Voltage Gated Carbon Nanotube Membranes - Langmuir (ACS

Jul 7, 2007 - Membranes composed of an array of aligned carbon nanotubes, functionalized with charged molecular tethers, show voltage gated control of...
13 downloads 12 Views 322KB Size
8624

Langmuir 2007, 23, 8624-8631

Voltage Gated Carbon Nanotube Membranes Mainak Majumder,† Xin Zhan,‡ Rodney Andrews,§ and Bruce J. Hinds*,†,‡ Department of Chemical & Materials Engineering, and Department of Chemistry, UniVersity of Kentucky, Lexington, Kentucky 40506-0046, and Center for Applied Energy Research, UniVersity of Kentucky, Lexington, Kentucky 40511-8410 ReceiVed March 9, 2007. In Final Form: May 4, 2007 Membranes composed of an array of aligned carbon nanotubes, functionalized with charged molecular tethers, show voltage gated control of ionic transport through the cores of carbon nanotubes. The functional density of tethered charge molecules is substantially increased by the use of electrochemical grafting of diazonium salts. Functionality can be forced to occur at the CNT tip entrances by fast fluid flow of an inert solvent through the core during electrochemical functionalization. The selectivity between Ru(bi-pyridine)32+ and methyl viologen2+ flux is found to be as high as 23 with -130 mV bias applied to the membrane as the working electrode. Changes in the flux and selectivity support a model where charged tethered molecules at the tips are drawn into the CNT core at positive bias. For molecules grafted along the CNT core, negative bias extends the tethered molecules into the core. Electrostatically actuated tethers induce steric hindrance in the CNT core to mimic voltage gated ion channels in a robust large area platform.

1. Introduction Controlled and selective chemical transport across cell walls through biological channels is of fundamental importance and offers unique opportunities to mimic functionality in robust manmade platforms. These biological protein channels are of remarkable complexity with highly selective chemical transport regulated by chemical binding or voltage response. Voltage gated K+ ion channels, the basis for neural transmissions, have been among the most widely studied, and the crystal structure of KcsA shows that charged protein segments induce “paddle like” conformation changes, thereby changing the transport through the channel.1 The critical essence for mimicking a voltage gated channel is the electrostatically induced motion, or actuation resulting in steric blocking of the channel.2-4 A primary approach to mimic ion channels is the synthesis of intricately designed macromolecule channels to span micelles that respond to input such as voltage, light, and pH.5 Such macromolecule/micelle structures can be incorporated into micro machined platforms to be utilized in sensors, chemical separations, or drug delivery applications. However, micelles are inherently fragile and extremely difficult to incorporate into robust large area applications. To form robust protein channel mimic, a promising approach is to use nanoporous materials as a scaffold to place selective gating chemistry on the pore. Ordered anodized alumina and track etch polycarbonate have been shown to perform size and chemically selective separations by the modification of the surface chemistry along the pore wall.6,7 Responsive pores with surface bound molecules that change conformation with pH can effectively act as gates.8 A promising system has been reported to change the conformation of surface molecules by * Corresponding author. E-mail: [email protected]. † Department of Chemical & Materials Engineering. ‡ Department of Chemistry. § Center for Applied Energy Research. (1) Jiang, Y.; Lee, A.; et al. Nature 2003, 423, 33-41. (2) Yellen, G. Q. ReV. Biophys. 1998, 31, 239-295. (3) Perozo, E.; Cortes, D. M.; Cuello, L. G. Science 1999, 285, 73-78. (4) del Camino, D.; Yellen, G. Neuron 2001, 32, 649-656. (5) Hector, R. S.; Gin, M. S. Supramol. Chem. 2005, 17, 129-134. (6) Jirage, K. B.; Hulteen, J. C.; Martin, C. R. Science 1997, 278, 655-658. (7) Martin, C. R.; Kohli, P. Nat. ReV. Drug DiscoVery 2003, 2, 29-37. (8) Hollman, A. M.; Bhattacharyya, D. Langmuir 2004, 20, 5418-5424.

