Continuous and Size-Dependent Sorting of Emulsion Droplets Using

On the other hand, Microdevice B was designed for the sorting of emulsion droplets prepared previously. It has two inlets and six outlets. The depth o...
0 downloads 0 Views 410KB Size
Download PDF
Langmuir 2008, 24, 4405-4410

4405

Continuous and Size-Dependent Sorting of Emulsion Droplets Using Hydrodynamics in Pinched Microchannels Hirosuke Maenaka,† Masumi Yamada,‡ Masahiro Yasuda,† and Minoru Seki*,†,§ Osaka Prefecture UniVersity, 1-1 Gakuen-cho, Naka-ku, Sakai, Osaka 599-8531, Japan, Tokyo Women’s Medical UniVersity, 8-1 Kawada-cho, Shinjuku-ku, Tokyo 162-8666, Japan, and Chiba UniVersity, 1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan ReceiVed NoVember 16, 2007. In Final Form: January 17, 2008 In this report, a microfluidic system is presented for continuous and size-dependent separation of droplets utilizing microscale hydrodynamics. The separation scheme is based on laminar-flow focusing and spreading in a pinched microchannel, referred to as “pinched flow fractionation (PFF)”, which was previously developed for the sizedependent separation of solid particles, such as polymer microparticles or cells. By simply introducing emulsion and the continuous phase into a microchannel, continuous separation could be achieved without using complicated operations or devices. We first examined whether this scheme could be applied for droplets, by using a pinched microchannel with one outlet, and observed the behaviors of monodisperse droplets generated at the upstream T-junction. Analysis via high-speed imaging revealed that the length of the pinched segment is critical for precise separation of droplets. Then, separation of a polydisperse oil-in-water emulsion that was prepared previously was demonstrated using a microfluidic device equipped with multiple outlets. These results showed the ability of the presented system to sort or select specific-sized droplets easily and accurately, which would be difficult to achieve using normal-scale schemes, such as centrifugation or filtration.

Introduction Preparation of monodisperse emulsions has been one of the important techniques in the field of precision chemical or pharmaceutical engineering, since uniform physical properties are essential to accurate control of chemical reactions or physical interactions with outer substrates. For example, polymeric microspheres with uniform size distribution are obtained from monodisperse emulsions, which have been utilized for the columns of high-performance chromatography and as a spacer for LCD devices. Also, monodisperse emulsions are advantageous for fundamental chemical/biological studies, since uniform characteristics give simpler results than polydisperse emulsions. However, when we prepare emulsions using a mechanical homogenizer or mixer, droplets with a wide size distribution are usually obtained. In order to produce monodisperse emulsions, specific techniques such as membrane emulsification1,2 must be adopted. In recent years, microfluidic devices have been developed as a new tool for conducting microscale reaction, analysis, and separation.3,4 Microfluidic devices are suitable for manipulating micrometer-size objects including cells, particles, droplets, and macromolecules, due to the inherently small size of their microchannel. It has been recognized that one of the most important applications of microfluidic devices is the production of monodisperse micrometer-size droplets/emulsions.5 A number * To whom correspondence should be addressed. Phone/Fax: +81-43290-3436. E-mail: [email protected]. † Osaka Prefecture University. ‡ Tokyo Women’s Medical University. § Chiba University. (1) Joscelyne, S. M.; Tragardh, G. J. Membr. Sci. 2000, 169, 107-117. (2) Nakashima, T.; Shimizu, M.; Kukizaki, M. AdV. Drug DeliVery ReV. 2000, 45, 47-56. (3) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623-2636. (4) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (5) Song, H.; Chen, D. L.; Ismagilov, R. F. Angew. Chem., Int. Ed. 2006, 45, 7336-7356.

