Temperature-Controlled Flow Switching in Nanocapillary Array

23 Aug 2006 - Franck Montagne , Jérome Polesel-Maris , Raphael Pugin and Harry Heinzelmann. Langmuir ..... Roger H. Terrill , Xuejun Wang , Paul W. B...
0 downloads 0 Views 285KB Size
Download PDF
Langmuir 2007, 23, 305-311

305

Temperature-Controlled Flow Switching in Nanocapillary Array Membranes Mediated by Poly(N-isopropylacrylamide) Polymer Brushes Grafted by Atom Transfer Radical Polymerization† Ishika Lokuge, Xuejun Wang, and Paul W. Bohn* Department of Chemistry, Beckman Institute for AdVanced Science and Technology and Frederick Seitz Materials Research Laboratory, UniVersity of Illinois at Urbana-Champaign, 600 S. Mathews AVenue, Urbana, Illinois 61801 ReceiVed March 27, 2006. In Final Form: June 6, 2006 We report actively controlled transport that is thermally switchable and size-selective in a nanocapillary array membrane (NCAM) prepared by grafting poly(N-isopropylacrylamide) (PNIPAAm) brushes onto the exterior surface of a Au-coated polycarbonate track-etched membrane. A smooth Au layer on the membrane surface, which is key to obtaining a uniform polymer film, was prepared by thermal evaporation of ∼50 nm Au on both exterior surfaces. After evaporation, the inner diameter of the pore is reduced slightly, but the NCAM retains a narrow pore size distribution. PNIPPAm brushes with 10-30 nm (dry film) thickness were grafted onto the Au surface through surfaceinitiated atom transfer radical polymerization (ATRP) using a disulfide initiator, (BrC(CH3)2COO(CH2)11S)2. Molecular transport through the PNIPAAm polymer brush-modified NCAMs was investigated by real-time fluorescence measurements using fluorescein isothiocyanate (FITC)-labeled dextrans ranging from 4.4 to 282 kDa in membranes with variable initial pore diameters (80, 100, and 200 nm) and different PNIPAAm thicknesses. Manipulating the temperature of the NCAM through the PNIPAAm lower critical solution temperature (LCST) causes large, sizedependent changes in the transport rates. Over specific ranges of probe size, transport is completely blocked below the LCST but strongly allowed above the LCST. The combination of the highly uniform PNIPAAm brush and the monodisperse pore size distribution is critical in producing highly reproducible switching behavior. Furthermore, the reversible nature of the switching raises the possibility of using them as actively controlled filtration devices.

Introduction There has been a great deal of recent interest in controlling the transport of molecular and macromolecular species through nanochannels, and a wide range of materials having channels of molecular-scale and nanoscale dimensions have been studied, such as track-etched polycarbonate membranes,1-6 porous alumina membranes,7 carbon nanotube arrays,8 Nafion,9 zeolite,10,11 fullerene tubules,12 porin proteins,13 and membranes formed from templating media, for example, latex spheres,14,15 and block copolymers.16 These nanopore-mediated transport structures have potential applications in drug delivery,17-19 † Part of the Stimuli-Responsive Materials: Polymers, Colloids, and Multicomponent Systems special issue. * Corresponding author. E-mail: [email protected].

(1) Kuo, T. C.; Sloan, L. A.; Sweedler, J. V.; Bohn, P. W. Langmuir 2001, 17, 6298-6303. (2) Martin, C. R. Science 1994, 266, 1961-1966. (3) Martin, C. R. Acc. Chem. Res. 1995, 28, 61-68. (4) Martin, C. R. Chem. Mater. 1996, 8, 1739-1746. (5) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655-658. (6) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Anal. Chem. 1999, 71, 49134918. (7) Dalvie, S. K.; Baltus, R. E. J. Membr. Sci. 1992, 71, 247-255. (8) Miller, S. A.; Young, V. Y.; Martin, C. R. J. Am. Chem. Soc. 2001, 123, 12335-12342. (9) Koval, C. A.; Spontarelli, T. J. Am. Chem. Soc. 1988, 110, 293-295. (10) Davis, M. E. Microporous Mesoporous Mater. 1998, 21, 173-182. (11) Beck, J. S.; Vartuli, J. C.; Roth, W. J.; Leonowicz, M. E.; Kresge, C. T.; Schmitt, K. D.; Chu, C. T. W.; Olson, D. H.; Sheppard, E. W.; McCullen, S. B.; Higgins, J. B.; Schlenker, J. L. J. Am. Chem. Soc. 1992, 114, 10834-10843. (12) Dujardin, E.; Ebbesen, T. W.; Hiura, H.; Tanigaki, K. Science 1994, 265, 1850-1852. (13) Hofnung, M. Science 1995, 267, 473-474. (14) Holland, B. T.; Blanford, C. F.; Stein, A. Science 1998, 281, 538-540. (15) Antonietti, M.; Berton, B.; Goltner, C.; Hentze, H. P. AdV. Mater. 1998, 10, 154-157. (16) Templin, M.; Franck, A.; DuChesne, A.; Leist, H.; Zhang, Y. M.; Ulrich, R.; Schadler, V.; Wiesner, U. Science 1997, 278, 1795-1798.

