INSTITUTE OF PHYSICS PUBLISHING

JOURNAL OF MICROMECHANICS AND MICROENGINEERING doi:10.1088/0960-1317/15/1/002

J. Micromech. Microeng. 15 (2005) 6–10

PMMA-based capillary electrophoresis electrochemical detection microchip fabrication Ray-Hua Horng1, Pin Han1, Hung-Yu Chen1, Kuan-Wen Lin2, Tung-Mung Tsai2 and Jyh-Myng Zen2 1 Institute of Precision Engineering, National Chung Hsing University, Taichung, Taiwan 402, Republic of China 2 Department of Chemistry, National Chung Hsing University, Taichung, Taiwan 402, Republic of China

E-mail: [email protected]

Received 6 July 2004, in final form 23 August 2004 Published 23 September 2004 Online at stacks.iop.org/JMM/15/6 Abstract In this paper, a 50 µm (depth) × 50 µm (width) microfluidic channel is made on a poly(methyl methacrylate) (PMMA) substrate using thick photoresist. Openings were drilled for buffer reservoirs on an additional piece of PMMA. A final PMMA/patterned photoresist/PMMA sandwich configuration was completed using a bonding process. The thick photoresist was used as the adhesion layer and also as the microfluidic system. Using screen-printed technology for carbon and silver electrode fabrication, the microchip electrophoretic device functions were demonstrated. Successful detection of uric acid and L-ascorbic acid (the main components in human urine) validates the functionality of the proposed system. Successful ascorbic and uric acid separation in a sample from a urine donor who had consumed 500 mg of vitamins verified the proposed biochip. (Some figures in this article are in colour only in the electronic version)

1. Introduction Microfluidics (the manipulation of liquids and gases in channels) have become viable platforms for performing analytical tasks typically done with bench top instruments [1–3]. Conventional methods for fabricating microfluidic devices have focused mainly on etching quartz [4] and silicon [5] to form a replica molding of the microfluidic device. This is followed by molding using polymers, such as poly(methyl methacrylate) (PMMA), polycarbonate, polysthylene, or poly (dimethylsiloxane) (PDMS) [6–9]. Molding provides a replica that contains three of the four walls necessary for enclosed channels. Sealing the replica to a flat surface provides the fourth wall. This flat material can be PDMS, producing channels in which all four walls are made from the same material, or another comparable material can be used. The main drawback of conventional replication methods is that the etching in Si and glass is expensive and time0960-1317/05/010006+05$30.00

consuming. Recently, polymer-based microfabricated devices have increasingly been developed instead of Si- and glassbased replica technique. The main advantages for polymerbased systems are the good material compatibility with chemical and biochemical assays and low cost. Especially, some polymer material can be made for photoresist; that is, the polymer consists of photo-sensitizer. This can directly form the micro-channel after the photolithography process [10, 11]. On the other hand, fabricating detection electrodes inside a micro-channel is difficult because the detection electrodes must be inserted into the micro-channel without a gap between the channel and the substrate. A sealing problem exists that could allow the micro-channel to leak the test sample solution. In this paper another good impermeable micro-channel and detection electrode contact will be proposed. Inexpensive PMMA substrates are used to sandwich the patterned thick photoresist micro-channel using simple fabrication processes for microchip electrophoresis

© 2005 IOP Publishing Ltd Printed in the UK

6

PMMA-based capillary electrophoresis electrochemical detection microchip fabrication