of studies have been conducted on microscale emulsification techniques using T-shaped microchannel confluence,6,7 flowfocusing,8,9 and microchannel array.10-12 These systems have been applied to production of monodisperse particles,12,13 dropletbased reactions in time,14 confinement of biological entities,15,16 and so on. The obvious alternative to the direct production of monodisperse emulsions is the development of methods for separating droplets according to size. Such methods would potentially be useful not only for the production of monodisperse emulsions but also for the selection/removal of specific-size droplets from complex mixtures, such as biological emulsions and vesicles. On a normal scale, it would be difficult to separate emulsion droplets according to size, since techniques like filtration or centrifugation could not be applied to fragile materials. On the other hand, hydrodynamics in the microscale is considered to be a means to achieve the separation of droplets in a mild condition. Recently, several researchers have reported on the possibility for size-dependent separation of droplets using microfluidic devices.17,18 For instance, when a microchannel emulsification technique is employed using a T-shaped microchannel confluence, (6) Tice, J. D.; Song, H.; Lyon, A. D.; Ismagilov, R. F. Langmuir 2003, 19, 9127-9133. (7) Nisisako, T.; Torii, T.; Higuchi, T. Lab Chip 2002, 2, 24-26. (8) Xu, Q. Y.; Nakajima, M. Appl. Phys. Lett. 2004, 85, 3726-3728. (9) Garstecki, P.; Fuerstman, M. J.; Stone, H. A.; Whitesides, G. M. Lab Chip 2006, 6, 437-446. (10) Sugiura, S.; Nakajima, M.; Tong, J. H.; Nabetani, H.; Seki, M. J. Colloid Interface Sci. 2000, 227, 95-103. (11) Sugiura, S.; Nakajima, M.; Iwamoto, S.; Seki, M. Langmuir 2001, 17, 5562-5566. (12) Sugiura, S.; Nakajima, M.; Itou, H.; Seki, M. Macromol. Rapid Commun. 2001, 22, 773-778. (13) Seo, M.; Nie, Z. H.; Xu, S. Q.; Mok, M.; Lewis, P. C.; Graham, R.; Kumacheva, E. Langmuir 2005, 21, 11614-11622. (14) Gerdts, C. J.; Sharoyan, D. E.; Ismagilov, R. F. J. Am. Chem. Soc. 2004, 126, 6327-6331. (15) He, M. Y.; Edgar, J. S.; Jeffries, G. D. M.; Lorenz, R. M.; Shelby, J. P.; Chiu, D. T. Anal. Chem. 2005, 77, 1539-1544. (16) Tan, Y. C.; Hettiarachchi, K.; Siu, M.; Pan, Y. R.; Lee, A. P. J. Am. Chem. Soc. 2006, 128, 5656-5658. (17) Tan, Y. C.; Lee, A. P. Lab Chip 2005, 5, 1178-1183.

10.1021/la703581j CCC: $40.75 © 2008 American Chemical Society Published on Web 03/08/2008

4406 Langmuir, Vol. 24, No. 8, 2008

Figure 1. Schematic diagram showing the basic principle of “pinched flow fractionation” for size-dependent separation of particles/droplets. The initial region of the flow containing particles/droplets is graycolored.

