5708
Langmuir 2002, 18, 5708-5712
Effect of Channel Structure on Microchannel Emulsification Shinji Sugiura,†,‡ Mitsutoshi Nakajima,*,† and Minoru Seki‡ National Food Research Institute, Kannondai 2-1-12, Tsukuba, Ibaraki 305-8642, Japan, and Department of Chemistry and Biotechnology, The University of Tokyo, Hongo 7-3-1, Bunkyo-ku, Tokyo 113-8656, Japan Received April 7, 2002. In Final Form: May 7, 2002 Microchannel (MC) emulsification is a novel technique for producing monodispersed emulsions in which droplets are formed by spontaneous transformation caused by interfacial tension. An MC structure consists of a narrow channel and a slitlike terrace. In this study, to investigate the effect of the channel structure on the dynamic behavior of droplet formation from an MC, we performed MC emulsification using MCs with different channel widths and lengths and observed emulsification behaviors using a microscope equipped with a high-speed camera. The formed droplet diameter was independent of the channel width and length and was constant below the critical flow velocity, over which the droplet diameter drastically increased. The time for droplet detachment also drastically increased above the critical velocity. It was found that the critical velocity was high for narrow and long channels. It means that droplet formation is based on the spontaneous transformation caused by interfacial tension at high flow velocity using narrow and long channels. Narrow and long channels produce a large pressure drop in the channel. The large pressure drop in the channel may disturb the flow of the dispersed phase in the channel into the well through the terrace during detachment. Therefore, the dispersed phase on the terrace detaches in less time using narrow and long channels. As a result, an MC structure with narrower and longer channels is better for high emulsion production rates.
1. Introduction Emulsions have been utilized in various industries, including food, cosmetics, and pharmaceutical. Many of the most important properties of emulsion-base products (e.g., shelf life, appearance, texture, and flavor) are determined by the size of the droplets they contain.1 The stability, rheology, chemical reactivity, and physical properties depend on their droplet size and size distribution.1-3 Resistance to creaming and Ostwald ripening of emulsions are influenced by their size and size distribution.2,3 Colloidal interactions between emulsion droplets are also affected by the droplet size. Monodispersed emulsions are useful for fundamental studies because the interpretation of experimental results is much simpler compared to polydispersed emulsions.2 They can also serve as useful systems for measuring important properties of emulsions. For example, the stability of droplets can be monitored very simply, since changes in the droplet size are easily studied when using monodispersed droplets. Monodispersed emulsions can greatly reduce Ostwald ripening by reducing the effective Laplace pressure difference because the droplets are identical in size. Various systems have been used to produce emulsions on industrial and laboratory scales, including high-speed blenders, colloid mills, and high-pressure homogenizers.4,5 * To whom correspondence should be addressed. Phone: +81298-38-7997. Fax: +81-298-38-8122. E-mail:
[email protected]. † National Food Research Institute. ‡ The University of Tokyo. (1) Dickinson, E. An Introduction to Food Colloids; Oxford University Press: Oxford, 1992. (2) McClements D. J. Food Emulsions: Principles, Practice, and Techniques; CRC Press: Boca Raton, FL, 1999; Chapter 1. (3) Mason, T. G.; Krall, A. H.; Gang, H.; Bibette, J.; Weitz, D. A. In Encyclopedia of emulsion technology, Becher, P., Eds.; Marcel Dekker: New York, 1996; Vol. 4, Chapter 6. (4) McClements D. J. Food Emulsions: Principles, Practice, and Techniques; CRC Press: Boca Raton, FL, 1999; Chapter 6.
