Anal. Chem. 2008, 80, 3035-3039
Sheathless Hydrophoretic Particle Focusing in a Microchannel with Exponentially Increasing Obstacle Arrays Sungyoung Choi and Je-Kyun Park*
Department of Bio and Brain Engineering, KAIST, 335 Gwahangno, Yuseong-gu, Daejeon 305-701, Republic of Korea
We present a novel microfluidic device with exponentially increasing obstacle arrays to enable sheathless particle focusing. The anisotropic fluidic resistance of slant obstacles generates transverse flows, along which particles are focused to one sidewall. In the successive channel with exponentially increasing widths, bent obstacles extended from the slant obstacles increase the focusing efficiency of the particles. With the device, we achieved the focusing efficiency of 76%, 94%, and 98% for 6, 10, and 15 µm beads, respectively. The focusing efficiency of the particles can be further improved in the devices with more extension steps. In addition, using the microfluidic devices with the symmetric structure of the slant and bent obstacles, we achieved complete focusing of biological cells to the centerline of a channel within 1.7% coefficient of variation. The results demonstrated the sheathless hydrophoretic focusing of microparticles and cells with the advantages of a sheathless method, passive operation, single channel, and flow rate independence. The development of microfluidic methods for controlling the movement of particles is a key requirement for the miniaturization and integration of chemical and biological tools in a microchip. Functional micro- or nanoparticles as a mobile substrate for biomolecules are routinely handled for the measurement and quantification of biological or chemical species.1,2 In applications of flow cytometry, the focusing of particles in a flow chamber enables a more accurate analysis because the particles are located in an identical streamline.3,4 Many promising approaches for particle focusing and manipulation through microchannels have been demonstrated, using hydrodynamic,5-7 inertial,8 dielectrophoretic,9,10 electrokinetic,11 acoustic focusing,12 and hydrodynamic * To whom correspondence should be addressed. E-mail:
[email protected]. Phone: +82 42 869 4315. Fax: +82 42 869 4310. (1) Kawaguchi, H. Prog. Polym. Sci. 2000, 25, 1171-1210. (2) Tan, W.; Takeuchi, S. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 1146-1151. (3) Crosland-Taylor, P. J. Nature 1953, 171, 37-38. (4) Tung, Y.; Torisawa, Y.; Futai, N.; Takayama, S. Lab Chip 2007, 7, 14971503. (5) (a) Simonnet, C.; Groisman, A. Appl. Phys. Lett. 2005, 87, 114104. (b) Simonnet, C.; Groisman, A. Anal. Chem. 2006, 78, 5653-5663. (6) Sundararajan, N.; Pio, M. S.; Lee, L. P.; Berlin, A. A. J. Microelectromech. Syst. 2004, 13, 559-567. (7) Mao, X.; Waldeisen, J. R.; Huang, T. J. Lab Chip 2007, 7, 1260-1262. (8) Carlo, D. D.; Irimia, D.; Tompkins, R. G.; Toner, M. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 18892-18897. (9) Choi, S.; Park, J.-K. Lab Chip 2005, 5, 1161-1167. 10.1021/ac8001319 CCC: $40.75 Published on Web 03/21/2008
© 2008 American Chemical Society
filtration.13 The particle focusing and manipulation methods based on sheath flows depend on accurate flow control between sample and sheath flows. Therefore, it can be difficult to apply the methods to the development of integrated microfluidic devices which require an operation on mobile and generate irregular flows. In comparison with flow-assisted methods, active methods require relatively complex fabrication processes such as patterning of metal electrodes and piezoceramic plates. There is a still challenge for the development of a simple, low-cost, sheathless, and passive method for particle focusing and manipulation. In this paper, we describe a microfluidic device for sheathless hydrophoretic particle focusing using exponentially increasing obstacle arrays. We recently introduced a hydrophoretic separation principle that eliminated the needs of sheath flows and accurate pumping.14 The slant obstacle