applied bias9,10 and could have application in conductive membranes. However, to have rapid chemical transport across the membrane and thus to closely mimic protein channels, it is necessary to have a short gate path length, ideally only at the pore entrance. Recent advances in track-ion etch fabrication allow for conical pores that have very narrow apertures (∼2 nm) on one side of the membrane. This has an advantage of concentrating electric field to allow rectification of ionic current.11 This has been further enhanced by the grafting of charged DNA along the pore walls, with the DNA conformation at the narrow aperture being most sensitive to a potential gradient across the membrane.12 The effect of charged molecules at narrow apertures has also been seen with conical electrode wells.13 However, it remains a challenge to place responsive gating molecules directly in the pore entrance in a robust large area nanoporous material. Membranes composed of aligned multiwalled carbon nanotubes (CNTs) crossing a solid polymer film14 can be an alternative platform for creating artificial chemical channels over a large mechanically robust area. Importantly, CNTs offer the following unique physical properties unknown to other membrane systems: (1) atomically flat graphite planes with large van der Waals distance allowing for extremely fast fluid flow;15,16 (2) the act of cutting CNTs places reactive broken carbon at the entrance to the CNT core for a well-defined geometry of ligands at the pore entrance; and (3) the CNTs are conductors inside an insulating matrix allowing for the concentration of electric field at the tips of the CNTs for voltage gating or use as an electrode. The last step of CNT membrane fabrication requires a plasma cutting process that removes iron catalyst particles17 and forms carboxyl end groups at the entrances to inner core cores of CNTs (7 nm i.d.). These can be readily chemically functionalized by a standard carbodiimide reaction and have shown reversible (9) Lahann, J.; Mitragotri, S.; et al. Science 2003, 299, 371-374. (10) Rant, U. K.; et al. Nano Lett. 2004, 4, 2441-2445. (11) Siwy, Z.; Gu, Y.; Spohr, H. A.; Baur, D.; Wolf-Reber, A.; Spohr, R.; Apel, P.; Korchev, Y. E. Europhys. Lett. 2002, 60, 349-355. (12) Harrell, C. C.; Kohli, P.; et al. J. Am. Chem. Soc. 2004, 126, 1564615647. (13) Wang, G.; Zhang, B.; et al. J. Am. Chem. Soc. 2006, 128, 7679-7686. (14) Hinds, B. J.; et al. Science 2004, 303, 62. (15) Majumder, M.; Chopra, N.; Andrews, R.; Hinds, B. J. Nature 2005, 38, 44. (16) Holt, J. K.; et al. Science 2006, 312, 1034-1037. (17) Huang, S. M.; Dai, L. M. J. Phys. Chem. B 2002, 106, 3543-3545.

10.1021/la700686k CCC: $37.00 © 2007 American Chemical Society Published on Web 07/07/2007

Voltage Gated Carbon Nanotube Membranes

biochemical reactions to ligand gated ionic flux across the membrane.18 The effect of the molecular tether length, hydrophobicity, and charge on the transport of differently sized ions through CNT membranes was systematically studied.19 The principle conclusions from this study were: (1) tethered molecules at the entrance to CNTs can sterically hinder larger permeates; (2) negatively charged ligands enhance the diffusion of positively charged permeates, but this enhancement can be screened by the short Debye length in high ionic strength solutions; (3) ligands are located near the tips of CNTs, but a hindered diffusion model is consistent with functionality tens of nanometers down the central core possibly due to plasma treatment; and (4) the functional density of carboxyl groups on CNTs is relatively small with only modest changes in separation between large and small permeate (20 at -130 mV.

Figure 7. (a) Change in separation coefficient with voltage applied to the CNT-SG-spacer(polypeptide)-dye membrane (b) Fluxes of the two permeates at applied voltages across CNT-FG membrane. Membrane area is 0.3 cm2. The line joining the data points is to aid visual clarity.

