Anal. Chem. 2009, 81, 8213–8218
Formation of a Water-Xylene Interface in a Microchannel without Sidewalls Masashi Watanabe* Faculty of Textile Science and Technology, Shinshu University, 3-15-1 Tokida, Ueda, Nagano 386-8567, Japan Channels in a microfluidic device were rapidly created using an office inkjet printer. The process was easy enough for common end users to carry it out by themselves. A water-xylene interface was formed by combining two parallel channels that were filled with water and xylene, respectively. Such an interface can be important in analytical and synthetic uses because various chemical processes, such as reaction, extraction, and separation, can be performed through the interface. The water-filled channel was 0.5-1.3 mm wide, and the xylene-filled one was 1.3 mm wide. Both channels were 0.09 mm deep and 20 mm long. The continuous flow of these fluids was successfully carried out using a syringe pump at a volume flow rate of 1 µL/min for 5 min. Microfluidic devices for chemical/biochemical analysis and synthesis have attracted much attention since the early 1990s.1 These devices normally have microchannels in which various chemical processes are carried out. Because the earlier concept was to utilize conventional microfabrication technologies, e.g., technologies for microelectromechanical systems (MEMS), as a miniaturization method in analytical chemistry, producing such devices has required expensive equipment and a relatively long time.2 These disadvantages do not seem to be a big problem when the devices are used as mass-produced and ready-made ones; these may be used for routine analyses in laboratories and hospitals. However, when an end user requires devices designed for his/her unique purpose (moreover, required devices may be few), these disadvantages will become obvious. Such a situation may be seen in laboratories because novel and unique experiments are often necessary to create something new in research and development. However, various alternative rapid and inexpensive production methods have also been proposed during the past decade.3-5 Although many of these methods used polydimethylsiloxane (PDMS) as the material, PDMS can swell in various organic * To whom correspondence should be addressed. E-mail: mwatana@ shinshu-u.ac.jp. (1) Manz, A.; Graber, N.; Widmer, H. M. Sens. Actuators, B 1990, 1, 244– 248. (2) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895–897. (3) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H.; Schueller, O. J. A.; Whitesides, G. M. Electrophoresis 2000, 21, 27–40. (4) Kobayashi, J.; Yamato, M.; Itoga, K.; Kikuchi, A.; Okano, T. Adv. Mater. 2004, 16, 1997–2001. (5) Grimes, A.; Breslauer, D. N.; Long, M.; Pegan, J.; Lee, L. P.; Khine, M. Lab Chip 2008, 8, 170–172. 10.1021/ac901511p CCC: $40.75 2009 American Chemical Society Published on Web 08/28/2009
solvents such as hydrocarbons,6 dichloromethane, and tetrahydrofuran; these are often used in organic syntheses. Therefore, microfluidic devices made from PDMS do not seem to be very suitable for organic syntheses. On the other hand, glass is a stable material both in organic solvents and aqueous solutions.7 However, producing glass microfluidic devices often requires an etching process with hydrofluoric acid, which is a harmful chemical.8 Recently, Watanabe reported on microfluidic devices that were made of glass slides but could be rapidly (for a few minutes) and inexpensively produced without any etching process.9 Because the surface of the glass slide used had a rough texture, the contact angle hysteresis, i.e., the difference between the advancing and receding contact angles, was enhanced.10 This enabled the liquid fluid, such as water or an organic solvent, to exist along a line, though there was no groove or chemical line pattern on the surface. This situation can be regarded as a channel that was filled with the fluid. Because this type of channel can be recreated after one erased the channel by rinsing the surface with an appropriate solvent, it is called a refreshable microfluidic channel.11 Similar, but partially different, channels often called “surfacedirected channels” have been reported by some groups.12-21 These channels could confine water along a hydrophilic line pattern that was chemically bonded to the surface, though the aforementioned refreshable microfluidic channels did not have such a chemically bonded pattern. Because the surface-directed channels did not have physical sidewalls, they were different from normal microchannels that could be regarded as pipes with a rectangular cross section. (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