However, the emulsions produced by these systems exhibit considerable polydispersity, with typical droplet size distributions being 2-10 µm with high-speed blenders,1 1-5 µm with colloid mills,1 and 0.05-1 µm with highpressure homogenizers. Membrane emulsification, in which the pressurized dispersed phase passes through a microporous membrane and forms emulsion droplets, is a promising technique for producing monodispersed emulsions with a coefficient of variation of approximately 10%.6-8 The emulsion droplet size is controlled by the membrane pore size. This technique can be used to produce emulsions without strong mechanical stress.9 Another method for producing monodispersed emulsion has been proposed, in which shear rupturing in a couette flow is applied to produce monodispersed emulsion.10-12 Recently, we proposed a novel method for making monodispersed emulsion droplets from a microfabricated channel array.13 This emulsification technique is called microchannel (MC) emulsification. Emulsions with a coefficient of variation of less than 5% and droplet sizes of 3-100 µm have been successfully prepared using this technique.14-16 The droplet diameter is determined by the MC geometry.17 This technique is promising for (5) Walstra, P. In Encyclopedia of Emulsion Technology; Becher, P., Eds.; Marcel Dekker: New York, 1983; Vol. 1, Chapter 2. (6) Nakashima, T.; Shimizu, M.; Kukizaki, M. Key Eng. Mater. 1991, 61, 62, 513-516. (7) Joscelyne, S. M.; and Tra¨gårdh, G. J. Membrane Sci. 2000, 169, 107-117. (8) Abrahamse, A. J.; van der Padt, A.; Boom, R. M.; and de Heij, W. B. C. AIChE J. 2001, 47, 1285-1291. (9) Schro¨der, V.; Schubert, H. Colloids Surf. A 1999, 152, 103-109. (10) Mason, T. G.; Bibette, J.; Weitz, D. A. Phys. Rev. Lett. 1996, 77, 3481-3484. (11) Mason, T. G.; Bibette, J. Langmuir 1997, 13, 4600-4613. (12) Mabille, C.; Schmitt, V.; Gorria, Ph.; Calderon, F. L.; Faye, V.; Deminie`re, B.; Bibette, J. Langmuir 2000, 16, 422-429. (13) Kawakatsu, T.; Kikuchi, Y.; Nakajima, M. J. Am. Oil Chem. Soc. 1997, 74, 317-321.
10.1021/la025813a CCC: $22.00 © 2002 American Chemical Society Published on Web 06/22/2002
Effect of Channel Structure on Microchannel Emulsification
preparing not only monodispersed emulsions but also monodispersed microspheres, which are composed of various materials. We have applied it to the preparation of several types of oil-in-water emulsions, water-in-oil emulsions,18 lipid microparticles,19 and polymer microparticles.20 Monodispersed emulsions prepared by MC emulsification have also been used for fundamental studies.21 MC emulsification exploits the interfacial tension, which is the dominating force on micrometer scales, as the driving force for droplet formation.14 The energy input for MC emulsification is very low compared to the conventional emulsification technique because droplet formation from MC is based on spontaneous transformation.14 The microfabricated structure of an MC consists of a narrow channel and a slitlike terrace. The previous study demonstrated that the droplet size is affected by the terrace structure.17,19,22 In this study, we investigated the effect of the channel structure on dynamic behavior of droplet formation from an MC. Droplet formation was investigated at different flow velocities of the dispersed phase through the channel using MCs with different channel widths and lengths.
Langmuir, Vol. 18, No. 15, 2002 5709
Figure 1. Schematic of silicon MC plate.
