Quantitatively controlled nanoliter liquid manipulation using ...

INSTITUTE OF PHYSICS PUBLISHING

JOURNAL OF MICROMECHANICS AND MICROENGINEERING

J. Micromech. Microeng. 13 (2003) 89–97

PII: S0960-1317(03)37800-3

Quantitatively controlled nanoliter liquid manipulation using hydrophobic valving and control of surface wettability Sang-Ho Lee1, Chang-Soo Lee2, Byung-Gee Kim2 and Yong-Kweon Kim1,3 1

School of Electrical Engineering and Computer Science, Seoul National University, ENG 420-007, Seoul 151-742, Korea 2 School of Chemical Engineering, Seoul National University, Seoul 151-742, Korea E-mail: [email protected]

Received 7 June 2002, in final form 4 November 2002 Published 4 December 2002 Online at stacks.iop.org/JMM/13/89 Abstract In this paper, we discuss nanoliter metering, transportation, merging, and biochemical reaction on a nanoliter fluidic chip. The proposed nanoliter fluidic handling is based on passive fluidic control using hydrophobic valving and liquid flow control by selective wettability patterning. For the selective patterning of the wettability, a hydrophobic fluorocarbon (FC) film (a mixture of FluoradTM from 3M, USA) was spin-coated on a hydrophilic glass wafer and patterned using a lift-off process. The patterned FC film showed strong hydrophobicity, indicated by a high water contact angle (108–110◦ ). For the fabrication of the nanoliter fluidic chip, polydimethysiloxane microchannel networks were aligned and bonded onto the glass wafer, along with the FC patterns. The proposed nanoliter metering technique showed an accuracy of 4% using a 20 nL criterion. A biochemical reaction on a chip was performed between β-galactosiadase (enzyme) and a fluorescein di-β-galactopyranoside (FDG) substrate. The FDG and the enzyme solution were manipulated on the nanoliter fluidic chip in desired volumes and mixing ratios. The reactions between different ratios of the enzyme and FDG substrate were monitored using the fluorescence intensity as a function of the reaction time. From the enzymatic reaction on the chip, we confirmed that the proposed fluidic handling was successfully performed on the chip, and that the reaction depended on the enzyme concentration.

1. Introduction Lab-on-a-chip (LOC) devices that perform multiple reactions, both in series and in parallel, have been realized in microfabricated devices for biochemical analysis [1, 2]. Passive fluidic manipulation is considered an important strategy in nano/microliter fluidic handling on an LOC. This method can offer several advantages, such as having no active fluidic component, having a minimal control system, ease of fabrication, and having low power consumption [3–5]. Passive fluidic control can perform its discrete fluidic operations in 3

Author to whom any correspondence should be addressed.

0960-1317/03/010089+09$30.00 © 2003 IOP Publishing Ltd

microchannel networks by employing a hydrophobic surface and an external pneumatic control. In this case, hydrophobic surfaces are used as a valve to control the flow direction of a fluid reagent [3–7]. In our study, in the fluidics design we also employed passive fluidics using the hydrophobic surface and pneumatic control. Major operations in LOCs for biochemical analysis involve the preparation of chemical or biochemical reagents, and their separation and detection. The sample preparation step includes various fluidic operations, such as injection, metering, mixing, and reaction [2, 3]. A precise quantitative liquid handling technique for biochemical analysis is one of the essential fluidic operations

