Anal. Chem. 2004, 76, 4495-4500
Liquid Membrane Operations in a Microfluidic Device for Selective Separation of Metal Ions Tatsuo Maruyama, Hironari Matsushita, Jun-ichi Uchida, Fukiko Kubota, Noriho Kamiya, and Masahiro Goto*
Department of Applied Chemistry, Graduate School of Engineering, Kyushu University, 6-10-1 Hakozaki, Fukuoka 812-8581, Japan
A three-phase flow, water/n-heptane/water, was constructed in a microchannel (100-µm width, 25-µm depth) on a glass microchip (3 cm × 7 cm) and was used as a liquid membrane for separation of metal ions. Surface modification of the microchannel by octadecylsilane groups induced spontaneous phase separation of the three-phase flow in the microfluidic device, which allows control of interfacial contact time and off-chip analysis using conventional analytical apparatus. Prior to the selective transport of a metal ion through the liquid membrane in the microchannel, the forward and backward extraction of yttrium and zinc ions was investigated in a two-phase flow on a microfluidic device using 2-ethylhexyl phosphonic acid mono-2-ethylhexyl ester (commercial name, PC88A) as an extractant. The extraction conditions (contact time of the two phases, pH, extractant concentration) in the microfluidic device were examined. These investigations demonstrated that the conventional methodology for solvent extraction of metal ions is applicable to solvent extraction in a microchannel. Finally, we employed the three-phase flow in the microchannel as a liquid membrane and observed the selective transport of Y ion through the liquid membrane. In the present study, we succeeded, for the first time, in the selective separation of a targeted metal ion from an aqueous feed solution to a receiving phase within a few seconds by employing a liquid membrane formed in a microfluidic device. Liquid membranes, comprising an organic phase with two aqueous phases, have a wide variety of industrial and analytical applications, including separation, concentration, and removal of analytes from wastewater, environmental, and biomedical samples. The attractive features of the liquid membrane are high separation selectivity and simultaneous forward and backward extraction. In particular, the separation and preconcentration of metal ions by liquid membranes have been widely studied.1,2 This is because various oil-soluble extractants selective for targeted metal ions have been developed for solvent extraction. The selectivity in the liquid membrane is tuned for targeted analytes by choosing the proper compositions for the three phases. * Corresponding author: (e-mail)
[email protected]; (phone/ fax) +81-(0)92-642-3575. (1) de Gyves, J.; de San Miguel, E. R. Ind. Eng. Chem. Res. 1999, 38, 21822202. 10.1021/ac049844h CCC: $27.50 Published on Web 07/08/2004
© 2004 American Chemical Society
There are several forms of liquid membrane that can be useds bulk liquid membranes, supported liquid membranes, emulsion liquid membranes, and hollow-fiber liquid membranes. Surmeian et al. have recently reported a novel form of liquid membrane, a three-phase flow stably formed in a microchannel (192-µm width and 22-µm depth), and they showed, for the first time, that the three-phase flow serves as a liquid membrane for the permeation of an organic compound.3 This microchip technology is of major interest to analytical chemists. There are many advantages to the development of solvent extraction and liquid membrane techniques on miniaturized devices.3,4 In the past decade, many researchers have described chemical and biochemical reactions on microfluidic devices.5,6 A microchannel in a microfluidic device provides large surface and interface area-to-volume ratios, which are beneficial to chemical and biochemical processes involving interface phenomena, such as solvent extraction. Indeed, several groups reported effective solvent extraction of a metal ion in a two-phase flow on a glass microchip.7-12 From an analytical point of view, miniaturization of the liquid membrane technique reduces the amount of sample required, compared with that of conventional bulk reactors. However, the reduction in the amount of sample prevents on-chip detection using a common UV-visible spectrophotometer. Many research papers on microfluidic devices have employed laser-induced fluorescence for monitoring molecular behavior or chemical reactions. Fluorescent measurement, unfortunately, limits the application of the microfluidic device to fluorescent chemistry.14,15 (2) Guyon, F.; Parthasarathy, N.; Buffle, J. Anal. Chem. 