2. Materials and Methods 2.1. Materials. Triolein (purity >90%) obtained from Nippon Lever B. V. (Tokyo, Japan) was used as the dispersed oil phase. MilliQ water was used as the continuous water phase. Sodium dodecyl sulfate (SDS) was purchased from Wako Pure Chemical Ind. (Osaka, Japan) and used as the surfactant for emulsification. 2.2. MC Emulsification. The laboratory-scale apparatus for MC emulsification was described previously.13 The emulsification behavior was observed through a glass plate using a microscope video system with a total magnification of ×1000. In this study, a high-speed camera (FASTCAM ultima 1024; Photoron Ltd., Tokyo, Japan), which can capture 16 000 frames/s, was equipped with a microscope to observe the dynamic droplet formation behavior. The MC module was initially filled with the continuous phase. The dispersed phase was pressurized and caused to flow into the module by lifting the liquid chamber filled with the dispersed phase. The dispersed phase supplied from the liquid chamber entered the space between the silicon MC plate and the glass plate, and droplets were formed from the MC. The prepared emulsions were recovered by a continuous phase flow. Figure 1 shows the schematic of the MC plate used in this study. It was fabricated by photolithography and orientationdependent etching.23 The MC structure consists of a channel and a terrace. Over the terrace end, there is a deeply etched well. In the present study, MC plates with different channel lengths and widths are used as shown in Table 1. MCs A, B, and C have similar terrace structures and different channel widths. MCs A, D, and E have similar terrace structures and different channel lengths. Droplet formation in MC emulsification is also depicted in Figure 2. During droplet formation, the distorted dispersed phase (14) Sugiura, S.; Nakajima, M.; Iwamoto, S.; Seki, M. Langmuir 2001, 17, 5562-5566. (15) Kobayashi, I.; Nakajima, M.; Nabetani, H.; Kikuchi, Y.; Shohno, A,; Satoh, K. J. Am. Oil Chem. Soc. 2001, 78, 797-802. (16) Sugiura, S.; Nakajima, M.; Seki, M. J. Am. Oil Chem. Soc. 2002, in press. (17) Sugiura, S.; Nakajima, M.; Seki, M. Langmuir 2002, 18, 38543859. (18) Kawakatsu, T.; Tra¨gårdh, G.; Tra¨gårdh, C.; Nakajima, M.; Oda, N.; Yonemoto, T. Colloids Surf. A 2001, 179, 29-37. (19) Sugiura, S.; Nakajima, M.; Tong, J.; Nabetani, H.; Seki, M. J. Colloid Interface Sci. 2000, 227, 95-103. (20) Sugiura, S.; Nakajima, M.; Itou, H.; Seki, M. Macromol. Rapid Commun. 2001, 22, 773-778. (21) Liu, X. Q.; Nakajima, M.; Nabetani, H.; Xu, Q. Y.; Ichikawa, S.; Sano, Y. J. Colloid Interface Sci. 2001, 233, 23-30. (22) Kawakatsu, T.; Tra¨gårdh, G.; Kikuchi, Y.; Nakajima, M.; Komori, H.; Yonemoto T. J. Surfactants Deterg. 2000, 3, 295-302. (23) Kikuchi, Y.; Sato, K.; Ohki, H.; Kaneko, T. Microvasc. Res. 1992, 44, 226-240.
Figure 2. Droplet formation process during MC emulsification. Table 1. Dimensions of MC Plates Used in This Study
MC plate
MC depth (µm)
channel length (µm)
channel width (µm)
terrace length (µm)
terrace width (µm)
MC-A MC-B MC-C MC-D MC-E
7.0 7.0 7.0 7.0 7.0
53.1 55.6 53.8 24.1 136.1
11.6 5.9 14.3 11.5 11.3
39.3 35.6 37.8 33.8 38.4
38.0 36.6 36.4 37.6 37.7
is spontaneously transformed into spherical droplets by interfacial tension.14 The dispersed phase passed through the channel inflates on the terrace in a disklike shape (inflation process). This distorted disklike shape is the essential point for spontaneous transformation, because the disklike shape has a higher interface area than the spherical shape, which results in the instability from the viewpoint of the interface free energy. When the dispersed phase reaches the end of the terrace, the dispersed phase detaches from the terraces and spontaneously transforms into spherical droplets (detachment process). 2.3. Measurement and Analytical Method. The droplet diameters were determined from pictures taken with the microscope video system described above. Winroof (Mitani
5710
Langmuir, Vol. 18, No. 15, 2002
Sugiura et al.
Figure 4. Microscope photographs of MC emulsification process below (A) and above (B) the critical velocity. The flow velocity of the dispersed phase is 1.1 (A) and 2.4 (B) mm/s.