catalytic reactions,10 chromatographic sampling,20 selective transport of analyte species,6,21-23 and separation science. Of particular note is the explosive growth of research in membrane separation technology due to the special advantages of high selectivity and resolution and the ability to separate lowconcentration analytes with high efficiency and handle biological macromolecules without denaturation. Furthermore, membrane separations can be implemented more economically than an energy-intensive method such as distillation. Of special interest in this laboratory is the use of nanocapillary array membranes (NCAMs) composed of discrete parallel cylindrical nanopores in the size range of 10 nm e d e 200 nm with in-plane pore densities in the range of 108 cm-2 < Np < 109 cm-2 as interconnects to establish controllable fluidic communication between micrometer-scale channels operating in vertically separated planes, thereby making possible three-dimensional integration of fluidic microstructures. Previously, NCAMs have been used to accomplish electrically switchable transport among microchannels,24-26 but it would be of significant interest to (17) Nolan, L. M. A.; Corish, J.; Corrigan, O. I. J. Chem. Soc., Faraday Trans. 1993, 89, 2839-2845. (18) Schwendeman, S. P.; Amidon, G. L.; Levy, R. J. Macromolecules 1993, 26, 2264-2272. (19) Keister, J. C.; Kasting, G. B. J. Membr. Sci. 1992, 71, 257-271. (20) Schwendeman, S. P.; Amidon, G. L.; Meyerhoff, M. E.; Levy, R. J. Macromolecules 1992, 25, 2531-2540. (21) Hou, Z. Z.; Abbott, N. L.; Stroeve, P. Langmuir 2000, 16, 2401-2404. (22) Nishizawa, M.; Menon, V. P.; Martin, C. R. Science 1995, 268, 700-702. (23) Hulteen, J. C.; Jirage, K. B.; Martin, C. R. J. Am. Chem. Soc. 1998, 120, 6603-6604. (24) Cannon, D. M.; Kuo, T. C.; Bohn, P. W.; Sweedler, J. V. Anal. Chem. 2003, 75, 2224-2230. (25) Kuo, T. C.; Cannon, D. M.; Chen, Y. N.; Tulock, J. J.; Shannon, M. A.; Sweedler, J. V.; Bohn, P. W. Anal. Chem. 2003, 75, 1861-1867. (26) Kuo, T. C.; Cannon, D. M.; Shannon, M. A.; Bohn, P. W.; Sweedler, J. V. Sens. Actuators, A 2003, 102, 223-233.

10.1021/la060813m CCC: $37.00 © 2007 American Chemical Society Published on Web 08/23/2006

306 Langmuir, Vol. 23, No. 1, 2007

develop NCAM switching schemes that allow the transport of a specific fraction of the molecular population. In this context, it is interesting that both physical and chemical methods have been exploited to separate molecules using membranes based on their charge,22,23 chemical interaction,6,23,27-29 and size.5,30-33 In this work, we describe an NCAM capable of changing its inner pore diameter with temperature, thereby effecting differential transport based on the size of the molecule. Of course, membranes have been used extensively in size-selective separations, for example, in dialysis and in ultrafiltration, but in these applications, the membrane is designed for a specific molecular size threshold.5,34 Grafting membrane surfaces with polymers with tunable properties is an effective route to obtaining environmentally sensitive composite membranes. Physical changes in the grafted film can be triggered by pH,35-37 ionic strength,38-40 solvent quality,41,42 electric field,43-47 light,48 redox potential,49 and temperature,27,34,50-57 resulting in stimuli-responsive film-grafted membranes with high mechanical strength, quick response to external signals, the ability to transport both neutral and charged molecules, and strong resistance to degradation due to covalent bonding between substrate and polymer. Of particular interest in this work are temperature-induced volume changes in the grafted polymer in and around the nanopore opening to alter the effective pore diameter. The contraction and (27) Kobayashi, Y.; Martin, C. R. Anal. Chem. 1999, 71, 3665-3672. (28) Martin, C. R.; Kohli, P. Nat. ReV. Drug DiscoVery 2003, 2, 29-37. (29) Lee, S. B.; Martin, C. R. Chem. Mater. 2001, 13, 3236-3244. (30) Hope, M. J.; Bally, M. B.; Webb, G.; Cullis, P. R. Biochim. Biophys. Acta 1985, 812, 55-65. (31) Hernandez, A.; Martinezvilla, F.; Ibanez, J. A.; Arribas, J. I.; Tejerina, A. F. Sep. Sci. Technol. 1986, 21, 665-677. (32) Nystrom, M.; Lindstrom, M.; Matthiasson, E. Colloids Surf. 1989, 36, 297-312. (33) Wakeman, R. J.; Tarleton, E. S. Chem. Eng. Res. Des. 1991, 69, 386-397. (34) Liang, L.; Feng, X. D.; Peurrung, L.; Viswanathan, V. J. Membr. Sci. 1999, 162, 235-246. (35) Beebe, D. J.; Moore, J. S.; Bauer, J. M.; Yu, Q.; Liu, R. H.; Devadoss, C.; Jo, B. H. Nature 2000, 404, 588-591. (36) Crook, C. J.; Smith, A.; Jones, R. A. L.; Ryan, A. J. Phys. Chem. Chem. Phys. 2002, 4, 1367-1369. (37) Velada, J. L.; Liu, Y.; Huglin, M. B. Macromol. Chem. Phys. 1998, 199, 1127-1134. (38) English, A. E.; Tanaka, T.; Edelman, E. R. J. Chem. Phys. 1996, 105, 10606-10613. (39) Ohmine, I.; Tanaka, T. J. Chem. Phys. 1982, 77, 5725-5729. (40) Puppo, M. C.; Anon, M. C. J. Agric. Food Chem. 1998, 46, 3583-3589. (41) Lee, J. W.; Kim, E. H.; Jhon, M. S. Bull. Korean Chem. Soc. 1983, 4, 162-169. (42) Buehler, K. L.; Anderson, J. L. Ind. Eng. Chem. Res. 2002, 41, 464-472. (43) Kishi, R.; Osada, Y. J. Chem. Soc., Faraday Trans. 1 1989, 85, 655-659. (44) Lokuge, I. S.; Bohn, P. W. Langmuir 2005, 21, 1979-1985. (45) Ivan, K.; Kirschner, N.; Wittmann, M.; Simon, P. L.; Jakab, V.; Noszticzius, Z.; Merkin, J. H.; Scott, S. K. Phys. Chem. Chem. Phys. 2002, 4, 1339-1347. (46) Osada, Y.; Gong, J. P. AdV. Mater. 1998, 10, 827-837. (47) Osada, Y.; Hasebe, M. Chem. Lett. 1985, 1285-1288. (48) Chung, D. J.; Ito, Y.; Imanishi, Y. J. Appl. Polym. Sci. 1994, 51, 20272033. (49) Ito, Y.; Nishi, S.; Park, Y. S.; Imanishi, Y. Macromolecules 1997, 30, 5856-5859. (50) Schilli, C. M.; Zhang, M. F.; Rizzardo, E.; Thang, S. H.; Chong, Y. K.; Edwards, K.; Karlsson, G.; Muller, A. H. E. Macromolecules 2004, 37, 78617866. (51) Rao, G. V. R.; Krug, M. E.; Balamurugan, S.; Xu, H. F.; Xu, Q.; Lopez, G. P. Chem. Mater. 2002, 14, 5075-5080. (52) Liu, S. Q.; Yang, Y. Y.; Liu, X. M.; Tong, Y. W. Biomacromolecules 2003, 4, 1784-1793. (53) Kanazawa, H.; Kashiwase, Y.; Yamamoto, K.; Matsushima, Y.; Kikuchi, A.; Sakurai, Y.; Okano, T. Anal. Chem. 1997, 69, 823-830. (54) Li, C. M.; Gunari, N.; Fischer, K.; Janshoff, A.; Schmidt, M. Angew. Chem., Int. Ed. 2004, 43, 1101-1104. (55) Okano, T.; Yamada, N.; Okuhara, M.; Sakai, H.; Sakurai, Y. Biomaterials 1995, 16, 297-303. (56) Okano, T.; Yamada, N.; Sakai, H.; Sakurai, Y. J. Biomed. Mater. Res. 1993, 27, 1243-1251. (57) Uchiyama, S.; Ohshima, A.; Nakamura, S.; Hasegawa, J.; Terui, N.; Takayama, S. I. J.; Yamamoto, Y.; Sambongi, Y.; Kobayashi, Y. J. Am. Chem. Soc. 2004, 126, 14684-14685.

