Langmuir 2006, 22, 10909-10911
10909
Carbon Nanotube Supported Single Phospholipid Bilayer Jennifer Gagner,†,| Hannah Johnson,†,⊥ Erik Watkins,† Qingwen Li,‡ Mauricio Terrones,§ and Jaroslaw Majewski*,† Manuel Lujan Neutron Scattering Center and MST STC, Los Alamos National Laboratory, Los Alamos, New Mexico, and Instituto Potosino de InVestigacio´ n Cientifica y Tecnolo´ gica, San Luis Potosi, Me´ xico ReceiVed July 14, 2006. In Final Form: September 12, 2006 Single bilayer membranes of 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) were formed on micron thin-films of hydrophilized carbon nanotubes (CNT) by fusion of small unilamellar vesicles. The structure of the membrane was investigated using neutron reflectivity (NR). The underlying thin film of CNT was formed by chemical vapor deposition (CVD) in the presence of Fe catalyst, followed by reaction with 5 M nitric acid to render the film hydrophilic. We demonstrate that this platform lends support to homogeneous and continuous bilayer membranes that have promising applications in the fields of biomaterials, biosensors, and biophysics.
As we move toward seamless integration of technology and biology, the first obstacle is the functional consolidation of organic and inorganic components into a single device. One of the most basic building blocks of a living system is the cell membrane, composed of phospholipids in a self-assembling bilayer conformation. Its primary functions are to maintain various gradients between the interior and exterior of the cell and serve as a matrix to host trans-membrane proteins, which control flux through the membrane and provide mechanisms for cell recognition. It is possible to produce a cell membrane on a solid substrate,1 but the proximity of a flat support prevents the demonstration of key characteristics that would be seen in a biological setting.2 The novel mechanical, electrical, chemical, and thermal properties of carbon nanotubes (CNT) show potential for development of groundbreaking technologies.3 One such property is the inherent hydrophobicity of CNT;4 this can be modified through a variety of methods, such as nitrogen doping through acid treatment.5 Increased hydrophilicity allows for greater penetration of water into the nanotube network. A variety of network conformations are possible for nanotube synthesis on a substrate, including growth as an aligned array of monodispersive height and density6 or as an irregular, randomly distributed and intercalating CNT network7 (Figure 1). Both arrangements create a mechanically strong and nanoporous * To whom correspondence should be addressed. E-mail:
[email protected]. † Manuel Lujan Neutron Scattering Center. ‡ MST STC. § Instituto Potosino de Investigacio ´ n Cientifica y Tecnolo´gica. | Department of Materials Science & Engineering, Massachusetts Institute of Technology, Cambridge, MA. ⊥ University of California-Santa Barabara, Santa Barbara CA. (1) (a) Reviakine, I.; Brisson, A. Langmuir 2000, 16, 1806-1815. (b) Jass, J.; Tja¨rnhage, T.; Puu, G. Biophys. J. 2000, 79, 3153-3163. (c) Vacklin, H.; Tiberg, F.; Fragnet, G.; Thomas, R. Langmuir 2005, 21 (7), 2827-2837. (2) Doshi, D.; Dattelbaum, A.; Watkins, E.; Brinker, C.; Swanson, B.; Shereve, A.; Parikh, A.; Majewski, J. Langmuir 2005, 21, 2865-2870. (3) (a) Terrones, M. Int. Mater. ReV. 2004, 49, 325-377. (b) Nakayama, Y.; Akita, S.; Shimada, Y. Jpn. J. Appl. Phys. 1995, 34, L10-L12. (c) Tans, S.; Devoret, M.; Dai, H.; Thess, A.; Smalley, R.; Geerligsand, L.; Dekker, C. Nature, 1997, 386, 474-477. (d) Hones, J.; Llaguno, M.; Blercuk, M.; Johnson, A.; Batlogg, B.; Benes, Z.; Fischer, J. Appl. Phys. A: Mater. Sci Proc. 2002, 74, 339-343. (e) Fan, S.; Liang, W.; Dang, H.; Franklin, N.; Tombler, T.; Chapline, M.; Dai, H. Physica E 2000, 8, 179-183. (4) Li, S.; Li, H.; Wang, X.; Song, Y.; Liu, Y.; Zhu, D. J. Phys. Chem. B 2002, 106, 9274-9276. (5) Niyogi, S.; Hamon, M.; Hu, H.; Zhao, B.; Bhowmik, P.; Sen, R.; Itkis, M.; Haddon, R. Acc. Chem. Res. 2002, 35, 1105-1113. (b) Hennrich, F.; Wellmann, R.; Malik, S.; Ledebkin, S.; Kappes, M. Phys. Chem. Chem. Phys. 2003, 5, 178183. (c) Hirsch, A. Angew. Chem., Int. Ed. 2002, 41, 1853. (d) Sun, Y.-P.; Fu, K.; Lin, Y.; Huang, W. Acc. Chem. Res. 2002, 35, 1096. (6) Wang, X.; Liu, Q.; Zhu, D. AdV. Mater. 2002, 14, 165-167.
