obtained by comparing both peak widths (at half peak height) in the electropherograms amounts to around 50% for the 5 mM MES/His sample. The extent of the peak broadening reveals valuable information that can be used to, e.g., optimize the influence of different channel surface finishes with regard to separation efficiency. The peak width measured at the upstream detector also gives insight into time-independent dispersion effects if compared with the initially injected sample plug length. Because of the close proximity of this detector to the injector, the temporal peak width obtained at this position nearly reflects the initial spatial plug length (with units of length) of the injected sample. The injected plug length is defined by the offset of the double-T injector (450 µm). Before the measured (temporal) and injected (spatial) plug lengths can be directly compared with each other, a straightforward mathematical transformation which involves the migration velocity of the analyte has to be performed.39 This velocity is 0.38 mm/s and was calculated from the graph of Figure 3. The calculated lateral extension of the peak is approximately 1.4 mm at the position of the upstream detector, which represents an increase of 300% compared to the initially injected sample plug length. The zone dispersion inflicted by the sample injection is, therefore, by about a factor of 6 higher than the one caused by (38) Foret, F.; Deml, M.; Bocek, P. J. Chromatogr. 1988, 452, 601–613. (39) Huang, X.; Coleman, W. F.; Zare, R. N. J. Chromatogr. 1989, 480, 95–110. (40) Jones, H. K.; Nguyen, N. T.; Smith, R. D. J. Chromatogr. 1990, 504, 1–19.

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3273

the subsequent separation process, which is 50%. This demonstrates that in fact injection is the major reason of dispersion in this case. Hydrodynamic differences are suspected to cause leakage from the side channels into the double-T injector, effectively prolonging the sample plug. Such effects could be reduced using pinched injection and/or a smaller cross-section of the separation channel for higher hydrodynamic back-pressure in the sample reservoirs. Another potential reason for zone dispersion are bulging effects of the electric field near the detector electrodes.10,25 In this case, the field stretches out in lateral direction of the separation channel. A rise in the output signal is already measured even though the separated zones have not yet fully entered the detector cell. According to Figure 3, the value of the baseline level is 20 mV rather than zero, as is expected in case of full compensation. One reason is electronic crosstalk between the two measurement branches. It is the remaining offset of the baseline which could not be compensated by the balancing circuit. This offset amounts to about 3 mV. Negative peaks exceeding this value would cause negative differential output signals. This is, however, currently unsupported by the electronics. The baseline level was, therefore, manually increased to a value of 20 mV by the balancing circuit so that even greater negative peaks can be measured. The lockin amplifier is set to an input sensitivity of 30 mV, which covers the expected output range of the differential signals with the factor of postamplification set to 20. Figure 4 presents electropherograms of samples containing different inorganic cations measured with the differential method. Because the samples have a higher conductivity than the surrounding background buffer, positive peaks now follow negative ones. This is contrary to Figure 3, where the sample had a lower conductivity than the buffer. Average time-dependent peak dispersion that occurs during the separation is 45%, 40%, and 35% for potassium, sodium, and lithium, respectively. The time-independent zone dispersion inflicted by the injection process is again 300%, confirming that the extent of the sample leakage is independent of the sample composition. Figure 5 shows successful separations of three different ions with conventional single-end C4M (A) and the novel end-to-end differential measurement technique (B). The electronic circuit is designed in a way that allows easy switching between these two measurement modes. A very high baseline offset of about 260 mV is present if only the downstream detector is utilized. Furthermore, considerable noise is visible in the electropherogram in Figure 5A. The measured peak heights only amount to about 0.2% of the total signal because of the high baseline level. In this detector configuration, the input stage of the lock-in amplifier needs to be increased to a dynamic range of 300 mV for appropriate sampling. The calculated LODs are 59, 75, and 91 µM for K+, Na+ and Li+, respectively. If the electronic circuit is switched to differential mode, i.e., the upstream detector is utilized as well, the baseline drops to 20 mV. The injected sample plug first moves past the upstream detector, where the three ions are still grouped together as visible in Figure 5B. Interestingly, the sample zone has already separated from the uncharged system plug at this time. All sample ions reach the downstream detector after about 60 s, whereas the system plug migrating due to the EOF arrives at the downstream detector 3274

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

Figure 4. Electropherograms of single ions at a concentration of 200 µM. Separation parameters are as follows: 10 mM MES/His buffer at pH 6; separation field strength is 100 V/cm. Excitation voltage is 10 Vp-p. The factor of postamplification is set to 20.

