Journal of Micromechanics and Microengineering You may also like PAPER Microfluidic hydrodynamic focusing for highthroughput applications - A simple approach for the fabrication of 3D microelectrodes for impedimetric sensing Mustafa Tahsin Guler, Ismail Bilican, Sedat Agan et al. To cite this article: Jingjing Zhao and Zheng You 2015 J. Micromech. Microeng. 25 125006 - A cell counting/sorting system incorporated with a microfabricated flow cytometer chip Sung-Yi Yang, Suz-Kai Hsiung, YungChing Hung et al. View the article online for updates and enhancements. - Sheath-assisted hydrodynamic particle focusing in higher Reynolds number flows Nishtha Panwar, Peiyi Song, Swee Chuan Tjin et al. This content was downloaded from IP address 140.226.13.85 on 07/01/2025 at 17:06 Journal of Micromechanics and Microengineering J. Micromech. Microeng. 25 (2015) 125006 (12pp) doi:10.1088/0960-1317/25/12/125006 Microfluidic hydrodynamic focusing for high-throughput applications Jingjing Zhao1,2,3,4 and Zheng You1,2,3,4 1 Collaborative Innovation Center for Micro/Nano Fabrication, Device and System, Tsinghua University, Beijing 100084, People’s Republic of China 2 State Key Laboratory of Precision Measurement Technology and Instrument, Tsinghua University, Beijing 100084, People’s Republic of China 3 Department of Precision Instrument, Tsinghua University, Beijing 100084, People’s Republic of China 4 Beijing Laboratory for Biomedical Detection Technology and Instrument, Tsinghua University, Beijing 100084, People’s Republic of China E-mail: Yz-dpi@mail.tsinghua.edu.cn Received 17 July 2015 Accepted for publication 2 September 2015 Published 20 October 2015 Abstract Microfluidic hydrodynamic focusing is critical for chip-based bioanalytical systems to increase throughput and sensitivity, especially for microflow cytometers, enabling a sample flow to be confined to the center of a microchannel with a narrow cross-section. Current microfluidic hydrodynamic focusing designs are usually unable to maintain stable focusing in high flow velocity conditions, resulting in a large cross-section or even failed focusing. To overcome this challenge, this paper aims to develop a design that can achieve effective microfluidic hydrodynamic focusing at high velocity with favorable performance. For this purpose, specially designed structures and arc-shaped channels are used. Two focusing regions are modeled and optimized mathematically, and flow behavior is investigated using numerical simulations. The functional relationship between flow rates and the cross-sectional dimensions of the focused sample flow is explored, and a measurement method for testing the dimensions is developed. The design is implemented in glass chips and characterized experimentally. In a rectangular channel with a cross-section of 300 μm × 150 μm the sample flow can be focused down to 5–11 μm horizontally and 10–21 μm vertically at a roughly constant velocity of 4.4 m s−1 when the sample flow rate varies between 10 and 60 μl min−1. Effective focusing is accessible within a wide velocity range from 0.7 to 10 m s−1. The experimental results validate that the focusing design performs better than existing microfluidic designs at high velocities, while its performance is close to that of the designs used in conventional flow cytometers with much less volume and a simpler structure. The focusing design can serve as the basis for microflow cytometers or it can be integrated into various microfluidic systems where complete focusing is needed. Keywords: hydrodynamic focusing, microflow cytometer, microfluidics (Some figures may appear in colour only in the online journal) 1. Introduction of their large size, complicated structure and high cost, as well as the high-level competence requirements for operation personnel. Recently, advances in microfluidic technology have raised the possibility of developing compact, user friendly, portable and inexpensive microflow cytometers. Microfluidic focusing of biological particles or a sample flow is an essential functionality for microflow cytometers and is critical to As powerful instruments for biomedical research and clinical diagnosis, conventional flow cytometers have been widely used and developed for more than 50 years [1]. Though capable of analyzing biological particles accurately and rapidly, conventional flow cytometers are not easy to use because 0960-1317/15/125006+12$33.00 1 © 2015 IOP Publishing Ltd Printed in the UK J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 horizontal focusing regions. The sheath flow rates for vertical and horizontal focusing are controlled separately, mainly to enhance the ability to adjust the vertical and horizontal dimensions of the focused sample flow. Because of the limitations of planar microfabrication technology, channels can be curvy in the horizontal direction