Figure 1. SEM image of the hydrophilic CNT network used to support a phospholipid bilayer for our NR experiment. The intercalated nature of the CNT network makes a nanoporous thin film. Inspection of the CNT prior to exposure to D2O and the lipid membrane revealed similar topography.
surface, though the intercalating network has greater structural stability. Though it is possible to produce aligned arrays of carbon nanotubes, alignment is lost after exposure to water and the subsequent evaporative drying process. When regular CNT arrays were examined after drying, it was observed that the nanotubes were no longer spaced at intervals but instead had unevenly aggregated over the substrate surface. It is possible that this effect is the product of capillary forces between the tubes as evaporation takes place.8 Intercalated networks of CNT are less vulnerable to the destructive forces of water, which facilitates work in in vitro conditions. Utilizing neutron reflectometry (NR) as a characterization method, we have demonstrated that it is possible to fuse a single phospholipid bilayer onto such an irregular network of hydrophilic CNT. CNT were prepared using a chemical vapor deposition (CVD) process with a Lindberg/Blue tube furnace and a gas flow control (7) Sinnott, S.; Andrews, R.; Qian, D.; Rao, A.; Mao, Z.; Dickey, E.; Derbyshire, F. Chem. Phys. Lett. 1999 315, 25-30. (8) Lau, K.; Bico, J.; Teo, K.; Chhowalia, M.; Amaratunga, G.; Milne, W.; McKinley, G.; Gleason, K. Nano Lett. 2003, 3, 1701-1705.
10.1021/la062038g CCC: $33.50 © 2006 American Chemical Society Published on Web 11/16/2006
10910 Langmuir, Vol. 22, No. 26, 2006
unit. Monocrystalline quartz substrates were sputtered with 2 nm-size Fe catalyst, which was allowed to oxidize in air over a period of 2 days in order to avoid severe agglomeration of Fe film under high pressure. The substrate was placed inside the quartz tube and heated to 750 °C over a period of 15 min. After allowing the system to equilibrate, ethylene gas was passed together with Ar at a 3:2 ratio with a total rate of 500 cm3/min for 5 min. To make the CNT hydrophilic, the sample was immersed in 5 M nitric acid for 3 h and rinsed with deionized water. The CNT films were characterized using a JEOL 6300FXV high-resolution SEM at 1 kV (Figure 1). Vesicles composed of 1-palmitoyl-2-oleoyl-sn-glycero-3phosphocholine (POPC; Avanti Polar Lipids) dissolved in H2O in a concentration of 1 mg/mL were passed through a 0.1 micron extruder and injected into the solid-liquid interface cell at room temperature, which is above the phase transition temperature (Tm ) -2 °C) of POPC. Vesicle fusion onto the CNT was allowed to progress for 15 min, and then the cell was rinsed with deuterated water (D2O). The sample was then characterized using neutron reflectivity. Following the NR characterization, the substrate was removed from the cell and dried, and SEM was conducted again to verify the presence of nanotubes on the surface of the quartz. NR was performed utilizing the Surface Profile Analysis Reflectometer (SPEAR) at the Los Alamos Neutron Science Center9 (Figure 2). NR was chosen for the ability to penetrate through the thick layer of supporting quartz crystal and probe the buried interfaces. Further, D2O was used to provide strong neutron scattering contrast against hydrogenated POPC. Reflectivity, R, is defined as the intensity ratio of neutrons specularly scattered from a surface relative to the incident neutron beam intensity. When measured as a function of wave-vector transfer, qz (qz )|kout - kin| ) 4π sin θ/λ, where θ is the angle of incidence and λ is the wavelength of the neutron beam), the reflectivity curve contains information regarding the sample-normal of the in-plane average of the scattering length density (SLD) profile. Detailed information on the SLD distribution in the direction normal to the interface can be determined by modeling the deviation of the measured reflectivity from Fresnel’s law for a perfect interface.10 Analysis of the NR data was performed using the Parratt formalism.11 Our philosophy was to implement the simplest possible model with physical relevance. A two-box model was used, describing the hydrated CNT film and the POPC bilayer with parameters corresponding to the thickness D, SLD, and roughness between adjacent layers (Table 1). In all cases, the interfacial roughness, σ, was described by an error function centered at the interface. Selected parameters were adjusted using a Levenberg-Marquardt least-squares fitting algorithm to attain the best possible fit. For the fit described here, parameters corresponding to the quartz substrate were constrained to known values, the thickness of the CNT network was fixed to a value greater than the instrumental resolution, and the SLD of the D2O subphase was calculated to match the measured NR critical edge position. Prior to vesicle fusion, the reflectivity from the CNT network in D2O was measured (Figure 2a). The resulting reflectivity did not deviate significantly from the Fresnel curve. The increase of intensity observed after vesicle fusion is shown in the top curve of Figure 2a, providing clear evidence of a hydrogenated layer bordered on both sides by highly scattering regions of bulk D2O and a hydrated (D2O) CNT film. Using the Parratt formalism,11 (9) http://www.lansce.lanl.gov/lujan/index.html. (10) Als-Nielsen, J. Physica A 1986, 140, 376-389. (11) Parratt, L. Phys. ReV. 1954, 95, 359-369.