Figure 5. Electropherograms of three different ions at a concentration of 300 µM using conventional single-end measurement (A) and at a concentration of 100 µM by differential measurement (B). Other separation parameters as of Figure 4.

after about 115 s. The calculated LODs are 4, 5, and 6 µM for K+, Na+, and Li+, respectively. This represents a significant

Figure 6. Electropherograms of three different ions at a concentration of 40 µM (A) and histogram of the baseline noise (B). Separation parameters as of Figure 4. The factor of postamplification is set to 100.

increase in measurement sensitivity by a factor of about 15 compared to the conventional single-end C4M approach. The lower noise level on the baseline of the differential signal is explained by the subtraction of common mode noise terms incident on both detectors. Such components could be caused by poor high-frequency filtering of the HV power unit which enters the shielded measurement box via connecting cables. A further rise in measurement sensitivity of the differential setup is achieved by increasing the factor of postamplification to 100. Ferrite cores are now placed around all connecting cables leading to the measurement box to reduce a preload of highfrequency components on the system. Successful separations of inorganic ions at a concentration of 40 µM are shown in Figure 6A. LODs are 1.9, 1.8, and 1.5 µM for K+, Na+, and Li+, respectively. An accurate analysis of the baseline noise is of utter importance for meaningful calculation of LODs. Noise must be Gaussian distributed; only then does the standard deviation represent a reliable value to correctly derive the measurement sensitivity. Figure 6B presents the normal noise distribution of the unfiltered baseline, taken over a time span of 10 s. The standard deviation representing the reliable magnitude of noise of the system is 4.3 µV. CONCLUSION An end-to-end differential contactless conductivity sensor for on-chip capillary electrophoresis is introduced in this work. The novel measurement scheme offers several advantages which were pointed out in this paper: (a) Higher peak response because of

improved signal-to-noise ratio. The enhancement in detection sensitivity amounts to a factor of 30-60 compared to conventional single-end detectors. (b) Compensation of baseline drifts caused by Joule heating effects and external electromagnetic fields. (c) Determination of time-dependent dispersion effects that occur during the CE separation. (d) Determination of time-independent dispersion effects that occur at the time of injection. The information obtained about dispersion allows for the optimization of separation and design parameters of the opaque LTCC module. In the current setup, sample leakage occurring at the time of injection dominates over dispersion effects caused during the separation. Future optimization strategies will, therefore, aim at an improved sample introduction procedure. Since the measurement mode is differential, the resulting output signal will be scrambled and unreadable if analytes are present at both detector areas simultaneously. Such a situation could appear at a high velocity of the EOF where slowly moving anions and fast migrating cations are dragged into the same direction. This is not observed with the buffer composition used in this work. If, however, different buffers and/or chip materials exhibiting very high EOF velocities are used, the system needs to be designed accordingly to avoid overlapping peaks of the differential signal. All experiments were conducted on a CE device fabricated using LTCC tape material. The novel measurement technique is, however, also suitable for transparent substrate materials to improve LODs and/or substitute optical means of dispersion control. ACKNOWLEDGMENT The authors are very grateful to Franz Kohl and Franz Keplinger for discussions about the differential measurement setup. Special thanks go to Johannes Steurer for valuable information regarding the electronic realization. The authors would like to thank Michael Unger, Bruno Balluch, Ibrahim Atassi, and Heinz Homolka for their precious support in the realization of the device. We wish to extend special thanks to Rupert Chabicovsky for proofreading the manuscript. This work has been funded by the Austrian Federal Ministry of Transport, Innovation and Technology (BMVIT), and the province of Lower Austria in the framework program “Industrial Competence Centers and Networks (Kind)”.

Received for review January 7, 2010. Accepted March 10, 2010. AC100041P

Analytical Chemistry, Vol. 82, No. 8, April 15, 2010

3275