as a source of hydrophoresis was used to separate particles by size. Because of the sizedependent behaviors of particles, the obstacle is more effective for particle separation than for particle focusing. To achieve sheathless hydrophoretic particle focusing, a novel scheme is designed as shown in Figure 1. The channel area around the slant obstacle has less resistance to flow along the slant obstacle than along the direction of the fluid flow.14,15 Such anisotropic fluidic resistance between the top or bottom area and side area of the obstacle generates lateral pressure gradients, which induce helical recirculation (see the insets in Figure 1c). The obstacle structure at the bottom and top of the channel generates helical recirculations in clockwise and counterclockwise directions, respectively. When particles pass through the slant obstacles alternately placed on the bottom and top of the channel, they are exposed to the focusing flows, a dominant flow of the rotating flows (see the left insets in Figure 1c) and focused within the width of a in the channel width of x1. The lateral position of particles from the bottom sidewall depends on their size.14 Particles that have a comparable size to the gaps of the obstacles can be aligned at the center of z-axis by the obstacle (Figure 1c). By the alignment (10) Yu, C.; Vykoukal, J.; Vykoukal, D. M.; Schwartz, J. A.; Shi, L.; Gascoyne, P. R. C. J. Microelectromech. Syst. 2005, 14, 480-487. (11) Schrum, D. P.; Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1999, 71, 4173-4177. (12) Goddard, G.; Martin, J. C.; Graves, S. W.; Kaduchak, G. Cytometry, Part A 2006, 69A, 66-74. (13) Yamada, M.; Seki, M. Lab Chip 2005, 5, 1233-1239. (14) (a) Choi, S.; Park, J.-K. Lab Chip 2007, 7, 890-897. (b) Choi, S.; Song, S.; Choi, C.; Park, J.-K. Lab Chip 2007, 7, 1532-1538. (15) Stroock, A. D.; Dertinger, S. K. W.; Ajdari, A.; Mezic´, I.; Stone, H. A.; Whitesides, G. M. Science 2002, 295, 647-651.
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008 3035
where xn is the channel width for the nth step, x1 is the channel width for the first step, a is the width of a focused particle stream in the slanted region, En is the focusing efficiency for the nth step, and D is the diameter of the particle. The channel area around the extended region has the isotropic resistance to the perpendicular direction to the fluid flow. There are no significant variations of the pressure fields in the transverse direction. Therefore, the lateral flow in that region is absent (see the right insets in Figure 1c). For the complete focusing, particles should be within the slanted region.
Figure 1. Hydrophoretic focusing device with slant and bent obstacles alternately placed on the bottom and top of a channel. (a) Optical micrograph of the microfluidic channel with five step extensions. (b) Magnified image of slant and bent obstacles. (c) Schematic diagram showing the focusing procedures of particles. Colored and lined areas denote lower and upper PDMS layers, respectively.
effect, the particles are mainly exposed to upward, focusing, and downward flows, not deviation flows. Therefore, they can be completely focused to the bottom sidewall. The alignment effect depends on the size difference between the gap of the obstacle and a particle. The smaller particle can be exposed to deviation flows and deviate from the bottom sidewall. Therefore, the lateral position increases as the size of the particles decreases. Here, when the channel width increases from x1 to x2, satisfying that the width of the focused particle stream (a) is expanded to the width of the slanted region (x1), particles passing through the slanted region of the bent obstacles experience the same focusing flows (see the right insets in Figure 1c) and focus again within the width of a in the wider channel, x2. By repeating this extension step, the ratio of the width of the focused particle stream to the channel width becomes smaller, while the focusing efficiency increases.