Letters
Figure 2. (a) Neutron reflectivity measured for a CNT film (bottom) and the same film supporting a fused POPC bilayer (top) against D2O. The measured reflectivity is plotted as R/RFresnel versus the magnitude of momentum transfer vector qz. Error bars for the reflectivity data represent statistical errors in these measurements. Measured data is represented as symbols and the line represents the fit to the top curve, which corresponds to the smeared SLD profile (solid line) shown in (b). In panel b, the solid and dashed lines show the SLD profile smeared by interfacial roughness and unsmeared, respectively. The unsmeared profile is presented to elucidate the length scale associated with the POPC membrane. Superimposed is a schematic of the CNT network and lipid bilayer tails. The model indicates a lipid membrane with a 30 Å thickness and an approximate surface coverage of 66%. The CNT film was too thick to be resolved by NR (greater than ∼5000 Å) due to instrumental resolution. Table 1. Two-Box Model Fit Parameters for a Lipid Bilayer on an Intercalated CNT Network
quartz CNT network POPC (HC tails) D2O subphase a
thickness (D), Å
SLD, Å-2 x 10-6
roughness (σ), Å
10 000fa 30.2
4.18fa 4.82 1.99 6.18fa
3fa 10 10
f indicates that the parameter was fixed during the χ2 refinement.
the reflectivity was modeled to provide a SLD profile normal to the surface (Figure 2b). The resulting profile describes a hydrogenated lipid membrane resting on a CNT film hydrated with D2O. The CNT layer was too thick to be resolved by NR (greater than ∼5000 Å), so for modeling purposes, this parameter was fixed to 10 000 Å. Although the thickness of the hydrated
Letters
CNT film cannot be resolved, its SLD can be obtained accurately because the contrast between it and the adjacent hydrogenous layer determines the height of the interference maximum (Figure 2a). The phospholipid membrane was modeled using a single layer. Given the limited qz range of the measurement, including the POPC headgroups in our model did not improve the χ2 of the fit and over-parametrized the problem. The thickness of the layer describing the phospholipids was approximately 30 Å, which is comparable to the length of the two hydrocarbon tails of POPC. This suggests the formation of a lamellar bilayer structure as opposed to an adsorbed layer of micelles/hemimicelles or other conformations. Previous studies have shown that a well packed POPC12 membrane has an approximate area per molecule of 68.3 Å2. Such a packing density of the lipid tails yields a neutron scattering length density of -0.13 × 10-6 Å-2. The higher SLD value modeled here indicates significant penetration of D2O into the phospholipid layer. By comparing the measured SLD of the POPC film with the theoretical value, we were able to approximate the surface coverage to be 66%. Although SEM indicates a rough, nanoporous CNT network, after immersion in D2O and vesicle fusion, the lipid bilayer did (12) Kucerka, J.; Tristam-Nagle, S.; Nagle, J. F. J. Membr. Biol. 2006, 208 (3), 193-202.
Langmuir, Vol. 22, No. 26, 2006 10911
not conform to the surface structure of the CNT film. We hypothesize that hydration of the CNT network and rigidity of the membrane yields a flat and a lamella-like conformation with the membrane bridging the basins of the CNT network. This is confirmed by a SLD profile (Figure 2b) that describes a supported membrane with an approximate RMS roughness of only 10 Å. We have shown that it is possible to fuse a phospholipid bilayer onto a network of hydrophilic CNT. By doping the nanotubes to enhance hydrophilicity, it was possible to create a structure that may act as a nanoporous support for a single lipid bilayer. Future development of this system may allow the insertion of trans-membrane proteins and the measurement of ionic and molecular fluxes through the membrane. The production of membranes supported by nanoporous networks is a step toward the creation of functional biomimetic systems which behave similarly to a living membrane in a biological environment. This may have important implications for the manufacture of functional biomaterials, biosensors, and bionanoelectromechanical devices. Acknowledgment. We thank Dr. Andrew Dattelbaum from LANL for help and discussion. This work was supported by Los Alamos National Laboratory under DOE Contract W7405-ENG36, and by the DOE Office of Basic Energy Science. LA062038G