Printed in the UK

89

S-H Lee et al

on a chip. Several researchers have reported on metering techniques using a hydrophobic surface and pneumatic control to obtain liquids in a desired volume for given reactions [3–7]. However, nano/microliter fluidic devices developed during the course of these research projects have had limited metering, because the volume was commonly fixed in advance, according to the dimensions of the microchannel or reservoir, which were fixed by the requirements of a particular reaction. If a sample reagent could be manipulated in a desired volume in a microchannel, then biochemical analysis using passive fluidic control will confer added advantages, such as the ability to vary the mixing ratios, and carry out diverse kinetic studies on a single chip. Recently, we have developed a variable metering technique that can meter biochemical reagents in a desired volume on a fluidic chip. We have also developed a simple and effective hydrophobic surface preparation method using a spin-coated fluorocarbon film and lift-off process to control the surface wettability for nanoliter fluidic handling. Using the selective preparation of hydrophobic patterns on a hydrophilic surface, we can manipulate the flow direction to a given region in a microchannel network [8]. Enzymatic reactions are used in the diagnosis of many diseases, as amplifiers in immunoassays, and in the detection of gene markers [9, 10]. To test the feasibility of the proposed nanoliter manipulation technique, we have carried out known enzyme reactions and product detection experiments on a chip. The biochemical reaction on the chip was between the enzyme β-galactosiadase and a fluorescein di-β-galactopyranoside (FDG) substrate. The enzyme and the FDG were metered using a desired volume in a proposed variable metering technique. After the FDG substrate had reacted with the enzyme in different volume ratios, the in situ monitoring of the fluorescence intensity was carried out as a function of the reaction time to determine the reactivity between the FDG and the enzyme.

2. Materials, instruments and microfabrication 2.1. Chemicals and analytical instruments The β-galactosiadase was purchased from Roche Biochemicals (Basel, Switzerland). The FDG and phosphate buffer were purchased from Sigma Chemicals (St Louis, USA). The buffer solution used in the experiments was 50 mM potassium phosphate buffer at pH = 7.5. The FDG solution was freshly prepared before each experiment. Fluorescence images were obtained using a Zeiss LSM5 Pascal confocal microscope. Samples were scanned using the 488 nm wavelength from an Ar laser. To measure a pattern area in the image, we used the Kontron Elektronik KS Lite image analysis program v2.0. A Kr¨uss G10 contact angle analyzer was used to measure the water contact angles of the sample surface. A PSIA Corp. (Korea) auto probe M5 atomic force microscope (AFM) was used to observe the surface morphology of the samples. The surface scans were performed using a Si3N4 tip in non-contact mode. The applied pressure was measured using a Konics Co (Korea) PT-3300 electronic pressure transmitter, with an accuracy of ±0.3% FS. A Kyence Inc. (Japan) VF-7510 profile micrometer was used to obtain optical micrographs of the prepared samples. 90

N-PMER

AZ 4330 silicon glass

(e)

Metal coupler

(a) glass

silicon

(b)

(f) PDMS

glass (c)

silicon (g)

glass with FC patterns

PDMS

(d)

(h)

PDMS A

glass with FC patterns (i) PR

PDMS

FC film

Al

Figure 1. The fabrication process flow of a nanoliter fluidic chip for a biochemical reaction: (a) PR patterning and Al wet etching in a phosphoric acid based solution; (b) PR patterning for the definition of a hydrophobic region; (c) spin-coating of an FC solution and baking in a convection oven; (d) FC patterning using the lift-off process in a sonicated acetone solution; (e) replica master mold fabrication; (f ) aligning and adhering metal couplers for connection hole formation; (g) the PDMS pouring and curing process; (h) peeling the PDMS replica from the master mold; and (i) alignment and bonding of the PDMS replica microchannels on the glass wafer containing the hydrophobic patterns. (Note that all the three-dimensional illustrations in this paper show only cross section ‘A’ of (i), and the remnant upper cross section of PDMS is not shown.)

