Design and fabrication of a passive microfluidic ... - SPIE Digital Library

J. Micro/Nanolith. MEMS MOEMS 8共2兲, 021105 共Apr–Jun 2009兲

Design and fabrication of a passive microfluidic mixer with dynamic disturbance produced by catalytic chemical reactions Jin-Cherng Shyu National Kaohsiung University of Applied Sciences Department of Mechanical Engineering Kaohsiung, Taiwan 80778 E-mail: [email protected]

Ching-Jiun Lee National Taiwan University Institute of Applied Mechanics Taipei, Taiwan 10617

Chung-Sheng Wei National Taiwan University Nano-Electro-Mechanical-System Research Center Taipei, Taiwan 10617

Abstract. We report a novel passive microfluidic mixer design for mixing enhancement based on dynamic disturbances due to bubble generation produced by catalytic decomposition of hydrogen peroxide. A Y-shaped passive microfluidic mixer with a width and depth of 1 mm and 50 ␮m, respectively, with platinum deposition on the partial undersurface of the mixing channel for H2O2 catalytic decomposition, is demonstrated. Various Reynolds numbers 共0.06 to 63.5兲 and H2O2 concentrations are tested to investigate their effects on the mixing. The experimental results show that mixing can be significantly improved either with the decrease of volumetric flow rate at a given H2O2 concentration or with the increase of H2O2 concentration at intermediate Reynolds numbers based on the present design. The mixing index scatters between 0.8 and 1.0 at x 艌 15 mm for all H2O2 concentrations if Re= 0.06 in the mixing channel. However, the H2O2 concentration has no significant effect on mixing provided Re艌 63.5. In addition, the maximum mixing enhancement for QL = 1, 10, 100, and 1000 ␮L / min 共Re= 0.06, 0.63, 6.35, and 63.5兲 at x 艌 15 mm are 5.7, 11.85, 6.27, and 4.8, respectively, with 0.1 M ⬍ 关H2O2兴 ⬍ 8.8 M in this study. © 2009 Society of Photo-Optical Instrumentation Engineers. 关DOI: 10.1117/1.3122367兴

Subject terms: microfluidic mixer; passive; dynamic disturbance; chemical reaction. Paper 08119SSR received Aug. 15, 2008; revised manuscript received Feb. 9, 2009; accepted for publication Mar. 2, 2009; published online Apr. 27, 2009.

1

Introduction

Since rapid mixing is essential in many microfluidic systems used in biochemistry analysis, in addition to the micropump, the micromixer is also considered an important component in a microfluidic system. For example, cell activation and enzyme reactions, as well as protein folding during biological processes, often involve reactions that require thorough reactant mixing for initiation. Furthermore, mixing is also important in lab-on-a-chip 共LOC兲 platforms for complex chemical reactions. Based on the operation mode, micromixers are generally divided into two types. One is the passive micromixer, whose mixing processes do not consume external energy but rely entirely on molecular diffusion and chaotic advection. Therefore, this type of micromixer is usually robust, stable in operation, and easily integrated in a more complex system because there is no need for external actuators. The most basic design of a passive micromixer is the T-shaped or Y-shaped micromixer,1–4 which employs diffusion as the major mixing mechanism. In order to enhance the mixing efficiency in such T-shaped or Y-shaped micromixers, different modifications of the mixing channel of T-shaped or Y-shaped micromixers have also been proposed5–13 for the introduction of another form of mass transfer, namely, chaotic advection. However, most of the aforementioned results indicate that the mixing process would be faster at 1932-5150/2009/$25.00 © 2009 SPIE

J. Micro/Nanolith. MEMS MOEMS

higher Reynolds numbers even if the chaotic advection could be successfully initiated for more complex mixing channel designs.5,6,10 The other type is the active micromixer, in which mixing processes are carried out with a disturbance generated by an external field, such as pressure, temperature, acoustics, or electrohydrodynamics. Better mixing performances have been found in active micromixers. However, compared with passive micromixers, the fabrication of active micromixers is usually more difficult. In addition, integrating an active micromixer into a microfluidic system is both challenging and expensive. In all types of active micromixers in the published literature, it has been verified that bubble initiation and growth in a microchannel is able to enhance the mixing performance.14 The authors used a heater made of copper on a glass wafer to generate thermal bubbles in the microchannel. It was observed that one stream penetrated into the other and was sucked back as the thermal bubble expanded and collapsed, respectively. These phenomena caused a wavy interface and increased the contact area between the two liquids. The mixing efficiency was thus enhanced for thermal bubble agitation. If one can incorporate a dynamic perturbation caused by bubble growth and movement into a Y-shaped micromixer without any external energy, then in addition to the simpler structure and easier fabrication process, a more efficient mixing performance can also be expected.

