A new two-chip concept for continuous measurements on PMMA-microchips

Abstract

A new concept for continuous measurements on microchips is presented. A PMMA (polymethylmethacrylate) based capillary electrophoresis chip with integrated conductivity detection is combined with a second chip, which undertakes the task of fluid handling and electrical connections. The combination of electrokinetic and hydrodynamic flows allows long-term continuous stable analyses with good reproducibilities of migration time and peak heights of analytes. The two-chip system is characterized in terms of stability and reproducibility of separation and detection of small ions. Relative standard deviations of <1% and 3% respectively for retention times and peak heights during long-term measurements can be achieved. The new system combines simple handling and automated analysis without the need for refilling, cleaning or removal of the separation chip after one or several measurements.

---

Introduction

Since the introduction of micro total analysis systems (µ-TAS) in the 1990's, many applications of lab-on-a-chip have been published, mainly focused on the analysis, separation and detection of biomolecules such as DNA and oligonucleotides, proteins and peptides.^1–6 Commercial systems for these purposes are already available.

The microchip based separation techniques include chromatographic⁷ and in particular different electrokinetic techniques like capillary electrophoresis (CE), isotachophoresis (ITP) and isoelectric focusing (IEF).^8–10 Detection methods mainly applied in µ-TAS, can be divided into optical and electrochemical techniques with laser induced fluorescence (LIF),¹¹ refractive index¹² luminescence¹³ on the one hand and amperometry,¹⁴ conductivity¹⁵ and voltammetry¹⁶ on the other.

But all of these techniques applied to microchips are faced with the problem of fluid handling. As a rule microchip based systems, including commercial ones, are used discontinuously with sample introduction, cleaning and refilling steps, or even removal of the chip after one or several analyses. Because of this effort the measurement frequency is limited and thus so is the possible field of applications. In order to solve this so called “world-to-chip” interface problem and to integrate microchips into measurement and control systems, a continuous and automated high throughput sampling system is necessary.

There are few publications dealing with this problem so far.^17–26 Chen and co-workers^21,24 described a microchip system for CE-separation with electrokinetic injection from a bypass flow into the separation channel as an approach for coupling microchips to flow systems. Although they improved the total operating time for several consecutive injections up to 25 minutes and demonstrate the applicability for protein separation, they were still faced with the problem of changing reservoir levels of buffer and waste, which results in a hydrostatic backpressure and a decreasing electroosmotic flow (EOF). From that followed a decreasing injection volume and decreasing peaks.

Attiya and co-workers²³ introduced a two-chip system which enables semi-continuous analysis with automated refilling and cleaning steps. They coupled the separation chip to a second chip, in which high voltage electrodes are embedded and where a continuous sample supply through a sample bypass channel is possible. Refilling and cleaning are realized by applying a combination of electrokinetic and pressure forces. Due to the refilling steps needed the system can only be operated semi-continuously.

In both, the two-chip system of Attiya and the chip of Chen, electrical contact for the separation voltage is applied through metal wires in reservoirs which have to be levelled, refilled and cleaned.

Besides this, the liquids in open reservoirs can evaporate or can be contaminated. Finally high voltage electrodes placed in static reservoirs can cause heating or even electrochemical processes, leading to pH variations or chemical redox processes.

To overcome these disadvantages and limitations a two-chip based microchip system was developed, consisting of the so-called “world-to-chip” interface, with all fluidic and electrical connections, that can easily be coupled to an exchangeable analysis chip—in this case a CE microchip made of PMMA with integrated on-chip-electrodes for detection based on conductivity measurements. The two-chip analysis system provides maximum flexibility with regard to changing analytical requirements. The “world-to-chip”-interface combines the basic principle for a fully automated chip based monitoring system without constraining the variety of chip operation modes. For e.g. sample introduction can be easily switched from electrokinetic to hydrodynamic injection. In principle, the interface can be coupled with microchips of any design, electrokinetic separation techniques and detection methods, provided that the fluid connection of the microchip is adjusted to the interface.

The presented “world-to-chip” interface is operated with a PMMA CE microchip using conductivity detection to demonstrate long term stability of the system, characterized by the reproducibility of injection, separation and detection of long term measurements of inorganic ions.