Lokuge et al.

dilation of polymer films can be based on the reversible ionization of the functional groups, solvent intrusion and extrusion into the polymer network, and phase separation. Most investigations have centered on temperature-induced transitions because of their importance in physiological systems and both positive temperature-sensitive polymers, which shrink when cooled below their upper critical solution temperature, and negative temperaturesensitive polymers, which contract when heated above their lower critical solution temperature (LCST), have been identified. Temperature-sensitive gels that exhibit a LCST phase transition can be prepared from N-substituted acrylamide derivatives (e.g. N-isopropylacrylamide, N,N′-diethylacrylamide, Nacryloylpyrrolidine, N-vinylisobuytramide, and N-acryloylpiperidine58,59). The thermodynamic properties of poly(N-isopropylacrylamide) (PNIPAAm), which undergoes a dramatic volume transition in water at its LCST near 32 °C, were first reported by Heskins et al. using a Flory-Huggins analysis.60,61 Later, discontinuous phase-transition properties were reported by Tanaka et al.,62 and Lopez and co-workers described how the degree of polymerization and surface coverage of PNIPAAm brushes tethered onto planar surfaces affect temperature-manipulated volume transitions.63-65 This distinctive behavior of PNIPAAm has been attributed to the rapid alteration of its hydrophilicity.66,67 At temperatures lower than the LCST, PNIPAAm expands because of hydrogen bond formation between hydrophilic segments in the side chains of the polymer and water. The hydrogen bonds form a stable hydration shell around the hydrophobic groups, which leads to large water uptake at low temperatures. Above the LCST, the polymer-solvent interactions are disrupted, hydrogen bonding is weakened, and polymer-polymer hydrophobic interactions dominate, causing an abrupt collapse in the polymer free volume due to the release of entrapped water. In exploiting these dramatic volume changes, for example, to construct size-selective NCAMs, the response time is critical. In contrast to conventional network structures, which have relatively rigid chain ends, these terminally grafted polymer brushes exhibit rapid conformational changes,67-70 the speed being attributed to the mobility of free chain ends.67,71-73 In this study, the temperature sensitivity of PNIPAAm is exploited to achieve actively controlled thermoresponsive, size-selective transport switching (Scheme 1) by grafting PNIPAAm brushes onto a Au-coated NCAM using atom transfer radical polymerization (ATRP). (58) Beltran, S.; Baker, J. P.; Hooper, H. H.; Blanch, H. W.; Prausnitz, J. M. Macromolecules 1991, 24, 549-551. (59) Pelton, R. AdV. Colloid Interface Sci. 2000, 85, 1-33. (60) Heskins, M.; Guillet, J. E. J. Macromol. Sci. Chem. 1968, A2, 1441. (61) Lee, W. F.; Huang, Y. L. J. Appl. Polym. Sci. 2000, 77, 1769-1781. (62) Hirokawa, Y.; Tanaka, T. J. Chem. Phys. 1984, 81, 6379-6380. (63) Yim, H.; Kent, M. S.; Satija, S.; Mendez, S.; Balamurugan, S. S.; Balamurugan, S.; Lopez, G. P. Phys. ReV. E 2005, 72. (64) Mendez, S.; Curro, J. G.; McCoy, J. D.; Lopez, G. P. Macromolecules 2005, 38, 174-181. (65) Balamurugan, S.; Mendez, S.; Balamurugan, S. S.; O’Brien, M. J.; Lopez, G. P. Langmuir 2003, 19, 2545-2549. (66) Kim, S. J.; Lee, C. K.; Lee, Y. M.; Kim, S. I. J. Appl. Polym. Sci. 2003, 90, 3032-3036. (67) Yoshida, R.; Uchida, K.; Kaneko, Y.; Sakai, K.; Kikuchi, A.; Sakurai, Y.; Okano, T. Nature 1995, 374, 240-242. (68) Kaneko, Y.; Nakamura, S.; Sakai, K.; Kikuchi, A.; Aoyagi, T.; Sakurai, Y.; Okano, T. Polym. Gels Networks 1998, 6, 333-345. (69) Kaneko, Y.; Sakai, K.; Kikuchi, A.; Yoshida, R.; Sakurai, Y.; Okano, T. Macromolecules 1995, 28, 7717-7723. (70) Annaka, M.; Sugiyama, M.; Kasai, M.; Nakahira, T.; Matsuura, T.; Seki, H.; Aoyagi, T.; Okano, T. Langmuir 2002, 18, 7377-7383. (71) Chu, L. Y.; Niitsuma, T.; Yamaguchi, T.; Nakao, S. AIChE J. 2003, 49, 896-909. (72) Yamaguchi, T.; Ito, T.; Sato, T.; Shinbo, T.; Nakao, S. J. Am. Chem. Soc. 1999, 121, 4078-4079. (73) Peng, T.; Cheng, Y. L. J. Appl. Polym. Sci. 1998, 70, 2133-2142.