2.2. Fabrication of the nanoliter fluidic chip The nanoliter fluidic chip was fabricated by bonding polydimethysiloxane (PDMS) microchannels on a glass substrate in a hydrophobic pattern. A hydrophobic fluorocarbon (FC) film patterning technique was developed to control the surface wettability. Figures 1(a)–(d) show the process flow for FC film patterning ˚ aluminum film using the lift-off method. First, a 2000 A was thermally evaporated and patterned using a phosphoric acid-based etchant and a photoresist (PR) mask (figure 1(a)). The defined aluminum patterns were used as an alignment mark during chip assembly; otherwise, the alignment and assembly process would be very laborious, because the FC films and PDMS microstructures are both transparent. To define the hydrophobic surface pattern, a 3 µm thick film of AZ 4330 positive photoresist (Clariant Co, USA) was spincoated onto a glass wafer and patterned using photolithography (figure 1(b)). Then, a mixture of FluoradTM FC 722 and FC 40 (3M, USA) [11] was spin-coated onto the glass wafer along with the photoresist patterns, and then baked in a convection oven at 110 ◦ C for 10 min (figure 1(c)). After baking, any excess FC film and PR patterns were removed using an acetone solution in the lift-off process (figure 1(d)). This removal step was performed in an ultrasonic bath employing water as the contact liquid. In the lift-off patterning test,

Quantitatively controlled nanoliter liquid manipulation

FC surface Glass surface

Al pattern FC surface

Glass surface (b)

(a)

Figure 2. An FC film patterned using the lift-off process: (a) a negative patterning test result; (b) a positive FC pattern on a glass wafer including the aluminum patterns underneath. The FC pattern is the transparent boomerang-shaped film enclosed by the dotted line. Air venting channel 3 Pneumatic control channel 1

PDMS channel network Metering region

Injection channel 2

Air venting channel 4

Hydrophilic glass

Reaction region Scale bar

Hydrophobic FC patterns (a)

Air venting channel 3 Air venting channel 4 Pneumatic control channel 1 Injection channel 2

(b)

Figure 3. The final nanoliter manipulation fluidic chip: (a) a schematic illustration; (b) the fabricated nanoliter fluidic chip. (To obtain a clear photograph, the nanoliter fluidic chip was fabricated on an opaque silicon substrate. However, we used a nanoliter fluidic chip fabricated on transparent glass in the fluidic manipulation experiments and in the enzymatic reaction study.)

we could obtain 10 × 10 µm2 square negative patterns as a minimum size, as shown in figure 2(a). Figure 2(b) shows a boomerang-shaped positive hydrophobic FC film on a silicon wafer that has aluminum patterns underneath. To create microchannel networks on a glass wafer with FC patterns, PDMS microchannel networks were fabricated via a replica molding process according to the fabrication process flow chart shown in figures 1(e)–(h). A master mold was fabricated using a 50 µm thick N-PMER negative photoresist (TOK Co., Japan) (figure 1(e)). To facilitate the release of the cured PDMS from the master mold, the FC thin film was deposited on the surface of the master mold in C4F8 gas plasma using a Plasma-Therm Inc. (USA) SLR inductively coupled plasma (ICP) reactor. Metal couplers were pasted by

brushing photoresist onto the region to form tube connection holes (figure 1(f )). For the replica mold formation process, a mixture of PDMS prepolymer and Sylgard 184 curing agent (Dow Corning, USA) was thoroughly stirred and then degassed in a vacuum oven. The degassed PDMS mixture was poured onto the silicon master mold and cured for 4 h at 60 ◦ C [4, 12] (figure 1(g)). After curing, the PDMS replica was peeled away from the master mold (figure 1(h)). The PDMS structure was then manually aligned and bonded to the glass wafer and the FC patterns using aluminum and the PDMS microchannel patterns (figure 1(i)). The alignment was checked using an optical microscope linked to a laptop computer. Figure 3 shows the completed microchip (28.3 × 28.3 mm2), along with a schematic illustration that describes 91

