SEARCH

SEARCH BY CITATION

Keywords:

  • lab-on-a-chip;
  • microflow cytometry;
  • sheatheless focusing;
  • microfluidic

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Cell focusing into a narrow stream is an essential step prior to counting and sorting cells in microfluidic devices for flow cytometry and cell sorting applications. Hydrodynamic focusing techniques, however, rely on the need for large volumes of sheath liquid and complex mechanical setup for flow control, preventing miniaturization of the systems. Although microfluidic methods based on active or passive particle control offer sheathless and efficient focusing, they often accompany fabrication complexities or bulky external setups, and operate in a certain range of flow rates. We present here a microfluidic device to focus cells into a narrow stream. The device employs hydrophoresis to guide cells by locally patterned slanted grooves, and channel expansion to improve focusing efficiency and produce a narrow stream of cells. This device principle allows easy improvement of focusing efficiency by adding more expansion steps. Adjusting channel expansion also ensures successful cell focusing without defocusing by inertial effects even at high Reynolds numbers. Using this device, we successfully produced a narrow stream of cells having size variation of >11% in a coefficient of variation (CV), achieving a narrow cell stream with a focusing variation below CV of 3.0%. © 2013 International Society for Advancement of Cytometry


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Flow cytometry is a standard analytical tool to measure multiple characteristics of individual cells as well as identify specific cell types in heterogeneous populations [1-5]. Although it provides reliable multiparametric analysis, it is still expensive, mechanically complex, and has a large instrument footprint. A regular and portable examination of blood cells is required for uses in emergency rooms as well as potential uses by everyday health care practitioners [5]. The needs of point-of-care testing have led to development of inexpensive, portable microflow cytometry that incorporates small optical components and microfluidic technologies [5].

One of the challenges for miniaturization of flow cytometry is to eliminate the need for large volumes of sheath liquid and complex mechanical setup for flow control. Modern conventional flow cytometry has been designed to have a flow tube that allows focusing of particles in a tight streamline based on the principle of hydrodynamic focusing with sheath flows [6, 7]. This mechanism enables an accurate measurement without clogging of the tubing and impairment of signal collection efficiency by precisely positioning particles in an optical interrogation region, but leads to large instrument footprint including containers for sheath and cleaning fluids.

To address the shortcomings, microfluidic approaches have been developed with simpler mechanical setups or without the use of sheath liquid. Hydrodynamic focusing has been achieved with a simple, single sheath control by utilizing transverse flows generated by chevron-shaped grooves [8, 9] and dean flows induced in a contraction-expansion array microchannel [10] that fully wrap a sheath stream around a sample stream. These approaches enable three-dimensional focusing, but still require a volume of sheath liquid at least ten times larger than a sample volume. Although the use of air as a sheath fluid can remove sheath liquid consumption [11, 12], it requires an additional vacuum setup for air supply. For sheathless focusing, external physical fields such as electric [13, 14] and acoustic fields [15-17] have been employed to exert physical forces upon particles and induce focusing at an interrogation position. These approaches, however, accompany fabrication complexities (i.e. metal patterning and device assembly) that often determine the cost of the device. Field-free, sheathless focusing schemes have been recently demonstrated based on passive principles for particle ordering and sorting. Bifurcating channels for hydrodynamic filtration have been used for sheathless focusing by utilizing selectively drained, particle-free flows as sheath flows [18]. However, the utility of this approach to focus cells having large size variation (>10% in a coefficient of variation (CV)) has not been demonstrated and the overall volume throughput remains low (1 - 4 µL/min). Inertial focusing techniques have demonstrated the ability to generate a single-stream cell train in a high-throughput manner (≈300 µL/min) [19, 20], however focusing efficiency was varied by particle size and flow condition because induced secondary flows can perturb a focusing stream. Grooved channels for hydrophoresis have been implemented for sheathless cell focusing but thus far are limited by low throughput capacity (4 µL/min) and non-trivial assembly process (i.e. alignment of polydimethysiloxane (PDMS) substrates on a microscale) [21, 22]. Taken together, the developed microfluidic technologies for cell focusing offer high focusing performance under well-controlled conditions. In active platforms, particle focusing can be achieved by precise exposure of external forces to cells and adjustment of the susceptibility of the suspending medium [13, 14, 16]. In passive platforms, focusing efficiency can be highly affected by flow rate. For example, inertial focusers, a state of the art technology for sheathless cell focusing work at very high Reynolds number (Re 50-80) [10, 19, 20], which might not be versatilely applicable for some microfluidic systems working at moderate Re.