Flow Switching in Nanocapillary Array Membranes Scheme 1

Experimental Section Materials. Deionized (DI) water (18.2 MΩ cm) from a MilliQ UV-Plus system (Millipore) was used to prepare all solutions unless otherwise specified. Polycarbonate track-etched membranes having pore diameters in the range of 30 nm e dp e 200 nm were purchased from Fisher and used as NCAMs. The membranes were typically 6-10 µm thick with pore densities in the range of 3 × 108 cm-2 e Np e 6 × 108 cm-2, depending on the pore diameter. Au stock was cut into small pieces and thermally evaporated onto polycarbonate NCAMs. FITC dextran (4.4, 10, 19.4, 40, 77, and 282 kDa molecular weight), 11-mercapto-1-undecanol, 2-bromo-2-methylpropionyl bromide (98%), copper(I) bromide (99.999%) and 1,1,4,7,7-pentamethyldiethylenetriamine (PMDETA, 99%) were purchased from Aldrich and used as received. NIPAAm monomer (99%) was purchased from Acros and purified by passing through an inhibitor removal column using a mixture of dichloromethane and hexane (∼1:1 v/v) as the solvent and then recrystallized in hexane. Inhibitor removal columns were purchased from Aldrich and used directly. Substrate Preparation. An adhesion layer of Cr (2.5 nm) was deposited prior to all Au depositions. Au films ca. 45 nm thick were vapor deposited sequentially onto both sides of NCAMs with initial pore diameters of 80-200 nm. Another set of NCAMs was Au coated using an electroless Au deposition method.6,27,74 The plating time and pH of the Au solution were varied to determine the optimum conditions required to achieve the desired final pore diameter. For polymer characterization experiments, silicon wafers were cleaned in piranha (4:1 H2SO4/30% H2O2) cleaning solution for 30 min, followed by thorough rinsing with DI water. (CAUTION: Piranha is a Vigorous oxidant and should be used with extreme caution.) After cleaning, the silicon wafers were rinsed with isopropyl alcohol and dried under flowing N2 before Au evaporation. All Au-coated samples were stored under N2 before surface grafting. Formation of Initiator Self-Assembled Monolayers (SAMs). A disulfide initiator, (BrC(CH3)2COO(CH2)11S)2, was synthesized by adapting a reported procedure.75 Au-coated NCAMs or Au-coated silicon wafers were soaked in a 1 mM ethanolic solution of the initiator overnight to form uniform SAMs. After formation, the SAM was rinsed in ethanol, dried in flowing N2, and either used immediately or stored in dry N2 until ready for use. Surface-Initiated Polymerization of NIPAAm. NIPAAm was polymerized in a controlled atmosphere box. Methanol (MeOH) and H2O were degassed through four freeze-pump-thaw cycles before being introduced into the controlled atmosphere box. Typically, NIPAAm (3.15 g, 27.5 mmol), CuBr (40.0 mg, 0.278 mmol), and PMDETA (175 µL, 0.835 mmol) were dissolved in 60 mL of MeOH/ H2O (1:1 v/v) or MeOH/H2O (3:1 v/v), the resulting solution was transferred into a vial containing Au-coated NCAMs previously coated with uniform SAMs of the disulfide initiator, and the vial was sealed and kept at room temperature for 15 min. For comparison, Au-coated silicon wafers with uniform initiator SAMs were treated in the same manner. After polymerization, the substrates were removed from the vial and rinsed with copious amounts of DI water, followed by sonication in EtOH and then H2O. After drying with a stream of N2, the samples were stored under dry N2. Scanning Electron Microscopy. SEM images of the Au-deposited NCAMs were obtained using a high-resolution field emission SEM (Philips XL30) equipped with an Everhart-Thornley secondary electron detector. Samples were first sputter coated with Au/Pd for 65 s using a Denton Desk II TCS turbo-pumped sputter coater, and images were obtained with an accelerating voltage of 5 or 10 kV. (74) Menon, V. P.; Martin, C. R. Anal. Chem. 1995, 67, 1920-1928. (75) Shah, R. R.; Merreceyes, D.; Husemann, M.; Rees, I.; Abbott, N. L.; Hawker, C. J.; Hedrick, J. L. Macromolecules 2000, 33, 597-605.