021105-1

Apr–Jun 2009/Vol. 8共2兲

Shyu, Lee, and Wei: Design and fabrication of a passive microfluidic mixer…

(a)

(e)

(b)

(f)

(c)

(g)

(a) Inlet for dyed DI water

Platinum deposition Outlet

(d) (h)

0

Fig. 1 Fabrication process of the Y-shaped microfluidic mixer: 共a兲 photolithography of Y-shaped microchannel; 共b兲 forming of Y-shaped microchannel by ICP-RIE; 共c兲 photolithography of through-holes on backside of the wafer; 共d兲 formation of through-holes by ICP-RIE; 共e兲 photolithography of catalyst; 共f兲 formation of Pt catalyst by liftoff process; 共g兲 anodic bonding; and 共h兲 PDMS sealing of through-holes for reactant inlets and outlet.

Instead of a thermal bubble, a bubble initiated by a catalytic chemical reaction will be presented in this study to eliminate the external energy consumption. Since liquid hydrogen peroxide 共H2O2兲 can be self-decomposed into water and oxygen via the catalytic chemical reaction by platinum,15 rapid mixing at a relatively low Reynolds number could be expected if hydrogen peroxide is present and decomposes in the micromixer, since the mixing mechanism employing dynamic disturbance induced by bubble initiation is similar to that of active micromixers. Therefore, the present study aims to develop a Y-shaped passive micromixer with dynamic perturbation resulting from bubble initiation due to catalytic chemical reaction, not additional actuators. Both the effects of hydrogen peroxide concentration and the volumetric flow rate on the mixing performance will also be quantitatively analyzed for further improvement of such a passive micromixer. 2

Micromixer Fabrication and Experimental Setup A schematic fabrication process for the microfluidic mixer is shown in Fig. 1. A single-sided polished silicon wafer with a thickness of 500 ␮m was used in the fabrication of the microfluidic reactor with a Y-shaped microchannel. Three photolithography and etching steps were required to produce the wafer with the essential features, as shown in Fig. 1. The first photolithography step transferred the pattern of the Y-shaped microchannel to the wafer with photoresist AZ-P4620 by conventional i-line aligner. A Y-shaped microchannel was formed by the ICP-RIE 共inductively coupled plasma reactive ion etching, MESC Multiplex ICP, Surface Technology Systems兲 to a depth of 50 ␮m and a width of 1000 ␮m. The next lithography process, followed by ICP etching, was carried out on the back side of the J. Micro/Nanolith. MEMS MOEMS

3

13

30

x

Unit: mm Inlet for H2O2 solution

(b)

Fig. 2 共a兲 Schematic illustration of the present experimental system. 共b兲 Y-shaped microfluidic mixer in the present study.

wafer to introduce three through-holes with diameters of 3 mm as the fluid inlets and outlet. Subsequently, in order to form a platinum catalyst on the bottom surface of the Y-shaped microchannel, the final lithography step involved patterning two-layer photoresist with a total thickness of ⬃20 ␮m 共Clariant, AZ-P4620兲. E-beam evaporation was performed to deposit a layer of platinum adhered to a titanium layer on the silicon wafer. The photoresist for all lithography processes mentioned here was removed by soaking in acetone followed by DI water rinsing. After the fabrication processes of the Y-shaped microfluidic reactor on a silicon wafer, anodic bonding of the wafer with a 500-␮m-thick Pyrex 7740 glass wafer was carried out for a closed microchannel. The anodic bonding process was carried out at 400°C with an applied voltage of 1 kV across the silicon and glass wafers. The assembly was then diced to produce each microfluidic device. Each throughhole was sealed by poly-dimethyl-siloxane 共PDMS兲, as shown in Fig. 1. The working fluids could flow in and out through the needles connected to the fluid delivery system. In addition to the microfluidic mixer, an image acquisition system and a syringe pump with two syringes 共KDS 210兲 were also employed. The image acquisition system consisted of an inverted optical microscope 共Nikon TE2000兲, a digital camera 共Nikon D70兲, and a mercury lamp. The schematic illustration of the present experiment is shown in Fig. 2. Both dyed DI water 共fluorescent red dye, Rhodamine 6G, C28H31C1N2O3兲 and hydrogen peroxide were forced into the mixer via individual inlets at the same volumetric flow rate and discharged via an outlet, as shown in Fig. 2共b兲. Once the H2O2 stream contacted the platinum on the microchannel undersurface, oxygen bubbles were generated. The experimental conditions are listed in Table 1. In addition, 11 visualization positions were prescribed, at x

