NANO LETTERS
Properties of One-Dimensional Molybdenum Nanowires in a Confined Environment
2009 Vol. 9, No. 4 1487-1492
Vincent Meunier,*,† Hiroyuki Muramatsu,‡ Takuya Hayashi,§ Yoong Ahm Kim,§ Daisuke Shimamoto,§ Humberto Terrones,| Mildred S. Dresselhaus,⊥ Mauricio Terrones,| Morinobu Endo,§ and B. G. Sumpter† Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6367 USA, Institute of Carbon Science and Technology, Shinshu UniVersity, 4-17-1 Wakasato, Nagano-shi 380-8553, Japan, Faculty of Engineering, Shinshu UniVersity, 4-17-1 Wakasato, Nagano-shi 380-8553, Japan, Laboratory for Nanoscience and Nanotechnology Research (LINAN) and AdVanced Materials Department, Instituto Potosino de InVestigación Científica y Tecnológica, camino a la presa San José 2055, Col. Lomas 4. Sección, 78216 San Luis Potosí, México, and Department of Electrical Engineering and Computer Science and Department of Physics, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139-4307 USA Received November 13, 2008; Revised Manuscript Received January 12, 2009
ABSTRACT The atomistic mechanism for the self-assembly of molybdenum into one-dimensional metallic nanowires in a confined environment such as a carbon nanotube is investigated using quantum mechanical calculations. We find that Mo does not organize into linear chains but rather prefers to form four atom per unit cell nanowires that consist of a subunit of a Mo body-centered cubic crystal. Our model explains the 0.3 nm separation between features measured by high-resolution transmission electron microscopy and why the nanotube diameter must be in the 0.70-1.0 nm range to accommodate the smallest stable one-dimensional wire. We also computed the electronic band structure of the Mo wires inside a nanotube and found significant hybridization with the nanotube states, thereby explaining the experimentally observed quenching of fluorescence and the damping of the radial breathing modes as well as an increased resistance to oxidation.
The controlled assembly of metallic nanowires with an atomic-scale cross section is a formidable task to overcome for further developments in a number of key areas of nanoscience and nanotechnology. These systems are of fundamental importance as they represent the ultimate dimensional reduction for functional electronic device fabrication. At such reduced dimensions, quantum processes governing behavior are of special significance for the technological quest toward harnessing the full range of practical applications. The construction of stable atomic wires that are resistant to rapid oxidation and other types of structural degradation has not been easy to achieve.1 To date, filling carbon * Corresponding author,
[email protected]. † Oak Ridge National Laboratory. ‡ Institute of Carbon Science and Technology, Shinshu University. § Faculty of Engineering, Shinshu University. | Laboratory for Nanoscience and Nanotechnology Research (LINAN) and Advanced Materials Department, Instituto Potosino de Investigación Científica y Tecnológica. ⊥ Department of Electrical Engineering and Computer Science and Department of Physics, Massachusetts Institute of Technology. 10.1021/nl803438x CCC: $40.75 Published on Web 03/18/2009
2009 American Chemical Society
nanotubes with a chosen metallic host has foiled the most successful attempts. Because their diameter can be tuned by finely tweaking growth parameters and thanks to their accessible inner core, carbon nanotubes are ideal candidates as nanoconfiners for a chosen metallic host. Bismuth-filled multiwalled carbon nanotubes were the first reported success,2 and numerous attempts to encapsulate other elements such as Pb, Y, Mn, Fe, Co, Ru, La, Gd, and I as well as a number of metal halides have also yielded some encouraging results.3 Despite these notable successes, most of the reported core-shell wirelike systems display cross sections that are much larger than the desired atomic chain and in fact tend to form three-dimensional (3D) crystal-like structures rather than nanowires. By further exploiting the confining environment of small diameter carbon nanotubes, a number of groups have recently reported the formation of nanoscale objects emerging from the encapsulation of individual atoms, such as Mo,4 I,5 and La.6 Iodine was found to form diameterdependent polymorphic structures while La was shown to yield long chains of dimers. Mo appears to be the only
Figure 1. HRTEM images of nine double-walled carbon nanotubes of differing inner diameter (d), as defined on micrograph a. The values of the inner diameter are 0.53 nm (a), 0.57 nm (b), 0.62 nm (c), 0.71 nm (d), 0.73 nm (e), 0.78 nm (f), 0.95 nm (g), 1.32 nm (h), and 1.06 nm (i). In (a-c), the tube diameter is too small to allow formation of a Mo structure. In (d-f), a narrow Mo structure with feature separation of about 0.3 nm is observed at the center of the tube, while in (g-i), the inner tubes are large enough to allow for the development of small crystals with cross sections larger than 0.5 nm in width.
