Microfluidics and Nanofluidics

, Volume 3, Issue 5, pp 561–570

Development of sorting, aligning, and orienting motile sperm using microfluidic device operated by hydrostatic pressure


  • Duck-bong Seo
    • Department of Mechanical and Aerospace EngineeringUniversity of Missouri-Columbia
  • Yuksel Agca
    • Department of Veterinary Pathobiology, Comparative Medicine CenterUniversity of Missouri-Columbia
  • Z. C. Feng
    • Department of Mechanical and Aerospace EngineeringUniversity of Missouri-Columbia
    • Department of Veterinary Pathobiology, Comparative Medicine CenterUniversity of Missouri-Columbia
Research Paper

DOI: 10.1007/s10404-006-0142-3

Cite this article as:
Seo, D., Agca, Y., Feng, Z.C. et al. Microfluid Nanofluid (2007) 3: 561. doi:10.1007/s10404-006-0142-3


In vitro fertilization (IVF) and intracytoplasmic sperm injection (ICSI) are the most commonly used assisted reproductive technologies to overcome male infertility problems. One of the obstacles of IVF and ICSI procedures is separating motile sperm from non-motile sperm to select the most competent sperm population from any given sperm sample. In addition, orientation and separation of the head from the tail is another obstacle for ICSI. Using the self-movement of sperm against flow direction, motile and non-motile sperm can be separated with an inexpensive polymeric microfluidic system. In this paper, we describe the development of a microfluidic system obtained through low-cost fabrication processes. We report experimental results of sperm sorting using hydrostatic pressure of three different species: bull, mouse, and human. The movement of cells in these channels was observed under a microscope and recorded with a digital camera. It is shown that the hydrostatic pressure and self-movement of motile sperm can be used to solve separating, aligning and orienting sperm in the microchannel.


MicrofluidicsMotile-sperm sortingIntracytoplasmic sperm injection

1 Introduction

The microfabrication technologies of the integrated circuit and semiconductor industry that started in the 1960s have made it possible to integrate complex electronic and mechanical functions, providing us with even smaller and less expensive sensors and devices. With the development of micromachining technologies for fluidic systems in recent years (DeLos Santos 1999; Nadim 1999; Duffy et al. 1998; McDonald et al. 2000; Collins 2003), microfluidics have been widely applied to biomedical fields because very small amounts of fluid transport is useful for DNA analysis systems (Burns et al. 1998), chemical analysis systems (Harrison et al. 1993; Manz et al. 1994; Becker and Manz, 1997; Buettgenbach and Robohm 1998; Darlin et al. 2002; Sudarsan et al. 2005), assisted reproductive technologies (Beebe et al. 2002; Suh et al. 2006), and drug delivery systems (Chan et al. 1999). The microfluidic pumps and valves in particular have witnessed extensive research activities (Xia et al. 2006; Khoo and Liu 2000). Furthermore, research efforts are also focused on integrating many functions on one chip, the so-called labs-on-chip, to achieve two main goals: (1) to simplify the working mechanism and (2) to reduce labor and cost.

One of the biological applications of microfluidic devices is cell sorting. Bull, mouse, and human spermatozoa are widely used in biomedical research, agriculture application, and human reproductive medicine related procedures. Separating motile sperm from non-motile sperm is critical for successful intracytoplasmic sperm injection (ICSI) and in vitro fertilization (IVF). Cell sorting microfluidic devices have been previously reported (Krüger et al. 2002; Huh et al. 2002). Sperm sorting systems using microfluidic devices have also been recently reported (Schuster et al. 2003; Horsman et al. 2005). Among the sperm sorting microfluidic systems, Cho et al. (2003) introduced a motile human sperm sorting device. This system uses the self-movement of motile sperm to escape from the initial inlet streamline, which is generated by passive pumping systems using hydrostatic pressure, and then collects motile sperm in one of two outlet reservoirs. Even though their system works well, there are a few limitations to its usage. The pressure has to be stable during sorting and flow velocity has to be large enough to prevent motile sperm swimming against flow direction. In addition, they reported sorting results for human sperm only.