xn ) En ) 3036
(
() x1 a
n-1
x1
(1)
)
(2)
xn - D/2 - a × 100 xn - D
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
EXPERIMENTAL SECTION Device Design Principle. The microfluidic device is a stacked structure in which two poly(dimethylsiloxane) (PDMS)-channel layers with upper and lower obstacles face each other. Accordingly, the height of the obstacles and their gap between the obstacle and the top or bottom of a channel were defined as the half of the channel height. The channel consists of slant obstacles and bent obstacles with exponentially increasing widths every step. The microfluidic devices were designed by applying eq 1 for two different types: one with an extension ratio (x1/a) of 2 and the other with the ratio of 1.25. The channel width for the first step was 100 µm; the corresponding focusing criteria were 50 and 80 µm for the ratio of 2 and 1.25, respectively. When the width of a focused particle stream (a) is smaller than the criteria, they can work well on the focusing principle. The increasing channel widths every step (in µm) were 100, 200, 400, 800, and 1600 for the ratio of 2 (Figure 1a), and 100, 125, 156, 195, and 244 for the ratio of 1.25. The obstacles were geometrically defined with LS ) 70 µm, Dob ) 70 µm, WS ) 100 µm, and θ ) 55° (Figure 1b). To confirm the focusing enhancement every step, we fabricated microfluidic devices which include the first, second, third, fourth, and all the steps. For the cell focusing to the channel center, we fabricated microfluidic devices with the symmetric structure of slant and bent obstacles. Two PDMS-channel layers with upper and lower obstacles were fabricated using a standard soft lithography method. They were aligned and bonded facing each other after their brief treatment with oxygen plasma. The obstacle was 19.51 ( 0.32 µm in height. The whole channel with five steps was ∼13.5 mm in length. Particle Preparation. Polystyrene beads with 6, 10, and 15 µm nominal diameters (6.59, 10.09, and 14.78 µm in exact diameter) were used for the demonstration of the focusing principle. The beads had coefficient of variation (CV) less than 9% for particle size. The beads were prepared in 0.2% Tween 20 aqueous buffer with concentrations of ∼725, 173, and 148 µL-1 for 6, 10, and 15 µm beads, respectively. We then added 13% (w/ v) sucrose to the solution for preventing particle sedimentation. The beads were introduced into the microfluidic devices using a syringe pump. The positions of particles were measured from images captured in the outlet region of 1 mm width. The images were taken with a CCD camera (DS-2MBWc; Nikon Co., Japan) attached to an inverted optical microscope (TS100; Nikon Co.). Cell Culture. Jurkat cells (human T-cell lymphoblast-like cell line) were maintained in RPMI 1640 (JBI, Korea) medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum (Invitrogen, CA), penicillin G (100 U/mL), streptomycin (100 µg/ mL), and L-glutamine (2 mM) at 37 °C in a humidified atmosphere containing 5% CO2 and 95% air. Jurkat cells in suspension had a
Figure 2. (a and b) Optical micrographs showing trajectories of 6 and 15 µm beads passing through the bent obstacles with an extension ratio of 2, respectively. Panel a shows the trajectories of three different beads. The particles passing through the extended region flowed following the direction of the fluid flow (1 and 2 in panel a). In contrast, the particle passing through the slanted region was exposed to the focusing or deviation flows and oscillated following the flows (3 in panel a). (c and d) Focusing behaviors of 6 and 15 µm beads passing through the bent obstacles with an extension ratio of 1.25, respectively. White, black, and green circles in the images of 6 µm beads (a and c) are plotted on the positions of the small beads for clear illustration. The applied flow rate was 1 µL/min.
diameter of 11.0 ( 1.4 µm (n ) 120). For cell focusing experiments, Jurkat cells were centrifuged and resuspended in RPMI medium at a concentration of ∼530 cells/µL. Position Measurement. The images to measure the lateral positions of microbeads were acquired in a resolution of 800 × 600 pixels. The obtained images were converted into binary images and adjusted to the proper threshold value using a MATLAB program. Then, the MATLAB program was used to automatically count the bead positions each image. The lateral positions of cells were measured from the images captured in the expanded outlet region of 1 mm. A commercial image analyzing program, i-Solution (IMT i-Solution Inc., Korea) was used to measure cell positions and cell sizes. RESULTS AND DISCUSSION We tested whether 6 and 15 µm-sized beads can work well on the principle. For this test, we injected the beads into two different microfluidic devices: one with an extension ratio (x1/a) of 2 (Figure 2, parts a and b) and the other with the ratio of 1.25 (Figure 2, parts c and d). The applied flow rate was 1 µL/min. Figure 2 shows overlaid trajectories of 6 and 15 µm-sized beads in the third and fourth steps of the devices. After passing the first step (100 µm in width, x1), 6 and 15 µm-sized beads were focused within 781 and 245 µm in the outlet region (1000 µm in width) from the bottom sidewall. The corresponding widths of the focused particle streams (a) in the first step were 78 and 25 µm for 6 and 15 µm beads, respectively. The 15 µm-sized beads satisfied the focusing criteria of two devices (a < 50 µm) and well focused to the bottom sidewall in the third and fourth steps (Figure 2, parts b and d). The focusing width of 6 µm-sized beads was between the focusing criteria of two devices (50 < a < 80 µm). They satisfy the principle only in the device with an extension ratio of 1.25 (Figure 2c). In the device with the ratio of 2, some particles flow in the extended region without focusing (Figure
Figure 3. (a) Simulated pressure fields and streamlines around lower and upper bent obstacles. The geometric conditions were identical with the channel region, A, in Figure 2b. (b) Plot of the calculated pressure gradients along the line from A to B in (a).