applications. Here, the thick photoresist is a commercially available polymer from JSR Corporation. Because the photoresist material has liquid-like behavior before processing, it can conform to the detecting electrodes. After the photolithography process, the patterned photoresist is used as the micro-channel and the sealing material. Note that this is the first report of using a photoresist layer as part of the microfludic channel. Microchannel fabrication using photoresist provides better control of channel dimension than molding fabrication [12]. Because the photoresist layer is part of the microfluidic channel, it is essential that the surface properties of photoresist after exposure to UV light remain similar to PMMA. In this way the hydrodynamics of sample flow through the micro-channel can easily be calculated [13, 14]. More importantly, the patterned photoresist micro-channel shape in the PMMA/patterned photoresist/PMMA structure should remain unchanged after the bonding process. This latter concern is confirmed true through SEM measurement. Screen-printing fabrication technology has been shown to offer the possibility of large-scale and inexpensive mass production of a capillary electrophoresis detector [15]. For successful electrochemical detection, the presented microchip design alleviates the need to realign the working electrode with the separation electrode with a separate extrinsic reference electrode or counter electrode. After drilling holes for buffer reservoirs in another piece of PMMA, a final device having a PMMA/photoresist/PMMA sandwich configuration was completed by a bonding process. This is as effective as a UVglue bonding process for microchip as reported by Landers and co-workers [16]. The coupling of low-cost PMMA-based microchips with mass-produced screen-printed technology holds great promise for creating single-use disposable capillary electrophoresis electrochemical microsystems for a wide range of applications.

2. Experimental details The fabrication flowchart is shown in figure 1. In this study, one mask to pattern the micro-channel was required as shown in figure 2. The PMMA substrate was cleaned first using ethyl alcohol. The carbon and silver electrodes were then screen-printed onto the PMMA substrate and baked in an oven at 100 ◦ C for 30 min. A three-electrode system (W, C, R in figure 3(a)) is used as the detection electrodes. The Ag electrode (R) was treated with 15% H2O2 for 15 min followed by 3 M KCl for 15 min to form stable AgCl as the reference electrode. A 50 µm thick photoresist was spun onto the PMMA substrate and soft baked at 90 ◦ C for 7 min. A 50 × 50 µm2 microfluidic system was formed using a photolithographic process. In photolithography, the wavelength and power of the light source (mercury– xenon lamp) used for exposure of the photoresist-coat substrate are 436 nm and 500 W, respectively. The patterned photoresist was then hard baked at 100 ◦ C for 10 min to strengthen the microfluidic structure. The upper PMMA layer has four openings aligned with the ends of the micro-channels on the lower PMMA layer housing the screen-printed three-electrode system. The lower PMMA (the one with patterned photoresist) was then bonded onto

Cleaning PMMA substrate

Screen-printed carbon and silver electrodes

Making the Ag/AgCl reference electrode

Spin coating thick photoresist on PMMA

Exposing and developing the thick photoresist

Forming micro-channels by photolithography process

Strengthen the micro-channel structure by hard baking

Bonding the lower PMMA with the upper one to form the complete set Figure 1. PMMA/patterned photoresist/PMMA microchip fabrication process flowchart.

Figure 2. Micro-channel device photomask.

the upper layer using a bonding machine. The bonding pressure was about 15 Psi. The bonding temperature was 120 ◦ C. A microfluidic system with three detection electrodes was obtained. Figure 3(a) shows the capillary electrophoresis microchip layout for electrochemical detection. As illustrated in figure 3(a), I represents the high voltage (HV) for injection, in which I1 is the HV input and I2 is the ground. S represents the 7

R-H Horng et al

(a )

S2

I1

C

r1 r3

r4

O

S1

r2

I2 (b )

W

R

10 cm

5 cm (b) Figure 2.

Figure 4.

Figure 3. Microchip device with electrochemical detector. (a) S represents the high voltage (HV) for separation, in which S1 is the HV input and S2 is the ground. I represents the high voltage (HV) for injection, in which I1 is the HV input and I2 is the ground. W is the working electrode (carbon). C is the counter electrode (carbon). R is the reference electrode (Ag/AgCl). r1, r2, r3 and r4 are the reservoirs, in which r1 r2 is the injection channel and r3 r4 is the separation channel. (b) Photograph of the real sample.