In view of the limitations of the sorting devices reported in the literature, we developed a novel sperm sorting device. Additionally, controlling the orientation of the sperm cell while being sorted is important to integrate the sperm dissection, which is another important step during ICSI. This device is based on the self-movement of the sperm in a flow. The sperm self-movement achieves sorting and orientation control at the same time. Equally significant, the fabrication procedure is based on the “soft-lithography” method proposed by Duffy et al. (1998) using an elastometric material such as poly-dimenthylsioxane (PDMS). We carried out the procedures without expensive facilities such as clean room (Seo 2002; Collins 2003). A HCl bonding method also helps to simplify the filling of the fluidic channels.

In this paper, we report the development of a novel microfluidic device. This microfluidic device design has been developed with the long-term goals of : (a) being able to detect sperm DNA with sufficient resolution to enable sorting of X and Y chromosome-bearing sperm, and (b) being able to add a “cutting laser” component to separate sperm heads and tails for ICSI procedures. Based on these goals, the a multi purpose device may be designed to allow high resolution sorting system at the expense of the throughout. We provide the mechanical principles governing the design of the device. Sperm motion in the device is observed and recorded using an inverted microscope (ECLIPSE TE200, Nikon) and a digital camera (COOLPIX 5000, Nikon). We also report the experimental results on the variability among different species’s sperm: bull, mouse, and human.

2 Motile sperm sorting microfluidic system

The development of the motile sperm sorting microfluidic system (MSMS) is based on two well-known observations: (1) motile sperm orient themselves against the flow and (2) motile sperm can swim against the flow with specific flow velocity ranges. To quantify these observations, a microchannel with a cross section of 30 μm by 25 μm was fabricated. Human sperm were introduced into the inlet reservoir (Fig. 1). A hydrostatic pressure, about Δh = 4.5 mm (45 N/m2 pressure), was applied to the inlet reservoir. The movement of the solution and sperm from inlet to outlet takes place and shows that over 80% of the motile sperm has a tendency to swim and orient their heads against flow direction (Fig. 2).
Fig. 1

Orientation of motile cells in a channel with flow driven by hydrostatic pressure

Fig. 2

Orientation of human motile sperm head in flow

Based on the above observations, the MSMS consists of four channels and three reservoirs as shown in Fig. 3a. The flow is driven by hydrostatic pressure created by the height of liquid columns at the reservoirs. These four channels meet at one point called a ‘junction’. Reservoir 2 contains the sperm to be sorted and reservoir 3 collects sorted sperm. The liquid in channel C flows from the junction to reservoir 3. By controlling the hydrostatic pressure carefully, the liquid in channel B flows from the junction to reservoir 2. By swimming against the flow in channel B, motile sperm in reservoir 2 reach the junction. Once they reach the junction, they are transported to reservoir 3 by a much faster flow in channel C as shown in Fig. 4.
Fig. 3

The motile sperm sorting microfluidic system (MSMS)

3 Analysis and fabrication of the channel

3.1 Analysis of the fluid flow in the microchannel

Careful control of the flow speed in channel B and channel C is crucial to the proper operation of the device. Since the flow velocity is determined by the pressures in the reservoirs and the flow resistance, we calculated the relationship between the flow rate and the channel geometry in the following. The flow model is for a single rectangular channel which has 2b height along the z-direction, 2a width and length of l as shown in Fig. 4.
Fig. 4