Langmuir, Vol. 23, No. 1, 2007 307 FTIR External Reflection Spectroscopy. FTIR external reflection spectra were acquired with p-polarized radiation using a Digilab FTS-60A spectrometer (Bio-Rad, Cambridge, MA) equipped with a Harrick Scientific “Seagull” reflection accessory set to a 70° angle of incidence and a liquid-N2-cooled HgCdTe detector. Typically, spectra were acquired by co-adding 512 scans at a resolution of 4 cm-1. All spectra were baseline corrected for accurate comparison. Ellipsometric Thickness Measurement. Ellipsometric measurements were made at an incident angle of 70° with a Gaertner model L116C ellipsometer running LGEMP software (Gaertner Scientific). The optical constants (refractive index n, absorption coefficient k) for bare Au-coated silicon were measured immediately after evaporation. Film thicknesses in air were then measured with the previously determined values (n, k) for the Au substrate and an assumed refractive index of 1.50 for the adsorbate. Transport Measurements. Transport experiments were carried out in a standard membrane transport bicell described previously.1 The NCAM, held in place via an O-ring, served to separate two identical compartments (2.5 mL capacity) in a custom-built cell made of Delrin. Magnetic stir bars were placed in both the source (initially containing the fluorescent probe) and receiving (initially containing buffer solution only) compartments and were rotated at a constant speed to ensure thorough mixing. Fluorescence from the receiving side was monitored with a fluorimeter (Spex) operating in 90° excitation/detection orientation, and data were collected by a computer running custom LabVIEW (National Instruments) acquisition software. Fluorescence measurements were obtained at excitation and emission wavelengths of 491 and 515 nm, respectively. For each permeation experiment with PNIPAAm-grafted membranes, control experiments were carried out using ungrafted Au membranes and bare NCAMs. The temperature of the polymer brushes was manipulated by changing the temperature of the probe solution using a stainless steel sheath cartridge heater (Omega Engineering Inc.) inserted into the source side of the cell. An Agilent HP dc power supply (50 V/0.2 A) was used to control the temperature. The temperatures of the source and receiving sides were recorded using thermocouples inserted into each compartment. When increasing the temperature, 40 V was applied initially. Once the temperature of the feed solution reached the target temperature of 40 °C, a constant temperature was maintained by reducing the voltage to 20 V. To cool the modified NCAM rapidly, probe solution was replaced on the source side with a fresh aliquot of feed solution, reducing the temperature of the polymer brushes to room temperature. Permeation experiments were conducted with a temporal program of 0-7 min, 25 °C; 7-16 min, 40 °C; 16-17 min, replace the feed solution; and 17-23 min, 25 °C.

Results and Discussion Substrate Characterization. The approach outlined here depends on having a uniform thickness polymer brush layer in conjunction with a monodisperse distribution of high aspect ratio cylindrical nanopores. With a smooth Au substrate being key to obtaining a highly uniform polymer film thickness, SEM and transmission electron microscopy (TEM) were used to evaluate the efficacy of electroless deposition versus thermal evaporation to produce Au films capable of supporting ATRP synthesis. Electroless deposition is attractive because it makes possible the formation of Au films on the interior surfaces of the nanopores as well as the exposed exterior surfaces. However, TEM images showed that Au films produced by electroless deposition are heterogeneous, even after extended (∼6 h) plating time. Typically, some nanopores are completely filled with Au whereas others exhibit no Au at all on the interior nanopore surface. Electroless deposition is also very sensitive to the pH of the Au plating solution, with slight variations in pH producing major changes in deposition rates. Additionally, SEM images indicate that the top surface of an electrolessly deposited Au membrane has an

308 Langmuir, Vol. 23, No. 1, 2007

Figure 1. SEM images showing a comparison of Au films produced on polycarbonate NCAMs by (a) electroless plating (6 h, pH 12) and (b) thermal evaporation. (Below) Schematic cross-sectional diagram of an individual nanopore, emphasizing that electroless deposition produces a Au coating on the interior of the nanopore, whereas thermal evaporation is largely confined to the exterior surface.

Figure 2. SEM image of a Au-coated NCAM with the top Au layer partially removed. Image acquired near the boundary between coated (light, dp ≈ 80 nm) and uncoated (dark, dp ≈ 100 nm) regions.

uneven granular structure, which can block pores and forms a poor platform from which to graft a uniform PNIPAAm film. Smuleac and co-workers have obtained smooth Au surfaces on a polycarbonate track-etched membrane by implementing electroless Au plating under convective flow conditions.76 They demonstrated that the aggregated Au islands, which cover a substantial number of pores, can be reduced in this manner. However, this method is applicable only for membranes with large pore diameters. To overcome the problems associated with electroless plating, the polycarbonate NCAMs were coated with Au by thermal evaporation. SEM micrographs indicate that thermally evaporated Au on NCAMs is smooth with minimal defects, and the monodisperse character of the pore size distribution is retained (Figure 1). Characteristically thermal evaporation produces Au films only on the outer surface of the membrane with no significant Au coating inside the nanochannel. Also, even thin Au coatings, such as the 45 nm films used here, reduce the pore diameter by ∼10-20%. Figure 2 shows an SEM image in a region where the Au layer has been removed from the NCAM by mechanical polishing with ethanol. Comparison of the pores in the Au-coated region (lighter area) to the original pores in the adjacent polycarbonate region (darker area) shows a significant reduction in the pore diameter. (76) Smuleac, V.; Butterfield, D. A.; Bhattacharyya, D. Chem. Mater. 2004, 16, 2762-2771.

Lokuge et al.