element that has been shown to form linear atomic chains inside carbon nanotubes.4 The reason why Mo has a special behavior is unclear, even if it is known to have a much larger bond enthalpy and smaller crystal lattice size than iodine or lanthanum, for instance. These properties should promote 3D growth given adequate space (as is observed in the experiments where nanotubes >1 nm in diameter are used) or dimer structures, since Mo forms a very strong triple bond that leaves no possibility for further bonding. The fact that single atom chains are observed is therefore puzzling. Because of the short-range properties of the bonding, considerable rotational and translational diffusion is however possible and could lead to a misinterpretation of experimental data. In fact, available experimental evidence reveals a high mobility of Mo encapsulated inside the carbon nanotubes. In this work we use quantum mechanical calculations and experimental evidence from high-resolution transmission electron microscopy (HRTEM) measurements to demonstrate that Mo does not form linear single-atom wide chains but organizes into small cross-section nanowires in multiwalled carbon nanotubes with inner diameters of at least 0.7 nm. The synthesis of double-walled carbon nanotubes (DWCNTs) was achieved by utilizing catalytic (an iron catalyst) chemical vapor deposition followed by oxidative purification.4 This process yields highly pure and crystalline DWCNTs which are then used as templates for encapsulating Mo structures 1488
(see Figure 1). To prepare self-assembled Mo structures inside the DWCNTs, air oxidation at 550 °C for 1 h was used to remove amorphous carbon, chemically active singlewalled nanotubes, and defective multiwalled nanotubes. This procedure also opens the ends of the purified DWCNTs, and subsequent treatment of these DWCNTs with heptamolybdate tetrahydrate (0.8 g) and a hydrochloric acid (18 wt %) solution at 100 °C for 24 h followed by air oxidation at 500 °C for 30 min leads to the generation of Mo structures encapsulated inside certain DWCNTs (see Figure 1d-i). The experimental data do not indicate any preference for the encapsulation of Mo on the carbon nanotube type (metallic or semiconducting) or helicity. An important aspect is that the nanotubes are composed of at least two layers.4 The principal factor that leads to small-diameter Mo nanowires was a small nanotube inner diameter ( 9, the nanowire Mo-NW1 is found to be stable. When n > 14, the size of the inner cavity can accommodate a larger structure, such as Mo-NW2, built from a bcc crystal (3D crystallites). The experimental micrographs were obtained using techniques described in the text and in ref 4. The simulated HRTEM image corresponds to Mo-NW1 encapsulated in a (9,0)@(18,0) double-walled carbon nanotube (see text for further discussion of the distorted cross section).
Figure 5. DFT (LDA) electronic band structures for (a) a pristine (9,0) tube, (b) an isolated Mo-nanowire model Mo-NW1, and (c) for the system made up of a Mo nanowire encapsulated within a (9,0) nanotube. The wavevector is expressed in reduced units of π/a where a ) 1.262 nm is the unit cell along the nanowire axis. The Fermi level is at E ) 0.0 eV in each case. The structures shown in the inserts correspond to one unit cell used in each calculation.
of the fluorescence for Mo-filled tubes as well as the damping of the radial breathing modes.4 The hybridization also accounts for the increased resistance of the Mo wire to oxidation. Consistent results are found using GGA (not shown). Nano Lett., Vol. 9, No. 4, 2009
Experimental evidence indicates that the wires are very mobile inside the tube. To computationaly address the Mo wire diffusion inside a nanotube, we used the adiabatic trajectory method,12 where the wire is moved along the tube axis with small increments, while monitoring changes in the 1491
The proposed model explains all the experimental observations presented in ref 4 and in Figure 1, including the confinement effect, the hybridization of orbitals between the wire and the tube, the high mobility, and the details of the experimental HRTEM observations. Our work further emphasizes the key role of simulations in gaining an understanding and an interpretation of the experimental HRTEM images.
Figure 6. Potential energy curve calculated (DFT-LDA) for a Mo nanowire that has been moved along the axis of a fixed (9,0) carbon nanotube. The time needed for the wire to be displaced by 1 µm was evaluated using standard diffusion theory in one-dimensional systems and is represented by the excursion time t vs temperature T in the insert.