high voltage (HV) for separation, in which S1 is the HV input and S2 is the ground. W is the working electrode (carbon). C is the counter electrode (carbon) and R is the reference electrode (Ag/AgCl). Similarly to the preparation of working electrodes on capillary electrophoresis microchips, S2 also serves as the decoupler. During electrophoretic separation, the high electric field is sufficient for the electrolysis of water, which may cause gas evolution at the ground electrode and interfere with the electrochemical signal. For the decoupler onto which hydrogen ions are reduced and adsorbed, the unique property is that hydrogen can diffuse relatively fast. Therefore, before gas bubbles can develop, hydrogen is removed from the decoupler by the solution due to the electroosmotic flow. Taking advantage of the decoupler, single or multiple working electrodes can be placed across the separation channel. The problem of gas bubbles in the channel disrupting the process does not exist in this three-electrode system. When the microchip is in use, the reservoir r1 is filled with the sample being tested. The reservoirs r2, r3 and r4 are filled with a running buffer solution. The injection portion of the microchannel length is 2 cm (r1r2) and the separation channel (r3r4) has a 5 cm effective separation length. The size of the entire microchip assembly is 5 cm × 10 cm. A photograph of the microchip with a sample is shown in figure 3(b). 8

Figure 4. SEM micrograph of the patterned photoresist material observed from the channel intersection (as indicated). The device is shown before the bonding process.

The proposed microfludic system was evaluated by detecting uric acid and L-ascorbic acid. Before detection, the chip channels were rinsed with de-ionized water (resistivity > 18 M
cm) and buffer solution for 30 min. A running buffer solution, pH 7.0, was prepared by mixing 0.01 M Na2HPO4 and 0.01 M H3PO4. The test uric acid and L-ascorbic acid standard solution samples (10 mM) were prepared using DI water and diluted by running buffer to the required concentration. To evaluate the detection function, reservoirs r2, r3 and r4 were filled with the running buffer solution. Reservoir r1 was then filled with the analytical sample. To make the sample run through the injection channel r1r2, 100 V cm−1 electric field intensity was applied to I1 with I2 grounded for 20 s. The applied electric voltage to S1 and ground to S2 produce an electric field causing the central part of the sample to run into the separation channel Or4 . These two samples were separated in the micro-channel by electrophoresis.

3. Results and discussion Because the micro-channel is formed after the thick photoresist is photolithographed, it is important to observe the developing pattern to make sure whether the developing pattern is the same as the designed shape or not. Figure 4 shows the patterned photoresist microfludic channels before bonding examined by scanning electron microscope (SEM). The measured

PMMA-based capillary electrophoresis electrochemical detection microchip fabrication

(b )

(a )

Figure 5. Photomicrographs of the micro-channels after the bonding process. (a) Micro-channel intersection. (b) The micro-channel near the reservoirs.

dimensions (depth × width) 50 × 50 µm2 are nearly consistent with the designed dimensions. The microchip vertical side walls have a smooth surface and sharp pattern that confirms the reliability of the patterned photoresist for fabricating microchannels. The final PMMA/patterned photoresist/PMMA structure microfluidic channel shape is formed after the bonding process, even though the pattern is defined by the photo mask before the bonding process. The bonding process (with some pressure and temperature) could deform the channel structure. After carefully examining the final product by optical microscope, we observed that the main channel body and cross section were complete and intact after the bonding process, as shown in figure 5(a). However, the channel end shape near the reservoirs (r1–r4) was too narrow or closed, as shown in figure 5(b). This will affect the performance or negate the function of the microchip system. To overcome this problem, the photomask shape at these points was modified as shown in figure 6(a). The good microfluidic channel shape was maintained (shown in figure 6(b)) and the fluid was unhindered in the channel in the test. Therefore a stable micro-channel system was obtained after the bonding process. The microchip was subsequently used to detect uric acid and L-ascorbic acid with an electrochemical method. The detection potential at the W electrode was +0.7 V with respect to the R electrode (Ag/AgCl). Various electrical field intensities (200 V cm−1, 100 V cm−1 and 50 V cm−1) were applied to the sample (S1) reservoirs with S2 grounded. The experimental results are shown in figure 7. Good sample separation signals were achieved within a short time. The proposed analysis applications were successful. The detecting time decreased as the applied voltage was increased. However, a shortened detection time will result in lowered resolution. The appropriate applied electric field was determined as 100 V cm−1. To further characterize the performance of the proposed microchip, a 10-fold buffer diluted urine sample from a human

(a )

(b )

Figure 6. (a) Modified microfluidic channel photomask. (b) Photomicrograph of the micro-channel near the reservoirs after the bonding process fabricated with the modified photomask.

