Droplet gradient array for parallel experimentation via microfluidic device
Lindsay Pickett
Eckerd College
Faculty Mentor: Abraham P. Lee
Department of Biomedical Engineering
Graduate Mentor: Rob Lin
Department of Biomedical Engineering
Droplet gradient array for parallel experimentation via microfluidic device
Lindsay Pickett
Eckerd College
Faculty Mentor: Abraham P. Lee
Graduate Mentor: Rob Lin
ABSTRACT
The aim of this study is to generate a gradient array consisting of monodispersed aqueous droplets in oil as a platform for parallel experimentation in a microfluidic device. The droplets are approximately 50µm in diameter and serve as picoliter reaction chambers. Formation of droplets is accomplished by shearing two aqueous reagent streams (dispersed phase) with a constant oil stream (continuous phase) at a junction of high shear stress. The flow rates of the reagents and the oil phase are in the range of 1-2µl/min. The aqueous inlet at the shearing junction is 12µm and the shearing orifice is 7.5µm. The droplet array will be captured in serpentine channels after the shearing junction to allow imaging and characterization. This platform not only offers the ability to reduce reagent consumption during parallel experimentation; it can also be easily adapted for various applications.
Key Terms: Droplet, gradient, microfluidic device, parallel experimentation.
INTRODUCTION
Fluid phenomena at the microscale have opened up novel techniques and experimental procedures in the research world that allow for execution of studies that are impossible at the macroscale. The field of microfluidics is rapidly expanding, as its applications range across many fields of research from biology and chemistry to engineering and biotechnology. The use of microfluidics offers numerous advantages to researchers such as reduced consumption of reagents and analytes, reduced production of harmful by-products, increased portability of the system and minimization of laboratory space required to perform studies (Duffy et al., 2000). The applications of microfluidic devices are virtually limited only by the creativity of those developing this form of technology. For example, microfluidic channels that approach the size and flow conditions of blood vessels have been engineered to serve as instruments for physiological modeling and study (Runyon et al., 2004). So far, microfluidics has been mainly applied for use in biomedical devices, analytical systems and biochemical, chemical and other fundamental research tools (Duffy et al., 2000). The microfluidic device designed in this experiment can potentially be applied to all of these areas with the use of picoliter-sized droplets. The droplets serve as micro-reaction vessels where chemical reactions and syntheses as well as biomolecular and cellular analyses can be performed (He et al., 2005). He et al. have demonstrated controlled encapsulation of cells and mitochondria in aqueous droplets, Tan et al. (2006) have done the same inside liposomes generated by a similar device; protein crystallization conditions were monitored in droplets by Zheng et al. and Srinivasan et al. have used droplets as analysis chambers for clinical assays of biological fluids. Droplets have also been used as molds for microsphere generation and platforms for self-assembly of nanoparticles (Nisisako et al., 2004; Yi et al., 2003). Droplet microfluidic devices are especially useful due to the variability of droplet volume, radii, composition and concentrations. These properties can be controlled actively by outside forces (pressure, electrokinetics, gravity) or passively by altering channel geometry and wetting patterns on the channel surfaces of the device (Squires et al., 2005). In this study, droplet properties will be governed both actively and passively by use of hydrostatic pressure and varying channel geometries. The general design of the microfluidic devices used for droplet formation consists of inlet channel(s) for aqueous reagents that are intersected by an oil phase channel at a shear junction. The microfluidic design used in this experiment has two aqueous inlet channels equipped with a filter system and serpentine channels for resistance to prevent back-flow. The junction where oil shears off the aqueous streams into droplets is designed for controlled droplet size and position. The junction has a pressure chamber (widening of the oil channel at the oil/aqueous interface) and at the beginning of the outlet (droplet) channel, there is a small orifice followed by an expanding region of channel serving to focus the highest amount of shear stress at the orifice for stable, uniform droplet formation (Tan et al., 2005).
MATERIALS AND METHODS
The device design was created in a CAD program (Freehand 10.0, Macromedia, San Francisco, CA). The channels of the design were 50 µm in height and width, with the exception of the serpentine channels following the shear junction. The serpentine channels were increased to 100 µm in width to reduce the resistance and backflow of the system and to allow for better imaging of the droplets for characterization. The junction dimensions of the first device design were 50 µm at the aqueous inlet and orifice, while the second device’s dimensions were 70 (aqueous inlet) and 30 µm (orifice). The final junction design had dimensions of 12 and 7.5 µm as is shown in the figure below to allow for reduced droplet sizes.
Fig. 1. Final junction 3-D design created in Cobalt 7.0 software. The aqueous inlet is 12 µm in width; the orifice width is 7.5 µm, and channel depth is 50 µm.
The filtration system of the device started out with a branch channel width of 20 µm and was changed to a debris-trapping design with flow-through channels of 6 µm in width. The newer system is also equipped with alternate flow channels of 20 µm in width. All the changes made to the device design were based on experimentation.