S-H Lee et al

CCD

PC

Fluidic chip

Microscope

Sample and air loading system

Sample X-Y stage (a) Syringes for pneumatic control

Spring Piston

Micrometer

whereas a high contact angle (>90◦ ) implies a poor wettability and a low surface energy [13, 14]. Therefore, to characterize the surface, the water contact angles on the prepared surfaces were measured using a contact angle analyzer. Contact angle analysis is a simple but sensitive method that measures the changes in surfaces at the monolayer level [11, 12]. From the contact angle measurements, the PDMS surface was estimated to have a fairly hydrophobic nature in that it exhibited a high water contact angle of 110–113◦ . The lift-off process developed for the patterning of the FC films can give rise to mechanical damage to the FC film surface from the sonication used. Therefore, using AFM analysis and contact angle analysis, we characterized the surface of the FC film both before and after the lift-off process using sonication. In the AFM analysis, the FC films after the lift-off process showed an increase in the root mean square roughness ˚ when compared to the (Rrms) from Rrms = 4.42 to 21.7 A, FC film surface before the lift-off process. The increase in roughness may imply that the FC film was damaged by sonication. However, we have confirmed that the FC film surface maintained its high hydrophobic nature, indicated by the high water contact angle of 108–110◦ after sonication.

4. Nanoliter liquid manipulations Syringes for sample loading (b) Figure 4. Schematic illustrations of (a) the entire experimental setup, and (b) the apparatus for liquid sample injection and pneumatic control.

each part of the designed microchip. As shown in figure 3(a), the fabricated fluidic chip included two metering regions and a reaction region. The scale bars were used to measure that the metered droplet was the same as the desired volume. 2.3. Apparatus for sample injection and pneumatic control Figure 4(a) shows a schematic illustration of the entire experimental setup. The apparatus was designed to execute liquid sample injection and pneumatic control using an external air pressure. As shown in figure 4(b), the apparatus consisted of plastic syringes, springs, a micrometer, and acrylic plastic housing. The rotational motion of the micrometer was converted into the linear motion of a piston and, thereby, R the sample was either injected or withdrawn through Tygon microbore tubing plastic tubes (SAINT-GOBAIN PPL Corp.), which were connected to the microchip. The external air was supplied by pushing the piston of the syringe.

3. Surface characterization of patterned FC films and PDMS In the design of the fluidic chip, we had to investigate the surface properties of PDMS and the FC film. The wettability of a liquid is related to the contact angle of a droplet of the liquid on a solid surface. A low contact angle implies a high wettability (or a hydrophilic nature) and a high surface energy, 92

4.1. Metering using a T-shaped microinjector with a sidewall microchannel array The capillary pressure of liquid in a microchannel can be expressed by equation (1) showing the relationship between the work performed by the pressure (dUp), the surface free energy of the system (dUs), and Young’s equation [3–7]: 
 w  2h + w + γ cos θGlass or FC P = γ cos θPDMS . (1) hw hw Here, w is the width, h is the height, θ is the contact angle, and γ is the surface tension of H2O (0.073 N m−1). A positive pressure indicates capillary filling, and a negative pressure indicates a repellent force at the solid–liquid–gas interface. A microchannel is maintained in the ‘valve-close (OFF)’ state if an external pressure is less than the theoretical negative pressure resulting from equation (1). To turn the microchannel to a ‘valve-open (ON)’ state against the liquid flow, the external pressure has to be higher than the theoretical pressure needed to push the fluid in a desired direction. In this study, the capillary pressure in the microchannels was designed according to equation (1). The difference in capillary pressure created the hydrophobic valving or capillary filling, which was used to manipulate the liquid in the microchannel network. A prototype design for nanoliter metering used a T-shaped injector, as shown in figure 5(a). However, the injected liquid rapidly escaped on both sides of the metering microchannel, so that it was hard to perform any metering and fluidic manipulation using this type of T-shaped injector, as shown in figure 5(b). From this experiment, we assumed that the inability to control the outflow of the liquid was caused by an abrupt volume expansion in the microchannel. When the liquid is injected through opening ‘A’ as shown in figure 5(a) using the T-shaped microinjector, then the liquid

Quantitatively controlled nanoliter liquid manipulation Hydrophobic PDMS Hydrophobic FC film

Metering channel

Injection channel A Hydrophilic glass wafer Liquid injection (a)