sterically hindered. At negative bias, the maximum separation factor (∼6) is observed at -50 and -100 mV. This is consistent with limited functional molecules lining the cores of the CNT. At less negative bias, (-100, -150 mV), the flux of the smaller cationic species appears to increase. This again is likely the result of the tethered molecule causing steric hindrance of the larger permeate but allow electrostatic attraction of the smaller species. A dramatic increase in the separation factor (∼22) was seen in the case of the completely functionalized CNTs (static grafted), at a negative bias of 130 mV (Figure 7). The flux of Ru(bipy)32+ decreases by an order of magnitude to 0.017 ((0.003) nmol/h at an applied bias of -130 mV, as compared to 0.19 ((0.006) nmol/h, when no bias applied. The flux of the smaller MV2+ molecule decreases to 0.37 ((0.034) nmol/h from 0.53 ((0.03) nmol/h. Interestingly, increases in separation are seen for both positive and negative bias. A separation factor of ∼14 is observed at an applied bias of +100 mV. For positive bias, a similar mechanism to the tip-only functionalized sample would be expected. That is, any tip functionality would be drawn into the membrane entrance and thus introduces steric hindrance. At negative bias, the ligands lining the core are repelled, thus introducing steric hindrance and the observed size-based separation. Unlike the case of flow grafted membranes, in CNT-SG the fluxes of both the cations decrease with bias, which is also consistent with steric hindrance at both CNT tips and along the cores.

Figure 8 shows the UV-vis spectra of the permeates obtained during the transport measurement through CNT-CG-spacer dye membranes with strong voltage dependence. Interestingly, the steric effects of the dye tether are modest unless bias is applied, presumable with the electric field strongly placing the dye molecules in the channel. In the case of non-grafted membranes, there is a very low density of chemical functionality (vida infra) resulting from just the H2O plasma cutting process. The voltage dependence on separation factor is very modest (∼15% increase) as compared to grafted membranes (as high as ∼800%). Thus, the use of the efficient electrochemical diazonium grafting is critical to have enough functional density on the CNT for “voltage gated” transport. The control experiment with non-functionalized membranes showed a similarly modest change in separation factor or flux with applied voltage up to (300 mV.

4. Conclusions Aligned carbon nanotube membranes are a promising platform to mimic natural protein channels with a well-defined scaffold to place selective gatekeeper chemistry at the channel entrances. The conformation of the large anionically charged molecules tethered to the CNT surfaces changes with applied bias and dramatically affects diffusional transport though the cores of the CNTs. The unique properties of the CNT cores, fast fluid flow and electrically conductive tubes, allow preferential chemical functionality at the core entrance. The application of bias directly to the conductive CNT composite allows for concentration of an electric field directly at the conductive CNT tips. With the precise location of tethered functional groups, modest voltages (∼100 mV) can be applied to control the steric environment of the channel entrance. This is a subtle, but important, distinction from the more common geometry of a cross-membrane potential drop (voltage drop applied to solutions on each side of the membrane) where the electric field is essentially uniform down the length of the pore. In that case, the difference in energy for the conformation of charged ligands away from or toward the pore wall is small. In this case, high electric fields effectively pin charged ligands position in the channel. Because these membranes can be produced over a large area with mechanically robust polymers, it is possible to apply the principles of natural channels to large-scale chemical separations and active drug delivery. For example, ongoing research efforts in this laboratory are to use the electrostatic gatekeeping to control the diffusion flux of addictive substances such as fentenyl or nicotine. This will allow

Voltage Gated Carbon Nanotube Membranes

programmed transdermal drug delivery over large areas with metabolically useful fluxes. The ability to electrochemically graft diazonium salts to CNT entrances also demonstrates the ability to use CNT tips as electrodes for electrochemical reactions with the advantage of forced mass transport of products across the membrane. A broader implication for these membranes, with tethered ligands only at the tips of conductive pores, is that applied bias can control the conformation of the tethered molecules, thus acting as an actuator for the selective transport of permeate molecules across the membrane. This mimics the fundamental mechanism of ion channels that can be applied to macroscopic chemical applications.

Langmuir, Vol. 23, No. 16, 2007 8631

Acknowledgment. Funding support was provided by the NSF CAREER award (CTS), NIH (NIDA), and the ARO Advanced Carbon Nanotechnology Program. Critical infrastructure was provided by the Center for Nanoscale Science and Engineering, University of Kentucky. We acknowledge Uma Prasad Mullick for chemical synthesis advice, undergraduates Wendy Satterwhite, Corey Meadows, and Jeggan Cole for supporting experiments, and D. Bhattacharyya for helpful discussions. M.M. acknowledges a final year dissertation fellowship from the University of Kentucky. LA700686K