We describe a microfluidic device containing successive arrays of slanted obstacles with exponentially increasing widths to focus cells without requiring sheath control, external fields, and non-trivial fabrication process (Fig. 1). The obstacles placed on the channel top alter the streamlines and direct the trajectories of cells that result in sheathless cell focusing. While previous studies presented the possibility of focusing cells based on this effect [21, 22] these approaches rely on cumbersome alignment process of two PDMS substrates, and no high-throughput operation has been implemented. The proposed microfluidic device is specifically designed to allow the obstacles to be formed on the top wall, eliminating the alignment process. This design feature allows easy improvement of focusing efficiency by adding more steps. In the previous hydrophoretic focusing methods, the difficulty of aligning two PDMS substrates for device fabrication limits the enhancement of focusing efficiency [22]. Adjusting the expansion ratio also ensures successful cell focusing without defocusing by inertial effects even at high Re. We demonstrated the utility of our device for focusing of polystyrene particles and mammalian cells having size variation over 11% in CV, achieving focusing efficiencies over 99.3% in a throughput of 80 µL/min or approximately 344 cells/s.

image

Figure 1. Schematics of a microfluidic device to create a narrow particle stream. (a) Randomly distributed particles at the inlet focus to a narrow stream by a combination of hydrophoretic ordering and channel expansion. (b) Particle focusing autonomously occurs by hydrophoresis within a width of wf and focusing enhancement is further achieved by channel expansion. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com.]

Download figure to PowerPoint

Materials and Method

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Device Design and Fabrication

Master molds for the microfluidic channels were made by SU8 (MicroChem Corp., Newton, MA) micropatterning on silicon wafers using two-step lithography techniques [23]. The device comprises of 50 grooves each step that were defined with θ = 140°, pg = 50 µm, pt = 50 µm, hg = 19 µm, and ht = 25 µm (Figs. 2 and 3). The channel width is exponentially increased every step in a ratio of ≈1.69 and expanded from 200 µm (0 step) to 4,660 µm (6 steps). PDMS was thermally cured on the master and bonded with glass slides using plasma bonding. The channels were blocked with 1% BSA to prevent potential particle adhesion on channel surfaces.

image

Figure 2. Optical micrograph images of the focusing device with 6 steps. The channel width is exponentially increased every step in a ratio of ≈1.69 and expanded from 200 µm (0 step) to 4,660 µm (6 steps). The total length of the channel is ≈39 mm. Cell guiding through locally patterned slanted grooves autonomously occurs by hydrophoresis. The design enables effective cell focusing in the controlled area, the slanted grooves. Scale bars, 200 µm.

Download figure to PowerPoint

image

Figure 3. Flow simulations. (a) Normalized pressure plots on yz cross-sections in 2 steps. Because of geometry symmetry, only half of the channel is shown here. Scale bar, 50 µm (b) Simulated streamlines below the grooves. The flow streamlines are going upward at the center that induces steric interactions between particles and grooves. Scale bar, 20 µm (c). The lateral pressure gradient is locally generated in the slanted region and attenuated as going to the extended region. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Download figure to PowerPoint

Cell and Particle Preparation

HL60 and K562 cell lines were cultured in Iscove's modified Dulbecco's medium (Welgene Corp., Korea) supplemented with 10% fetal bovine serum (Invitrogen, CA) and penicillin-streptomycin (Welgene). Cells were washed with fresh media before each experiment to remove cell debris. Polystyrene microparticles (Invitrogen) were resuspended in 0.1% Tween solution. The cells and particles were flowed into the device using a syringe pump (KD scientific Inc., Holliston, MA) in a concentration of ≈40 particles/µL, unless specified. Focusing behavior was recorded at 1,200 frames per second using a high-speed camera (EX-F1; CASIO, Japan) mounted on an inverted microscope (TS100; Nikon, Japan). The positions of cells at the end of the channel were measured using ImageJ software (NIH). Cell size was manually measured using ImageJ software. Cell diameter was calculated from a measured cell area. More than 179 cells were measured for each cell type.