(b)

Figure 5. Behavior of the liquid flow in a T-shaped microinjector: (a) schematic illustration of the T-shaped injector; (b) non-controllable out-going liquid flow caused by an abrupt volume expansion.

pressure can accumulate until liquid flows into the metering channel, overcoming the negative capillary pressure in opening ‘A’. The accumulated liquid pressure inside the microchannel can result in an abrupt volume expansion, and this abrupt volume expansion induces the uncontrollable outflow of liquid. Moreover, the liquid volume expansion may result in a sudden increase in the surrounding air pressure at the liquid–solid–gas interface system. The modified design was for the situation where the liquid droplet grows at a stable rate at the end of the syringe needle in free open space. If more additional pathways for airflow are formed to diffuse the surrounding air pressure in the T-shaped microinjector, then the rate of increase in pressure inside the microchannel will be retarded. In this case, an abrupt expansion in the liquid volume will be relieved, and the liquid droplet will grow in a stable manner in the microinjector. In the modified T-shaped microinjector, a sidewall microchannel array (with unit channel dimensions of 20 (w) × 50 (h) µm2 was devised, to provide the additional pathways to the airflow, similar to the case of an open space liquid dispensing mechanism using a syringe needle. As shown in figure 6(a), the sidewall microchannel array provides the airflow with an additional pathway in the Y-direction of the metering channel and in the ±X-direction. Figure 6(b) shows the stable liquid growth achieved without an abrupt volume expansion in the T-shaped microinjector when using the sidewall microchannel array. When liquid was injected through the injection channel, we applied an external pressure of 1.2 kPa to the liquid, which was higher than the minimum pressure of 1.19 kPa calculated using equation (1) for θ PDMS or θ FC = 110◦ and a crosssectional area of the microchannel of 260 (w) × 50 (h) µm2. The sidewall microchannel array enables a ‘valve-close (OFF)’ state against a pressurized liquid flow, and a ‘valve-open (ON)’ state against the airflow. When the contact angle (θ PDMS or θ FC) was 110◦ for a geometry with unit channel dimensions of 20 (w) × 50 (h) µm2, the theoretical pressure was calculated to be –3.5 kPa from equation (1). The pressurized liquid could not stream out through the sidewall microchannel array if the inside pressure was less than 3.5 kPa. As the liquid droplet grew in the metering channel, we could measure the length of the liquid corresponding to a desired volume using the scale bars on the chip. The measurement was performed via the control micrometer of the sample loading system. After the measurement was complete, and the liquid injection had stopped, the measured liquid droplet did not move in any direction while forming the immobile boundary A–A and C–C , as shown in figure 6(c). A negative pressure of

−1.14 kPa exited at the boundary of A–A and C–C in the metering channel for θ PDMS or θ FC = 110◦ and a cross-sectional area of the microchannel of 340 (w) × 50 (h) µm2 after the liquid injection stopped. Therefore, the liquid could not flow as long as there was no pressure applied above 1.14 kPa after the liquid injection stopped. Asymmetric liquid growth occurred, which was thought to be attributable to the asymmetric conductance of the microchannel. In our design, airflow was more favorable to the right-hand side than the lefthand side connection into the pneumatic tube line, as shown in figure 3(b). Figures 6(d ) and (e) show the cutting step formed by external air pressure injection. Until the liquid droplet had been thoroughly cut, the immobile line A–A hardly moved, and the liquid volume corresponding to B –C was returned to the liquid injection channel. The convex and concave meniscus at A–A and B–B of figure 6(d) can be understood by contact angle hysteresis. The contact angle hysteresis is defined by the difference between advancing and receding angles and it is a function of the surface polarity, heterogeneity and surface roughness [13, 15]. In the contact angle measurements using captive drop method, the advancing angles of the PDMS and FC film surfaces were within the range of 112.17–116.82◦ and the receding angles were within the range of 76.20– 82.80◦ . The resulting contact angle hysteresis was around 35◦ . The measured contact angle hysteresis supports the fact that the liquid droplet driven by air pressure includes the inevitable convex or concave meniscus at the solid–liquid–gas interface of the fabricated microchannels. In the measuring and cutting step, we found that the metered volume was determined by the length A –B , regardless of the length of B –C in figure 6(c). Figure 6(f ) shows the final metered liquid droplet using a criterion of 20 nL. After the external air injection had stopped, the liquid droplet was maintained in an immobile state, and the return liquid was caught in the injection channel with the hydrophilic bottom surface. To investigate the accuracy of the proposed metering, the microscope images of the final metered droplet were captured using the experimental setup shown in figure 4. The images were captured and manipulated using the Kontron KS Lite image analysis software to determine the droplet area. The measured area was converted into a volume by multiplying the value by the microchannel depth, 50 µm. In 30 metering runs using a criterion of 20 nL, the accuracy of the metering showed about a 4% error (mean = 20.18 nL, standard deviation = 0.74 nL). 93