2000, 72, 1328-1333. (3) Surmeian, M.; Slyadnev, M. N.; Hisamoto, H.; Hibara, A.; Uchiyama, K.; Kitamori, T. Anal. Chem. 2002, 74, 2014-2020. (4) Thordarson, E.; Palmarsdottir, S.; Mathiasson, L.; Jonsson, J. A. Anal. Chem. 1996, 68, 2559-2563. (5) Ismagilov, R. F. Angew. Chem., Int. Ed. 2003, 42, 4130-4132. (6) Sato, K.; Hibara, A.; Tokeshi, M.; Hisamoto, H.; Kitamori, T. Adv. Drug Delivery Rev. 2003, 55, 379-391. (7) Tokeshi, M.; Minagawa, T.; Kitamori, T. J. Chromatogr., A 2000, 894, 1923. (8) Hisamoto, H.; Horiuchi, T.; Tokeshi, M.; Hibara, A.; Kitamori, T. Anal. Chem. 2001, 73, 1382-1386. (9) Kim, H. B.; Ueno, K.; Chiba, M.; Kogi, O.; Kitamura, N. Anal. Sci. 2000, 16, 871-876. (10) Kerby, M. B.; Spaid, M.; Wu, S.; Parce, J. W.; Chien, R. L. Anal. Chem. 2002, 74, 5175-5183. (11) Kuban, P.; Berg, J.; Dasgupta, P. K. Anal. Chem. 2003, 75, 3549-3556. (12) Kubota, F.; Uchida, J.; Goto, M. Solv. Extr. Res. Dev. Jpn. 2003, 10, 93102. (13) Wenn, D. A.; Shaw, J. E. A.; Mackenzie, B. Lab Chip 2003, 3, 180-186.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004 4495
Thermal lens microscopy is also a powerful tool for monitoring reactions in microfluidic devices.16,17 This technique, however, cannot selectively detect a targeted analyte. Collection of reacted medium from a microfluidic device allows the use of conventional analyses such as spectroscopy, but the chemical reactions should not progress outside the microfluidic device during or after sample collection. Since interface-based reactions proceed only in a confluent microchannel,18 solvent extraction and a liquid membrane in a microfluidic device are suitable for “off-chip” analysis. In the present study, we achieved the separation and collection of each phase from a three-phase flow in a microfluidic device by modifying the partial surface of a microchannel, which allows offchip analysis by an ICP-atomic emission spectrometer. Then we investigated the forward and backward extraction of yttrium and zinc ions in a n-heptane/water two-phase flow on the microfluidic device. Based on these investigations, we carried out the simultaneous forward and backward extraction of Y3+ employing a liquid membrane (water/n-heptane/water) in a microfluidic device and succeeded in the selective transport of Y3+ through the liquid membrane. MATERIALS AND METHODS Materials. Extractant, PC-88A (2-ethylhexyl phosphonic acid mono-2-ethylhexyl ester), was generously supplied by Daihachi Chemical Industry Co. Ltd. and used without further purification. Trichloro(octadecyl)silane was purchased from Aldrich. All other chemicals were purchased from Wako Pure Chemicals (Osaka, Japan). The microchannel was fabricated by a photolithographic wet etching method18,19 on a Pyrex glass plate (38 mm × 70 mm). The glass plate was initially coated with a 20-nm Cr layer and then with a 100-nm Au layer. A positive photoresist was spin-coated on the Au layer. UV light was exposed through a photomask using a mask aligner to transfer the microchannel pattern onto the photoresist. The Cr and Au layers were etched with Ce(NH4)2(NO3)6 and I2/NH4I solutions, respectively. The bare glass surface was etched with an HF solution. The remaining Cr and Au layers were thoroughly removed in I2/NH4I and Ce(NH4)2(NO3)6 solutions. Inlet and outlet holes were mechanically drilled on another Pyrex glass plate (cover glass plate). The microfabricated glass plate was covered by the cover glass plate and thermally laminated at 650 °C. Figure 1 shows the layout and dimensions of the microchannels (for a three-phase flow and a two-phase flow) fabricated on glass plates. The confluent parts of the microchannels were typically 3 cm long, 150 µm wide, and 25 µm deep for a three-phase flow microchip and 3 cm long, 150 µm wide, and 25 µm deep for a two-phase flow microchip. The inlet and outlet channels were 50 µm wide and 25 µm deep for the three-phase flow microchip and 75 µm wide and 25 µm deep for the two-phase flow microchip. (14) Bruin, G. J. M. Electrophoresis 2000, 21, 3931-3951. (15) Verpoorte, E. Electrophoresis 2002, 23, 677-712. (16) Tokeshi, M.; Uchida, M.; Uchiyama, K.; Sawada, T.; Kitamori, T. J. Lumin. 1999, 83-4, 261-264. (17) Sato, K.; Kawanishi, H.; Tokeshi, M.; Kitamori, T.; Sawada, T. Anal. Sci. 1999, 15, 525-529. (18) Maruyama, T.; Uchida, J.; Ohkawa, T.; Futami, T.; Katayama, K.; Nishizawa, K.; Sotowa, K.; Kubota, F.; Kamiya, F.; Goto, M. Lab Chip 2003, 3, 308312. (19) Hibara, A.; Tokeshi, M.; Uchiyama, K.; Hisamoto, H.; Kitamori, T. Anal. Sci. 2001, 17, 89-94.
