NANO LETTERS
Atomic-Scale Structure of Mo6S6 Nanowires
2008 Vol. 8, No. 11 3928-3931
Jakob Kibsgaard,† Anders Tuxen,† Martin Levisen,† Erik Lægsgaard,† Sibylle Gemming,‡ Gotthard Seifert,§ Jeppe V. Lauritsen,*,† and Flemming Besenbacher*,† Interdisciplinary Nanoscience Center (iNANO) and Department of Physics and Astronomy, UniVersity of Aarhus, Denmark, Institute of Ion Beam Physics and Materials Research Forschungszentrum Dresden-Rossendorf, Dresden, Germany, Physikalische Chemie, Technische UniVersita¨t Dresden, Germany Received August 6, 2008; Revised Manuscript Received September 16, 2008
ABSTRACT We have studied the atomic-scale structure of the Mo6S6 nanowires using scanning tunneling microscopy and spectroscopy (STM and STS) and density functional theory (DFT). A novel synthesis route based on metallic Mo precursors is presented for the selective formation of elementary pure Mo6S6 nanowires. The Mo6S6 nanowires selectively organize as trimer bundles, and each of the Mo6S6 nanowires consists of an electrically conducting Mo backbone dressed with a sulfur exterior cap. The Mo6S6 nanowires may thus be of interest as novel building blocks in nanoelectronics because the Mo6S6 nanowires exist in a robust, singular structural conformation with uniquely defined electrical (metallic) properties.
Chalcogenide derivatives of molybdenum (Mo) and tungsten (W) are among the most promising and versatile building blocks for the design of novel nanomaterials.1-5 A variety of stable one- or two-dimensional chalcogenide sulfide nanostructures have been discovered with unique combinations of mechanical, electronic, magnetic, and chemical properties completely different from those observed for the 3D bulk phase of these materials.6-10 The common 3D bulk phase of MoS2 is a layered compound consisting of S-Mo-S sheets held together by van der Waals forces, and in analogy to graphene sheets, they may wrap up into spectacular hollow fullerenelike structures and nanotubes.6 Recently however, novel onedimensional molybdenum-sulfide nanowire structures that are not constructed from MoS2 layers but instead from Mo6Sx subunits (x ) 6 or 9) have been proposed.11-13 Quasi-1D batches of molybdenum sulfide nanowires and chalcogenide (Se, Te) analogues have been synthesized previously in bulk inorganic synthesis using various dopants such as alkali metals14,15 or iodine.11 The counterions have been considered necessary to stabilize the nanowires in the previous synthesis protocols and, e.g., in the case of iodine, the dopant atoms are physically incorporated in random concentrations, leading * To whom correspondence should be addressed. E-mail:
[email protected] (J.V.L.) and
[email protected] (F.B.). † Interdisciplinary Nanoscience Center (iNANO) and Department of Physics and Astronomy, University of Aarhus. ‡ Institute of Ion Beam Physics and Materials Research Forschungszentrum Dresden-Rossendorf. § Physikalische Chemie, Technische Universita ¨ t Dresden. 10.1021/nl802384n CCC: $40.75 Published on Web 10/25/2008
2008 American Chemical Society
to a range of chalcohalide isomers with a Mo6Sx-zIz configuration (x ) 6 or 9 and z ) random). It has so far not been possible to experimentally examine the atomic structure of isolated nondoped, rigid Mo6S6-type nanowires in a pure well-defined conformation and relate the structure to the detailed electronic properties of the nanowires. The dopants will most likely affect the conductivity of the individual wires, and from a nanoelectronics perspective, it is of paramount importance to be able to synthesize wires with well-defined electrical, and hence, structural properties. The present synthesis, however, demonstrates that counterions are not required to stabilize Mo6S6 wires. The key to the selective synthesis of the Mo6S6 nanowires versus other sulfided Mo structures is a careful control of the sulfiding environment during synthesis. The synthesis of the Mo6S6 nanowires was performed by physical vapor deposition of Mo onto a highly ordered pyrolytic graphite (HOPG) substrate followed by high-temperature sulfidation in a low-pressure H2S environment (PH2S ) 1 × 10-6 mbar) at 1000 K (see Supporting Information). The H2S partial pressure during the synthesis was observed to have a very pronounced effect on the synthesized molybdenum sulfide nanostructures. The selectivity toward Mo6S6 nanowire formation is explained by the fact that low-pressure sulfiding conditions do not stabilize the sulfur-terminated edges of the S-Mo-S layers required for layered MoS2 nanocrystals, which we have observed previously.16 Indeed, if the H2S pressure is increased by 1 order of magnitude, the synthesis
Figure 1. Selectve formation of trimer bundles of Mo6S6 nanowires. (a) STM image (50 nm × 50 nm, It ) 0.230 nA, Vt ) -1250 mV) of Mo6S6 wires on HOPG. (b) Width distribution of the wires. (c) The Mo6S6 building block. (d) End view of a single Mo6S6 nanowire. (e) Side view of a single Mo6S6 nanowire. Color code: Mo, blue; S, yellow.