S-H Lee et al

Hydrophobic PDMS Sidewall microchannel array (Hydrophobic valve)

Hydrophobic FC pattern

Sidewall microchannel array Hydrophobic valve open to airflow and closed to liquid flow

Metering channel Airflow direction

Injection channel y

Hydrophilic glass -x

+x Liquid injection (b)

(a) C

A

B

A

External air injection

C´

A´

B´

A´

B´ Liquid injection stop

Liquid injection stop

(d)

(c) A Air injection

External air injection stop

A´ Liquid injection stop

Figure 3

Hydrophilic bottom surface Liquid injection stop

(e)

20 nL

(f)

Figure 6. Liquid droplet growth and metering in a T-shaped microinjector with a sidewall microchannel array: (a) schematic illustration of the T-shaped microinjector with a sidewall microchannel array; (b) initial droplet growth; (c) measurement using scale bars; (d), (e) cutting by external air injection; (f ) a final metered droplet using a 20 nL criterion.

4.2. Liquid flow control by the surface wettability In the microchannel networks, the liquid flow could be controlled using the wettability of the liquid on a surface. For a highly wettable (hydrophilic) surface in the hydrophobic (non-wettable) microchannel networks, the liquid flow would be guided to the hydrophilic surface by the difference in the wettability. As shown in figure 7(a), a hydrophilic surface was designed to be on the bottom surface of the hydrophobicdiagonal microchannel. The PDMS microchannels and FC patterns of the bottom surface created a non-wettable surface, whereas the glass wafer provided a wettable bottom surface to the nanoliter fluidic device. A liquid droplet flowed from left to right by the external air pressure in the metering channel, as shown in figure 7(b). The liquid droplet met the hydrophilic bottom surface of the reaction channel, turned right, and flowed into the reaction channel, as shown in figures 7(c) and (d ). This phenomenon arises from the difference in wettability. The boomerang-shaped hydrophobic FC pattern played the role of surface guide for the flow control onto the hydrophilic glass surface. 94

5. In situ monitoring of a biochemical reaction on the chip In order to estimate the feasibility of the proposed nanoliter handling approach for a biochemical reaction, an enzymatic reaction on a chip was performed, monitoring the β-galactosiadase rendered fluorescent product from an FDG substrate. The β-galactosiadase enzyme hydrolyzes the nonfluorescent FDG to fluorescein via the intermediate compound, fluorescein mono-β-galactopyranoside (FMG) [16, 17]. To investigate the non-fluorescence of FDG before the enzyme reaction, we carried out a control experiment of the reaction of a phosphate buffer with the FDG. We did not observe any fluorescence while monitoring the reaction. The green fluorescence emission was detected only in the reaction between the enzyme and FDG. A whole sequence of the metering and the reaction scheme are depicted in figure 8. The FDG was metered in the first metering channel (figure 8(a)), and was cut and transported into the reaction region (figures 8(b) and (c)). The enzyme was