021105-2


Shyu, Lee, and Wei: Design and fabrication of a passive microfluidic mixer… 1

Table 1 Experimental conditions.

50 ␮m

Width of the microchannel

1000 ␮m

Length of the mixing channel

Mixing index

0.8

Depth of the microchannel

30 mm

Catalyst dimensions

0.5 mm⫻ 10 mm

Measured positions, x

3, 5, 8, 10, 13, 15, 18, 20, 23, 25, 28 mm

Inlet volumetric flow rates 共␮L / min, QL兲

1, 10, 100, 1000

Corresponding Re number

0.06, 0.63, 6.35, 63.5

Liquids used

DI water and fluorescent dye Hydrogen peroxide

H2O2 concentration

0 M, 0.1 M, 1.0 M, 5.0 M, 8.8 M

0.6

0.4

0M

0.1 M

1.0 M

5.0 M

8.8 M

curve_0 M

curve_0.1 M

curve_1.0 M

curve_5.0 M

curve_8.8 M

0.2

0 0

10

15

20

25

30

Position (mm)

Fig. 4 Mixing index 共MI兲 variation along the mixing channel and curves used to fit those data in the present microfluidic mixer with various H2O2 concentrations at QL = 10 ␮L / min.

= 3, 5, 8, 10, 13, 15, 18, 20, 23, 25, and 28 mm, as shown in Fig. 2共b兲. Note that the lengths of the main channel of the Y-shaped microchannel and the platinum are 30 mm and 10 mm, respectively, in the present experiment, as shown in Fig. 2共b兲. All pictures, taken at 3008共H兲 ⫻ 2000共V兲 pixels by a 10⫻ object lens, were analyzed for mixing index estimation based on the following equation:16

再

1 兺关I共j兲 − ¯I兴2 MI = 1 − ¯I N

冎

1/2

共1兲

,

where N is the number of total pixels, I is the intensity of each pixel, and ¯I is the mean intensity value of all pixels.

3 Results and Discussion Before the mixing experiment, two water streams were simultaneously injected into the Y-shaped microchannel. Pictures were taken and analyzed based on Eq. 共1兲 at x = 3, 5, 8, 10, 13, 15, 18, 20, 23, 25, and 28 mm of the main channel for various inlet volumetric flow rates. The mixing indices obtained from the analysis of those pictures could be regarded as the indication of the mixing of the two streams in the present study owing to diffusion only. For the sake of comparison with the present measured mixing index with catalytic chemical reaction, those mixing indices are shown in Figs. 3–6 for different inlet volumetric flow rates. Based on the observation of Figs. 3–6, it can be found that the mixing index variation along the main channel without addition of H2O2, denoted as 0 M, becomes insignificant with the increase of the inlet volumetric flow rates. Figures 3–6 show the mixing index 共M I兲 variation along the mixing channel at different inlet volumetric flow rates 共QL兲 with various H2O2 concentrations of 0.1 M, 1.0 M, 5.0 M, and 8.8 M, respectively, as well as without H2O2 addition. When the volumetric flow rate of the liquid was 1 ␮L / min, i.e., Re= 0.06, in the mixing channel, the mixing 1