but not in the vertical direction, meaning that the vertical focusing region has a larger local loss coefficient than the horizontal focusing region. Taking into consideration the fact that the flow rate and velocity in the first focusing region are much lower than those in the second, vertical focusing is implemented before horizontal focusing in order to maintain smooth flows and improve focusing stability. There are four symmetrical sheath flow channels and a main channel in the vertical focusing region, where the sample flow is confined vertically by the four vertical focusing sheath flows. Then, in the horizontal focusing region two symmetrical sheath flow channels merge into the main channel, all of the channels within the same layer, and the sample flow is squeezed between the two horizontal focusing sheath flows. After being vertically and then horizontally focused, the sample flow is completely encircled by sheath flows and confined in the center of the main channel, and it becomes very narrow. Laminar flows are critical for hydrodynamic focusing and smooth transitions are necessary to lead the sheath flows and the sample flow into the main channel. Hence microfluidic channels are designed with curved profiles, which are useful for keeping flows stable, typically at high velocities. The details of the design are presented in the following sections. The relationship between the flow rates and the dimensions of the focused sample flow is investigated. According to the hydrodynamics principle [34], the velocity profile of a fully developed laminar flow in a straight rectangular channel can be formulated as equation (1) (see figures 1(b) and (c)), the performance of cytometers for counting, analyzing and sorting biological particles [2]. To date, a number of focusing methods based on different physical mechanisms have been developed, with the main ones being hydrodynamic focusing; electrokinetic focusing (electrophoresis [3], electroomosis [4, 5] and dielectrophoresis [6–8]); acoustic focusing [9, 10]; inertial focusing [11–14]; and microstructure focusing (micro grooves [15], stepped channels [16, 17], micro weirs or obstacles [18–20]). In comparison to other methods, hydrodynamic focusing benefits from its low sensitivity to flow rates and from having no requirement for particular electrokinetic or acoustic properties for the biological particles or fluids, while it can operate effectively at high flow velocities, easily achieving a high throughput. Due to sheath flow demands and the limitations of planar microfabrication techniques, hydrodynamic focusing in microfluidic chips is usually implemented in the horizontal direction [21–23]. It is difficult to accomplish vertical focusing or three-dimensional (3D) focusing (vertical and horizontal). Even so, some 3D hydrodynamic focusing designs have been established in microfluidic chips by manipulating multiple sheath flows to fully envelop the sample flow [24–33]. But the focusing performance of these designs is inferior to that of conventional flow cytometers with slower sample flow rates and larger focused dimensions, as listed in table 1. In addition, these designs generally require complex fabrication processes, some of which are incompatible with conventional microfabrication processes. To overcome these challenges, this work studies a 3D microfluidic hydrodynamic focusing design for high-throughput applications, in particular microflow cytometers. It aims to achieve a similar focusing performance to conventional flow cytometers. The design mainly comprises two parts: vertical and horizontal focusing regions. The regions are modeled, analyzed, simulated and optimized for the purpose of keeping flows laminar over a wide velocity range, particularly at high velocities. High velocities are essential for high throughput. In addition, the functional relationship between flow rates and the cross-sectional dimensions of the focused sample flow is explored, which can be used to control or predict the focusing results, and a measurement method for testing the dimensions is developed. For experimental testing, the design is established in a microfluidic chip formed by laminating five micromachined glass plates. The chip then exhibits a focusing performance close to that of BD Accuri C6, a conventional flow cytometer. It is proved that this 3D focusing design is effective and powerful. Furthermore, the design process, the method for predicting focusing performance and the measurement method can be useful for designing other microfluidic systems. ∞ cosh[(2n + 1)πy /b] ⎫ 4b 2 ⎛⎜ dP ⎞⎟ (−1)n ⎧ ⎨1 − ⎬ − ∑ 3⎝ 3 cosh[(2n + 1)πa /2b] ⎭ μπ dx ⎠ n = 0 (2n + 1) ⎩ (2n + 1)πz × cos , (1) b v= where a is the channel width, b is the channel height and P is the pressure. Obviously, the velocity is along the channel direction and the velocity component