Quantitatively controlled nanoliter liquid manipulation

PDMS Hydrophobic FC surface External air injection Liquid Hydrophilic glass surface Hydrophilic glass (b)

(a)

External air injection

External air injection stop

(d)

(c)

Figure 7. Surface guided flow controlled by the surface wettability: (a) conceptual illustration of the surface guided flow control; (b), (c) liquid flow guided by the hydrophobic FC surface on the hydrophilic glass surface; (d) the trapped liquid droplet in the reaction region. Hydrophilic glass surface is the bottom surface of the diagonal microchannel enclosed by the dotted line.

1st Metering

(a) 2nd Metering and Cutting

1st Transportation

1st Cutting

(c)

(b) Merging

FDG (d) Reaction

(e) 30 s

(g)

Enzyme side

(h)

7 min 30 s

A

Enzyme (f)

A´

(i)

FDG side Figure 8. Nanoliter handling for the enzymatic reaction study: (a) first metering; (b) first cutting; (c) first transportation into the reaction region; (d) second metering and cutting; (e) merging; (f ) initial contact between the FDG and enzyme; (g) reaction; (h), (i) fluorescence images taken during the reaction between enzyme and FDG (30 nL/30 nL) at room temperature.

95

S-H Lee et al 300

30 min

7min 30s

Fluorescence (A.U.)

250 200

30s

150 100 50 0

FDG0 side (A)100

200

300 Enzyme 4005side 00 (A')

Distance

Intensity x Volume (A.U. nL)

Figure 9. Fluorescence intensity variation with reaction time from the centerline profile analysis of the merged droplet.

6. Conclusions

12000 10000 Enzyme:Substrate=2:1 Enzyme:Substrate=1:1 Enzyme:Substrate=1:2

8000 6000 4000 2000 0 0

2

4

6

8

10 12 14 16 18 20

Reaction Time (min) Figure 10. The variation of fluorescence intensity multiplied by the volume as a function of the reaction time and mixing ratio of the enzyme (E) and FDG (S).

metered, cut and transported in the second metering channel (figure 8(d)). When the enzyme met the FDG substrate in the reaction region, the chemical reaction was initiated, with the FDG side emitting a green fluorescence, as shown in figure 8(f ). Ongoing external air pressure makes the enzyme droplet flow into the reaction region, and then the FDG and enzyme droplets merge into a single droplet, as shown in figure 8(g). In this reaction step, the merged droplet appears to maintain its initial state that had a bright fluorescence on the FDG side, and was darker on the enzyme side, as shown in figure 8(f ). This appears to show a poor diffusion reaction between the FDG and the enzyme. However, in contrast, after a reaction time of 7 min 30 s, the fluorescence was emitted more brightly at the enzyme side. Incomplete mixing may be the cause of this heterogeneous reaction, since no mixing technique was adopted, only merging the FDG and enzyme streams into a single droplet. After 30 min reaction time, the green fluorescence was relatively uniform over the completely merged droplet. Further reaction time seems to enhance the diffusion reaction between FDG and the enzyme. As shown in figure 9, these fluorescence intensity variations during the reaction were reconfirmed from the centerline (A–A ) profile analysis of the merged droplet in figure 8(i). To investigate the reactivity at different enzyme and substrate ratios, we changed the mixing ratio using different 96

metered volumes of the enzyme and the FDG. Figure 10 shows the variation of the fluorescence as a function of the reaction time during the enzyme reaction. During the on-chip reaction, evaporation occurred because the total reaction volume (60 nL) was small compared with the high surface area. The total volume diminished linearly, and was around half the initial volume after a reaction time of 30 min. Therefore, to consider the effect of evaporation on the fluorescence intensity, we normalized the fluorescence intensity by multiplying by the volume at each data acquisition step. As the ratio of the enzyme increased, the fluorescence intensity increased rapidly, and reached a saturation value. The trend of the reaction profile was similar, regardless of the mixing ratio. However, faster reaction conditions, such as found for ratios of 1:1 and 2:1, showed a slight decrease in intensity after 10 min.