Experimental

A schematic diagram of the system used for continuous sample introduction into a CE chip is shown in Fig. 1. This system consists of a separation (CE) chip and a system chip, both made of PMMA.

Fig. 1 Scheme of the new two-chip system. The CE separation chip and system chip are compressed by a spring device and sealed through PDMS gaskets. SP: sample port; RB: running buffer port; SW sample waste port; RBW running buffer waste port. (1) one of the four bypass channels (arrows = flow direction); (2) one of four bores for the high voltage electrodes; (3) one of four 100 µm bores for the fluid connection with the separation chip; (4) bores for spring connectors used for conductivity detection; (5) bores for alignment spring connectors (6) sample inlet, (6′) sample outlet.

The CE separation chip

The CE separation chip consists of two parts, the substrate and the cover plate. The substrate of the CE separation chip (70 × 20 × 0.5 mm) contains crossed liquid channels in the well known cross design, with a length of 60 and 10 mm, respectively, made by hot embossing. The long channel is divided by the 10 mm channel (Running Buffer, RB, Sample Waste, SW) in a 10 mm injection part (Sample Port, SP, cross) and a 50 mm separation part (cross, Running Buffer Waste, RBW). The cross section has a square geometry of 50 ± 3 µm. The substrate is bonded (by gluing) to a cover plate (3 mm thick), which contains two pairs of gold electrodes for conductivity detection crossing the separation channel (not shown in Fig. 1). These sputtered gold electrodes cross the channel 5 mm before the outlet (RBW) and are specified as follows: 150 nm thick, 23–30 µm wide, with an interelectrode spacing of 20–30 µm. As illustrated in Fig. 1 four 2 mm holes in the substrate enable the contact of the electrodes to the conductivity meter and two others enable alignment to the system chip. The liquid contact to the system chip is allowed by 100 µm bores at the end of the channels.

The system chip

The CE separation chip described above is combined with a second chip, which is designed as a “world-to-chip” interface. This so called system chip (SC) consists of a substrate (85 × 40 × 0.5 mm) with micromilled channels bonded to a cover of the same area but 10 mm thick containing all fluidic and electrical connections (Fig. 1).

As can be seen from Fig. 1 analyte and running buffer are supplied via bypass channels (four thick solid lines, for e.g. 6,6′), which are either coupled to a sampling device (via inlet 6) or to a buffer supply (three other bypass inlets). The solutions are pumped through the micromilled channels (200 µm wide and deep) with flow rates from 5 to 30 µL min⁻¹. This flux prevents formation of interfering big bubbles at each of the four HV electrodes (e.g. 2), which are positioned in a circularly widened part of each bypass channel (Pt wire d = 0.5 mm; enlarged view in Fig. 2) just at the intersection between the junction channel (100 µm wide) and the bypass channel.

Fig. 2 Enlarged view of the high voltage electrode positioning in the four bypass channels.

The junction channels end in each case in 100 µm bores as outlets (3) positioned opposite those in the CE separation chip, when both chips are compressed together by a spring device. The tightness is ensured by inserting home-made PDMS gaskets (0.4 mm thick, 0.5 mm i.d., 2 mm o.d.) positioned in an appropriate space around the 100 µm bore of the system chip.

This connection amounts to a small dead volume (<0.1 µL) but helps to adjust for any imprecision in the positioning of the bores of both chips.

Spring-mounted pins act as contacts for the conductivity electrodes and as positioning guides for the separation chip (Fig. 1, (4),(5)).

When both chips are compressed, the whole system is fluidic-tight and all fluid supplies are realised over standard HPLC-connections from the back side of the system chip.

Injection modes

As mentioned before, the two-chip system enables an electrokinetic as well as hydrodynamic injection mode. Gated injection can be performed for continuous measurements according to a CE-chip system described by Lin et al.²⁴

The hydrodynamic mode—shown in Fig. 3—could be easily realised simply by closing the bypass outlet of the analyte (Fig. 1 (6′)), thereby directing the flow into the sample port, and switching the high voltages as follows.