4496 Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 1. Microfluidic devices used in the present study. (a) Cross section of the confluent part of the three-phase flow microchannel. (b) Top view of the three-phase flow microchannel. (c) Experimental setup. (d) Cross section of the confluent part of the two-phase flow. (e) Top view of the two-phase flow microchannel. Channel length was 3 cm.
The surface of the outlet channel (for an organic phase) was chemically modified with octadecylsilane groups.18 Toluene solution containing 10 vol % octadecyltrichlorosilane from the central outlet and pure toluene from the side outlets were fed in parallel. The flow rates of both solutions were kept at 0.5 mL/h for 2 h. The flow in the microchannel was laminar, so that the octadecyltrichlorosilane solution tended to flow in the center of the confluent microchannel, resulting in chemical modification of the center of the confluent microchannel. After the chemical modification, the microchannels were washed with pure toluene. Thus, the central channel at the end junction was chemically modified to be hydrophobic. All the extraction experiments were carried out at 20 °C. For the forward extraction in a two-phase flow microchannel, an organic solution was prepared by dissolving PC-88A as an extractant in n-heptane. An aqueous solution containing Y3+ and Zn2+ (each 0.5 mM) was prepared by dissolving their nitrate salt hexahydrates in a nitric acid solution. The pH of the aqueous solution was adjusted with 0.1 M nitric acid and 0.1 M 4-aminobutyric acid solutions. For the backward extraction, the organic solution containing PC-88A (20 mM) and metal complexes (Y3+ and Zn2+, 0.5 mM each) was prepared by extraction of the metals in a screw cap bottle using an aqueous buffer (pH 3.5) containing the metal ions (0.5 mM each) and a n-heptane solution containing 20 mM PC-88A with 24-h incubation. The aqueous and organic solutions were fed separately into the microfluidic device using syringe pumps (IC3210, KD Scientific). The aqueous phase was collected from the outlet and was subjected to analysis by an ICP-atomic emission spectrometer (Perkin-Elmer Co., Optima 3100 RL) to determine the concentrations of Y and Zn ions.
Figure 2. Photographs of the three-phase flow in the microchannel. (a) Junction near the inlets of the microchannel without surface modification. (b) Center of the confluent microchannel without surface modification. (c) Junction near the outlets of the microchannel without surface modification. (d) Junction near the inlets of the microchannel modified with octadecylsilane groups. (e) Center of the confluent microchannel modified with octadecylsilane groups. (f) Junction near the outlets of the microchannel modified with octadecylsilane groups. Flow rates of the aqueous and organic phases in (a-c) were 0.4 and 0.7 mL/h, respectively Those in (d-f) were 0.4 mL/h.
Microfluidic devices with 3-, 8-, and 12-cm-long confluent microchannels were used for liquid membrane experiments. The liquid membrane consisted of a feed aqueous phase (pH 2.0, Y3+ and Zn2+ 0.5 mM each), an organic phase (n-heptane containing 10 mM PC-88A), and a receiving aqueous phase (1.0 M nitric acid). All the flow rates of the aqueous and organic phases were 0.3 mL/h. RESULTS AND DISCUSSION Formation of a Three-Phase Flow and Phase Separation in the Microchannel. Figure 2 shows the three-phase (water/ n-heptane/water) flow in a 3-cm-long microchannel. The aqueous phase was dyed using orange G. The three-phase flow and the clear liquid-liquid interface formed over the entire microchannel. In the microchannel without any surface modification, the aqueous flow invaded the central channel for the organic phase at the end junction of the microchannel (Figure 2c), even though the flow rate of the organic phase was much higher than that of the aqueous phase. The naked glass surface was hydrophilic and likely to be wet by water, which explains the invasion of the aqueous phase. On the other hand, Figure 2f exhibits the three-phase flow formed in the microchannel partially modified with trichloro(octadecyl)silane. Neither the organic nor the aqueous flow invaded the other channel at the end junction of the microchannel,
and each phase was collected from each channel without contamination by another phase. Hibara et al. also reported that the octadecylsilane modification of a microchannel allowed stable phase confluence and separation in a microchannel.20 Recently, we also succeeded in phase separation of a two-phase (water/ isooctane) flow in the same manner.18 These results show that the surface hydrophobicity of the microchannel plays a critical role in the phase separation in a microfluidic device. It should be noted that the width of the n-heptane phase was much smaller than those of the aqueous phases due to the high viscosity of water. Prior to the study of the three-phase flow as a liquid membrane for metal ion separation, the forward and backward extraction of metal ions (Y3+ and Zn2+) using PC-88A as an extractant was investigated in a two-phase (water/n-heptane) flow in a microchannel (Figure 1d and e). One of the attractive features of a microfluidic device is the ability to precisely control the contact time between an organic phase and an aqueous phase. In general, molecular diffusion governs the mass transfer in a microchannel because of the laminar flow (the Reynolds number in our system was less than (20) Hibara, A.; Nonaka, M.; Hisamoto, H.; Uchiyama, K.; Kikutani, Y.; Tokeshi, M.; Kitamori, T. Anal. Chem. 2002, 74, 1724-1728.