The schematic of motile sperm sorting in MSMS

Fig. 5

Geometry of a micro-channel with rectangular cross-section

The equation that governs the motion of the Newtonian liquid is the Navier–Stokes equation described as:
$$\rho\left[\frac{{\partial \vec u}}{{\partial t}}+\vec{u}\cdot\nabla\vec{u}\right]=-\nabla P+\mu\nabla^{2}\vec{u}+\vec{g}.$$
For flow in the microchannels, viscous effects dominates; the inertia so the gravity effect can be ignored; and the equation can be written as:
$$-\nabla P+\mu\nabla^{2}\vec{u}=0$$
where \({\vec{u}}\) is the velocity, P is pressure, \({\vec{g}}\) is gravitation acceleration, ρ is liquid density, and μ is liquid viscosity. For flow in a long channel as shown in Fig. 5, if the length of the channel is larger than the cross section dimension, the flow can be considered unidirectional. Therefore, the velocity only has a component in the streamwise direction. The equation of motion can be written as:
$$\frac{\partial^{2}u}{\partial y^{2}}+\frac{\partial^{2}u}{\partial z^{2}}=-\frac{G}{\mu}$$
where −G is the pressure gradient in the streamwise direction \((\hbox{i.e.} {G = - {\partial P} \mathord{\left/{\vphantom {{\partial P} {\partial x}}} \right.\kern-\nulldelimiterspace} {\partial x}}).\) Corresponding to the rectangular channel in Fig. 5 the no-slip boundary conditions are:
$$u(y=\pm a)=u(z=\pm b)=0$$
The partial differential equation (3) is known as Poisson’s equation. The fluid velocity and flow rate in a channel with rectangular cross section are described by Seo (2002) and Pozrikidis (1997) as:
$$u(y,z) = \frac{G}{{2\mu}}{\left[ {b^{2} - z^{2} + 4b^{2} {\sum\limits_{n = 1}^\infty {\frac{{(- 1)^{n}}}{{\alpha^{3}_{n}}}}}\frac{{\cosh (\alpha_{n} y/b)}}{{\cosh (\alpha_{n} a/b)}}\cos \left(\alpha_{n} \frac{z}{b}\right)} \right]}$$
The flow rate obtained from integration of velocity over the channel cross section is:
$$Q=\frac{4Gab^{3}}{3\mu}\left[1-\frac{6b}{a}\sum\limits_{N- 1}^{\infty}\frac{\hbox{tanh}(\alpha_{N}a/b)}{\alpha_{N}^{5}}\right]$$
For each channel, n = 1, 2, 3, we have:
$$Q_{n}=\frac{4G_{n}a_{n}b^{3}}{3\mu}\left[1- \frac{6b}{a_{n}}\sum\limits_{N= 1}^{\infty}\frac{\hbox{tanh}(\alpha_{N}a_{n}/b)}{\alpha_{N}^{5}}\right] $$
where n = 1,2 and 3 denote channel A, B, and C, respectively. Parameters an denote the widths of the three channels; they all have the same height b.
Let P1, P2, and P3 be the pressure at each reservoir and P0 be the pressure at the junction. The pressure gradient can be expressed in terms of the pressure at the reservoir if the pressure at the junction can be found:
$$G_{n}=\Delta P_{n}/l_{n}$$
where ln is the length of each channel and
$$\Delta P_{1} = P_{1} - P_{0},\quad \Delta P_{2} = P_{2} - P_{0},\quad \Delta P_{3} = P_{0} - P_{3}$$
Therefore, Eq. 8 is simplified in terms of pressure difference as:
$$Q_{n} = B_{n} \Delta P_{n}\quad (n=1, 2, 3)$$
$$B_{n} = \frac{{4a_{n} b^{3}}}{{3\mu l_{n}}}{\left[{1 - \frac{{6b}}{{a_{n}}}{\sum\limits_{N = 1}^\infty} \frac{\tanh (\alpha_{N} a_{n} /b)}{\alpha^{5}_{N}}} \right]}$$
To determine the junction pressure P0, referring to Fig. 6, we obtain the continuity equation as follows:
$$2Q_{1} + Q_{2} = Q_{3}$$
Substituting Eqs. 10 and 11 into the above, we obtain the junction pressure:
$$P_{0} = \frac{{2B_{1} P_{1} + B_{2} P_{2} + B_{3} P_{3}}}{{2B_{1} + B_{2} + B_{3}}}$$
Fig. 6