Fig. 3 Hydrodynamic injection modes. During Injection mode 1 only ground SW is switched to floating, separation voltage remains, while during injection mode 2 high voltage and ground are switched to floating. HV: high voltage; GND: ground. Other acronyms see Fig. 1.

During the separation mode HV electrode RB (running buffer bypass) is set to high voltage while HV electrode SW (sample waste bypass) and HV electrode RBW (running buffer waste bypass) are set to ground. The analyte is continuously pumped via sample port (SP) into the sample channel. The electroosmotic flow from RB to SW (sample waste) prevents the penetration of analyte into the separation channel. To inject a definite volume of analyte into the separation channel the electroosmotic gate flow between RB and SW is interrupted by switching SW from ground to floating (injection mode 1) for a variable but definite time interval. Another injection characteristic follows from the second option given in Fig. 3 (mode 2), when ground and high voltage are set to floating.

Hydrodynamic injection²⁷ combines the particular advantages of both commonly used electrokinetic injection modes: bias-free²⁸ pinched injection and a variable injection volume, offered by the gated injection.²⁶ In the case of hydrodynamic injection the injection volume is determined by the switching interval of the high voltage or by the flow rate of the hydrodynamically pumped analyte.

Both injection modes can easily be automated with a function generator, which injects with a definite frequency and for a definite time.

Apparatus and reagents

All reagents were analytical grade and used as delivered from Sigma-Aldrich (Germany) and Merck KgA (Darmstadt, Germany). Stock solutions were prepared by weighing the appropriate amount of MOPSO (3-[N-morpholinol])-2-hydroxy-propane sulfonic acid) and histidine or alkali chlorides, respectively. Standard solutions were obtained by dilution with buffer. For all solutions deionized water (Millipore) was used. Separation and gating voltage was generated using a home-made high voltage power supply (max. 3 kV). Hydrodynamic flows are controlled by syringe-pumps (TSE Systems Germany). Conductivity detection is performed using a home-made conductivity detector (2 kHz, 6 V amplitude).

Separation chips are delivered from Forschungszentrum Karlsruhe (Germany) and from the University of Dortmund (Germany). The system chip is made by Forschungszentrum Karlsruhe (Germany).

Results and discussion

The presented work focuses on the development of a “world-to-chip” interface, which enables microchip analysis for continuous online monitoring. In their approach Lin et al.²⁴ coupled a hydrodynamic flow to an electrophoretic microchip separation based on a sample bypass from which the sample was electrokinetically or hydrodynamically injected. As already described a real continuous analysis failed because of the chip based buffer reservoirs. This problem was solved by the two-chip concept and the consequential separation of the fluidic and high voltage supply (system chip) and analysis (CE separation chip, detection).

Long-term stability of the two-chip-system

One problem of the Lin's chip system is the open reservoirs, which influence the separation due to the changing fluid levels. This was overcome by the presented system by using a closed flow system instead of open reservoirs. An additional advantage is that an exchange of analytes with the surrounding atmosphere is prevented.

On the other hand one is faced with a severe disadvantage, which occurs when high voltage electrodes are integrated into a closed system: the formation of gas bubbles by electrolysis and hence an interference of flows and an interruption of the current. Therefore the main challenge was to prevent interferences of measurements by generated bubbles.

To meet this requirement, the four high voltage electrodes were embedded in bypass channels and flushed with an appropriate flow of buffer or sample solution.

Fig. 4a shows four electropherogramms of a continuous analysis applying an electrokinetic injection (2 s) of an ion standard solution (potassium, sodium, lithium 1 mM each) after different times. Flow rates of 10 µl min⁻¹ for the sample bypass and 30 µl min⁻¹ for the three buffer bypasses were found to be sufficient to remove small bubbles out of the system through the bypass wastes.

Fig. 4 (a) Results for a long-term analysis of a standard solution of potassium, sodium and lithium (0.5 mM of chlorides each): injection time 2 s, injection interval 100 s, separation voltage 1.0 kV, analyte voltage 1.2 kV, 5 mM MOPSO–histidine pH 6.7 as carrier electrolyte. (b) Long-term measurement of potassium (1 mM) in hydrodynamic mode, 250 nL sample flow, 0.8 kV separation voltage, 10 mM MOPSO–histidine pH 6.7 as carrier electrolyte, 0.5 s injection time.