in the cross-section is zero. In addition, the maximum velocity occurs at the center. Equations (2) and (3) respectively present the formulas of the maximum velocity VMAX and the average velocity VAVR, as shown below: 4b 2 ⎛⎜ dP ⎞⎟ (−1)n − ∑ 3⎝ dx ⎠ n = 0 (2n + 1)3 μπ ⎧ ⎫ 1 ⎬ × ⎨1 − ⎩ cosh[(2n + 1)πa /2b] ⎭ ∞ VMAX = v y = z = 0 = 2. Design 2.1. Principal design VAVR = The hydrodynamic focusing design is composed of three layers of rectangular microfluidic channels, whose concept is illustrated in figure 1(a). In terms of function, the focusing design is divided into two major parts: the vertical and 2 ⎛ ⎞ ∬A vdAA = 12bμπ 3 ⎝− ddPx ⎠ ⎜ ⎟ ∞ ⎧ 192b tanh[(2n + 1)πa /2b] ⎫ ⎬, × ⎨1 − 5 ∑ π a n=0 (2n + 1)5 ⎩ ⎭ 2 (2) (3) J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Table 1. 3D hydrodynamic focusing performance of microfluidic chips and conventional flow cytometers. Focusing channel a b Dimensions (μm) Focused sample flow Cross-section Hydraulic diameter (μm) Flow rate (μl min−1) Velocity (m s−1) Shape Dimensions (μm) Ellipse Circle 23b × 8.5b 30 (Designed value) 350b × 80b 13b × 25b 20b × 30 10 × 10 4 × 6, 7 × 6, 12 × 10, 14 × 13 5 × 5, 5 × 6, 9 × 8, 10 × 13, 22 × 11 6.5a, 11a, 15a 10, 16, 22 Type Shape Micro 1 [26] Micro 2 [27] Micro 3 [28] Micro 4 [29] Micro 5 [30] Micro 6 [31] Micro 7 [32] Rectangle 100 × 95 Square 300 Rectangle 1000 × 500 Rectangle 100 × 50 Rectangle 100 × 110 Square 100 × 100 Quasi-circle 125 × 125 97 300 670 67 105 100 125 3.3 60 0.16 3 2.5 1 up to 3 0.1, 2, 16, 20 0.13a–0.15a Rectangle 0.15 Spindle Square 3 Quasi-circle Micro 8 [33] Quasi-circle 100 × 100 100 0.2, 1, 10, 24, 48 3 Ellipse BD Calibur BD Accuri C6 Rectangle Circle 430 × 180 200 254 200 12, 35, 60 14, 35, 66 Circle Circle 6 3a The values are calculated roughly, using the relationship between the flow rate, velocity and dimensions. The dimensions are measured from the published microscopic images. Figure 1. (a) Schematic diagram of the 3D hydrodynamic focusing design. (b) Flow in a rectangular channel. (c) Velocity profile in a straight rectangular channel, the blue-to-red rainbow represents the flow-velocity magnitude distribution. (d) Flows distribution in a channel cross-section. ⎧ ⎪QS = A v dA = VSwh S ⎪ ⎪ v dA ⎨Q VS = , (4) AVS ⎪ ⎪ v dA ⎪QHS = ⎩ AHS where A is the cross-sectional area of the channel. It is assumed that after 3D focusing the sample flow and sheath flows are represented by the rectangular cross-sections in figure 1(d). Combining this assumption, the velocity profile and mass conservation, the relationship between the flow rates and the cross-sectional dimensions is given by ∬ ∬ ∬ 3 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 where w and h are respectively the width and height of the focused sample flow, QS is the sample flow rate, VS is the sample flow velocity after 3D focusing, QVS is the total flow rate of the vertical focusing sheath flows, QHS is the total flow rate of the horizontal focusing sheath flows, and AS, AVS and AHS represent the cross-sectional areas of the sample flow, vertical focusing sheath flows and horizontal focusing sheath flows respectively. Absolutely, the distributions of the sample flow and the sheath flows described in figure 1(d) are very simple, without taking account of details of the interfaces between neighboring flows. However, the relationship can give an approximation to the above parameters and predict the focusing performance. This design aims to achieve a similar focusing performance to conventional flow cytometers, specifically in terms of sample flow rate, sample flow velocity and the dimensions of the focused sample flow. Referring to conventional flow cytometers, the main channel is designed to be 300 μm in width and 150 μm in height, with a hydrodynamic diameter of 200 μm. For fabrication convenience, all the microchannels are 150 μm in height. According to equations (2) and (3), the maximum velocity VMAX at the center of the main channel is 2.05 times the average velocity VAVR of the fluid. In addition, the sample flow velocity VS is almost equal to the maximum velocity VMAX, because of the central position and the narrow profile of the focused sample flow. To achieve high throughput, the sample flow rate QS is set within a range of 10–75 μl min−1. For a narrow flow, the cross-sectional dimensions of the focused sample flow are expected to be below 22 μm in both the horizontal and vertical directions. Assuming the focused sample flow is with a square cross-section, several sets of parameters are designed (as listed in table 2) that conform to the functional relationship established above. Sets 1–3 are chosen according to the rates of conventional flow cytometers in table 1, to test the focusing performance when the sample flow rate varies between 10 and 60 μl min−1 with roughly constant velocities. In sets 1, 4 and 