satellite small droplets are produced coincidently with the main droplets; hence, the use of branched channels for the separation of satellite droplets from main droplets has been reported.17 Also, there is a report on a microfluidic gravity-driven system for sorting droplets according to size, which achieved continuous fractionation of droplets.18 In addition, split-flow thin (SPLITT) fractionation was utilized for the separation of emulsion droplets, using a channel with a depth of several hundred micrometers.19 In our previous study, we developed a method for separating particles according to size, named “pinched flow fractionation (PFF)”, using hydrodynamics in a microchannel with two inlets and a pinched segment.20-22 The basic principle is shown in Figure 1. By continuously introducing liquid flows with and without particles from each inlet, and by controlling the inlet flow rates, the positions of the introduced particles can be focused on one sidewall in the pinched segment. Then, by utilizing the spreading flow profile at the boundary between the pinched segment and the following broadened segment, the slight difference between the positions of large and small particles could be amplified, and these particles could then be separated perpendicularly to the flow direction according to size. By employing microchannels with multiple outlets, the separated particles could be individually recovered. This method is advantageous for particle sorting, since continuous processing can be achieved without complex structures or operations. In this study, we applied the PFF method to the size-dependent separation of emulsion droplets using microfluidic channels. If it is possible to sort droplets using the PFF method without employing complex devices or schemes, the system will be useful in various industrial and research applications. Also, such a system will play an important role as a technique to sort droplets in an integrated droplet-base microsystem. In the experiment, we first fabricated a microchannel with three inlets, a T-junction, a pinched segment, a broadened segment, and an outlet, to produce uniformsize droplets at the T-junction and observe the droplet behavior. Then, a microchannel with six outlets was designed and fabricated, in order to sort emulsion droplets, which were prepared outside using a conventional homogenizer. Experimental Section Materials. Silicon wafer (100) was obtained from Furuuchi Chemical Crop. (Tokyo, Japan). Negative photoresists, SU-8 2025, 2050, and 2075, were obtained from MicroChem Corp. (MI). (18) Huh, D.; Bahng, J. H.; Ling, Y. B.; Wei, H. H.; Kripfgans, O. D.; Fowlkes, J. B.; Grotberg, J. B.; Takayama, S. Anal. Chem. 2007, 79, 1369-1376. (19) Fuh, C. B.; Giddings, J. C. J. Microcolumn Sep. 1997, 9, 205-211. (20) Yamada, M.; Nakashima, M.; Seki, M. Anal. Chem. 2004, 76, 54655471. (21) Takagi, J.; Yamada, M.; Yasuda, M.; Seki, M. Lab Chip 2005, 5, 778784.

Maenaka et al. Polydimethylsiloxane (PDMS; Sylpot 184) was obtained from Dow Corning Asia (Tokyo, Japan). Triolein was obtained from Kanto Chemical Co., Inc. (Tokyo, Japan). Sodium dodecyl sulfate (SDS) and lecithin from soybeans were obtained from Wako Pure Chemical Ind. (Osaka, Japan). Microdevice Fabrication and Design. Microdevices were fabricated using rapid prototyping and replica molding techniques.23 Each silicon wafer was coated with SU-8 using a spin-coater, and it was prebaked, irradiated with UV under a photo mask, postbaked, and finally developed to obtain a mold structure. PDMS prepolymer was then cast on the master and cured. The PDMS replica with microgroove structures was peeled off from the master and was bonded with another PDMS plate after O2 plasma treatment. In this study, two types of microdevices (Microdevices A and B) were designed and fabricated, as shown in Figure 2. Microdevice A was designed to produce monodisperse droplets at the T-junction, and to facilitate observation of droplet behavior. To widely vary the size of the produced droplets through the T-junction, several types of microchannels with different T-junction geometries were fabricated. For this purpose, PDMS plates with the pinched and broadened segments were produced from an identical master, while PDMS plates with the T-junction were produced from different masters, and then, these two types of plate were bonded. Downstream of the pinched segment, there is a droplet-observing point with a width of 12 mm, for the ease of accurate analysis of the droplet size due to reduced flow speed. The depth of the T-junction was ∼20 µm, while that of the pinched segment was ∼90 µm. The width of the horizontal channel at the T-junction varied from 20 to 100 µm. On the other hand, Microdevice B was designed for the sorting of emulsion droplets prepared previously. It has two inlets and six outlets. The depth of the microchannel was almost uniform, ∼50 µm. Behavior of Monodisperse Droplets. In order to verify the applicability of the PFF method for size-dependent separation of droplets, we first examined whether the positions of flowing droplets in the broadened segment differ when the droplet size is varied, by using Microdevice A with a pinched-segment length of 320 µm. Monodisperse aqueous droplets were produced in an oil phase through the T-junction. For the continuous oil phase, triolein containing 1% (w/w) lecithin was used, while distilled water was used as the dispersed phase. Continuous and dispersed phases were respectively introduced from inlets 2 and 3 and inlet 1, using three syringe pumps (KDS200, KDScientific Corp., MA). The introduction of continuousphase flow from inlet 3 is essential to focus the droplet positions onto the upper sidewall regardless of size. The flow rates from inlets 1, 2, and 3 were 5-25, 75-95, and 900 µL/h; the sum of flow rates from inlets 1 and 2 was constant, 100 µL/h. Droplet behavior was monitored using an optical microscope (IX70, Olympus Corp., Tokyo, Japan) and a CCD camera (DXC-151, Sony Corp., Tokyo, Japan). The positions of flowing droplets in the broadened segment were measured at the detection line, which was located 1000 µm downstream from the boundary of the pinched and broadened segments. Also, the droplet size was measured by capturing images of flowing droplets at the droplet-observing point, and by analyzing via image processing. In addition, we used other two types of Microdevice A, with pinched-segment lengths of 100 or 200 µm, to examine the effect of the pinched-segment length on the droplet behavior. Droplet behavior in the pinched segment was observed using the microscope and a high-speed camera (VCC-H1000, Digimo Corp., Osaka, Japan). Photographs were captured at a rate of 1000 frames per second. Separation of Polydisperse Emulsions. As a practical application of the presented scheme, we tried to separate an emulsion, which was prepared previously, and collect droplets according to size, using Microdevice B (Figure 2b). An oil-in-water emulsion was prepared using pure water containing 1% (w/v) SDS as the continuous (22) Sai, Y.; Yamada, M.; Yasuda, M.; Seki, M. J. Chromatogr., Sect. A 2006, 1127, 214-220. (23) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H. K.; Schueller, O. J. A.; Whitesides, G. M. Electrophoresis 2000, 21, 27-40.