Figure 3. Effect of the channel width on the formed droplet diameters and detachment times at different flow velocities of the dispersed phase. The channel widths are 5.9 µm for MC-B (O), 11.6 µm for MC-A (2), and 14.3 µm for MC-C (0). Critical velocities are the flow velocities of the dispersed phase, over which the droplet formation behavior changed drastically. Corporation, Fukui, Japan) software was used to analyze the captured pictures. The number-average diameters of the prepared emulsions were determined from pictures of 20 droplets taken with the microscope video system described above.
3. Results and Discussion MC emulsification was carried out using triolein as the dispersed phase and 1% SDS aqueous solution as the continuous phase. To investigate the effect of the channel structure on the dynamic behavior of the droplet formation, we investigated the droplet diameters and detachment times at different flow velocities of the dispersed phase using MC plates with different channel widths and lengths. The detachment time is defined as the time from the point that the dispersed phase reaches the terrace end to the point that the dispersed phase detaches from the MC and forms a separate droplet (Figure 2). The detachment time corresponds to the time for spontaneous transformation and represents the dynamic behavior during droplet formation. The detachment time was measured from the observed image with a high-speed camera. The droplet formation behavior from a single channel located at the center of the terrace line was observed. Flow velocities of the dispersed phase through the channel were calculated from the droplet diameters and droplet formation rates. Figure 3 shows the effect of the channel width on the formed droplet diameters and detachment times. The formed droplet diameters were independent of the channel width and were constant in low flow velocity ranges. These results are reasonable from the viewpoint of the droplet formation mechanism described in a previous article.14 The dispersed phase that has passed through the channel
inflates on the terrace in a disklike shape. When the dispersed phase reaches the end of the terrace, the dispersed phase flows into the well, and spontaneously transforms into spherical droplets. The droplet diameter was determined by the volume of the dispersed phase, which flows into the well during detachment.17 Therefore, the droplet diameter was determined by the volume of the dispersed phase inflated on the terrace, which is determined by the structure of the terrace. Consequently, the droplet diameter was independent of the channel structure and flow velocity. Different critical flow velocities were found for each channel width, over which the droplet formation behavior changed drastically. The formed droplet diameters were almost constant below the critical flow velocity and suddenly increased over the critical velocity. Detachment time also increased over the critical velocity. These results indicate that the droplet formation behavior changed above the critical velocity. Figure 4 shows microscope photographs of MC emulsification below and above the critical velocities using MC-A. Below the critical velocity, the interfacial tension, which is the driving force of droplet formation in MC emulsification, dominates other forces at this flow velocity range (Figure 4A). Consequently, monodispersed droplets were formed. Droplet formation is thus based on the spontaneous transformation caused by interfacial tension below the critical velocity. Above the critical velocity, the interfacial tension is not dominant. Consequently, larger droplets were formed (Figure 4B). The critical velocities were high for narrow channel. The critical velocity corresponds to the droplet formation rate from a single MC because formed droplet diameters were constant below the critical velocity. Therefore, a narrow channel provides a better emulsion production rate. Detachment times were also short for narrow channels. The narrower channel disturbed the flow of the dispersed phase from the channel into the well part during detachment because the narrow channel causes a large pressure drop in the channel. This prevents outflow of the
Effect of Channel Structure on Microchannel Emulsification
Langmuir, Vol. 18, No. 15, 2002 5711
the well by the spontaneous transformation caused by Laplace pressure expressed by the following equation14
(
PL ) γOW
)
1 1 + R1 R2
(1)
where PL is the pressure difference across the oil and water phases, γOW is the interfacial tension between the oil and water phases, and R1 and R2 are two principal radii of curvature of the interface. The pressure gradient between the terrace and well, which is based on the Laplace pressure, causes the dispersed phase on the terrace to flow into the well. At this moment, the dispersed phase in the channel also flows into the well through the terrace because the pressure in the channel is higher than in the well. This phenomenon can be observed optically by a microscope equipped with high-speed camera. The pressure drop in the channel disturbs the flow in the channel. Reynold’s number in the channel can be estimated by the following equation
Re )
Figure 5. Effect of the channel length on the formed droplet diameters and detachment times at different flow velocities of the dispersed phase. The channel lengths are 136.1 µm for MC-E (O), 53.1 µm for MC-A (2), and 24.1 µm for MC-D (0). Critical velocities are the flow velocities of the dispersed phase, over which the droplet formation behavior changed drastically.