5, the flow rate ratios among QS, QVS and QHS are the same, theoretically leading to the same dimensions of the focused sample flow. The three sets will demonstrate that the flow rate ratios are available to control the sample-flow cross-section dimensions. Set 6 is to further explore the potential of effective focusing at high velocities. These sets will be used to investigate the focusing characteristics through numerical simulations and practical experiments. Table 2. Five sets of flow rates to test focusing performance. Flow rate (μl min−1) Flow velocity (m s−1) Dimensions No. QS QVS QHS VS VAVR w = h (μm) wh (μm2) 1 2 3 4 5 6 60 30 10 30 10 75 360 260 160 180 60 650 5800 5900 6000 2900 967 14750 4.44 4.13 4.63 2.22 0.74 10.33 2.30 2.29 2.29 1.15 0.38 5.73 15 11 6 15 15 11 225 121 36 225 225 121 The curved profile of sheath flow channels is the key to keeping flows stable, so it is modeled and optimized. The inner and outer curves of one sheath flow channel are two circular arcs, and circles of the arcs are tangent to the main channel at points Ci and Co respectively. The radii of the two arcs are Ri and Ro, the central angles are α and (α − αo), and the centers are points Oi and Oo (Oi is out of sight). Lo is the distance between the points Ci and Co. Ls is the length of the downstream main channel. Ws approximately represents the width of the sheath flow channel at the confluence of the three flows. Wsh represents the width of the sheath flow channel at the entrance. θ is the angle between the y-direction and the cross-section of the sheath flow channel at the entrance. W, d, β and αo are auxiliary parameters, defined as shown in figure 2. The parameters mentioned above can be expressed by equation (5). It shows that Ri, Ro, α, αo and Lo are a set of independent geometrical parameters for completely describing the design. ⎧ W ⎪ β =arctan Lo ⎪ Lo ⎪ ⎪ Ri = sin 2β ⎪ ⎪ d = R (1 − cos α ) o o ⎨ . ⎪Ws = W2 + (Ro sin αo )2 ⎪ ⎪Wsh = [L o + (Ro − Ri ) sin α]2 + [(Ri − Ro )(1 − cos α)]2 ⎪ ⎪ θ = arctan L o + (Ro − Ri ) sin α (5) ⎪ (Ri − Ro )(1 − cos α) ⎩ For a more reasonable design, the effects of the parameters on focusing performance are analyzed. Numerical simulations (using COMSOL Multiphysics software) prove that increasing Ri and Ro and keeping α below 45° are practical measures to reduce secondary flows, which can aggravate flow disturbances and deviate the sample flow from the central position. Specifically, Ri and Ro are at least three times as large as Ws. In addition, increasing Lo and decreasing β can improve the stability and eliminate vortices in the main channel. Where the sample flow and the two sheath flows meet, the pressure difference between them should be reduced in order to prevent mixing and achieve effective focusing. Thus the velocities of the three flows should be similar and the ratio of QHS/2Ws to QA/b close to one. Wsh and θ need to be close to Ws and α respectively with respect to smooth flowing. When considering glass 2.2. Horizontal focusing design The microfluidic structure of the horizontal focusing region is simpler than that of the vertical focusing region, and it is discussed first. Figure 2 shows the schematic of the horizontal focusing region. Along the two symmetrical curved channels, sheath flows turn smoothly to the direction parallel to the main channel and gradually confine flow QA horizontally. The flow rate of each horizontal focusing sheath flow is QHS/2 and the rate of the fluid flowing into the main channel is QA (QA = QS + QVS). 4 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 2. Schematic of the horizontal focusing region. horizontal focusing design featuring straight sheath flow channels, as shown in figure 3. The cross-sectional dimensions of the channels of the commonly used design in figure 3(a) are all 300 μm × 150 μm, and are the same as those for the main channel in figure 3(b). The flow rates in the two conditions are the same and the average velocity after focusing is 2.3 m s−1 for both. The focusing fails in the commonly used design (effective focusing can be achieved when the average velocity is lower than 0.3 m s−1). This is mainly because the sample flow is severely affected by the two sheath flows, and the sample flow can even be broken up into several small flows as the velocity increases. There are two factors involved: first, the straight sheath channels cannot reduce the y-direction momenta brought in by the sheath flows; second, the velocity (dynamic pressure) difference between the sample flow and the sheath flows is large. In our design, the symmetric momenta and the velocity difference are reduced respectively by the curved shapes and the larger widths of the sheath channels. Obviously, the design developed here has advantages of smooth flow, effective focusing at high flow velocity, a more uniform pressure