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4497
Figure 3. Effect of contact time on the forward extraction of metal ions in a two-phase microfluidic device. Filled circles represent yttrium ion, and open squares represent zinc ion. The solid line shows the calculated value based on our numerical model of mass transfer from ref 18. The flow velocity was varied from 0.15 to 0.8 mL/h. The pH of the aqueous phase was 3.0. The organic phase was n-heptane containing 20 mM PC-88A.
10). This means that the interfacial contact time plays a significant role in solvent extraction in the microchannel. Figure 3 depicts the effect of the contact time between the n-heptane and aqueous phases on the forward extraction in a two-phase flow. The extraction ratios of Y3+ and Zn2+ increased with increasing contact time. At a contact time of 1 s, we obtained extraction ratios of approximately 40 and 35% for Y3+ and Zn2+, respectively. In our previous report, we demonstrated that the extraction of an organic compound from an organic phase into an aqueous phase in a two-phase flow on a microfluidic device could be predicted using a simple numerical model based on the diffusion of the organic compound in the microchannel.18 Here, the numerical model was applied to the Y3+ extraction in the microchannel, assuming that the concentration of Y3+ in the aqueous phase very close to the liquid-liquid interface is zero due to the excess amount of the extractant (See our discussion in Figure 5), and the mass transfer was governed by the molecular diffusion in the laminar flow in the microchannel.
C(x,t) ) C0 [erf[x/(4Dt)0.5]]
(1)
where C(x,t) is the concentration of Y3+ at position x (a vertical distance from the liquid-liquid interface in the confluent microchannel) with contact time t and C0 is the initial concentration of Y3+ (0.5 mM). The diffusion constant (D) of Y3+ is assumed to be the same as that of Cu2+ in an aqueous buffer (7.2 × 10-10 m2/ s).21 The extraction ratio R(t) with a contact time t in the microchannel can be written as follows:
(
R(t) ) 1 -
1 C0d
∫
d
0
)
C(x,t) dx × 100
(2)
where d is the width of the aqueous phase (75 µm). The solid line in Figure 3 represents the Y3+ extraction ratio calculated from eq 2. The calculated values agree well with the experimental data (R2 ) 0.943). This shows that the mass transfer of metal ions in the microchannel can be explained by the molecular diffusion and that the extraction of the metal ions in the microchannel is readily predicted by our numerical model. It (21) Sato, Y.; Akiyoshi, Y.; Kondo, K.; Nakashio, F. J. Chem. Eng. Jpn. 1989, 22, 182-189.
4498
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
Figure 4. Effect of pH on the forward extraction. Filled circles represent Y3+, and open squares represent Zn2+. The aqueous phase contained nitric acid and 4-aminobutyric acid to adjust the pH. The concentrations of metal ions were 0.5 mM each. The organic phase was n-heptane containing 20 mM PC-88A. The flow velocities of the aqueous and organic phases were 0.3 mL/h. At pH 3.0, the extraction ratio of the Y ion was 37%.