Size-Dependent Sorting of Emulsion Droplets

Langmuir, Vol. 24, No. 8, 2008 4407

Figure 2. Microdevice designs. (a) Microdevice A for droplet generation and observation, and (b) Microdevice B for sorting of polydisperse emulsions. phase and triolein as the dispersed phase. Continuous and dispersed phases were mixed at a ratio of 19:1, using a homogenizer (Polytron PT3100, Rotor PT-DA 3020/2, Kinematica Inc., Switzerland) at 6000 rpm for 1 min. Before conducting separation, the microchannel surface was made hydrophilic by treating with O2 plasma. The prepared emulsion was continuously introduced from inlet 1 at a flow rate of 100 µL/h, while pure water containing 1% (w/v) SDS was introduced from inlet 2 at a flow rate of 900 µL/h. Micrographs of flowing droplets were captured near each outlet, and the size distributions of each fraction were measured via image processing.

Results and Discussion Effect of Droplet Size on the Effluent Position. To examine whether the droplet behaviors differ when droplet sizes are varied, monodisperse droplets with various sizes were produced through the T-junction. The effluent positions of droplets were observed in the broadened segment of Microdevice A with the pinchedsegment length of 320 µm. Aqueous droplets with diameters of 30-60 µm could be produced using several types of microdevices with T-junctions of various channel widths. The ratio of the volumetric flow rates from inlets 1 and 2 and inlet 3 was 1:9; this ratio dominates the droplet behavior in the pinched segment, in the sense that droplet positions could be focused onto the upper sidewall of the pinched segment when this ratio was sufficiently high. In this experiment, the pinched-segment width was 100 µm, and the diameter of the smallest droplets was ∼30 µm, so it was expected that these droplets would be completely focused onto the sidewall. Droplet behavior at the boundary between the pinched and broadened segments when the droplet size was ∼45 µm is shown in Figure 3a. The effluent position L was ∼660 µm, and the position of the flowing droplets was stable due to the uniform size of the generated droplets. The relation between the droplet size and the position L is shown in Figure 3b. It was confirmed that the effluent position shifted toward the center of the broadened segment when the droplet size increased. That is, small droplets passed near the sidewall, while large droplets passed near the center of the broadened segment, showing the possibility of droplet separation according to size. In addition, it is known that satellite droplets are also produced in the microchannel emulsification technique using T-shaped