Fig. 2. Filtration system modification. The modified system of filtration is pictured on the left; modification of the branch system on the right was made to comply with the decreased orifice width in the latest device design. The wider channel down the center of the new filtration system will catch any debris and support most of the fluid flow, while the alternate flow channels will allow debris-free fluid to flow around the center chamber if it were to become clogged.
Fabrication of the Microfluidic Devices was done by photolithography. The device design was sent out to CADART commercial printing company in Oregon, where the design was reversed (clear channels, ink background) and printed as a transparency at high resolution and sent back. Inside the clean room, a silicon wafer purchased from Silicon Inc. was spin-coated with a thin (50 µm) layer of SU-8 50 (negative photoresist purchased from MicroChem) at 500 rpm for 30 sec, at 2000 rpm for 1 min and soft-baked in a 65º oven for 6 min. The transparency was then used as a contact mask atop the wafer during UV exposure. The mask and wafer were exposed to 420 nm UV light for 42 seconds. Upon exposure to UV light, the photoresist polymerized in areas where channels were located, the wafer was then post-baked at 95º for 5 min. The wafer was then washed in SU-8 developer and rinsed in isopropyl alcohol (IPA) to remove soft SU-8 and dried with a nitrogen spray gun. The wafer was then removed from the clean room to be used as a mold for PDMS (polydimethylsiloxane).
Fig. 3. Microfluidic Device Fabrication. This Schematic outlines the fabrication process described above, after the device design is complete. The device pictured above has been filled with food dye to enhance feature visibility.
PDMS was chosen as the casting substrate in this experiment for its chemical inertness and optical transparency. The PDMS mixture was made by mixing prepolymer and curing agent at a 10:1 ratio, it was degassed in a vacuum chamber and poured onto the mold and cured for 12 hours at 70°. The PDMS was cut from the master; inlets were punched with a16 gauge blunt needle and oxygen plasma treated then bonded to a glass slide. Once bonded, the channels were filled with silicon oil via hand-pumped syringe and baked in a 120° oven overnight to convert the oxidized (treated) hydrophilic surfaces of the PDMS back to hydrophobic surfaces (Duffy et al., 1998). This entire fabrication process was repeated after device testing until an optimal device design was obtained.
Fig. 4. PDMS devices. This figure shows the PDMS devices mounted on the inverted microscope stage during experimentation.
The Experimental Setup consists of 3 syringe pumps (Pico Plus, Harvard Apparatus) and a high-speed imaging camera (Fastcam PCl-10K, Photron Ltd.) mounted on the inverted microscope (Nikon Eclipse TE2000-5).
Fig. 5. Experimental setup. This figure displays the system setup consisting of 3 syringe pumps (2 on the left, used for aqueous reagent flow rate control; 1 pump on the right for the oil flow rate control). The device is mounted on the stage of the inverted microscope to which the high-speed camera is connected on the left.
Silicon oil purchased from Fisher Scientific was used as the continuous phase. Deionized water and McCormick food and egg dye (contents: water, propylene glycol, color) was used in the aqueous continuous phase. The flow rate of the oil phase was within the range of 1-2µl/min and was kept constant during experiments. The flow rates of the aqueous reagents were kept within the same range as the oil but were varied during the experiments. External 3-way miniature solenoid valves purchased from LEE co. were incorporated into the system to allow for all phase flow rates to be stopped synchronously, enabling droplet array image capture. The fluid flow route of the external valves was changed so that fluids were run through the “diversion” chamber instead of the “inlet” chamber to reduce the instability of the system introduced by the valves.
Fig. 6. External valves. The three valves pictured on the left were purchased from LEE co. The figure on the right is a schematic representation of the fluid re-routing method. The left valve shows the direction of fluid flow when the valve is run normally, in the “off” position. The third valve chamber that does not have a direction of fluid flow is the “diversion” chamber open to atmospheric pressure only. When the switch is activated, the inlet flow is diverted to this chamber. The valve on the right shows how the fluid-flow was re-routed to make use of the more stable “diversion” chamber during stabilization of the device. When the switch is deactivated, the fluid is redirected to the “inlet” chamber.
A computer-controlled step gradient was attempted via creation of a step profile in a text-feed windows program (Hyper Terminal) connected to the syringe pumps.
RESULTS
Stable droplets of optimal size (~50 µm in diameter) were formed in the final device design at constant flow rates (oil at 2 µm/min and aqueous reagents at 0.2 µm/min).
Fig. 7. Stable droplet formation. This image was taken with the Fastcam high-speed camera and displays droplet formation at constant flow rates (oil at 2 µm/min and aqueous reagents at 0.2 µm/min).
The same results were obtained when one aqueous reagent’s flow rate (water) was increased manually over time while the other remained constant. This is shown in the figure below.
Fig. 8. Manual flow rate variation. These images were taken with the Fastcam high-speed camera. The oil was flowed at 2 µm/min continuously and the dye at 0.70µm/min, while the water’s flow rate was varied manually. The water flow rates from left to right are: 0.80µm/min, 0.90 µm/min and 1.25µm/min.