should be noted that the mass transfer of Y3+ is a rate-limiting step for the extraction under the present conditions.22 PC-88A extracts Y3+ into an organic phase preferentially to 2+ Zn .23 Y3+ and Zn2+ are known to be extracted as complexes, formed with the dimeric extractant PC-88A, (HR)2, as follows:23
Y3+aq + 3(HR)2, org a YR33(HR)org + 3H+
(3)
Zn2+aq + 2(HR)2, org a ZnR2(HR)2 org + 2H+
(4)
Sato et al. and our group reported the equilibrium constants for extracting Y3+ and Zn2+ using PC-88A to be 8.5 (-) and 7.1 × 10-2 mM.23,24 The equilibrium constants provide pH0.5 values of 1.7 and 3.1 for Y3+ and Zn2+, respectively (pH0.5 is the pH at which 50% extraction is achieved), which means that PC-88A has a stronger interaction with Y3+ than with Zn2+. Since the extraction of metal ions by PC-88A proceeds via a proton-exchange mechanism (eqs 3 and 4), adjusting the pH of an aqueous phase allows selective extraction of metal ions by PC-88A. We then conducted the extraction of Y3+ and Zn2+ in the microchannel with the aqueous phase adjusted to various pH values (Figure 4). As with their extraction in a separatory funnel,12 extraction in the microchannel depends on the pH of the aqueous phase. From pH 1.3 to pH 2.0, hardly any Zn2+ was extracted into the n-heptane phase, and the extraction ratio increased when the pH was higher than 2.0. In contrast, Y3+ was extracted at all the pH values tested, although its extraction ratio varied. These results mean that the extraction behavior of metal ions in a microchannel is the same as that in a bulk reactor and that the separation methodology used in a bulk reactor is applicable to selective extraction in a microchannel. Extractant concentration is also a dominant factor affecting extraction ratio. Figure 5 shows the effect of PC-88A concentration on the forward extraction of metal ions using an aqueous phase at pH 2.0. The extraction ratios of Y3+ and Zn2+ increased with increasing PC-88A concentration and reached a ceiling of 40% at (22) Kubota, F.; Goto, M.; Nakashio, F. Solvent Extr. Ion Exch. 1993, 11, 437453. (23) Sato, Y.; Kondo, K.; Nakashio, F. J. Chem. Eng. Jpn. 1989, 22, 686-689. (24) Kubota, F.; Shinohara, K.; Shimojo, K.; Oshima, T.; Goto, M.; Furusaki, S.; Hano, T. Sep. Purif. Technol. 2001, 24, 93-100.
Figure 5. Effect of PC-88A concentration on the extraction. Filled circles represent Y3+, and open squares represent Zn2+. The aqueous phase (pH 2.0) contained nitric acid (0.1 M), 4-aminobutyric acid (0.1 M), and metal ions (0.5 mM each). The flow velocities of the aqueous and organic phases were 0.3 mL/h. At the PC-88A concentration of 20 mM, the extraction ratios of Y3+ and Zn2+ were 36 and 10%.
∼20 mM PC-88A, which means that a PC-88A concentration of 20 mM was enough to extract Y3*. In a separatory funnel, Y2+ can be thoroughly extracted into an organic phase containing 8 mM PC-88A in a few hours. The mass transfer in the microchannel was governed by molecular diffusion, and the contact time between the aqueous and organic phases was limited to 0.68 s in the microchannel. The extraction of the metal ions by PC-88A requires six or four extractant molecules per one metal ion. Therefore, at a low extractant concentration, the diffusion of the extractant and the metal-extractant complex adjacent to the interface is essential for the extraction in the microchannel, resulting in the rather low extraction ratio at 8 mM PC-88A. Separation of the metal ions by a liquid membrane consists of forward extraction and backward extraction of targeted ions. Several papers describe the forward solvent extraction of a metal ion in a microfluidic device,7-12 but to the best of our knowledge, the backward extraction of a metal ion has only been reported by Tokeshi et al.25 In the present work, we studied the backward extraction of metal ions from an organic phase into an aqueous phase in a microfluidic device. The effect of the aqueous phase on the backward extraction was investigated (Figure 6). All the acidic solutions successfully extracted Zn2+ into the aqueous phase from an organic phase in 0.68 s. The backward extraction ratios of Zn2+ were almost the same as the forward extraction ratio of Y3+ with a contact time of 0.68 s, as shown in Figure 3. This implies that the diffusion of Zn2+ is also a rate-limiting step for the backward extraction in the microchannel. On the other hand, the backward extraction ratio of Y3+ was influenced by the concentration of nitric acid and by the kind of acid, suggesting that the dissociation of the metal-extractant complex is a rate-limiting step for this backward extraction. The effect of the concentration is consistent with that in a separatory funnel. The kind of acid, however, does not affect the backward extraction in a separatory funnel. This could be explained by the difference between the backward extraction rates for nitric acid and hydrochloric acid. The microfluidic device can precisely control the contact time between the aqueous and organic phases, while solvent extraction using a separatory funnel cannot control the extraction time. The microfluidic device allowed us to observe, for the first time, the (25) Tokeshi, M.; Minagawa, T.; Uchiyama, K.; Hibara, A.; Sato, K.; Hisamoto, H.; Kitamori, T. Anal. Chem. 2002, 74, 1565-1571.