dispersed phase even though the velocity of the dispersed phase is high. Consequently, droplet formation is based on the spontaneous transformation caused by interfacial tension at high flow velocity. In each MC, detachment time decreased slightly below the critical velocity. This phenomenon might be explained by dynamic interfacial tension. At a high flow velocity of the dispersed phase, the formation speed of new interface increased, leading to a higher dynamic interfacial tension than that of the equilibrium state. The higher dynamic interfacial tension, which is the driving force for droplet formation, would produce shorter detachment times. Figure 5 shows the effect of the channel length on the emulsification behavior. The formed droplet diameters were independent of the channel length, and they were constant below the critical flow velocity. The droplet diameter and detachment time drastically increased above the critical velocity, and the droplet formation behavior changed drastically. The critical velocities were high for long channels. This means that long channels yield better emulsion production rates. Droplet formation is based on the spontaneous transformation caused by interfacial tension at high flow velocity for long channels. Detachment times were also short for long channels because the long channel causes a large pressure drop in the channel. In Figure 5, detachment times decreased slightly below the critical velocity, probably due to dynamic interfacial tension. The experimental results show that narrow and long channels lead to high critical velocities, which subsequently lead to high droplet formation rates. During detachment, the dispersed phase on the terrace flows into
FUdeq ≈ 10-4 η
(2)
where F is the liquid density (F ) 1 × 103 (kg/m3)), deq is the equivalent diameter of the channel (deq ≈ 1 × 10-5 (m)), U is the characteristic velocity (U ) 1 × 10-3 (m/s)), and η is the liquid viscosity (η ) 5 × 10-2 (Pa‚s)). The low Reynold’s number means that the flow in the channel is laminar. The pressure drop of the flow in the channel can be estimated by Fanning’s equation24
∆Pchannel ) 4f
( )( ) l 1 FU2 deq 2
(3)
where ∆Pchannel is the pressure drop in the channel, f is Fanning’s friction factor, and l is the channel length. This equation is valid for steady flow of a liquid of constant F in the straight conduit and applicable to the system in the present study. Equation 3 shows that a narrow, long channel produces a large pressure drop in the channel. The large pressure drop in the channel may disturb the flow of the dispersed phase in the channel into the well through the terrace during detachment. Therefore, the dispersed phase on the terrace detaches in less time using narrow and long channels. Droplet formation is based on the spontaneous transformation caused by interfacial tension at high flow velocity. It results in the stable droplet formation at high velocities of the dispersed phase using narrow and long channels. 4. Conclusion To investigate the effect of the channel structure on the dynamic behavior of droplet formation from an MC, we performed MC emulsification using MCs with different channel widths and lengths. We measured droplet diameters and detachment times at different flow velocities of the dispersed phase. The experimental results show that the droplet diameters were almost independent of the channel width, channel length, and the flow velocity in lower flow velocity ranges, as interpreted from the viewpoint of the droplet formation mechanism. The results obtained in this study provide insight for the process design of MC emulsification. MC structures with narrower and longer channels lead to large pressure (24) Bird, R. B., Stewart, W. E., Lightfoot, E. N. In Transport Phenomena; John Wiley & Sons: New York, 1960; Chapter 6.
5712
Langmuir, Vol. 18, No. 15, 2002
drops in the channel. The large pressure drop in the channel disturbs the flow of the dispersed phase in the channel into the well through the terrace during detachment. This leads to shorter detachment times, resulting in stable droplet formation based on the spontaneous transformation caused by interfacial tension at high velocities of the dispersed phase. Thus, MC structures
Sugiura et al.
with narrower and longer channels are better for high emulsion production rates. Acknowledgment. This work was supported by the Nanotechnology Project, Ministry of Agriculture, Forestry and Fisheries, Japan. LA025813A