Numerical Simulation

Flow simulations were performed to calculate the pressure gradient along the lateral direction of the channel and visualize streamlines in the channel. Commercial computational fluid dynamics software (CFD-ACE+; ESI, Huntsville, AL) was used to solve a three-dimensional model in the “Flow Mode.” A structured grid in the same dimension with the channel was generated. No-slip boundary conditions were applied at the channel walls.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Microfluidic Device Design

Hydrophoresis employs a helical flow pattern that deflects particles laterally, induces particle-surface interactions, and finally results in particle ordering to a certain equilibrium position. For hydrophoretic focusing, the design rules are as follows:

  1. For particles to be deflected laterally in the Stokes flow regime (Re < 1), the grooves should have an oblique angle to the longitudinal direction and the trench must have sufficient depth (ht) [24, 25], Re is the Reynolds number, a dimensionless measure of the ratio of inertial forces to viscous forces. Re is defined as Re = ρνL/µ, where ρ is the density of the fluid, ν is the flow velocity, L is a characteristic length, and µ is the dynamic viscosity of fluid. The anisotropy of the grooves causes a lateral pressure gradient and lateral displacement of fluid streams, thereby influencing particle lateral migration. Since the intensity of the pressure gradient is proportional to ht at a given hg [25]; ht should satisfy hg < ht to create sufficient transverse flows, where hg is the groove gap (Figure 3).
  2. For particles (d in diameter) to be ordered to a certain lateral position, the groove gap should be typically in d < hg < 2d [25]. The gap must be greater than the particle diameter to prevent channel clogging, but not so large that the particles will follow individual streamlines. When a particle approaches the channel center, the streamline that it follows enters into the trench (Figure 3). However, the particle is displaced into another streamline below it, and thus keeps out of the trench. At the next ridge, the particle is further displaced into another streamline below the present one, and so on. Thus, the particle should have a certain minimum size to stay ordered by hydrophoresis. At a given hg of 19 µm, cells having a diameter range from 10 to 18 µm can be focused by hydrophoresis.
  3. For particles in different size ranges to be focused to the same lateral position, the channel centerline, the channel width should be gradually expanded as going downstream (Fig. 1) [22]. When the particles are ordered by hydrophoresis, their equilibrium positions are varied by size [25]. For focusing to the same position, the flow paths of the particles can be guided by locally patterned slanted grooves, while expanding the channel width. If the particles have a focusing width (wf) in the slanted region, the channel can be expanded to the width of xn = x0 × (x0/wf)n, xn is the channel width of the nth step. In the expanded channel, the particles are further focused within wf, and thus focusing improves with increasing number of steps [22].
  4. The oblique angle of the grooves is a minor consideration for design of hydrophoretic focuser. For particles to be deflected laterally, the grooves should have an oblique angle to the longitudinal direction. As decreasing the oblique angle, a lateral shift of the position distribution of particles was observed [25], mostly shifting outward. The angle can thus determine wf and the expansion ratio (x0/wf) for successful focusing, but hydrophoretic ordering is determined by the height of the groove gap [25].