1

0.8

0.6

0.4

0M

0.1 M

1.0 M

5.0 M

8.8 M

curve_0 M

curve_0.1 M

curve_1.0 M

curve_5.0 M

Mixing index

0.8

Mixing index

5

curve_8.8 M

0M

0.1 M

1.0 M

5.0 M

8.8 M

curve_0 M

curve_0.1 M

curve_1.0 M

curve_5.0 M

curve_8.8 M

0.6

0.4

0.2

0.2

0

0 0

5

10

15

20

25

0

30

Fig. 3 Mixing index 共MI兲 variation along the mixing channel and curves used to fit those data in the present microfluidic mixer with various H2O2 concentrations at volumetric flow rates of 1.0 ␮L / min. J. Micro/Nanolith. MEMS MOEMS

5

10

15

20

25

30

Position (mm)

Position (mm)


021105-3


Shyu, Lee, and Wei: Design and fabrication of a passive microfluidic mixer… 1

Mixing index

0.8

0M

0.1 M

1.0 M

5.0 M

8.8 M

curve_0 M

curve_0.1 M

curve_1.0 M

curve_5.0 M

for different H2O2 concentrations used in the present study, as shown in Fig. 3. Note that the unit of the channel length, x, in the foregoing correlations shown in Table 2 for different concentrations is mm. With the increase of the inlet volumetric flow rate, the mixing index is affected at certain experimental conditions. For example, when the inlet volumetric flow rate was increased to 10 ␮L / min 共Re= 0.63 in the main channel兲, the mixing index at 关H2O2兴 = 0.1 M was significantly reduced 共0.2⬍ M I ⬍ 0.4兲, whereas the maximum mixing index ranges from 0.8 to 0.9 for 关H2O2兴 = 1.0, 5.0, and 8.8 M at x 艌 15 mm, as shown in Fig. 4. The M I,em would be 11.85 at x 艌 15 mm for 关H2O2兴 = 1.0, 5.0, and 8.8 M, as shown in Fig. 4. Moreover, the relationship between measured mixing indices and channel length at 关H2O2兴 = 1.0, 5.0, and 8.8 M could also be correlated by natural logarithm due to the same trend of the variation of the mixing index observed in Fig. 3, while the fitting curve for the lowest concentration case, i.e., 关H2O2兴 = 0.1 M, would be a constant, as shown in Fig. 4 and Table 2. A further increase of QL 共100 ␮L / min, i.e., Re= 6.35 in the main channel兲 further decreased the mixing performance obtained at H2O2 concentrations of 0.1 M and 1.0 M, while there was an almost invisible effect on the mixing index obtained at H2O2 concentrations of 5.0 M and 8.8 M, as shown in Fig. 5. The maximum mixing index ranges from 0.7 to 0.87 for 关H2O2兴 = 5.0 and 8.8 M at x 艌 15 mm. In addition, M I,em would be 6.27 at x 艌 15 mm for 关H2O2兴 = 5.0 and 8.8 M, as shown in Fig. 5. The relationship between measured mixing indices and channel length at 关H2O2兴 = 5.0 and 8.8 M could also be expressed in the form of a natural logarithm, while the fitting curve for the lower concentration cases, i.e., 关H2O2兴 = 0.1 and 1.0 M, would be constants, as shown in Fig. 5 and Table 2. If the inlet volumetric flow rate is further increased to 1000 ␮L / min 共Re= 63.5 in the main channel兲 as shown in Fig. 6, less mixing enhancement is revealed, regardless of the concentration of H2O2. The mixing indices are approximately scattered between 0.15 and 0.25 for all H2O2 concentrations at all visualization positions. Compared with the measured results shown in Figs. 3–5, those values are fairly close to the mixing index without H2O2. Therefore, the relationship between the mixing index and the channel length would be correlated as different constants for differ-

curve_8.8 M

0.6

0.4

0.2

0 0

5

10

15

20

25

30

Position (mm)


index scattered between 0.8 and 1.0 at x 艌 15 mm for all H2O2 concentrations, as shown in Fig. 3. Note that the end of the platinum catalyst lies at x = 13 mm. Furthermore, based on the observation of the mixing indices obtained from the experiment without any H2O2 addition at QL = 1 ␮L / min in Fig. 3, it could be found that the mixing index increased gradually downstream due to diffusion. However, those mixing indices are much lower than those obtained with H2O2 added into one of the streams. The maximum mixing enhancement at a given volumetric flow rate, M I,em, is defined as, M I,em = Max关M I共x j兲/M I,n共x j兲兴,