Becker, H.; Gartner, C. Anal. Bioanal. Chem. 2008, 390, 89–111. Watts, P.; Haswell, S. J. Chem. Eng. Technol. 2005, 28, 290–301. McCreedy, T. Anal. Chim. Acta 2001, 427, 39–43. Watanabe, M. Lab Chip 2009, 9, 1143–1146. de Gennes, P.-G.; Brochard-Wyart, F.; Quere, D. Capillarity and Wetting Phenomena; Springer-Verlag: New York, 2004; Chapter 9. Watanabe, M. Sens. Actuators, B 2006, 114, 296–300. Oh, C. S. U.S. Pat. 1999, 5, 904–824. Lam, P.; Wynne, K. J.; Wnek, G. E. Langmuir 2002, 18, 948–951. Zhao, B.; Moore, J. S.; Beebe, D. J. Anal. Chem. 2002, 74, 4259–4268. Bouaidat, S.; Hansen, O.; Bruus, H.; Berendsen, C.; Bau-Madsen, N. K.; Thomsen, P.; Wolff, A.; Jonsmann, J. Lab Chip 2005, 5, 827–836. Lee, J. S. H.; Barbulovic-Nad, I.; Wu, Z.; Xuan, X.; Li, D. J. Appl. Phys. 2006, 99, 054905. West, J.; Michels, A.; Kittel, S.; Jacob, P.; Franzke, J. Lab Chip 2007, 7, 981–983. Suk, J. W.; Cho, J.-H. J. Micromech. Microeng. 2007, 17, N11–N15. Sridharamurthy, S. S.; Jiang, H. IEEE Sens. J. 2007, 7, 1315–1316. Swickrath, M. J.; Shenoy, S.; Mann, J. A., Jr.; Belcher, J.; Kovar, R.; Wnek, G. E. Microfluid. Nanofluid. 2008, 4, 601–611. Yang, S. Y.; DeFranco, J. A.; Sylvester, Y. A.; Gobert, T. J.; Macaya, D. J.; Owens, R. M.; Malliaras, G. G. Lab Chip 2009, 9, 704–708.
Analytical Chemistry, Vol. 81, No. 19, October 1, 2009
8213
Thus, surface-directed channels were created on flat surfaces using chemical surface patterns. Such patterns were also used to control the liquid flow inside physical microchannels.22,23 Zhao et al. realized the interface between water and an organic solvent by chemically patterning the inside surface of the physical channels.24 The interface between water and an organic solvent is useful in microfluidic devices because various unit operations, such as reaction,25,26 extraction,27-29 and separation,30 can be carried out through it. There have been many studies about the water-organic solvent interface using normal microchannels.31,32 Many of them used physical guide structures and/or chemical patterning inside the channel in order to obtain the interface.25,28,33 However, as far as the current author knows, such an interface has not been realized using surface-directed channels or refreshable microfluidic channels. Therefore, in this study, the author investigated the procedure to obtain such an interface using refreshable microfluidic channels. As stated above, refreshable microfluidic channels do not have either chemical patterns or physical guide structures on the surface. Therefore, a novel method had to be developed in order to obtain a water-organic solvent interface.
Figure 1. Geometry of the (a) U-shaped, (b) L-shaped, and (c) V-shaped channels.
The surface energy then changes by an amount dE as dE ) 4w dy(γSV′ - γLS) + 4|R1 | dx(γLS - γSV) - 4h dyγLV 2h dxγLV (2)
THEORY Consider two parallel glass slides slightly separated by distance h and assume that a liquid initially exists between these slides like a long string with width w. This situation can be regarded as a microchannel that has two “physical walls” (i.e., top and bottom glass slides) and two “virtual sidewalls” (i.e., air); this channel is filled with the liquid. Here, assume a U-shaped channel as shown in Figure 1a. If the contact angle of the liquid on the slide is less than 90°, the pressure difference of the Laplace equation, ∆P, is negative at the liquid-air interface inside the U shape, i.e., ∆P ) γ(1/R1 + 1/R1′) < 0, where γ is the surface tension of the liquid, and R1 and R1′ (R1, R1′ < 0) are the radii of curvature inside the U shape along the planes parallel and vertical to the slide surface, respectively.14 Therefore, the interface inside the U shape will move right by dx under an appropriate condition. At the same time, both ends of the channel will also move left by dy because the volume of the liquid must be constant; this can be expressed as