The specifications and junction of four channels in Fig. 3

Once the junction pressure is obtained, the pressure gradients are known and the flow velocities can be determined using Eq. 5. In particular, the flow velocity in channel B is found to be:
$$u_{2} (y,z) = \frac{{G_{2}}}{{2\mu}} A_{2} (y,z)$$
$$A_{2} (y,z) = [b^{2} - z^{2} + 4b^{2} {\sum\limits_{N = 1}^\infty {\frac{{(- 1)^{N} \cosh (\alpha_{N} y/b)}}{{\alpha^{3}_{N}\cosh (\alpha_{N} a_{2}/b)}}}}\cos \left(\alpha_{N} \frac{z}{b}\right)$$
$$G_{2} = \frac{{B_{3} (P_{2} - P_{3}) - 2B_{1} (P_{1} - P_{2})}}{{l_{2} (2B_{1} + B_{2} + B_{3})}}$$
Note also that the pressure at each reservoir is due to the hydrostatic pressure of the liquid columns at each reservoir, i.e.
$$P_{n} = \rho gh_{n}\quad (\hbox{n} =1,2,3).$$
where h1, h2, and h3 represent the height of the liquid columns at reservoirs 1, 2, and 3, respectively. Therefore, the pressures can be obtained by measuring the height of the liquid columns at the reservoirs. The flow velocity in channel B is thus expressed as:
$$u_{2} (y,z) = \frac{{\rho g}}{{2l_{2} \mu (2B_{1} + B_{2} + B_{3})}}(B_{3} \Delta h_{2} - 2B_{1} \Delta h_{1})A_{2} (y,z)$$
where Δh1 is the height difference of the liquid column between reservoirs 1 and 2 and Δh2 the difference between reservoir 2 and reservoir 3; see Fig. 3a.
From Eq. 19, the reverse flow occurs in channel B when:
$$\Delta h_{1} > \frac{{B_{3}}}{{2B_{1}}}\Delta h_{2}$$

3.2 Channel design

The design of the MSMS is obtained based on the following two considerations: (1) to allow easy control of flow direction and velocity using hydrostatic pressure, and (2) to allow enough space for the sperm to swim against the flow direction without “clogging” the channel. Geometric parameters such as height, width, and length all affect the flow velocity and direction in channels. Among these parameters, the channel widths (2an) and lengths (ln) are defined by the patterns on the optical mask. The channel height is defined by the thickness of the SU-8 coating on the wafer which is used as the mold for the PDMS channel. Large channel width is desired to maximize the throughput of the sorting device. However, since the microchannels are made of polymeric material, the channels could close-up under external pressure if they are too wide.

To generate fastest unidirectional flow in channel C, the length and width of channel C must be smaller than the other two channels. We set the lengths of each channel as l1 = 23.46 mm, l2 = 8 mm, and l3 = 5.5 mm. The widths were also set as 2a1 = 205 μm, 2a2 = 100 μm, and 2a3 = 80 μm. The width of channel A was selected as a middle value between entrance and exit width. The channel height is determined based on two considerations. First of all, we found that cells clog easily in very narrow channels. Clogging is most likely to occur at the channel entrance as shown in Fig. 7.
Fig. 7

Clogging of sperm cells at the entrance of channel B

Channels with 25 μm height was proven to be convenient in terms of ease of the flow velocity control using hydrostatic pressure. Without precision flow pumps, we can adjust the hydrostatic pressure by the height of the liquid columns. We can manipulate a 10 mm height liquid column with ease. Figure 8 shows the pressure combinations that will achieve given flow velocities for channels of different height. Using liquid columns about 10 mm in height, the maximum velocities in the channel vary significantly. Since the human motile sperm’s self-propulsion velocity is estimated to be approximately 70 μm/s [21], the channel with a 25 μm height allows us to easily vary the maximum flow velocity between −100 and + 100 μm/s when the heights of the liquid columns are varies within 10 mm.
Fig. 8