Figure 3. (a) Behavior of W/O droplets in Microdevice A with the pinched-segment length of 320 µm. The effluent positions of droplets are detected on the detection line. (b) Droplet effluent position L when the droplet sizes were varied.

confluence. For example, in this experiment, satellite droplets with a diameter of 4.4 µm were produced at the T-junction when the diameters of the main droplets were ∼30 µm. In the experiment, it was observed that the position of the flowing satellite droplets was nearer to the upper sidewall in the broadened segment than that of main droplets; the effluent position L of the main droplets on the detection line was ∼750 µm, while that of the satellite droplets was ∼960 µm. This result also shows the ability of the presented microdevice to separate main and satellite droplets according to size. In our previous study, it was shown that the effluent position of solid particles could be estimated using the following equation:20

4408 Langmuir, Vol. 24, No. 8, 2008

L ) wB -

D wB × 2 wP

Maenaka et al.

(1)

where L is the effluent position of particles in the broadened segment, wP and wB are the widths of the pinched and the broadened segments, respectively, and D is the particle diameter. This equation is based on the concept that the stream line in the pinched segment is uniformly amplified into the broadened segment, and the particle position is represented by its center. Also, it is postulated that the particle position is perfectly focused onto one sidewall in the pinched segment. On the basis of this equation, the theoretical values of the effluent positions of particles/droplets were calculated, as shown in Figure 3b. The experimental values of the effluent position were smaller than the theoretical values, and the difference between these values was ∼100 µm. One of the main reasons for this discrepancy seems to be the droplet behavior in the pinched segment, since it was observed that droplet positions possibly dissociated from the sidewall when passing through the pinched segment. Thus, the droplet behaviors were observed after changing the pinchedsegment length. Effect of the Pinched-Segment Length. To examine the effect of the pinched-segment length on the behavior of droplets, two other types of Microdevice A with different pinched-segment lengths (100 or 200 µm) were designed and fabricated. Droplets with constant diameters (∼44 µm) were produced, and their behaviors were observed. Droplet positions in the broadened segment are shown in Figure 4. By calculating using eq 1, the theoretical effluent position L of droplets with diameter of 44 µm is 772 µm. So, the effluent positions of flowing droplets were close to the theoretical value when the pinched-segment length LP was 100 µm. On the other hand, the effluent positions shifted toward the center of the microchannel when LP increased. This result suggests that a greater pinched-segment length caused the droplet positions to be closer to the center of the microchannel. To precisely visualize the positions of flowing droplets in the pinched segment, droplet behaviors were observed using a highspeed camera. High-speed images are shown in Figure 5. As can be seen, once the droplet position was focused onto the upper sidewall in the pinched segment, the droplet position gradually moved toward the center of the microchannel. The distances d between the sidewall and the center of the droplets at the exit of the pinched segment were 23.8, 28.6, and 30.2 µm, when the pinched-segment lengths LP were 100, 200, and 320 µm, respectively. By simply assuming that the stream line is uniformly amplified in the broadened segment, and that the droplet position can be represented by its center, the effluent positions of droplets L could be calculated: 754.4, 706.4, and 691.4 µm for LP of 100, 200, and 320 µm, respectively. These values corresponded well to the experimental results shown is Figure 4. The shift of the droplet positions in the pinched segment may be caused by the hydrodynamic drag force due to the rotation of the droplet motion, since the flow speed near the sidewalls is lower than that near the center of the microchannel. Thus, the focusing of the droplets onto the sidewall was compromised, and consequently, the separation efficiency was decreased when the pinched-segment length was larger than 200 µm. It could be suggested that more effective droplet separation is achieved when the pinched segment is relatively short. Separation of Emulsion Droplets Using a Microchannel with Multiple Outlets. To demonstrate the actual possibility of size-dependent separation of droplets using the concept of PFF, a polydisperse emulsion was separated and respectively collected

Figure 4. Droplet effluent position L when three types of Microdevice A with different pinched-segment lengths were used. The droplet diameter was ∼44 µm.