where γSV, γLS, and γLV are the surface energies per unit area at the solid-air, liquid-solid, and liquid-air interfaces, respectively. In addition, γSV′ is used as the energy for the solid surface that underwent wetting because the properties of the solid surface can be changed by such an experience. Considering the equilibrium of the forces acting on the threephase contact line, γSV and γSV′ are written as γSV ) γLS + γLV cos θ
(3)
γSV′ ) γLS + γLV cos θ′
(4)
where θ is the contact angle at the inside of the U shape and θ′ is the angle at the right end of the channel (see Figure 1a). Because dE/dx ) 0 at equilibrium and θ < θA, θ′ > θR, eqs 1-4 become |R1 | >
2|R1 | dxh ) 2w dyh
(1)
(22) Besson, E.; Gue, A.-M.; Sudor, J.; Korri-Youssoufi, H.; Jaffrezic, N.; Tardy, J. Langmuir 2006, 22, 8346–8352. (23) Zhao, B.; Moore, J. S.; Beebe, D. J. Science 2001, 291, 1023–1026. (24) Zhao, B.; Viernes, N. O. L.; Moore, J. S.; Beebe, D. J. J. Am. Chem. Soc. 2002, 124, 5284–5285. (25) Maruyama, T.; Uchida, J.; Ohkawa, T.; Futami, T.; Katayama, K.; Nishizawa, K.; Sotowa, K.; Kubota, F.; Kamiya, N.; Goto, M. Lab Chip 2003, 3, 308– 312. (26) Hisamoto, H.; Saito, T.; Tokeshi, M.; Hibara, A.; Kitamori, T. Chem. Commun. 2001, 2662–2663. (27) Reddy, V.; Zahn, J. D. J. Colloid Interface Sci. 2005, 286, 158–165. (28) Tokeshi, M.; Minagawa, T.; Uchiyama, K.; Hibara, A.; Sato, K.; Hisamoto, H.; Kitamori, T. Anal. Chem. 2002, 74, 1565–1571. (29) Fries, D. M.; Voitl, T.; von Rohr, P. R. Chem. Eng. Technol. 2008, 31, 1182– 1187. (30) Weigl, B. H.; Yager, P. Science 1999, 283, 346–347. (31) Atencia, J.; Beebe, D. J. Nature 2005, 437, 648–655. (32) Shui, L.; Eijkel, J. C. T.; van den Berg, A. Adv. Colloid Interface Sci. 2007, 133, 35–49. (33) Aota, A.; Hibara, A.; Kitamori, T. Anal. Chem. 2007, 79, 3919–3924.
8214
Analytical Chemistry, Vol. 81, No. 19, October 1, 2009
h 2(cos θR - cos θA - h/w)
(5)
where θA and θR are the advancing and receding contact angles, respectively. Thus, if |R1| is so small as not to satisfy eq 5, the liquid will fill the inside of the U-shaped channel. As another case, consider an L-shaped channel (Figure 1b). The liquid will flood the channel so as to make the inside of the corner round because when the inner radius of curvature of the liquid-air interface, |R2|, is small, |∆P| of the Laplace equation increases. Similar to eq 5, the radius at equilibrium can be written as |R2 | >
h 2(cos θR - cos θA - h/w)
(6)
As shown in Figure 1c, also consider a V-shaped channel. If two channels (Ch 1 and Ch 2) are filled with liquid and they are in contact with each other at point C, the space between these
Figure 3. Surface profiles of the glass slide (a) before and (b) after the slide was ground with an abrasive.
Figure 2. Illustration of the strategy for creating an interface between water and an organic solvent.
channels will be filled with the liquid. The liquid-air interface between the channels, i.e., the arc whose radius of curvature is |R3|, will move right. Simultaneously, another interface, line segment AB, will also move right, if the arc is sufficiently far from the right end of the channel. When L . a, w (see Figure 1c), similar to eq 5, the radius of curvature |R3| at equilibrium can be written as |R3 | >
h 2{cos θR - cos θA - h/(2w)}
(7)
In the above theory, the glass slides have no chemically bonded surface patterns. However, this theory can also apply to normal surface-directed channels that have such patterns.12-21 In this case, θA and θR are the advancing and receding contact angles in the hydrophobic and hydrophilic regions, respectively. STRATEGY On the basis of the above theory, the following is a strategy to form an interface between water and an organic solvent. First, two channels (Ch 3 and Ch 4 shown in Figure 2a) are filled with an organic solvent and water, respectively, though the organic solvent is immiscible in water. Although Ch 3 has a short projection, the corner DEF will become round according to eq 6. After arc DF touches point G, a U-shaped channel including arc DH will be formed (Figure 2b). If the radius of curvature of this arc is small enough not to satisfy eq 5, the organic solvent will fill the space between Ch 3 and Ch 4. (Note that water will not fill the space because the advancing contact angle of water on a hydrophobic substrate is normally larger than that of an organic solvent.) Therefore, the two channels are expected to come in contact with each other as if a zipper is closed (Figure 2c).