gradient and greater stability. micromachining and bonding, it is better that d be no less than 100 μm. In geometry, ∠AOiCi is larger than ∠BOiCi. Based on the above considerations, the constraint relations between the parameters and the value ranges are presented in equations (6) and (7). Furthermore, several solutions are obtained using MATLAB software and then simulated via COMSOL. Among them, it is found that the solution of Ri = 17500 μm, Ro = 6600 μm, α = 30°, αo = 10° and Lo = 6500 μm has good stability, few secondary flows and low pressure loss, so it is selected for the design. Additionally, the simulated results show that in the main channel a distance of at least 5–6 mm is necessary for the focused sample flow to fully accelerate and reach a stable state. Therefore, Ls is assigned at 10 mm for redundancy. ⎧ Ro ⩾ 3Ws ⎪Q ⎪ SH − 1.5 ⩽ Ws ⩽ QSH +1.5 b 2QA ⎪ 2QA ⎪α > 2β ⎪ ⎪ θ ⎨ 0.8 ⩽ ⩽ 1.2 (6) α ⎪ ⎪ Wsh ⎪1 ⩽ W ⩽ 1.25 s ⎪ ⎪ β < 20° ⎪ d ⩾ 100 μ m ⎩ 2.3. Vertical focusing design The structure of the vertical focusing region is composed of three layers of rectangular microfluidic channels. There are two curved channels in both the upper layer and the lower layer, and there is a straight main channel in the medium layer. The dimensions of the main channel have been discussed above and the upper and lower layers have the same microfluidic structure. Therefore, the design is focused on the upper/lower layers, for which the channel configuration is presented in figure 4(a). Two sheath flow channels combine into a straight channel. At the end of the straight channel, sheath ⎧ 4000 μ m⩽ L o ⩽ 8000 μ m ⎪ ⎪ 20° ⩽ α ⩽ 45° ⎨ 5 °⩽ α o ⩽ 15° . (7) ⎪ 3b ⩽ W ⩽ 6b ⎪ ⎩15b ⩽ Ro ⩽ 25b To show the improvement in focusing performance, this design is simulatively compared with a commonly used 5 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 3. The flow rates from the first parametric set in table 2. The rainbow-colored streamlines represent the flow QA and the green streamlines represent the sheath flows. (a) Simulated results for a commonly used horizontal focusing design. (b) Simulated results of the design developed in this paper. ⎧Ws = (Ri cos α o + Lo sin α o + Ro sin2 α o ) ⎪ ⎪ −[(Ri cos α o + Lo sin α o + Ro sin2 α o )2 − (Lo + Ro sin α o )2 ]1/2 ⎪ ⎪α =arccos⎛⎜1 − b ⎞⎟ ⎪ o ⎝ 2Ro ⎠ ⎪ Ws cos α o ⎪ β=arctan ⎨ Lo + (Ro − Ws ) sin α o ⎪ L + ⎪ Ri = o (Ro − Ws ) sin α o ⎪ sin 2β ⎪ ⎪Wsv = [L o + (Ro − Ri ) sin α ]2 + [(Ri − Ro )(1 − cos α ) − b /2]2 ⎪ L o + (Ro − Ri ) sin α ⎪θ = arctan (8) (Ri − Ro )(1 − cos α ) − b /2 ⎩ flows are pressed into the main channel and the sample flow in the main channel is focused vertically by the two sheath flows from above and below. The two symmetrical sheath channels of the upper/lower layers contribute to the reduction of the horizontal momentum caused by the two sheath flows, enhancing stability. Similar to the horizontal focusing design, the inner and outer curves of one sheath flow channel are circular arcs. The inner arc is tangent to the straight channel at point Ci, and the circle of the outer arc touches the middle line of the straight channel at tangent point Co. The channels are modeled as expressed by equation (8). Equations (9) and (10) are constraint relations and value ranges. Using the same method as for the horizontal focusing design, a relatively optimal solution is found, with Ri = 11300 μm, Ro = 9900 μm, α = 30°, Lo = 900 μm, Lsv = 1500 μm and Ls = 2000 μm. The focusing performance is predicted by the simulations in figures 4(b) and (c). The vertically confined sample flow features a flying-saucer shape. ⎧Ws ≈ b ⎪ ⎪α > 2β ⎨ 0.8α ⩽ θ ⩽ 1.2α (9) ⎪W ⩽ W ⩽ 1.5W sh s ⎪ s ⎩ β < 20° 6 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 4. (a) Schematic of the vertical focusing region and design parameters. (b) The simulated results for vertical focusing and the flow rates are from the first parametric set in table 2 (only the sample flow is visible). (c) The cross-section of the focused flows, the rainbowcolored streamlines represent the sample flow and the blue streamlines represent the sheath flows. Figure 5. Flows distribution in the channel cross-section: the rainbow-colored streamlines are for the sample flow, the blue streamlines are for the vertical focusing sheath flows and the green streamlines are for the horizontal focusing sheath flows. A rectangle approximates the cross-section of the focused sample flow. ⎧ 500 μ m⩽ L o ⩽ 5000 μm ⎨ . (10) ⎩ 20° ⩽ α ⩽ 45° 3. Experimental methods 3.1. Chip fabrication For practicality, the 3D focusing design developed above is furnished with inlet and outlet channels for the introduction of flows into and out of the focusing regions, as illustrated in figure 6(a). Microfluidic chips were fabricated to experimentally characterize the focusing performance of the design. To fabricate one chip, five micromachined