Figure 6. Effect of acids on the backward extraction. Flow rates of the aqueous and organic phases were 0.3 and 0.5 mL/h, respectively. Filled bars represent the results of Y3+ and open bars those for Zn2+. The organic phase was n-heptane containing 20 mM PC-88A and 0.5 mM Y3+ and Zn2+. Nitric acid (1.0 N) resulted in the backward extraction ratios of 21 and 36% for Y3+ and Zn2+, respectively.
Figure 7. Transport of metal ions through a liquid membrane in the microfluidic device. Filled triangles represent Y3+ in the feed phase, and filled circles represent Y3+ in the receiving phase. Open triangles represent Zn2+ in the feed phase, and open circles represent Zn2+ in the receiving phase. The conditions were an aqueous buffer (pH 2.0) containing Y3+ and Zn2+ (0.5 mM each) for the aqueous feed phase, n-heptane containing 10 mM PC-88A for the organic phase, 1.0 mol/L nitric acid for the receiving phase. The flow rates of the three phases were all 0.3 mL/h. For the contact time of 2.44 s, the flow rates were all 0.1 mL/h. Microfluidic devices with 3-, 8-, and 12-cm-long confluent microchannel were used.
difference between the backward extraction rates for nitric acid and hydrochloric acid. On the basis of the above investigations, we carried out the selective transport of Y3+ through a liquid membrane in the microfluidic device in the presence of Zn2+. The liquid membrane consisted of a feed aqueous phase (pH 2.0, Y3+ and Zn2+ 0.5 mM each), an organic phase (n-heptane containing 10 mM PC-88A), and a receiving aqueous phase (1.0 M nitric acid). The concentrations of the Y- and Zn-extractant complexes in the organic phase can be calculated from the mass balance. The Y3+ concentration in the feed phase decreased with the contact time (Figure 7). That in the receiving phase increased with the contact time. At 2.4 s of the contact time, half of the initial Y ion was transported to the receiving phase. In the absence of the extractant in the organic phase, there were no changes in the Y3+ concentrations in either the feed or the receiving phase (data not shown). These results revealed that Y3+ was extracted into the organic phase and simultaneously back-extracted into the receiving phase in the microchannel. Namely, Y3+ permeated through the liquid membrane. In contrast, the concentration of Zn2+ in the feed phase Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
4499
decreased slightly, coincidently with a slight increase in the receiving phase. Since the pH in the feed phase was set at 2.0, hardly any Zn2+ was extracted into the organic phase, resulting in little permeation of Zn2+ through the liquid membrane. In summary, Y3+ was selectively transported from the feed phase that also contained another metal ion to the receiving phase, within several seconds, through a liquid membrane, which consisted of the three-phase flow in a microfluidic device. CONCLUSIONS We constructed a three-phase (water/heptane/water) flow in a glass microfluidic device. Surface modification of the microchannel with octadecylsilane groups facilitated the spontaneous separation of the three phases in the microchannel and allowed off-chip analysis of the reaction carried out in the microfluidic device using conventional apparatus. In a two-phase flow microchannel, we found that an extraction ratio of ∼40% was obtained within a few seconds in both the forward and backward extraction of metal ions (Y3+ and Zn2+) using PC-88A as the extractant in
4500
Analytical Chemistry, Vol. 76, No. 15, August 1, 2004
the microchannel. In addition, a three-phase flow in the microchannel was utilized as a liquid membrane for the separation of Y3+. Selective separation of a targeted metal ion from an aqueous feed solution was achieved within several seconds. The present study demonstrates that the conventional methodology on solvent extraction of metal ions is applicable to a liquid membrane in a microchannel. ACKNOWLEDGMENT This research was supported by a Grant-in-Aid for Science Research (15560656) from the Ministry of Education, Science Sports & Culture of Japan. We thank Tosoh Co. for the microfabrication of the glass plate and Daihachi Chemical Industry Co. Ltd. for providing PC-88A.
Received for review January 27, 2004. Accepted May 27, 2004. AC049844H