Device Characterization

We first tested the focusing capability of the device with two different sizes of polystyrene beads (diameter of 15.2 µm/CV 6.8% and 10.0 µm/CV4.0%). The bead experiments were conducted at a flow rate of 20 µL/min (Re = 6.3) (Figs. 4 and 5). We determined focusing variation for all experiments by measuring the positions of individual particles in the observation region (width of 800 µm). The expansion ratio (x0/wf) of the channel width was set to 1.69, and thus the channel width was gradually expanded from x0 = 200 µm to x5 = 4,660 µm (Fig. 2). At this condition, particles of wf < 118.3 µm will follow the design guide and their focusing improves with increasing number of steps. Focusing in the hydrophoretic device occurs in the manner that two focusing streams merge into the channel center as increasing steps (Figs. 4 and 5). All beads were successfully focused in the channel center after 6 steps, achieving a narrow focusing stream with a variation below 3.0% in CV (Fig. 6). The focusing positions were 407.0 ± 12.3 and 402.4 ± 6.0 µm for 10.0 and 15.2 µm beads, respectively. The absolute width of the focused bead stream (i.e., the width in which >90% of cell are flowing in) is 27.4 and 18.2 µm for 10.0 and 15.2 µm beads, respectively. We note that even though 10.0 µm polystyrene beads have a wider focusing width of 138.5 µm than the critical width of 118.3 µm, they all were focused in the center tightly after 6 steps. This is likely resulted from that the pressure gradient is still present even over the boundary between the slanted and expanded regions (Fig. 3). The width where less than half maximum pressure gradient is attenuated is ∼10 µm away from the boundary (Fig. 3c). Within the range, the particles can laterally migrate and follow the design guide. The critical expansion ratio can be, therefore, corrected with x0/(wf - 10 µm).

image

Figure 4. Fluorescence streak images of 15-µm and 10 µm-diameter particles in the grooved microchannel. The particles are successfully guided by locally patterned slanted grooves and focusing efficiency is improved as increasing number of steps. Re is 6.3 for both particles. The channel walls are outlined with dotted lines for clarity. Scale bars, 200 µm.

Download figure to PowerPoint

image

Figure 5. Continuous particle focusing. Particle focusing improves with increasing number of steps. Particle measurement was performed in the outlet region for observation (width of 800 µm) at Re = 6.3 and plotted in the focusing histograms. Each focusing histogram was obtained from measurement of more than 401 particles. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Download figure to PowerPoint

image

Figure 6. Improvement of bead focusing with the stages of focusing. The CVs for particle distribution decrease with increasing number of steps at Re = 6.3. The CVs after 6 steps are 3.0 and 1.5% for 10.0 and 15.2 µm beads, respectively. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Download figure to PowerPoint

Sheathless Focusing of Mammalian Cells

To further validate the utility of our device for focusing of cells having large size variation, we examined the focusing ability of the device with two different cell types (K562: diameter of 14.8 µm/CV 11.0% and HL60: 11.7 µm/CV12.7%) (Fig. 7). As injected at a flow rate of 20 µL/min that corresponds to Re = 6.3, wf for both cells was smaller than the critical width of 118.3 µm for hydrophoretic focusing (wf = 88.2 µm for K562 cells and 105.9 µm for HL60 cells) such that all the cells can be focused by hydrophoresis in the device, satisfying the design criterion. All cells were found on the center of the microchannel after 6 steps; the equilibrium positions were 410.4 ± 7.4 and 409.1 ± 8.5 µm for HL60 and K562 cells, respectively (Fig. 7). The corresponding CVs were 1.8 and 2.1% for HL60 and K562 cells, respectively (Fig. 8). The absolute width of the focused cell stream (i.e. the width in which >90% of cell are flowing in) is 20.5 and 21.6 µm for HL60 and K562 cells, respectively.

image

Figure 7. Continuous focusing of cells having large size variation (> 11% in CV) at Re = 6.3 and 25.2. Focusing behavior was not affected by change in the flow rate. High focusing efficiency with a minimum CV of 1.8% after 6 steps was achieved for both cells and both Re. Each focusing histogram was obtained from measurement of more than 406 cells. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Download figure to PowerPoint

image

Figure 8. Improvement of cell focusing with the stages of focusing. The CVs for cell distribution decrease with increasing number of steps at Re = 6.3. The CVs after 6 steps are 1.8 and 2.1% for HL60 and K562 cells, respectively. [Color figure can be viewed in the online issue which is available at wileyonlinelibrary.com]