共2兲

where M I共x j兲 and M I,n共x j兲, respectively, denote the mixing index measured at x = x j with H2O2 and without H2O2 According to the definition expressed by Eq. 共2兲, M I,em would be 5.7 at x 艌 15 mm for various H2O2 concentrations, as shown in Fig. 3. Also, it could be observed that the mixing index variation along the main channel at QL = 1 ␮L / min could be expressed in the form of M I = a ln共x兲 + b, where a and b are constants determined by experiment, as shown in Fig. 3 and Table 2, since the increase of the mixing index is significant in the upstream of the main channel and is gradually alleviated downstream. The maximum mixing index ranges from 0.8 to almost 1.0

Table 2 Equations for all fitting curves corresponding to the present measured data obtained at various experimental conditions. QL共␮L / min兲 Concentration 共M兲

1

10

100

1000

0.1

MI = 0.09 ln x + 0.7

MI = 0.27

MI = 0.2

MI = 0.15

1.0

MI = 0.37 ln x − 0.24

MI = 0.3 ln x − 0.16

MI = 0.23

MI = 0.17

5.0

MI = 0.32 ln x − 0.08 MI = 0.16 ln x + 0.33 MI = 0.25 ln x − 0.07

8.8

MI = 0.26 ln x + 0.11 MI = 0.25 ln x + 0.12 MI = 0.31 ln x − 0.15 MI = 0.06 ln x + 0.06


021105-4

MI = 0.18



Slug bubble

(a)

(a)

Trajectory of smaller bubbles

(b)

(b)

Fig. 7 Bubble size at different flow rates in the present study taken with filter 共right兲 and without filter 共left兲 at x = 28 mm with 关H2O2兴 = 0.1 M. 共a兲 1 ␮L / min. 共b兲 1000 ␮L / min.

Fig. 8 Bubble size at different flow rates in the present study taken with filter 共right兲 and without filter 共left兲 at x = 28 mm with 关H2O2兴 = 8.8 M. 共a兲 1 ␮L / min. 共b兲 1000 ␮L / min.

ent H2O2 concentrations at QL = 1000 ␮L / min, except for the results obtained with 关H2O2兴 = 8.8 M, as shown in Fig. 6 and Table 2. Accordingly, the H2O2 concentration effect on the mixing performance in the present microfluidic mixer obviously depended on the QL. In general, it can be summarized that the mixing will usually be enhanced with the increase of the H2O2 concentration at an intermediate Reynolds number. Nevertheless, the H2O2 concentration, ranging from 0.1 M to 8.8 M, will have a minor effect on the mixing performance for both extremely small QL 共Re艋 0.06兲 and extremely high QL 共Re艌 63.5兲, as shown in Figs. 3 and 6. Furthermore, the larger QL employed, the higher the concentration of H2O2 that has to be added to the present microfluidic mixer for noticeable mixing improvement for a liquid stream with an intermediate Reynolds number. Once QL 艌 1000 ␮L / min 共Re艌 63.5兲, the H2O2 addition in the present microfluidic mixer would have little effect on the mixing performance if 关H2O2兴艋 8.8 M, as shown in Fig. 6. Based on the experimental results shown in Figs. 3–6, a trend could be observed: The mixing index could be increased significantly with the decrease of QL at a given H2O2 concentration. Hence, the QL yielding the best mixing performance is always found to be 1 ␮L / min for all H2O2 concentrations tested here. In fact, the mixing index for QL = 1 ␮L / min 共Re= 0.06兲 is observed to be 3.7 times higher than that of QL = 10 ␮L / min operation with 关H2O2兴 = 0.1 M at x = 15 mm 共0.97 versus 0.26兲. According to the aforementioned results and discussions, it can be concluded that better mixing efficiency can be achieved with a smaller flow rate or with a smaller Reynolds number in the present passive microfluidic mixer with disturbance due to H2O2 catalytic decomposition. One of the reasons could be the solubility of oxygen bubbles in water, with more oxygen dissolved at larger QL. Therefore, the oxygen bubble has less influence on the liquid stream with the operation of higher QL because of smaller bubbles formed in the microchannel, as shown in Fig. 7. In addition, operation at smaller flow rates indicates a slow velocity of the streams and thus results in a longer time for diffusion caused by concentration gradient. This could be a result J. Micro/Nanolith. MEMS MOEMS