Table 1 summarizes long term measurements up to 8 h with respect to migration times and peak heights under the same conditions as given above (Fig. 4a).

Table 1 Results of long term measurements with respect to migration time of K⁺ and peak height. Conditions see Fig. 4a

Operation time/h	Migration time/s (RSD, %); n = 10	Peak heights/MOhm (RSD, %); n = 10
1	52 (6.4)	−0.35 (6.3)
2	42 (5.4)	−0.38 (5.7)
5	46 (3.0)	−0.31 (3.1)
8	42 (2.5)	−0.33 (3.4)

---

The deviations of migration times and peak heights are mainly caused by pressure variations within the system given by bypass flows and therefore changing hydrodynamic influences on the electroosmotic flow. In addition alteration of PMMA properties of the channel surfaces can influence the EOF and can shift the migration time.

As written before the two-chip system can be switched easily from electrokinetic to hydrodynamic injection mode just by closing the sample bypass waste (Fig. 1, 6′), disconnection of the HV-electrode (SP) and reduction of the sample flow rate from 10 µL min⁻¹ in the open bypass to values between 100 and 250 nL min⁻¹ directly pumped into the sample port.

Fig. 4b shows the electropherogramms of a 1 mM potassium solution, hydrodynamically pumped and injected. It should be noted that the flow rate with which the analyte is injected into the separation channel is much lower than the sample flow rate (due to the flow resistances of the channels at the cross). Comparison of signals obtained by electrokinetic and hydrodynamic injection shows similar shapes and areas for 1 s injections, which implies similar injection volumes of a few hundred pL. Every single injection is visible by a small peak (marked as I in the figures).

The reproducibility of hydrodynamic injection corresponds also to that of electrokinetic injection as shown by the 30 minutes example in Fig. 4b. This time is by no means limited, neither by on chip reservoirs nor by bubble formation at the HV electrodes.

Reproducibility of injection and migration times

As demonstrated, the two-chip concept facilitates real continuous sampling, separation and determination of the desired species for online measurements, monitoring and controlling of processes. However, a few conditions have to be met to guarantee a good reproducibility of sample injection and migration times. Bypass flow should be sufficient to remove gas bubbles from HV electrodes but should prevent hydrostatic influence to the separation flow. In the case of hydrodynamic injection a stable electroosmotic gating flow is required to avoid sample penetration into the separation channel. From measured critical voltages by Lin et al. it can be inferred that, for flows in the range of 200 nL min⁻¹ an electric field of about 100 V cm⁻¹ is sufficient. This is always given by the coupled electric field between RB and RBW. The stability of this electric field allows good reproducibility of migration times. The reproducibility of the hydrodynamic injection obviously depends on the stability of the injection pump. In this hydrodynamic mode the two-chip system runs normally with a sample flow rate of about 200 nL min⁻¹. When the gating flow is interrupted the hydrodynamic sample flow of 3.3 nL s⁻¹ is split at the injection cross according to the flow resistances of the channels from the cross to SW, RB and RBW (separation channel). For identical cross sections the splitting ratio is given by the lengths of the channels. In the case where only SW is floating for one second, the injected volume is estimated to be 0.33 nL, which is slightly more than would be achieved by applying pinched injection. It should be noted that the injected sample is diluted by the continuous separation flow from RB to RBW. The degree of the dilution depends on the ratio of EOF and injection flow. It results in an elongation of the injection plug which can prevent the optimum separation efficiency being achieved. If this effect becomes an analytical problem, both potentials RB and SW can be set to floating. In this case the hydrodynamic injection flow is also split into channel RB and therefore the injected sample volume increases considerably under otherwise identical conditions (flow rate and gating time). On the other hand the complete shutdown of the high voltage can influence the system equilibrium and the signal shape of analytes in case of short migration times (high separation voltage).