total energy. For each position along the axis, the wire is constrained along the axis and free to move in the directions perpendicular to the axis, thereby enabling the wire to follow the optimum path. This method has been found to be a realistic alternative to a costly point-by-point determination of the potential energy surface, especially for one-dimensional systems.12 To simulate the presence of multiple shells, we kept the carbon atoms frozen during this calculation. The resulting energy curve (DFT-LDA) is shown in Figure 6. This energy curve can be used to estimate the rate at which the wire moves inside the inner core of the tube.13 The time for a diffusional hop is a02/D, where a0 ) 0.21 nm,14 and D is the diffusion constant given by D ) νa02 exp[-Eb/kBT] where Eb ) 58 meV and ν ) 1012 Hz, as estimated from the potential energy curve. If we assume that the wire undergoes a one-dimensional walk inside the channel, it undertakes a total of N2 hops in order to move a distance L ) Na0 and the time t for the wire to move a distance L along the channel is equal to t ) L2/D. In the insert of Figure 6, we show a plot of excursion time t as a function of temperature for L ) 1 µm. At room temperature, the low energy barrier indicates that the wire moves quasi-freely along the tube, making the HRTEM measurements tedious to perform, as shown experimentally.4 In summary, we have presented considerable theoretical and experimental evidence, demonstrating that Mo forms small cross-sectional metallic nanowires inside carbon nanotubes of diameters ranging from 0.7 to 1.0 nm. Metastable organized structures of Mo do not include monatomic uniformly separated chains but rather collections of interacting dimers and nonlinear trimers. Using DFT, the electronic and diffusion properties of the wire in the tube are also presented, in clear agreement with experiment.
1492
Acknowledgment. This work was supported in part by the CNMS, sponsored by the Division of Scientific User Facilities, US-DoE, and by the Division of Materials Science and Engineering. This work was also supported in part by CONACYT-Mexico Grants 56787 (Laboratory for Nanoscience and Nanotechnology Research-LINAN), 45762 (H.T.), 45772 (M.T.), 41464-Inter American Collaboration (M.T.), 42428-Inter American Collaboration (H.T.), 2004-01-013/ SALUD-CONACYT (M.T.), Fondo Mixto de San Luis Potosi 63001 S-3908 (M.T.). This work was also partially supported by the CLUSTER (the second stage) and a Grantin-Aid for Specially Promoted Research (No. 19002007) of Ministry of Education, Culture, Sports, Science and Technology of Japan. M.S.D. was supported by NSF/DMR Grant 07-04197. References (1) Zach, M. P.; Ng, K. H.; Penner, R. M. Science 2000, 290, 2120. (2) Ajayan, P. M.; Ebbesen, T. W.; Ichihashi, T.; et al. Nature 1993, 362, 522. (3) Bao, J.; Tie, C.; Xu, Z.; et al. AdV. Mater. 2002, 14, 1483. Flahaut, E.; Sloan, J.; Friedrichs, S.; et al. Chem. Mater. 2006, 18, 2059. Grobert, N.; Mayne, M.; Terrones, M.; et al. Chem. Commun. 2001, 471. Leonhardt, A.; Hampel, S.; Muller, C.; et al. Chem. Vap. Deposition 2006, 12, 380. Meyer, R. R.; Sloan, J.; Dunin-Bokowski, R. E.; et al. Science 2000, 289, 1324. Rahman, M. M.; Kisaku, M.; Kishi, T.; et al. J. Phys. Soc. Jpn. 2005, 74, 742. Sloan, J.; Hammer, J.; Zwiefka-Sibley, M.; et al. Chem. Commun. 1998, 3, 347. Sloan, J.; Kirkland, A. I.; Hutchinson, J. L.; et al. Chem. Commun. 2002, 1319. Tsang, S. C.; Chen, Y. K.; Harris, P. J. F.; et al. Nature 1994, 372, 159. (4) Muramatsu, H.; Hayashi, T.; Kim, Y. A.; et al. Nano Lett. 2008, 8, 237. (5) Guan, L.; Suenaga, K.; Okubo, S.; et al. J. Am. Chem. Soc. 2008, 130, 2162. (6) Guan, L.; Suenaga, K.; Shi, Z.; et al. Nano Lett. 2007, 7, 1532. (7) Kresse, G.; Joubert, D. Phys. ReV. B 1999, 59, 1758. Blochl, P. E. Phys. ReV. B 1994, 50, 17953. (8) Ceperley, D. M.; Alder, B. J. Phys. ReV. Lett. 1980, 45, 566. (9) Perdew, J. P. Phys. ReV. B 1992, 46, 6671. Perdew, J. P. Phys. ReV. B 1993, 48, 4979. (10) Hopkins, J. B.; Langridge-Smith, P. R. R.; Morse, M. D.; et al. J. Chem. Phys. 1983, 78, 1627. (11) Zhang, W.; Ran, X.; Zhao, H. J. Chem. Phys. 2004, 121, 7717. (12) Wang, C. Phys. ReV. Lett. 1992, 69, 3789. Zhang, Q.-M. Phys. ReV. Lett. 1995, 75, 101. (13) Hill, T. L. An introduction to Statistical Thermodynamics; Dover: New York, 1986. (14) a0 is the distance between equivalent adsorption sites, along the tube axis for a zigzag nanotube. Note that it is a coincidence that its numerical value is the same as the shortest distance between a Mo atom and the nanotube wall. (15) Ross, R. G.; Hume-Rothery, W. J. Less-Common Met. 1963, 5, 258.
NL803438X
Nano Lett., Vol. 9, No. 4, 2009