EXPERIMENT Preparation of Hydrophobic and Oil-Repellent Glass Slides. Two glass slide pieces (60 × 26 mm2 and 76 × 26 mm2, 1 mm thick) were ground with an abrasive (30 µm of diamond paste, Trusco, Tokyo, Japan) in order to make their surfaces rough. The surface texture parameters, Ra (the arithmetical mean deviation of the roughness profile), RSm (the mean width of the profile elements), and r (the actual surface area/ geometric surface area) were measured using a stylus instrument (SJ-201R, Mitutoyo, Kawasaki, Japan).34,35 On the basis of the measured surface profile (Figure 3), the values of these parameters were calculated to be Ra ) 1.22 µm, RSm ) 57 µm, and r ) 1.04. In order to make the slide surfaces hydrophobic and oil repellent, a neat liquid of (tridecafluoro-1,1,2,2-tetrahydrooctyl)trimethoxysilane (TFS, Gelest, Morrisville, PA) was applied to the slide, which was then cured in an oven at 110 °C for 1 h and finally rinsed with acetone. This TFS treatment was repeated three times. An additional treatment was carried out by immersing the slide in hexamethyldisilazane (Wako Pure Chemical Industries, Osaka, Japan) at room temperature for 10 min. The slides were then rinsed with acetone and dried in air. Spacers (0.09 mm thick) made of polypropylene adhesive tape were attached to the longer edges of the 60 mm long slide (Figure 4). In addition, small notches were created at both ends of this slide using a diamond file.36 These notches were used to fix the tip of a syringe needle, as shown in Figure 5. Creation of Channels between the Two Stacked Glass Slides. As shown in Figure 4, lines were printed on each slide using an office inkjet printer (PM-G730, Epson, Suwa, Japan), which was equipped with an ink cartridge filled with 2-(2ethoxyethoxy)ethanol (EE)/isopropyl alcohol (IPA) (75/25 wt %) as the ink. This ink easily dissolves in both water and various (34) International Organization for Standardization. ISO 4287: 1997. (35) Wenzel, R. N. J. Phys. Colloid Chem. 1949, 53, 1466–1467. (36) Watanabe, M. Chem. Eng. Technol. 2008, 31, 1196–1200.
Analytical Chemistry, Vol. 81, No. 19, October 1, 2009
8215
Figure 4. Glass slides on which lines were printed using an inkjet printer.
Figure 5. Experimental setup to inject fluids into the channels.
organic solvents, such as xylene. Although the details of the method of printing lines on glass slides have already reported by Watanabe,37 a refillable ink cartridge (ED-1, Daiko, Tokyo, Japan) and a modified CD/DVD loading tray were used. Cho et al. also used a similar method in order to fabricate microsensors.38 The software used to draw the lines was Microsoft PowerPoint 2002. The mass of the ink released from the print head was controlled by changing the transparency of the line drawings using this software. The slides were then stacked and clipped. Water (distilled water) and m-xylene as the fluids were injected into the notches using a syringe pump (ESP-64, Eicom, Kyoto, Japan), as shown in Figure 5. These fluids went ahead along the printed lines, i.e., two channels filled with these fluids were formed. To clearly visualize the interface between these fluids, the water and m-xylene were colored with brilliant blue FCF (0.04 wt %) and 1-amino-4-hydroxyanthraquinone (0.1 wt %), respectively. The used slides were cleaned by rinsing with water and methanol and repeatedly used. RESULTS AND DISCUSSION Experimental Verification of the Theory. Although three shapes (U, L, and V) were discussed in the Theory section, the V-shaped channel (Figure 1c) was selected to experimentally verify the theory because of experimental convenience. The channel size used in this experiment was w ) 1 mm, L ) 50 or 20 mm, and a ) 1 mm. According to the procedure shown in Figure 6, the V-shaped channel was filled with EE (100%) as the fluid. The ink (37) Watanabe, M. Sens. Actuators, B 2007, 122, 141–147. (38) Cho, H.; Parameswaran, M.; Yu, H.-Z. Sens. Actuators, B 2007, 123, 749– 756.