ATRP is an ideal method to surface graft polymer brushes to achieve control of molecular transport because it is reproducible, does not lead to polymerization in the bulk, and proceeds at low polymerization rates, leading to uniform coverage of membrane surfaces and low polydispersity.77,78 If the ATRP polymer brush can be produced with thickness gdp/2, then the use of a thermally evaporated Au layer should not hinder the molecular switching action of the membrane because the polymer brushes grafted at the edge of the pores are capable of covering the pores in their swelled state (Scheme 1). Even aside from the improved morphology there are several advantages of using partial porecovering thermally evaporated Au as opposed to pore-filling electrolessly deposited Au. Polymer brushes grafted outside the pores can respond to rapid temperature changes more effectively than polymer buried in the center of the polycarbonate membrane, leading to rapid switching action.79 Permeation rates should be greater than with Au-filled pores because of decreased masstransfer resistance, and a simple mechanism applies, whereas in the case of Au-filled pores, the degree of graft yield can lead to a dual-response mechanism.71,73 Furthermore, in those cases where it is desirable to achieve Au coverage on both outer and inner surfaces, a combination of both methods (i.e., electroless plating, removing the Au from the exterior surface, followed by evaporation) could be used to produce a Au layer inside the channel while maintaining a smooth surface on the membrane. Surface-Initiated Polymerization of PNIPAAm Polymer Brushes. PNIPAAm polymer brushes were prepared on Aucoated silicon wafers for comparison with Au-coated NCAMs using the method described previously.80 Because the formation of an initiator monolayer and subsequent templated growth of a PNIPAAm polymer brush have been previously studied by XPS, only ellipsometry and FTIR reflection-absorption measurements were used to follow the PNIPAAm brush growth. It is crucial to control the polymer brush thickness for the molecular gating application of interest here. Because the maximum initial pore diameter is 100 nm, the PNIPAAm dry film thickness must be e50 nm. Thicker films can be grown, but these completely block the pores under all experimental conditions. To demonstrate the principle of highly controlled growth with nanometer precision, polymer brushes with an ellipsometric thickness of ∼10 nm in air were prepared via ATRP for 15 min in 3:1 v/v MeOH/H2O by reducing the concentrations of monomer and catalyst. Under these conditions, the polymer brush thickness increases very slowly with time, reaching ∼13 nm after 4 h. However, polymer brushes with an ellipsometric thickness of ∼30 nm in air are achieved by replacing the solvent with 1:1 v/v MeOH/H2O, a result consistent with literature reports that increasing the water content accelerates the ATRP reaction rate.81 PNIPAAm polymer brushes were prepared on Au-coated NCAMs using the same conditions as for Au-coated silicon wafers. It is difficult to measure the thickness of PNIPAAm polymer brushes formed on the membrane directly because of the optical scattering associated with the nanopores. However, it is reasonable to assume that the average molecular weight of polymer brushes is the same on Au-coated NCAMs as on Aucoated Si and that the grafting density of PNIPAAm polymer brushes is determined by the initiator coverage on the Au surface.80 (77) Fu, Q.; Rao, G. V. R.; Basame, S. B.; Keller, D. J.; Artyushkova, K.; Fulghum, J. E.; Lopez, G. P. J. Am. Chem. Soc. 2004, 126, 8904-8905. (78) Fu, Q.; Rao, G. V. R.; Ista, L. K.; Wu, Y.; Andrzejewski, B. P.; Sklar, L. A.; Ward, T. L.; Lopez, G. P. AdV. Mater. 2003, 15, 1262-1265. (79) Yang, B.; Yang, W. T. J. Membr. Sci. 2003, 218, 247-255. (80) Wang, X. J.; Tu, H. L.; Braun, P. V.; Bohn, P. W. Langmuir 2006, 22, 817-823. (81) Kaholek, M.; Lee, W. K.; Ahn, S. J.; Ma, H. W.; Caster, K. C.; LaMattina, B.; Zauscher, S. Chem. Mater. 2004, 16, 3688-3696.

Flow Switching in Nanocapillary Array Membranes

Langmuir, Vol. 23, No. 1, 2007 309

Figure 3. External reflection FTIR spectra of PNIPAAm brushes grafted onto Au-coated NCAMs in the 1300-1800 cm-1 region. All spectra were baseline corrected against a bare (i.e., ungrafted) Aucoated NCAM. The vertical scale bar represents an equivalent absorbance of 0.002. Table 1. Average Molecular Weight and Stokes Radii of FITC-Labeled Dextrans molecular weight (kDa)

approximate Stokes radii (nm)

4.4 10 19.4 40 77

1.4 2.3 3.3 4.5 6.0

Because full monolayer initiator coverage is achieved for both systems, the thickness of PNIPAAm polymer brushes should be approximately the same for both systems. Furthermore, it is expected that PNIPAAm polymer brushes coat both the top surface and the side wall of the Au layer in the mouth of the nanopore, similar to surface-initiated polymerization on nanopatterns reported by Zauscher et al.82 Thus, the pore size of the PNIPAAm-grafted Au-coated membrane is further reduced by the presence of the PNIPAAm polymer brushes. External reflection FTIR spectroscopy was used to characterize the membrane surface after PNIPAAm grafting. Figure 3 illustrates the spectrum obtained for a PNIPAAm film grafted onto a 45 nm Au film evaporated onto a polycarbonate NCAM with an initial pore diameter of dp ≈ 100 nm. The vibrational spectrum provides strong evidence for the presence of the polymer. Peaks at 1653, 1558, and 1540 cm-1 due to amide carbonyl absorption, N-H bending, and C-H bending, respectively, are characteristic of PNIPAAm.50,83,84 C-(CH3)2 peaks at 1387 and 1364 cm-1 are characteristic of the N-isopropyl moiety,83 whereas the peak near 1718 cm-1 could result from ester carbonyl groups of unreacted initiator. Taken together, the vibrational spectroscopy supports the conclusions from XPS and ellipsometry that these reaction conditions lead to successful grafting of PNIPAAm films on Au-coated polycarbonate. Transport Experiments. Several methods have been used to characterize molecular transport through nanochannels, including conductivity,71,85 water flux,79,84 refractive index,51 absorbance,86 UV spectroscopy,71,73 and total carbon analysis.34 In this work, (82) Ahn, S. J.; Kaholek, M.; Lee, W. K.; LaMattina, B.; LaBean, T. H.; Zauscher, S. AdV. Mater. 2004, 16, 2141-2145. (83) Nonaka, T.; Ogata, T.; Kurihara, S. J. Appl. Polym. Sci. 1994, 52, 951957. (84) Xie, R.; Chu, L. Y.; Chen, W. M.; Xiao, W.; Wang, H. D.; Qu, J. B. J. Membr. Sci. 2005, 258, 157-166. (85) Reber, N.; Kuchel, A.; Spohr, R.; Wolf, A.; Yoshida, M. J. Membr. Sci. 2001, 193, 49-58. (86) Ito, Y.; Park, Y. S. Polym. AdV. Technol. 2000, 11, 136-144.