After the external valves were introduced to the system, stability of the system was lost and constant droplet formation was no longer accomplished. Droplet formation was established for only short periods of time, and during those times, the valves were successful in freezing the droplet arrays for image capture.
Fig. 9. Droplet array image capture. These images were taken with the Colorcam camera, immediately after valves were activated to shut off all fluid flow into the device. These results show the effectiveness of the valves in freezing the droplet array for image analysis and gradient characterization.
Once the fluid flow route was changed on the valves, stability of the system was restored at flow rates of 1µm/min (dispersed phase) and 2µm/min (continuous phase). Droplets have not been consistently generated at these flow rates, indicated that complete stability of fluid flow has not yet been acheived.
DISCUSSION
The results of this study indicate that the multiple improvements made in device design were successful in that, a stable droplet-forming microfluidic device was established that is able to withstand rate variations during droplet generation. Although the current devices work very well, there are occasional problems with debris and backflow during initial stabilization of the device. Additional design changes have been made and are still in the fabrication stage. These changes include better filtration systems and wider (100µm) array channels. The newer filtration designs should solve the current debris problem (clogging of the 7.5 µm wide orifice) by having flow-through channels of 6 µm in width. The debris-capturing design should also keep the filtration system itself from becoming clogged. The wider channels should decrease the potential for backflow of the system by providing a region of lower resistance for fluid to flow into during stabilization. The wider channels should also help slow droplets down for better image capture.
The results of this study also clearly demonstrate the effectiveness of the external valves in freezing the droplet array for imaging. The external valves, however, introduced unsystematic instability to the system. Stability of fluid delivery was regained by re-routing the fluid flow within the valves. Consistent droplets have not been generated at the flow rates demonstrating this stability. Further experimentation at lower aqueous reagent flow rates is required for conclusions to be drawn concerning the use of the external valves in this manner as a stable means of droplet generation and capture. In addition to the instability that still remains, the valves may introduce a potential problem in future applications of this device. Because the valves are run “backwards” during this approach, they are activated during the entire experiment; the solenoids that actuate the valves generate heat with the constantly applied current. The heat generated by the valves could potentially be harmful to reagents and/or solutions such as proteins used during experiments. A potential solution to this problem, if it were to manifest would be to develop a cooling bath for the valves to absorb the excess heat generated. Replacement of the valves may also be considered, perhaps with valves designed for more sensitive fluid flow studies. An internal valve system may also be taken into consideration, for this would simplify the experimental setup and allow for a more compact and portable device.
In the near future, additional control over the step gradient will hopefully be attained via creation of a step profile in a text-feed windows program (Hyper Terminal) connected to the syringe pumps. This program has been briefly experimented with, but must be further explored. In addition to this simplistic text-feed program, other programs such as LabView will also be explored so that more complex gradient shapes may be established for use in future studies performed within the device. Once a controlled, reproducible gradient has been established, extensive characterization and quantification of the gradient will be conducted. Characterization studies will be carried out by using GFP (green fluorescent protein) as a marker in one of the aqueous solutions, so that exact concentration ratios within droplets and exact droplet volumes can be determined with ImageJ image analysis software. Incorporating a computer-controlled flow rate program and conducting extensive characterization of the gradient array device, will allow for control over the droplet volumes and gradient characteristics so that this device can be utilized in experiments such as protein crystallization condition screening, protein unfolding/refolding, chemical synthesis studies and others in the future.
REFERENCES
[1] Duffy, D. C., Anderson, J. R., Chiu, D.T., Wu, H., Schueller, O. J. A., McDonald, J. C., and Whitesides, G. M. “Fabrication of microfluidic systems in poly(dimethylsiloxane)” Electrophoresis 21 (2000): 27-40.
[2] Duffy, D.C., McDonald, J.C., Schueller, J.A., and Whitesides, G.M. “Rapid Prototyping of Microfluidic Systems in Poly(dimethylsiloxane).” Analytical Chemistry 70.23 (1998): 4974-4984.
[3] He, M., Edgar, S.J., Gavin, D.M.J., Lorenz, R.M., Shelby, J.P., and Chiu, D.T. “Selective Encapsulation of Single Cells and Subcellular Organelles into Picoliter and Femtoliter-Volume Droplets.” Dept. Chem. UofW. Reviewed: 2005
[4]Nisisako, T., Torii, T. and Higuchi, T. “Novel microreactors for functional polymer beads.” Chemical Engineering Journal. 101.1-3. (2004): 23-29
[5] Runyon, M.K., Johnson-Kerner B.L., and Ismagilov, R.R. “Minimal Functional Model of Hemostasis in a Biomimetic Microfluidic System.” Angew. Chem. Int. Ed. 43 (2004): 1531-1536.
[6] Squires, T.M. and Quake, S.R. “Microfluidics: Fluid physics at the nanoliter scale.” Reviews of modern physics. 77 (2005): 978-1015.