8216
Analytical Chemistry, Vol. 81, No. 19, October 1, 2009
Figure 6. Procedure to fill the V-shaped channel with a fluid. (a) First, the corresponding V shape was printed on two glass slides using EE (100%) as the ink. These slides were then stacked using a spacer (0.09-0.23 mm thick). (b) The channels, Ch 1 and Ch 2, were partially filled with EE that was also used as the fluid. (c) A small amount of the fluid was then removed from the channels using small pieces of filter paper. (d) The stacked slides were carefully tilted until the fluid in one channel joined the other channel at the junction. (e) While the stacked slides were horizontally placed again, the junction spontaneously moves right while increasing the radius of curvature |R3| and finally stops at the equilibrium position.
Figure 7. Photograph of the V-shaped channel filled with EE as the fluid. The spacer thickness was 0.189 mm.
was also EE (100%). As shown in Figure 7, the radius of curvature |R3| was measured by fitting a circle to the inside of the V-shaped channel. Figure 8 shows the radii measured using spacers with various thicknesses. The advancing and receding contact angles of EE were also measured by photographing the EE drop (100 µL) sliding on the tilted glass slide (Table 1).39 On the basis of these contact angle values, the minimum values of |R3| for various spacer thicknesses were predicted using eq 7. This theoretical prediction was roughly consistent with the measured values (Figure 8). (39) Furmidge, C. G. L. J. Colloid Sci. 1962, 17, 309–324.
Figure 8. Radius of curvature inside the V-shaped channel as a function of the spacer thickness. The circles represent the measured radii. The broken line represents the theoretical prediction based on eq 7. Table 1. Advancing and Receding Contact Angles on the Glass Slide
sample 2-(2-ethoxyethoxy) ethanol m-xylene distilled water a
advancing contact angle, θA (°)
receding contact angle, θR (°)
65.8
∼0a
69.1 128.9
19.5 52.6
Figure 9. Time-series photographs during the formation of the water-xylene interface at (a) 0, (b) 7.2, (c) 7.5, and (d) 7.8 s.
The contact line did not recede when the slide was tilted.
Creation of an Interface between Water and m-Xylene in the Channel. An interface between water and m-xylene was created using the aforementioned strategy. Because m-xylene is immiscible in water, it was selected as an example of organic solvents. The ink used in this experiment was EE/IPA (75/25 wt %), which was miscible in both water and m-xylene. The advancing and receding contact angles of m-xylene were measured as shown in Table 1. On the basis of eq 5 and the measured contact angles, the distance between the two channels, d, must be smaller than 0.09 mm (see Figure 2) in order to form an interface between water and m-xylene. However, such a small distance practically made the alignment of the two glass slides very difficult. Therefore, the value used in this experiment was d ) 0.3 mm. Because of the internal pressure of the fluid during injection, this larger d value did not prevent the formation of an interface. Figure 9 shows time-series photographs during the formation of the interface. After one channel was filled with water, m-xylene was injected into the other channel (Figure 9a). The slight space between these channels was spontaneously filled with m-xylene and a water-xylene interface was formed, as if a zipper was closed (Figures 9b-d). After the injection of these fluids was stopped, the interface could stably exist for more than 10 min. The continuous flow of the fluids was caused by a syringe pump using the following method. One syringe injected a fluid into the inlet of the channel, and another syringe simultaneously sucked it from the outlet (Figure 5). Figure 10 shows the water-xylene interface after the continuous flow at a volume flow rate of (panel a) 1 µL/min for 5 min or (panel b) 10 µL/ min for 10 min. The width of the water-filled channel was set at 0.13 mm, which was the same as the width of the xylene-
Figure 10. Water-xylene interface after continuous flow at a volume flow rate of (a) 1 µL/min for 5 min and (b) 10 µL/min for 10 min.
filled one, in order to make the average flow velocities in these two channels even under a common volume flow rate. As shown in Figure 10a, the interface was not disordered under mild conditions. However, because it was not stable under severer conditions, further investigations are necessary to obtain stabler interfaces by optimizing the surface properties of the glass slides. Thus, because the interface is stable only for several minutes, it cannot be used for extraction in microreactors that continuously produce chemicals on a gram scale. However, it may be useful for extraction in analytical applications, because the sample merely passes once through the channel for a short time (e.g.,