(who had consumed 500 mg of vitamin C) was performed. The sampling conditions were the same as above. The electropherogram result is shown in figure 8. The ascorbic acid was detected about 30 s after the uric acid. The uric and ascorbic acid signals in the urine sample were clearly detected. 9

R-H Horng et al

4. Conclusions (a )

A novel method for fabricating a high performance electrophoresis microchip was proposed in this study. The microchip was made of patterned photoresist material sandwiched between PMMA substrates. The patterned photoresist was used as the microfluidic channel and as the adhesive layer that bonded the PMMA substrate to the other PMMA substrate. In combination with screen-printed photolithography and wafer bonding techniques, electrode insertion was easily achieved without using a molding process. The microfluidic structure depth can easily be controlled using thick photoresist. The channel pattern can be changed as needed using the appropriate photomask design. Screenprinted electrodes can easily be fabricated and modified on a smooth PMMA substrate. The advantages of the proposed fabrication process are: inexpensive (thus disposable), simple and fast acting. Other topics worth studying in the future include choosing different photoresist materials to fabricate the micro-channels, using other materials to make the detection electrodes and different channel pattern shapes to achieve various experimental results.

L-ascorbic acid uric acid

Signal (a.u.)

(b )

L-ascorbic acid

uric acid

(c )

L-ascorbic acid

References

uric acid 0

30

60

90

120

150

180

210

240

Time (s)

Figure 7. Electropherograms of uric and L-ascorbic acid samples with different separation electrical fields. The electrical fields are: (a) 200 V cm−1, (b) 100 V cm−1 and (c) 50 V cm−1.

2 0 -2

ascorbic acid

-4

Current (nA)

-6 -8 -10 -12 -14 -16

uric acid

-18 -20

0

15

30

45

60

75

90 105 120 135 150 165

Time (s)

Figure 8. Electropherogram of a 10-fold buffer diluted urine sample from a human (who had consumed 500 mg of vitamin C). The separation electric field is 100 V cm−1.

10

[1] Kutter J P 2000 Trends Anal. Chem. 19 352–63 [2] Nashat A H, Moronne M and Ferrai M 1998 Biotechnol. Bioeng. 60 137–46 [3] Harrison D J, Fluri K, Seiler K, Fan Z, Effenhauser C S and Manz A 1993 Science 261 895–7 [4] Dolnk V, Liu S and Jovanovich S 2000 Electrophoresis 21 41–54 [5] Hong J W, Fuji T, Seki M, Yamamoto T and Endo I 2001 Electrophoresis 22 328–33 [6] Martynova L, Locascio L E, Gaitan M, Kramer G K, Christensen R G and MacCrehan W A 1997 Anal. Chem. 69 4783–9 [7] Larsson O, Ohman O, Billmadn A, Lundbladh L, Lindell C and Palmskog G 1997 Transducers’97 9th Int. Conf. Solid-State Sensors and Actuators (Chicago, IL) pp 1415–8 [8] Robert M A, Rossier J S, Bercier P and Girault H 1997 Anal. Chem. 60 2035–42 [9] McDonald J C, Duffy D C, Anderson J R, Chiu D T, Wu H, Schueller O J and Whitesides G M 2000 Electrophoresis 21 27–40 [10] Carlier J, Arscott S, Thomy V, Fourrier J C, Caron F, Camart J C, Druon C and Tabourier P 2004 J. Micromech. Microeng. 14 619–24 [11] El-Ali J, Perch-Nielsen I R, Poulsen C R, Bang D D, Telleman P and Wolff A 2004 Sensors Actuators A 110 3–10 [12] Chen Z, Gao Y, Su R, Li C and Lin J 2003 Electrophoresis 24 3246–53 [13] Nicholas P C S, Qiu F, Gooch K A and Fisher A C 2000 J. Phys. Chem. B 104 7110–4 [14] Fu L-M, Yang R-J and Lee G-B 2002 Electrophoresis 23 602–12 [15] Wang J, Pumera M, Chatrathi M P, Escarpa A, Konrad R, Griebel A, D¨orner W and L¨owe H 2002 Electrophoresis 23 596–601 [16] Huang Z, Sanders J C, Dunsmor C, Ahmadzadeh H and Landers J P 2001 Electrophoresis 22 3924–9