Figure 4. (a) Permeation of 4.4, 10, 19.4, 40, and 77 kDa FITClabeled dextran through an ID100/Au45/PNIPAAm10 NCAM. (b) Permeation of 4.4, 10, 19.4, 40, 77, and 282 kDa FITC-labeled dextran through an ID200/Au45/PNIPAAm30 NCAM.

in situ fluorescence measurements on FITC-tagged dextrans with molecular weights of 4.4, 10, 19.4, 40, 77, and 282 kDa were used in order to observe permeability changes directly in real time. Dextran is a random coil polymer that has a well-defined Stokes radius in water (cf. Table 1).34 NCAMs with 80, 100, and 200 nm initial diameters were coated first with 45 nm Au and then with 10 or 30 nm (dry film) PNIPPAm films. These were subsequently used to investigate the flux as a function of the pore diameter below and above the LCST of PNIPAAm. In the following sections, NCAM composite structures are denoted using a designation specifying the initial pore diameter/evaporated Au thickness/dry PNIPAAm film thickness, with all thicknesses understood to be given in nanometers. For example, an NCAM prepared from a 100 nm initial pore diameter polycarbonate membrane modified with 45 nm Au layer and a 10 nm dry film thickness of PNIPAAm is denoted ID100/Au45/PNIPAAm10. Figure 4 illustrates the permeation of FITC-labeled dextrans of different sizes through NCAMs ID100/Au45/PNIPAAm10 (Figure 4a) or ID200/Au4/PNIPAAm30 (Figure 4b). It is important to note that these experiments measure the instantaneous concentration of the FITC-labeled probe on the receiving side of the membrane. The permeability is given by

P)

Jd CA

(1)

where J is the probe flux, d(A) is the membrane thickness (area), and C is the probe concentration (source side). Furthermore, the probe concentration is sufficiently large that it is not reduced significantly during an experiment, so the only variable quantity in a given experiment is the probe flux, which is proportional to the slope of the intensity-time curves. Unfortunately, because the FITC tagging density on different dextrans is not uniform, typically varying between 0.003 and 0.02 mol of FITC per mole of glucose monomer, an absolute comparison of permeation rates between dextrans of different molecular weights is precluded. Initially, when the temperature of the PNIPAAm film is below its LCST (0 < t < 10 min), none of the dextrans, except the

310 Langmuir, Vol. 23, No. 1, 2007

smallest 4.4 kDa probe, are transported through either membrane. Raising the temperature of the PNIPAAm film above its LCST causes large increases in the permeation of 10 and 19.4 kDa dextrans, and the permeation rate of the 4.4 kDa dextran increases. In contrast, the permeation of 40 and 77 kDa molecules through the 100 nm membrane is completely hindered at all temperatures. Permeation through the 200 nm NCAM is similar, except that the 40 and 77 kDa molecules show slight changes in the transport rate above the LCST. However, for this membrane the 282 kDa dextran is not transported across the membrane at either temperature. Once the temperature is decreased below the LCST of PNIPAAm, permeation of 10 and 19 kDa dextrans stops, and the rate of permeation of the 4 kDa dextran returns to its initial value, illustrating the reversibility of the molecular gating action of the membrane. These results can be understood within the context of the pore-blocking model depicted in Scheme 1. Tu et al. approximate the PNIPAAm film thickness as 4 times its dry film thickness below the LCST and 2 times its dry film thickness above the LCST,87 with the caveat that the extent of swelling varies with grafting density and polydispersity. Using the above concordance between dry and wet film thicknesses, the 100 nm pores (80 nm after Au evaporation) should exhibit a completely closed pore at T < TLCST because dp ≈ 80 - (2 × 40) nm ) 0 nm, whereas above the LCST, dp ≈ 80 - (2 × 20) nm ) 40 nm. Thus, on the basis of pore diameter alone one would expect complete blockage of all molecular weights below the LCST and some level of permeability above the LCST irrespective of the molecular size of the probes used. In fact, some transport is observed for the smallest probe under all conditions, and the transport is hindered above the LCST for probe molecules with Stokes radii much smaller than the effective pore diameter (e.g., 4.5 nm for the 40 kDa probe). The observation of size exclusion when the molecular size is within an order of magnitude of the effective pore diameter has precedent in the literature. Effects relating to penetrant size are typically addressed within the framework of hindered diffusion,88-90 where solute diffusivity is reduced both by thermodynamic and hydrodynamic effects. The thermodynamic effect has its origin in a concentration-based reduction in the driving force for diffusion compared to that observed in bulk solution, whereas the hydrodynamic effect arises from an enhanced viscous drag arising from a proximal pore wall. These have been effectively treated using a virial expansion for the concentration dependence of the partition coefficient and intrapore diffusivity.91,92 At the other end of the size range, 4.4 kDa dextrans permeate through the PNIPAAm coating, below the LCST, where the pore is supposedly blocked. Presumably, there is a significant diffusion coefficient for molecules of this size because small conformational changes in the grafted polymers mediate molecular hopping through the extended gel network. These transient pores should have less effect on bulky molecules, meaning that thermoresponsive free volume transitions in grafted polymer chains are more suitable for controlling the transport of larger molecules than smaller species.93 Figure 5 shows the permeation of 19.4 kDa dextran (2.6 µM) through pores of different initial diameters and different (87) Tu, H.; Heitzman, C. E.; Braun, P. V. Langmuir 2004, 20, 8313-8320. (88) Beerlage, M. A. M.; Peeters, J. M. M.; Nolten, J. A. M.; Mulder, M. H. V.; Strathmann, H. J. Appl. Polym. Sci. 2000, 75, 1180-1193. (89) Nitsche, J. M.; Balgi, G. Ind. Eng. Chem. Res. 1994, 33, 2242-2247. (90) Deen, W. M. AIChE J. 1987, 33, 1409-1425. (91) Shao, J. H.; Baltus, R. E. AIChE J. 2000, 46, 1307-1316. (92) Shao, J. H.; Baltus, R. E. AIChE J. 2000, 46, 1149-1156. (93) Israels, R.; Gersappe, D.; Fasolka, M.; Roberts, V. A.; Balazs, A. C. Macromolecules 1994, 27, 6679-6682.