Figure 5. Droplet behaviors in microchannels with different pinchedsegment lengths: (a) 100 µm, (b) 200 µm, and (c) 320 µm. The value d is the distance between the channel wall and the center of the droplets at the end of the pinched segment. The droplet diameter was ∼44 µm.

using Microdevice B with a pinched-segment length of 100 µm. In Microdevice B, one of the outlet branch channels connected to outlet 6 was composed of a relatively long wide segment (1 mm in width) and a short narrow segment (200 µm in width), in order to decrease the flow resistance of this branch channel and to introduce a large portion of the liquid flow into it. Theoretically, approximately one-third of the introduced liquid flow would be distributed to outlet 6, and the rest would be equally distributed into outlets 1 through 5, which were calculated regarding the microchannel structure as a resistive circuit. Also, it was estimated that particles or droplets with diameters of 0-13 µm, 13-26 µm, 26-39 µm, and 40 µm and above, would be, respectively, recovered from outlets 1, 2, 3, and 4. In the experiment, the O/W emulsion was introduced from inlet 1, while the aqueous phase was introduced from inlet 2, with flow rates of 300 and 2700 µL/h, respectively.

Size-Dependent Sorting of Emulsion Droplets

Langmuir, Vol. 24, No. 8, 2008 4409

Figure 6. Micrographs of emulsion droplets; (a) emulsion droplets before separation, and (b-d) droplets flowing through the microchannel near outlets 1, 2, and 3 of Microdevice B, respectively. Scale bar, 100 µm. See Supporting Information for the movie of separation of emulsion droplets in the microchannel.

sizes were less than ∼60 µm, and the average diameter ( SD was 12.5 ( 12.6 µm. The droplet diameters recovered from outlets 1, 2, and 3 were 3.8 ( 1.5, 28.8 ( 7.4, and 47.7 ( 7.4 µm, respectively. Although the peaks slightly overlapped, it was confirmed that the droplets could be accurately sorted using this microdevice. There was a difference in droplet sizes between the theoretical estimation and the experimental results, especially in the case of large droplets recovered from outlets 2 and 3. One of the reasons for this difference would be the deformation of droplets caused by the strong shear stress due to the relatively narrowed pinched segment (50 µm in width) and the high flow rate of the continuous phase from inlet 2. Consequently, large droplets showed behaviors similar to those of solid particles with smaller sizes. If the flow rate into outlet 6 is increased by changing the microchannel design in order to distribute the flow more effectively, or if a microchannel with a larger number of outlets is employed, the emulsion droplets would be separated into larger number of fractions, and the size range of the droplets in each fraction would be narrowed.

Conclusions Figure 7. Size distributions of droplets (a) before and (b) after separation using Microdevice B.

A photograph of droplets before separation and photographs of droplets flowing toward outlets 1, 2, and 3 are shown in Figure 6. The droplets mainly passed through outlets 1, 2, and 3, although there are six outlets in the microdevice. The size distributions of these droplets are shown in Figure 7. Initially, the droplet

In this study, continuous and size-dependent separation of droplets utilizing microscale hydrodynamics in microfluidic devices was demonstrated. Owing to its simplicity in operation, the presented system can be applied to various industrial and research fields. Although the throughput in the experiment (300 µL/h) is not sufficiently high for most real-world applications, the throughput would be increased by optimizing the microchannel structure and by arranging multiple microchannels in parallel. We expect that this method can be applied to the separation of

4410 Langmuir, Vol. 24, No. 8, 2008

especially fragile materials, such as biological vesicles or microbubbles, as well as other materials. Acknowledgment. This research was supported in part by Grants-in-aid for research fellows from Japan Society for Promotion of Science (JSPS) and for Scientific Research B

Maenaka et al.

(16310101) from the Ministry of Education, Sports, Science, and Culture of Japan. Supporting Information Available: Movie of separation of emulsion droplets in the microchannel. This material is available free of charge via the Internet at http://pubs.acs.org. LA703581J