A novel nanoliter fluidic manipulation technique has been developed that can perform fluidic operations, such as injection, metering, transportation, merging, and reactions on a single chip. A nanoliter fluidic chip was fabricated on a glass substrate using the microfabrication techniques that contained hydrophobic patterns and a PDMS microchannel network. The FC patterning using a lift-off process was a simple and effective method to control the selective wettability on a chip. AFM analysis showed that the patterned FC film surface had a slight increase in its surface roughness after the lift-off process. However, contact angle measurements confirmed that the FC patterns kept their hydrophobic nature, as indicated by their high water contact angle of 108–110◦ . Using a T-shaped injector with a sidewall microchannel array, we could perform variable metering of nanoliter liquid droplets with a desired volume. The proposed nanoliter metering techniques showed an accuracy of 4% using a 20 nL criterion. Through surface guided flow control from the surface wettability, we could merge two different metered liquid droplets into a reaction region in desired mixing ratios. The water repellent hydrophobic FC patterns played the role of surface guide for the flow control onto the hydrophilic glass surface. The feasibility test of the proposed nanoliter manipulation using enzymatic reaction demonstrated a biochemical reaction on a chip. The FDG and enzyme solutions were manipulated in a desired volume (20–40 nL) by the proposed nanoliter metering technique and controlled surface flow guide. From the results of reaction monitoring, the fluorescence intensity of fluorescein was directly proportional to the ratio of the enzyme to FDG.

Acknowledgment This paper was supported by the Nano Bioelectronics and Systems Research Center of Seoul National University, Korea.

References [1] Krishnan M, Namasivayam V, Lin R, Pal R and Burns M A 2001 J. Curr. Opin. Biotechnol. 12 92–8

Quantitatively controlled nanoliter liquid manipulation

[2] Khandurina J and Guttman A 2002 J. Chromatography A 943 159–83 [3] Handique K, Burke D T, Mastrangelo C H and Burns M A 2000 Anal. Chem. 72 4100–9 [4] Hosokawa K, Fujii T and Endo I 1999 Anal. Chem. 71 4781–5 [5] Puntambekar A, Cho H J, Hong C C, Choi J, Ahn C H, Kim S and Makhijani V 2000 Proc. Transducers’01 (Berlin: Springer) pp 1240–3 [6] Zhao B, Moore J S and Beebe D J 2001 Science 291 1023–6 [7] Andersson H, Wijngaart W V, Griss P, Niklaus F and Stemme G 2001 Sensors Actuators B 75 136–41 [8] Lee S H and Kim Y K 2001 Proc. Micro Total Analysis Systems’01 (Monterey, CA, 21–25 Oct.) (Dordecht: Kluwer) pp 205–6 [9] Sun B, Xie W, Yi G, Chen D, Zhou Y and Cheng J 2001 J. Immunol. Methods 249 85–89

[10] Moss D W and Rosalki S B 1996 Enzyme Tests in Diagnosis (New York: Oxford University Press) pp 7–23 [11] Jansen H V, Gardeniers J G E, Elders J, Tilmans H A C and Elwenspoek M 1994 Sensors Actuators A 41 136–40 [12] Xia Y, Kim E and Whitesides G M 1996 J. Chem. Mater. 8 1558–67 [13] Schrader M E and Loeb G I 1992 Modern Approaches to Wettability (New York: Plenum) pp 1–27 [14] Mittal K L 1993 Contact Angle, Wettability and Adhesion (Urecht, The Netherlands: VSP) pp 3–36 [15] Berg J C 1993 Wettability (New York: Marcel Dekker) pp 1–73 [16] Tsung E F and Tilton R D 1999 J. Colloid Interface Sci. 213 208–17 [17] Bruke B J and Regnier F E 2001 Electrophor. 22 3744–51

97