2a). As mentioned before, the extended region has the uniform fluidic resistance to the lateral direction to the fluid flow. The particles passing through the extended region flowed following the direction of the fluid flow (1 and 2 in Figure 2a). In contrast, the particle passing through the slanted region was exposed to the focusing or deviation flows and oscillated following the flows (3 in Figure 2a). This oscillation results from rotating flows (a sequence of deviation, upward, focusing, downward, and deviation flows) (Figure 1c). Following the deviation and focusing flows, the particles passing through the slanted region show the oscillation motions. Figure 3a shows the computed contour of the pressure magnitude using a commercial CFD solver (CFD-ACE+; CFD Research Co., Huntsville, AL). The value of the pressure intensity was normalized for clear illustration. The geometric conditions are identical with those of region A in Figure 2b (the slanted region of 100 µm and the extended region of 300 µm in width). The applied flow rate was 1 µL/min along the y-axis. The streamlines are plotted from the center of the z-axis at the channel inlet. The pressure intensity in the slanted region of the first cross section is higher at the bottom area, and it becomes lower going to the upper area. In the slanted region of the second cross section, the pressure intensity decreases going to the right sidewall. There are no significant variations of the pressure intensity in the extended region due to its isotropic fluidic resistance to the fluid flow. Therefore, the streamlines in the extended region are parallel to the y-axis. In contrast, the lateral pressure variations in the slanted region make the streamlines bend right and left. Especially, the streamlines at the top and bottom area of the obstacles are curved to the right sidewall. Following the lateral flows, particles can be focused to the right sidewall. To investigate the influence of the flow rate, we plotted the calculated pressure gradients along the line from A to B in Figure 3a at two different flow rates of 1 and 4 µL/min. As shown in Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
3037
Figure 4. Widths of the focused particle streams after each extension step in the devices. The extension ratios of the devices were (a) 2 and (b) 1.25. Each data point represents the position of the outermost particle from the bottom sidewall in the outlet channel of 1000 µm width. For each point, more than 1283 particles were measured. The applied flow rate was 4 µL/min.
Figure 3b, the intensities of the pressure gradients increase going to the boundary between the slanted and extended regions and rapidly decrease at the boundary. The intensities of the pressure gradients in the extended region are nearly zero. The intensity of the pressure gradient increases in proportion to the applied flow rate in low-Reynolds-number Poiseuille flow. Even at higher flow rates, particles are exposed to increased lateral flows and focused to the sidewall independent of the flow rate. We characterized the microfluidic devices fabricated for each step with several beads (6, 10, and 15 µm in diameter). The applied flow rate was 4 µL/min. As we increased the flow rate from 1 to 4 µL/min, the 6 and 15 µm-sized beads still kept their focusing widths. After passing the first step, 10 µm-sized beads were focused within 450 µm of the outlet channel (1000 µm in width) from the bottom sidewall. The corresponding width of the focused particle stream (a) in the first step (100 µm in width, x1) was 45 µm. The 10 and 15 µm-sized beads satisfied the focusing criteria of two devices (a < 50 µm) and well focused to the bottom sidewall in every step. The respective position statistics of 6, 10, and 15 µm-sized beads after passing the first step were 393.7 ( 187.2, 178.2 ( 86.1, and 53.5 ( 18.8 µm in the outlet region (1000 µm in width). In the device only with the slant obstacles, the respective focusing efficiencies of 10 and 15 µm-sized beads were 55% and 76% by applying eq 2. After the fifth step of the device with an extension ratio of 2, the focusing efficiencies of 10 and 15 µmsized beads were enhanced to 94% and 98%, respectively (Figure 4a). In the device with an extension ratio of 2, the position statistics of 10 µm-sized beads in the outlet region (1000 µm in width) were 133.3 ( 30.7, 70.5 ( 23.0, 40.9 ( 17.4, and 32.0 ( 8.2 µm after the second to fifth steps, respectively. For 15 µm-sized beads, the position statistics were 26.0 ( 10.2, 18.6 ( 8.5, 13.1 ( 5.1, and 11.6 ( 4.1 µm after the second to fifth steps, respectively. The position statistics of 6 µm-sized beads were 180.4 ( 113.3, 130.2 ( 97.7, 66.4 ( 63.8, and 64.8 ( 85.0 µm after the second to fifth steps, respectively. The 6 µm-sized beads satisfy the principle only in the device with an extension ratio of 1.25 in every step. Their focusing efficiency after the fifth step was enhanced by 3.5-fold compared with the efficiency after the first step (Figure 4b). The corresponding efficiencies were 22% and 76% for the first and fifth steps, respectively. In the device with an extension ratio of 1.25, the position statistics of 6 µm-sized beads in the outlet region (1000 µm in width) were 367.4 ( 129.7, 311.6 ( 103.6, 218.5 ( 71.6, and 108.2 ( 34.3 µm after the second to fifth steps, respectively. 3038 Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
Figure 5. (a) Optical micrographs of the microfluidic channel with five step extensions. The channel is the symmetric structure of the slant and bent obstacles in Figure 1a. (b) Optical micrographs showing the focusing stream of Jurkat cells in the outlet region of 1 mm width. (c) Measured focusing profiles of Jurkat cells. Each focusing plot was obtained from the measurement of more than 719 cells.
For 10 µm-sized beads, the position statistics were 165.0 ( 67.4, 151.0 ( 38.7, 214.5 ( 68.7, and 121.5 ( 29.1 µm after the second to fifth steps, respectively. The position statistics of 15 µm-sized beads were 38.0 ( 8.3, 34.0 ( 8.4, 46.5 ( 16.5, and 21.0 ( 8.5 µm after the second to fifth steps, respectively. The focusing efficiency of the particles can be further improved in the devices with more extension steps. For the cell focusing to the channel center, we fabricated microfluidic devices with the symmetric structure of slant and bent obstacles in Figure 1a (Figure 5a). The increasing channel widths every step in µm were 200, 400, 800, 1600, and 3200. Jurkat cells were introduced into the microfluidic devices at a flow rate of 4 µL/min. After passing the first step, the position statistics of Jurkat cells were 501.7 ( 31.3 µm in the outlet region (1000 µm in width). Jurkat cells (more than 10 µm in diameter) satisfied the focusing criteria of the devices and well focused to the center of the channel. After passing the fifth step, the position statistics of the cells were 502.9 ( 8.7 µm in the outlet region (1000 µm in width) (Figure 5, parts b and c). The focusing variations about the channel center were 6.2% and 1.7% CV for the first and fifth step, respectively. Conventional hydrodynamic focusing methods employ accurate sheath controls to focus cells. For the simultaneous
focusing of multiple cells, it is required to increase the number of sheath controls. Although the hydrophoretic focusing requires relatively long channel lengths, it does not depend on accurate flow controls. Therefore, it can be realized to plug and focus cells without any calibration of flow conditions. Also, the repeated symmetric patterns of slant obstacles will enable one to focus multiple cells in a microchannel without any addition of control units and external forces. CONCLUSION In conclusion, we demonstrated a microfluidic device for sheathless hydrophoretic focusing of microparticles with the advantages of a sheathless method, passive operation, single channel, and flow rate independence. The focusing efficiency of particles under the proposed principle was improved as the number of the extension steps increased. On the basis of this microstructure, the focusing efficiency of the particles can be further improved in the devices with more extension steps. In addition, microfluidic devices that include the symmetric structure
of slant and bent obstacles were successfully applied to focus biological cells to the center of the channel. The hydrophoretic focusing device does not require external potential fields and complex channel networks to induce particle movements. These features can facilitate the integration of the device with cell sorters and microflow cytometry. ACKNOWLEDGMENT This research was supported by the Chung Moon Soul Center for BioInformation and BioElectronics and by the Nano/Bio Science and Technology Program (2006-00955) of the Ministry of Science and Technology, Korea. We thank the Digital Nanolocomotion Center at KAIST for technical support and facility access.
Received for review January 18, 2008. Accepted February 18, 2008. AC8001319
Analytical Chemistry, Vol. 80, No. 8, April 15, 2008
3039