Download figure to PowerPoint

Next, we examined the Reynolds number to assess the influence of inertial effects on the focusing process. As increasing Re to 25.2, the equilibrium positions of K562 cells after 0 step is shifted more to the sidewalls, from a binary distribution of 282.8 ± 15.6 and 511.7 ± 12.6 µm (Re = 6.3) to 259.8 ± 13.4 and 520.4 ± 12.2 µm (Re = 25.2). This is attributed from that inertial effects such as inertial lift forces and Dean flow at high Re can affect the focusing process, and the positions of cells can be shifted toward new particular equilibrium positions, mostly toward the sidewalls [25]. Although such shift occurred at Re = 25.2 and 0 step, wf of 83.5 µm for the cells was still smaller than the critical width for focusing and all the cells were focused by hydrophoresis in the device. Re is calculated from a given flow condition in 0 step and further decreased by ∼20 times as going downstream. Inertial effects were thus negligible in the 6 steps. After 6 steps, there was no difference in the focusing position by Re and all the cells were focused well to the channel center: 409.1 ± 8.5 and 406.4 ± 11.0 for Re = 6.3 and 25.2, respectively. At Re = 25.2, the focusing variation after 6 steps were 2.2 and 2.7% for HL60 and K562 cells, respectively. We note that the CVs achieved here were calculated from the position data for cells.

High-throughput focusing was then performed by increasing the flow rate and cell concentration without influencing the focusing efficiency. At 87 µL/min and 230 cell/µL, a high throughput of 344 cells/s was achieved for HL60 cells, while achieving a narrow cell stream with CV of 2.9% (n = 410 cells; Movie S1). Even in the wide observation region (800 µm in width), the full width at half maximum (FWHM) calculated is 27.8 µm indicating a narrow focusing distribution. FWHM was calculated from the width between the positions at the half maximum of the Gaussian fit of the cell distribution. Since there is enough room on both sides of the cell stream, we can further decrease the width of the observation channel to produce a narrower cell stream, achieving higher focusing efficiency. Even though the throughput performance can be sufficient for microflow cytometry applications [26], taking 29 sec to count 10,000 cells, it is still lower than inertial focusers, a state of the art technology for microfluidic focusing.[19,20]. As increasing Re, wf can be widened by inertial effects that can lead to focusing failure. This challenge can be simply resolved by adjusting the expansion ratio. Adjusting the expansion ratio ensures successful cell focusing without defocusing by inertial effects even at high Reynolds numbers. The expansion ratio (x0/wf) of the device can be decreased to satisfy the design rule even at high Re. The focusing efficiency can be maintained by increasing number of steps. Since the cell-free streams on both sides of the focused cell stream can be utilized as sheath flows, the CV values achieved in a wide channel width can be maintained even at a narrow channel width.

Focusing in the vertical direction as well as the lateral direction is essential for optical interrogation by a tightly focused laser beam, since a broad distribution of cells in the vertical direction can impair illumination and signal collection efficiencies. Even though more work is required to characterize the performance for vertical focusing, an ability of our device for vertical focusing can be presumed as follows:

  1. The narrow depth of 19 µm in the observation region ensures confinement of HL60 cells in a detection volume with a depth of ≈7.3 µm.
  2. Hydrophoretic ordering also enables vertical focusing by tight alignment of cells to the groove surface [27] that results in flowing HL60 cells at the same velocity of 34.3 ± 1.5 mm/s (n = 52) at Re = 6.3.

Conclusions

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

In summary, we have demonstrated a microfluidic method that enables sheathless and field-free cell focusing in a relatively wide range of flow rates. The method employs hydrophoresis to guide cells by locally patterned slanted grooves, and channel expansion to improve the focusing efficiency and produce a narrow stream of cells. The entire focusing process is not significantly limited by flow conditions, and performed in a passive manner and single-layer PDMS device such that other microfluidic components (i.e. cell sorting and cell analysis) can be integrated with ease. The proposed focusing method can make easier to fabricate point-of-care devices due to its simple design. The degree of focusing achieved (≈30 µm in focused stream width) is also good enough for the applications that the design requirement for cell focusing is ≈100 µm in width.[26] We anticipate that the detailed design guidelines described here will bring the microfluidic device within the reach of any laboratory and facilitate implementation of controlling particle position in a continuous flow for point-of-care diagnostic applications.

Literature Cited

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and Method
  5. Results and Discussion
  6. Conclusions
  7. Literature Cited
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
cytoa22395-sup-0001-suppmovie1.MOV1213KSupporting Information Movie S1. Sheathless focusing of HL60 cells after 6 steps in the observation region at 87 μL/min and 230 cell/μL. This video was taken at 1200 fps (0.025× actual speed).

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.