In order to calculate reproducibilities of peak height and migration time, indicating stable potentials and sample injection volumes, a number of consecutive electrokinetic (Fig. 5a) and hydrodynamic (5b) injections were carried out. A home-made device was used to control injection period and injection frequency.

Fig. 5 Automated measurement of potassium: (a) 0.5 mM potassium, injection time 2 s, injection interval 108 s, 10 mM MOPSO–histidine pH 6.7 as carrier electrolyte, separation voltage 1 kV, analyte voltage 1.2 kV, (b) 180 nL min⁻¹ sample flow (1 mM potassium), separation voltage 0.8 kV, 1 s injection time.

Fig. 5 shows signals of an injected potassium solution (0.5 mM (a) and 1 mM (b)) over a period of 1500 s (a) and 400 s (b), respectively. From the results of these measurements relative standard deviations (RSD) for peak heights of (a) 3.1% (n = 12) and (b) 3.8% (n = 9) and migration times of (a) 0.82% (n = 12), (b) 0.40% (n = 9) were calculated.

Changing chip operation modes

In commonly used microchip CE systems the injected volume of the analytes is either given by the geometrical shape of the cross section (pinched injection) or by the floating time (gated injection). In the latter case the injected sample amount depends also on the separation voltage which controls the EOF as well as the electrophoretic injection of ions. Due to the different mobilities of ions in the electric field, the injected sample composition can differ considerably from the analyte bulk. This sample bias²⁸ can not occur using hydrodynamic injection.

The two-chip system provides the opportunity to use both injection modes alternatively. The sample is either injected electrokinetically from the sample bypass flow or hydrodynamically, when the bypass outlet is closed and a reduced sample flow is directed into the injection channel of the separation chip. Hence the sample injection volume can be controlled in two ways. One is the gating time, the other is the sample flow rate. Since the injected volume is independent of the composition, samples must not be mixed with running buffer, to provide a sufficient EOF.

Fig. 6a shows the example of an electrokinetically injected sample with varying gating times. The correlation of peak areas vs. gating time shows a linear behavior. This is not the case for peak height vs. injection time due to the conductivity electrode characteristics.

Fig. 6 (a) Electrokinetic injection of standard ion solution of 0.5 mM potassium and 0.2 mM sodium, injection times 0.5, 1.0, 1.5 s. Separation voltage 1.5 kV, analyte voltage 1.8 kV. (b) Hydrodynamic injection (1 s) of potassium (1 mM) with analyte flow rates of 0.03, 0.06 and 0.18 µL min⁻¹, separation voltage 1.2 kV, pH 6.7 as carrier electrolyte. (c) Variation of hydrodynamic injection of 1 mM LiCl in 10 mM MOPSO–histidine. Flow rate: 0.27 µL min⁻¹; separation voltage 2.2 kV.

A corresponding electropherogramm with hydrodynamic injection and constant flow rate is given in Fig. 6b. As regards to the correlation of peak height and peak area vs. injection time the same behavior was observed as for the electrokinetic injection.

If the gating time is kept constant the injection volume can be varied by changing the sample flow rate. As shown in Fig. 6c the migration time is shortened with the increasing flow rate caused by an addition of hydrodynamic pressure to the EOF and the electrophoretic migration. However, the hydrodynamic or pressure injection mode requires a very constant pumping rate. Although the peak shapes in Fig. 6c reflect more a problem of data processing than a pulsation of the pump.

Conclusions

A chip based system for long-term measurements is described, which enables continuous sample introduction for CE separation with conductivity detection, thus offering the basic principle for a fully automated chip based monitoring system. This was achieved using a two-chip concept. One part, the system chip, serves as a “world-to-chip” interface containing all necessary fluidic and electrical connections for high voltage supply and conductivity detection. The system chip is combined with an analysis chip via four vertical bore holes. Buffers and analytes are transported electrokinetically or—as an option for sample injection—hydrodynamically between the two chips. The long term stability of the two-chip system was demonstrated as well as the reproducibility of the sample injection and the migration times of cations. The two-chip concept offers an additional operating advantage since analytical chips can be easily exchanged within one minute. In future applications the system chip can be easily combined with analyses chips of various materials, designs and detection methods with the only requirement, that the position of vertical bore holes fit to each other.