unique to passive microfluidic mixers, since mixing efficiency would usually be better at higher Reynolds numbers in a passive microfluidic mixer.5,6,10 Increasing the H2O2 concentration causes a more violent chemical reaction and thus results in larger bubbles at the same QL, as shown in Fig. 8. Therefore, both the size of the bubbles from catalytic decomposition of H2O2 and the gas/ liquid two-phase flow pattern in the present passive microfluidic mixer could be two of the major factors that improve the mixing performance. It is worth noting that the favorable gas/liquid two-phase flow pattern in the present microfluidic mixer could be slug/bubbly flow, according to the flow visualization in Fig. 8共a兲. Excellent mixing could be performed in front of the large bubble shown in Fig. 8共a兲, which shows a place between two slug bubbles. 4 Conclusions This study provides a novel design to augment the performance of passive mixers. Based on the present experimental results performed for a passive Y-shaped microfluidic mixer with dynamic disturbance produced by catalytic chemical reactions, the mixing performances were obtained at various H2O2 concentrations and various inlet volumetric flow rates. It could be found that the mixing efficiency could be increased significantly, either with the decrease of volumetric flow rate at a given H2O2 concentration or with the increase of H2O2 concentration at intermediate Reynolds numbers, based on the present design. In fact, an excellent mixing performance at extremely low Reynolds numbers was possible, e.g., 0.8⬍ M I ⬍ 1.0 for Re= 0.06 and 关H2O2兴 ⬍ 8.8 M, with the aid of dynamic bubble behavior in the microfluidic channel due to H2O2 catalytic decomposition. Moreover, the mixing index variation along the main channel could be expressed either by a natural logarithm or as a constant for different experimental conditions tested in the present study. In addition, the M I,em for QL = 1 ␮L / min 共Re= 0.06兲, 10 ␮L / min 共Re= 0.63兲, 100 ␮L / min 共Re= 6.35兲, and 1000 ␮L / min 共Re= 63.5兲 based on Eq. 共2兲 at x 艌 15 mm are 5.7, 11.85, 6.27, and 4.8, respectively, with various H2O2 concentrations tested in the present study.

021105-5



The present novel design could have the potential for some microfluidic mixer applications, especially those related to a low Reynolds number. It is important to note that the smaller the QL employed, the less the amount of H2O2 that has to be added to the present microfluidic mixer for mixing enhancement. Since the mixing improvement for the present passive microfluidic mixer has been confirmed, future work will aim to modify the microfluidic mixer design to allow less platinum deposition and easy removal of the bubbles from the microchannel. Correlating the bubbling ratio to Reynolds numbers will also be a research topic in the future. Acknowledgments The financial support of this research from the Industrial Technology Research Institute of Taiwan and from the National Science Council of Taiwan 共NSC-98–2218–E-151– 001 and NSC-97–2218–E-002–037兲 is gratefully acknowledged. References 1. T. T. Veenstra, “Characterization method for a new diffusion mixer applicable in micro flow injection analysis systems,” J. Micromech. Microeng. 9, 199–202 共1999兲. 2. R. F. Ismagilov, A. D. Stroock, P. J. A. Kenis, G. Whitesides, and H. A. Stone, “Experimental and theoretical scaling laws for transverse diffusive broadening in two-phase laminar flows in microchannels,” Appl. Phys. Lett. 76, 2376–2378 共2000兲. 3. D. Gobby, P. Angeli, and A. Gavriilidis, “Mixing characteristics of T-type microfluidic mixers,” J. Micromech. Microeng. 11, 126–132 共2001兲. 4. S. H. Wong, M. C. L. Ward, and C. W. Wharton, “Micro T-mixer as a rapid mixing micromixer,” Sens. Actuators B 100, 365–385 共2004兲. 5. R. H. Liu, M. A. Stremler, K. V. Sharp, M. G. Olsen, J. G. Santiago, R. J. Adrian, H. Aref, and D. J. Beebe, “Passive mixing in a threedimensional serpentine microchannel,” J. Microelectromech. Syst. 9, 190–197 共2000兲. 6. H. Wang, P. Iovenitti, E. Harvey, and S. Masood, “Optimizing layout of obstacles for enhanced mixing in microchannels,” Smart Mater. Struct. 11, 662–667 共2002兲. 7. A. D. Stroock, S. K. W. Dertinger, A. Ajdar, I. Mezic´, H. A. Stone, and G. M. Whitesides, “Chaotic mixer for microchannels,” Science 295, 647–651 共2002兲. 8. V. Mengeaud, J. Josserand, and H. H. Girault, “Mixing processes in a zigzag microchannel: finite element simulation and optical study,” Anal. Chem. 74, 4279–4286 共2002兲. 9. S. H. Wong, P. Bryant, M. Ward, and C. Wharton, “Investigation of mixing in a cross-shaped micromixer with static mixing elements for reaction kinetics studies,” Sens. Actuators B 95, 414–424 共2003兲. 10. C. C. Hong, J. W. Choi, and C. H. Ahn, “A novel in-plane microfluidic mixer with modified Tesla structures,” Lab Chip 4, 109–113 共2004兲. 11. X. Fu, S. Liu, X. Ruan, and H. Yang, “Research on staggered oriented ridges static micromixers,” Sens. Actuators B 114, 618–624 共2006兲. 12. Y. C. Lin, Y. C. Chung, and C. Y. Wu, “Mixing enhancement of the passive microfluidic mixer with J-shaped baffles in the tee channel,” Biomed. Microdevices 9, 215–221 共2007兲.