Theoritical velocity contours

3.3 Microchannel fabrication procedures

The complete process for fabricating the microchannel consists of five steps: (1) coating, (2) exposing, (3) etching, (4) molding, and (5) bonding. These five steps are detailed in recent reports 3–6 and our fabrication processes are similar except the bonding. We created the channels inexpensively without using a cleanroom facility. The following instruments and supplies were used to complete fabrication. A spin-coater (Model P6700, A Specialty Coating System. INC) was used to coat the silicon wafer with a photoresist (SU-8). After spin coating, an UV exposure unit (KVB – 30.KINSTEN, 55 mW/cm2) was used to expose the coated wafer. A heater with temperature control (Model PC-220, Corning) was used to pre-bake and post-bake. We used a 6-inch diameter silicon wafer, which has a polished side and an oxide side as the substrate of the mold. The coated wafer is developed lithographically to form the mold. The mask for lithography was designed on AUTOCAD 2002. This design was then printed on a transparency by a commercial printing shop (University of Illinois at Urbana-Champaign, Urbana, IL, USA).

Without using a cleanroom facility, we have taken special steps in order to create defect-free molds. It is better to use the polished side of the wafer to easily bond PDMS (SYLGARD® 184 Silcone Elastomer Base, DOW CORNING) and glass plate. Furthermore, to avoid removing SU-8 mold on the wafer during etching process, the exposed SU-8 coated wafer is etched by applying SU-8 developer drops on the surface of the coated wafer instead of immersing it. Applying drops can also reduce the amount of SU-8 developer used in the fabrication procedure.

Once the PDMS molding was complete, the PDMS channel and microscope glass are boiled in 1/1,000 mole diluted hydrochloride (HCl) solution for 10 min. Once taken out from the solution, the glass and the PDMS channel are dried using compressed air then placed together. They adhere to each other and form a bond strong enough to complete the enclosed channels. The bonding is not a permanent bond as in plasma bonding. The PDMS can still be peeled off the glass but the bonding strength is sufficiently strong to withstand the pressures used in these experiments. To peel off the bond, considerable force is required. The PDMS treated by HCl remains hydrophilic for at least 2 h. This time is sufficient to have the channels filled with the liquid. Surface tension draws the liquid into the channel without the high pressure that would otherwise be needed. For this application, once the channel is filled, we are no longer concerned about the surface property of the channel. Figure 9 summarized the procedures of the channel fabrication.
Fig. 9

A schematic showing the steps involved in the fabrication procedures of the microchannel

4 Flow visualization in the channel

Using the MSMS as shown in Fig. 3, we were able to visualization the fluid flow in the channels by using 1 μm microbeads. The proper pressure range generating reverse flow in channel B can be estimated depending on experimental results. Additionally, streamlines at the junction were observed. The microbeads were introduced in reservoir 2. The liquid columns in reservoirs 1 and 2 were adjusted to be equal initially. The pressure in reservoir 3 was kept lower than the other two reservoirs. The height of liquid in reservoir 1 was increased using a 20 μl pipette to generate hydrostatic pressure and the flow in channel B is observed. As shown in Fig. 10a, the flow in all channels is from left to right when pressure is maintained within a specific range. However, if the pressure in reservoir 1 was increased gradually, the direction of flow in channel B was changed when the pressure reaches a certain threshold as shown in Fig. 10b. Although the flow direction in channel B was changed, the flow direction in channel C remained the same. Therefore, sperm can be collected in reservoir 3 as they pass the junction. The height in the reservoirs changes slightly as fluid flow takes place. For example, when the flow velocity reaches 1 mm/s, it takes over 30 min for the height to change about 0.1 mm.
Fig. 10