Acknowledgements

Financial support from Siemens AG (Karlsruhe, Germany) is gratefully acknowledged. The authors also thank Evelyn Klammt for experimental assistance.

References

1. A. Manz, N. Graber and H. M. Widmer, Sens. Actuators B, 1990, 1, 244–248.
2. A. Manz, J. C. Fettinger, E. Verpoorte, H. Lüti and D. J. Harrison, Trends Anal. Chem., 1991, 10, 144–149.
3. A. Manz, D. J. Harrison, E. M. Verpoorte, J. C. Fettinger, A. Paulus, H. Lüdi and H. M. Widmer, J. Chromatogr., 1992, 593, 253–258.
4. D. R. Reyes, D. Jossifidis, P. A. Arnoux and A. Manz, Anal. Chem., 2002, 74, 2623–2636.
5. T. Vilkner, D. Janasek and A. Manz, Anal. Chem., 2004, 76, 3373–7386.
6. J. P. Kutter, Trends Anal. Chem., 2000, 19, 353–363.
7. S. C. Jacobson, R. Hergenröder, L. B. Koutny and J. M. Ramsey, Anal. Chem., 1994, 66, 1114–1118.
8. B. Grass, A. Neyer, M. Jöhnck, D. Siepe, F. Eisenbeiss, G. Weber and R. Hergenröder, Sens. Actuators B, 2001, 72, 249–258.
9. O. Hofmann, D. Che, K. A. Cruickshank and U. R. Müller, Anal. Chem., 1999, 71, 678–686.
10. N. J. Munro, Z. Huang, D. N. Finegold and J. P. Landers, Anal. Chem., 2000, 72, 2765–2773.
11. S. C. Jakeway and A. J. de Mello, Analyst, 2001, 126, 1505–1510.
12. X.-J. Huang, S.-L. Wang and Z.-L. Fang, Anal. Chim. Acta, 2002, 456, 167–175.
13. J. Wang, Talanta, 2002, 56, 223–231.
14. B. Graß, D. Siepe, A. Neyer and R. Hergenröder, Fresenius J. Anal. Chem., 2001, 371, 228–233.
15. N. E. Hebert, W. G. Kuhr and S. A. Brazill, Anal. Chem., 2003, 75, 3301–3307.
16. Q. Fang, F. R. Wang, S. L. Wang, S. S. Liu, S. K. Xu and Z. L. Fang, Anal. Chim. Acta, 1999, 390, 27–37.
17. C. G. Fu and Z. L. Fang, Anal. Chim. Acta, 2000, 422, 71–79.
18. Z. L. Fang and Q. Fang, Fresenius' J. Anal. Chem., 2001, 370, 978–983.
19. X. J. Huang, Q. S. Pu and Z. L. Fang, Analyst, 2001, 126, 281–284.
20. X. J. Huang, S. L. Wang and Z. L. Fang, Anal. Chim. Acta, 2002, 456, 167–175.
21. S. H. Chen, Y. H. Lin, L. Y. Wang, C. C. Lin and G. B. Lee, Anal. Chem., 2002, 74, 5146–5153.
22. Q. Fang, G. Xu and Z. Fang, Anal. Chem., 2002, 74, 1223–1231.
23. S. Attiya, A. B. Jemere, T. Tang, G. Fitzpatrick, K. Seiler, N. Chiem and D. J. Harrison, Electrophoresis, 2001, 22, 318–327.
24. Y.-H. Lin, G.-B. Lee, C.-W. Li, G.-R. Huang and S.-H. Chen, J. Chromatogr. A, 2001, 937, 115–125.
25. D. Solignac and M. A. M. Gijs, Anal. Chem., 2003, 75, 1652–1657.
26. S. V. Ermakov, S. C. Jacobson and J. M. Ramsey, Anal. Chem., 2000, 72, 3512–3517.
27. U. Backofen, F.-M. Matysik and C. E. Lunte, Anal. Chem., 2002, 74, 4054–4059.
28. B. E. Slentz, N. A. Penner and F. Regnier, Anal. Chem., 2002, 74, 4835–4840.

---