Lokuge et al.

Figure 5. Permeation of 19.4 kDa dextran through ID80/Au45/ PNIPAAm10, ID100/Au45/PNIPAAm10, ID200/Au45/PNIPAAm10, and ID100/Au45/PNIPAAm10 NCAMs.

PNIPAAm (dry film) thicknesses, comparing transport through ID80/Au45/PNIPAAm10, ID100/Au45/PNIPAAm10, and ID200/ Au45/PNIPAAm10 as well as through ID200/Au45/PNIPAAm30 NCAMs. Three classes of behavior are observed. The smallest pore diameter membrane, ID80/Au45/PNIPAAm10, does not admit transport of the 19.4 kDa probe under any circumstance, whereas the largest effective pore diameter membrane, ID200/ Au45/PNIPAAm10, shows a barely measurable difference in transport rates below and above the LCST. For the 19.4 kDa probe, switchable behavior is obtained for the two intermediate pore size membranes (i.e., ID100/Au45/PNIPAAm10 and ID200/ Au45/PNIPAAm30). These membranes yield dp ≈ 80 - (2 × 40) nm ) 0 nm and 160 - (2 × 120) nm ) 0 nm, respectively, below the LCST and dp ≈ 80 - (2 × 20) nm ) 40 nm and 160 - (2 × 60) nm ) 40 nm, respectively, above the LCST. An obvious explanation for this observation is that the final pore diameter of the surface-decorated membrane is too small for the transport of 19.4 kDa dextran below the LCST of PNIPAAm yet admits transport above the LCST. However, the low polydispersity and the density of grafted chains could also play a role in determining the observed permeability.73 If the degree of polymerization and the graft spacing were comparable with the pore diameters, then the grafted polymers would collapse onto the pores, thereby partially blocking transport, even above the LCST. In contrast, if the pore sizes were larger than the graft spacing, then the collapsed polymer brushes could block only the edge of the pore wall, resulting in a change in the transport rate without completely blocking the pore.93 Independent of the exact mechanistic explanation for the switching action, it is clear that the molecular-size-dependent thermally responsive switching displayed in these NCAMs results from a balance between molecular size, the effective pore size of the NCAM, and the physical mechanisms giving rise to blocked transport. As a practical matter, it is critical that the switched transport be reversible. Figure 6 illustrates the reversible switching capability of an ID200/Au50/PNIPAAm10 NCAM probed by a 77 kDa dextran through several heating-cooling cycles. Clearly, the switching behavior is reversible over the time period investigated. This is not surprising, given the numerous examples of reversible thermal transitions of PNIPAAm in the literature. However, it is not self-evident that thin films of PNIPAAm grafted to nanostructured materials should exhibit the same reversible behavior observed in homogeneous solution. Moreover, these membranes are also robust over long periods of time, several months in our laboratory, adding to the level of interest in them as actively gateable, size-selective fluidic switching devices.

Flow Switching in Nanocapillary Array Membranes

Langmuir, Vol. 23, No. 1, 2007 311

device. Furthermore, in the PNIPAAm-grafted NCAMs studied here the switching can be intelligent in the sense that the on-off transition can be tuned to a given molecular size threshold. Thermal evaporation of Au onto polycarbonate NCAMs produces smooth metal surfaces, which is key to obtaining uniform polymer films. Real-time transport data, below and above the LCST of PNIPAAm, obtained using in situ fluorescence measurements, indicate that the molecular size cutoff for transport results from the interplay between the Au-coated pore diameter, the PNIPAAm thickness, and the degree of contraction upon raising the temperature through the LCST. Thus, the membranes used in these experiments are ideal candidates for controlled valving of molecules based on molecular size, with the size threshold being tailored by varying the initial pore size of the membrane as well as the degree of polymerization of the grafted PNIPAAm. Figure 6. Reversible switching capability of the PNIPAAm grafted membrane. Permeation of 77 kDa dextran through an ID200/Au50/ PNIPAAm10 NCAM over several heating-cooling cycles.

Conclusions Exploiting external stimuli to operate a size-selective switchable molecular gate to control transport has a variety of applications in microfluidic systems. For example, the cylindrical nanochannels of the polymer-grafted membrane can be implemented as an actively switchable nanofluidic interconnect to establish controllable fluidic communication between micrometer-scale channels operating in different planes of an integrated microfluidic

Acknowledgment. The work reported here was supported by the Department of Energy through grant DEFG02 88ER13949 and by the National Science Foundation through the Nanoscale Chemical-Electrical-Mechanical Manufacturing Systems Center operated under grant DMI-0328162. Ellipsometry and XPS data were obtained at the Center for Microanalysis of Materials at the Frederick Seitz Materials Research Laboratory, which is supported by the Department of Energy through grant DEFG02 91ER45439. LA060813M