glass plates were used, as shown in figure 6(b). The three inner plates were 150 μm thick, and in these the microfluidic channels were caved. The two outer plates served as covers 2.4. Simulation testing Simulations show that the design can achieve 3D focusing, as illustrated in figure 5. The focused sample flow features quasirectangular cross-sections, which is similar to the assumption shown in figure 1(d). And the cross-section is approximated to a rectangle with a height h and a width w. It indicates that numerical simulation is able to predict the dimensions of the focused sample flow. 7 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 6. (a) The focusing design is furnished with five inlets and an outlet. (b) Five glass plates for one chip. (c) Photograph of a chip; the chip is filled with blue dye solution for visualization. Figure 7. Experimental setup. for protection, optical observation, fluid connection and installation. The five plates were successfully bonded using two methods, thermal and UV adhesive bonding. Figure 6(c) is a photograph of the chip. 3.2. Experimental setup Figure 7 illustrates the experimental setup utilized to visualize flows in the chip and quantify the dimensions of the focused sample flow. The sheath flows were distilled water, 8 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 8. Schematic diagram of the measurement method. Figure 9. This experiment was run with the first set of flow rates listed in table 2. (a) Flow behaviors in some key regions. (b) Microscopic image of the focused sample flow with α = 0. (c) Microscopic image of the focused sample flow with α = 30°. and the sample flow was distilled water dyed with Remazol Brilliant Blue dye. Three syringe pumps were manipulated to drive flows, enabling individual control of the sample flow and the vertical and horizontal focusing sheath flows. The two inlets for the vertical/horizontal focusing sheath flows were connected to one syringe pump using an off-chip Y-shaped connector. Microscopic images were obtained from a microscope equipped with a CCD camera. The dimensions were measured using an image processing technique. 3.3. Measurement method As discussed above, the cross-section of the focused sample flow is approximated by a rectangle with a height h and a width w. A method is developed to measure h and w. It consists of two steps. First, keeping the flow rates constant, several microscopic images are captured when α (the angle between the microscope and the direction perpendicular to the chip) varies, and the sample flow width m at every angle is measured, as shown in figure 8. The functional relationship between α, h, 9 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Figure 10. The fitted curves and the measured data: (a) experiment 1, (b) experiment 2, (c) experiment 3, (d) experiment 4, (e) experiment 5 and (f) experiment 6. 4. Results and discussions w and m is expressed by equation (11), where nw is the refractive index of water. Second, a fitted curve of the independent variable α versus the dependent variable m is obtained using MATLAB. With the curve and equation (11), the values of h and w can be obtained. In experiments, α was set to 0°, 5°, 10°, 15°, 20°, 25°, 30° and nw was 4/3. By this method, the width and height of the main channel were calculated to be 306 μm and 148 μm respectively, which was very close to the actual values 313 μm and 153 μm observed directly by microscope. It proved that this method is reasonable. Experiments were implemented to test the focusing performance of the microfluidic chips. Figure 9(a) presents the actual flow behaviors in some key regions, which agreed well with the simulated results. There existed a sharp interface between the sample flow and the sheath flows, implying that the flows in the channels stayed laminar and no mixing occurred. It proved that the focusing design has good stability under high flow velocity conditions. Figures 9(b) and (c) are two typical microscopic images of the 3D focused sample flow in the main channel. The former is the top view and shows that the sample flow was confined to the horizontal center. The latter image ⎛ sin α ⎞ m = w cos α + h cos α tan⎜arcsin ⎟. (11) ⎝ nw ⎠ 10 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 Table 3. Dimensions of the focused sample flow. Designed valuea a b Experimental value Simulated value No. QS (μl min−1) VS (m s−1) wh (μm2) w (μm)b h (μm)b wh (μm2) VS (m s−1) w (μm) h (μm) 1 2 3 4 5 6 60 30 10 30 10 75 4.44 4.13 4.63 2.22 0.74 10.33 10.72 ± 0.34 6.04 ± 0.58 4.27 ± 0.23 10.85 ± 0.31 11.87 ± 0.51 6.89 ± 0.38 21.03 ± 1.49 18.86 ± 2.52 9.74 ± 1.00 20.2 ± 1.34 18.45 ± 2.21 17.74 ± 1.67 225 114 42 219 219 122 4.44 4.38 4.00 2.28 0.83 10.24 24 19 12 22 22 20 152 = 225 112 = 121 62 = 36 152 = 225 152 = 225 112 = 121 11 7 4 12 14 6 The designed values are coincident