Flow streamlines at the channel junction corresponding to different pressure

5 Experiments

Sorting experiments were performed using the MSMS. The MSMS were completely filled with solution (0.1μm filtered Dulbecco’s Phosphate–Buffered Saline, GIBCO) to control flow easily using hydrostatic pressure before conducting each experiment. The pressure at reservoirs 1 and 2 were initially set to be almost equal (8∼ 10 mm height liquid). Reservoir 3 was set at a lower pressure (less than 1 mm height) than the others. Twenty microliter of sperm sample was loaded into reservoir 2. The pressure at reservoir 1 was increased by adding 20∼ 40 μl solution (about 1.14∼ 2.28 mm height) each time, to observe sperm movement in the channel. When the flow in channel B moves from right to left, the sperm swim from left to right. When motile sperm reached the junction, they are swept into channel C which is at a much faster flow velocity, and gathered in reservoir 3 (Figs. 11, 12).
Fig. 11

The tendency of swimming motile a bull and b mouse sperm against the flow in channel B

Fig. 12

The movement of bull motile sperm at the channel junction

The MSMS was designed to maintain the desired flow direction for at least 1 h. However, in our most recent experiments, we collected sorting data after 20 min of inserting sample.

To study the variability among different species, we applied the MSMS to the sorting of bull, mouse, and human sperm. Each experiment was performed using sperm from only one species. In spite of respective differences in their own characteristics, the experimental results show that they all have similar characteristics in terms of aligning their heads opposite to the flow direction as well as swimming against the flow (Fig. 12).

In Fig. 13, we plot the velocities of the motile cell and the debris at different pressure. The pressure is indicated using the height difference of the liquid columns in reservoirs 1 and 2. Because the slight differences in the height of the PDMS channels resulting from the variation of the SU-8 coating thickness, the height difference of the liquid columns (Δh1) varies from one device to the other. Despite the large variability, we clearly observe that the flow and the sperm move in opposite directions.
Fig. 13

Velocity of motile sperm and non-motile sperm (or debris) in channel B

Table 1 shows the results of the experiments with bull sperm sorting efficiencies using the MSMS. As shown in Fig. 14, the motile bull sperm were sorted using the MSMS. Approximately 20 % initial sperm motility in reservoir 2 increased to 80% after 20 min sorting time.
Table 1

Sorting rate of bull sperm using the MSMS

Sample no.

Approx. inlet sperm no.

Approx. inlet motile rate

Approx. outlet motile rate

Average sorting rate (ea/min)


2.68 × 105

23.4% (33/141)

93.8% (15/16)



4.20 × 105

16.2% (60/370)

81% (30/37)



4.90 × 105

19.4% (21/108)

75.3% (70/93)




18.4% (114/619)

78.8% (115/146)


The inlet and outlet motile rates and average sorting rate were counted based on digital pictures

Fig. 14

Motile and non-motile bull sperm a before (reservoir 2) and b after sorting (reservoir 3)

6 Conclusions and discussions

A motile sperm sorting microfluidic system was developed using an inexpensive fabrication method. Bull, mouse, and human motile sperm can be sorted by controlling hydrostatic pressure in the microchannel. Non-motile sperm and debris can be separated and collected. Since the sorting takes place by sperm swimming against the flow in channel B, a single channel may appear to be sufficient. However, our multi-channel design allows a much lager pressure difference which may be easier to use in the clinical laboratory. For a single channel with a size similar to channel B, the pressure difference at the two ends would have to be controlled within less than 0.1 mm to avoid generating flows in the channel that would overwhelm the cell swimming ability. Either reducing the channel cross section or lengthening the channel will diminish the sorting efficiency of the channel.

Even though the current design has a relatively low throughput, it provides proof of concept that this approach works well. In addition, the current system has the advantage of orienting and aligning the sperm. The modification of the MSMS design to increase the throughput is straightforward by adding multiple channels. Moreover, the system can be applied to other integrated microfluidic devices, for example, a micro Coulter-counter or other cell sensing devices.


This work was funded by a grant from the NIH (RR1482) to JKC.

Copyright information

© Springer-Verlag 2007