13. A. A. S. Bhagat, E. T. K. Peterson, and I. Papautsky, “A passive planar micromixer with obstructions for mixing at low Reynolds numbers,” J. Micromech. Microeng. 17, 1017–1024 共2007兲. 14. J. H. Tsai and L. Lin, “Active microfluidic mixer and gas bubble filter driven by thermal bubble micropump,” Sens. Actuators, A 97–98, 665–671 共2002兲. 15. J. Xu, Y. Feng, and J. Cen, “Transient flow patterns and bubble slug lengths in parallel microchannels with oxygen gas bubbles produced by catalytic chemical reactions,” Int. J. Heat Mass Transfer 50, 857– 871 共2007兲. 16. Y. K. Lee, P. Tabeling, C. Shih, and C. M. Ho, “Characterization of a MEMS-fabricated mixing device,” in Proc. MEMS, ASME Int. Mechanical Engineering Congress & Exposition, Orlando, FL, November 2000, pp. 505–511, ASME, New York 共2000兲. Jin-Cherng Shyu received his PhD degree in mechanical engineering from National Taiwan University, Taipei, Taiwan, in 2002. After a three-month enrollment training, he worked as a researcher at the Industrial Technology Research Institute from 2003 to 2008. He is currently an assistant professor in the Department of Mechanical Engineering at National Kaohsiung University of Applied Sciences, Kaohsiung, Taiwan. His research interests include microfluidics design and fluid measurement and visualization in microfluidics. Ching-Jiun Lee is currently a postdoctoral researcher in the Nano-Electro-MechanicalSystem 共NEMS兲 Research Center, National Taiwan University. He received his MS degree and his PhD degree from the Institute of Applied Mechanics at National Taiwan University in 1999 and 2007, respectively. His research interests include microfluidics systems, micro-PIV techniques, and micrototal-analysis systems.

Chung-Sheng Wei received a PhD degree in mechanical engineering from Chung Cheng University, Taiwan, in 2002, where he studied residual stress analysis on laser hard-facing layers by numerical and experimental methods. In 2003, he did postdoctoral work at the Nano-Electro-MechanicalSystems 共NEMS兲 Research Center, National Taiwan University, where he focused primarily on the fabrication and design of pyroelectric sensors. He was an associate researcher at the center, and he focused on pyroelectric devices, the controlled growth of signal wall carbon nanotubes 共SWNTs兲, and carbon nanotube FETs fabricated by aligned SWNT. In 2007, he joined Nanya Technology Corporation, TaoYuan, Taiwan, where he has been devoted to probe development of front-end testing. His research interests are probe development at the front end and back end and yield modeling on memory devices.

021105-6