with the values in table 2. Mean value ± 95% confidence interval (CI). focused sample flow, and the performance is comparable to that of some conventional flow cytometers. was captured at α = 30° and shows that the sample flow width was much smaller than channel height, indicating that vertical focusing was accomplished successfully. To quantify the focusing performance, experiments were carried out according to the sets of flow rates listed in table 2, and the measurement method developed above was utilized. The measured data and the fitted curves relating to each set are plotted in figure 10. Referring to the fitted-curve formulas, the experimental values of w and h under every flow condition were obtained and listed in table 3. Simulations were utilized as an auxiliary method to qualitatively describe the dimensions, as presented in figure 5. The simulated values of w and h were also listed in table 3, showing good agreement with the experimental values for all the cases studied. Although both the experimental and the simulated results showed that the focused sample flow had a quasi-rectangular cross-section (w ≠ h), which was different from the square shape (w = h) designed in table 2, the sectional areas (w × h) calculated by the experimental values were similar to the designed values. It proved that the functional relationship between the flow rates and the cross-sectional dimensions of flows expressed by equation (4) was practical, offering a way to control the dimensions by adjusting flow rates. The sample flow velocities (VS = QS /wh) were calculated using the experimental results of w and h, which were in good agreement with the theoretical velocities, as shown in table 3. In table 3, experiments 1–3 show that this focusing design was able to confine the sample flow effectively down to 5–11 μm horizontally and 10–21 μm vertically, when the sample flow rate varied between 10 and 60 μl min−1 with a roughly constant velocity of 4.4 m s−1. In experiments 1, 4 and 5, the flow rate ratios among QS, QVS and QHS were the same, resulting in almost the same dimensions for the focused sample flow. It was noted that the flow rate ratios were available to control the sample-flow cross-section dimensions, which was in accordance with equation (4). Experiment 6 demonstrates that the design was able to focus effectively at high velocities of 10 m s−1, thus having the potential for a very high throughput. Effective focusing was accessible within a wide velocity range from 0.7 to 10 m s−1. Compared with the other microfluidic hydrodynamic focusing designs listed in table 1, this design features superior performance in terms of sample flow rate, velocity and the cross-sectional dimensions of the 5. Conclusions This paper describes a 3D hydrodynamic focusing design that is based on the microfluidic technique and is capable of effective focusing at high flow velocity. Multiple attempts are made to maintain stable flows and achieve better focusing performance, including modeling and optimizing the channel geometry and establishing simulating tests. The research studies two further aspects: the mathematical relationship between the flow rates and the focusing results, and a measurement method. Both of them are verified by experiments. The design is easily constructed in a multilayer glass microfluidic chip. In practical terms, this design is able to confine the sample flow down to 5–11 μm horizontally and 10–21 μm vertically within a sample flow rate range of 10 to 75 μl min−1 and a velocity range of 0.7 to 10 m s−1, which is necessary for high-throughput flow cytometry analysis. The design offers a similar hydrodynamic focusing performance to some conventional flow cytometers, while occupying a much smaller volume and using a simpler structure. The 3D hydrodynamic design proposed in this paper seems suitable for microflow cytometers, especially for those requiring high-throughput analysis. In addition, this design has the potential to be incorporated into other microfluidic systems where 3D focusing is needed. Acknowledgments This work was supported by the Beijing Municipal Education Commission (Fund for Joint Project of Beijing). References [1] Tuchin V V (ed) 2011 Advanced Optical Flow Cytometry: Methods and Disease Diagnoses (New York: Wiley) p 273 [2] Ligler F S and Kim J S 2010 The Microflow Cytometer (Singapore: Pan Stanford) p 89 [3] McClain M A, Culbertson C T, Jacobson S C and Ramsey J M 2001 Flow cytometry of Escherichia coli on microfluidic devices Anal. Chem. 73 5334–8 11 J Zhao and Z You J. Micromech. Microeng. 25 (2015) 125006 microflow cytometer using groove-generated sheath flow Lab Chip 9 1942–50 [20] Yang R J, Hou H H, Wang Y N, Lin C H and Fu L M 2012. A hydrodynamic focusing microchannel based on micro-weir shear lift force Biomicrofluidics 6 034110 [21] Lee G B, Hung C I, Ke B J, Huang G R, Hwei B H and Lai H F 2001 Hydrodynamic focusing for a micromachined flow cytometer J. Fluids Eng. 123 672–9 [22] Knight J B, Vishwanath A, Brody J P and Austin R H 1998 Hydrodynamic focusing on a silicon chip: mixing nanoliters in microseconds Phys. Rev. Lett. 80 3863 [23] Huh D, Tung Y C, Wei H H, Grotberg J B, Skerlos S J, Kurabayashi K and Takayama S 2002 Use of air–liquid two-phase flow in hydrophobic microfluidic channels for disposable flow cytometers Biomed. Microdevices 4 141–9 [24] Hong S, Tsou P H, Chou C K, Yamaguchi H, Su C B, Hung M C and Kameoka J 2012 Microfluidic 3D hydrodynamic flow focusing for the rapid protein concentration analysis Biomicrofluidics 6 024132 [25] Simonnet C and Groisman A 2005 2D hydrodynamic focusing in a simple microfluidic device Appl. Phys. Lett. 87 114104 [26] Chang C C, Huang Z X and Yang R J 2007 3D hydrodynamic focusing in two-layer polydimethylsiloxane (PDMS) microchannels J. Micromech. Microeng. 17 1479 [27] Miyake R, Ohki H, Yamazaki I and Takagi T 1997 Investigation of sheath flow chambers for flow cytometers (micro machined flow chamber with low pressure loss) JSME Int. J. Ser. B 40 106–13 [28] Yang R, Feeback D L and Wang W 2005 Microfabrication and test of a 3D polymer hydro-focusing unit for flow cytometry applications Sensors Actuators A 118 259–67 [29] Sundararajan N, Pio M S, Lee L P and Berlin A 2004 3D hydrodynamic focusing in polydimethylsiloxane (PDMS) microchannels J. Microelectromech. Syst. 13 559–67 [30] Chiu Y J, Cho S H, Mei Z, Lien V, Wu T F and Lo Y H 2013 Universally applicable 3D hydrodynamic microfluidic flow focusing Lab Chip 13 1803–9 [31] Paiè P, Bragheri F, Vazquez R M and Osellame R 2014 Straightforward 2D hydrodynamic focusing in femtosecond laser fabricated microfluidic channels Lab Chip 14 1826–33 [32] Kummrow A, Theisen J, Frankowski M, Tuchscheerer A, Yildirim H, Brattke K, and Neukammer J 2009 Microfluidic structures for flow cytometric analysis of hydrodynamically focused blood cells fabricated by ultraprecision micromachining Lab Chip 9 972–81 [33] Frankowski M, Theisen J, Kummrow A, Simon P, Ragusch H, Bock N and Neukammer J 2013 Microflow cytometers with integrated hydrodynamic focusing Sensors 13 4674–93 [34] Cornish R J 1928 Flow in a pipe of rectangular cross-section Proc. R. Soc. A 120 691–700 [4] Schrum D P, Culbertson C T, Jacobson S C and Ramsey J M 1999 Microchip flow cytometry using electrokinetic focusing Anal. Chem. 71 4173–7 [5] Jiang H, Weng X and Li D 2014 A novel microfluidic flow focusing method Biomicrofluidics 8 054120 [6] Cheng I F, Chang H C, Hou D and Chang H C 2007 An integrated dielectrophoretic chip for continuous bioparticle filtering, focusing, sorting, trapping, and detecting Biomicrofluidics 1 021503 [7] Church C, Zhu J, Wang G, Tzeng T R J and Xuan X 2009 Electrokinetic focusing and filtration of cells in a serpentine microchannel Biomicrofluidics 3 044109 [8] Holmes D, Morgan H and Green N G 2006 High throughput particle analysis: combining dielectrophoretic particle focussing with confocal optical detection Biosensors Bioelectron. 21 1621–30 [9] Nordin M and Laurell T 2012 Two-hundredfold volume concentration of dilute cell and particle suspensions using chip integrated multistage acoustophoresis Lab Chip 12 4610–6 [10] Shi J, Mao X, Ahmed D, Colletti A and Huang T J 2008 Focusing microparticles in a microfluidic channel with standing surface acoustic waves (SSAW) Lab Chip 8 221–3 [11] Mao X, Lin S C S, Dong C and Huang T J 2009 Single-layer planar on-chip flow cytometer using microfluidic drifting based three-dimensional (3D) hydrodynamic focusing Lab Chip 9 1583–9 [12] Oakey J, Applegate R W Jr, Arellano E, Carlo D D, Graves S W and Toner M 2010 Particle focusing in staged inertial microfluidic devices for flow cytometry Anal. Chem. 82 3862–7 [13] Ha B H, Lee K S, Jung J H and Sung H J 2014 3D hydrodynamic flow and particle focusing using four vortices Dean flow Microfluid. Nanofluid. 17 647–55 [14] Wang X, Zandi M, Ho C C, Kaval N and Papautsky I 2015 Single stream inertial focusing in a straight microchannel Lab Chip 15 1812–21 [15] Howell P B Jr 2008 Two simple and rugged designs for creating microfluidic sheath flow Lab Chip 8 1097–103 [16] Scott R, Sethu P and Harnett C K 2008 3D hydrodynamic focusing in a microfluidic Coulter counter Rev. Sci. Instrum. 79 046104 [17] Kennedy M J, Stelick S J, Perkins S L, Cao L and Batt C A 2009 Hydrodynamic focusing with a microlithographic manifold: controlling the vertical position of a focused sample Microfluid. Nanofluidics 7 569–78 [18] Choi S and Park J-K 2008 Sheathless hydrophoretic particle focusing in a microchannel with exponentially increasing obstacle arrays Anal. Chem. 80 3035–9 [19] Golden J P, Kim J S, Erickson J S, Hilliard L R, Howell P B, Anderson G P and Ligler F S 2009 Multi-wavelength 12