Figure 2. Atomic-scale structure of Mo6S6 nanowires. (a) Atom-resolved STM image (2.5 nm × 5.3 nm, It ) 0.210 nA Vt ) -9.8mV) of a trimer Mo6S6 nanowire bundle. (b) Line scan along the row of protrusions on the nanowire in (a) as indicated by the blue arrow. (c) Line scans across the nanowire in (a) as indicated by the black arrow. (d) Simulated STM image with the Mo6S6 structure embedded. (e,f) Simulated STM line scans equivalent to the line scans in (b) and (c).
results instead in single- or multilayer MoS2 nanoclusters with the regular 2H-MoS2 stacking.17 We can therefore conclude that the synthesis of the Mo6S6 nanowires from a metallic Mo precursor is successfully performed only in a mildly sulfiding environment. Figure 1a shows a large-scale STM image of Mo6S6 nanowires supported on the HOPG substrate. The Mo6S6 nanowires are distributed uniformly on the surface with no preferred orientation relative to the high-symmetry substrate directions, which indicates that the direct bonding of the Mo6S6 nanowires segments to the substrate is rather weak. The Mo6S6 nanowires are observed to consist of straight segments in between a number of bends and joints, which we attribute to an initial random distribution of defects in the HOPG surface. The defects were induced intentionally by an initial gentle ion bombardment of the graphite substrate and act as nucleation and anchoring sites for the Mo6S6 nanowires (see Supporting Information). The height and the width of the nanowires are measured from line scans in the STM images to be extremely uniform, with the apparent Nano Lett., Vol. 8, No. 11, 2008
height of the Mo6S6 wires measured to be 0.90 ( 0.01 nm relative to the substrate, while the width of a typical nanowire segment is measured to be 1.2 ( 0.2 nm. Figure 2a depicts an atomically resolved STM image of a nanowire segment. The Mo6S6 nanowire is observed to consist of regularly spaced protrusions in three parallel rows, reflecting that the structure of the Mo6S6 nanowires consists of three interlocked single Mo6S6 nanowires (a trimer). A histogram of the width distribution of the Mo6S6 nanowires (Figure 1b) clearly shows that predominantly wires corresponding to trimers are synthesized, revealing an unprecedented high selectivity (>90%) toward the formation of a structurally singular nanowire conformation. From the width distribution, two other discrete width values are also observed but with a much lower count. These are attributed to dimer and tetramer Mo6S6 nanowires, but no wider nanowires were observed in this study. Our gradient-corrected DFT calculations (see Supporting Information) of the formation energies Ef shown in Figure 3 reveal a clear preference for the trimer formation as 3929
Figure 3. Energies and distances calculated from DFT. The energy of formation, Ef, with respect to the most stable trimer structure 1, the energy barrier, Eb, against mechanical separation, and the spacing between the two outermost wires for selected low-energy dimer, trimer, and tetramer structures (all energies in meV/at, distances in nm).
depicted in Figure 1b. The DFT shows that the individual Mo6S6 nanowires undergo a structural deformation when the wires are combined into bundles. The deformation energy of a central Mo6S6 wire in, e.g., a trimer or tetramer, is at least 0.24 eV per Mo6S6 unit, higher than the value of the two outer Mo6S6 wires, because a central wire has to form two contacts and its geometry changes accordingly. Wider Mo6S6 bundles such as pentamers are thus expected to display an even higher formation energy, which explains why we do not observe wider bundles. In addition, Eb, the energy required for cleavage into unrelaxed free wires shows that the wires are mechanically stable. The DFT calculations of trimer structure models predict that a slightly larger spacing of 1.2 nm between the two outermost wires is preferred in the most stable trimer arrangement (structure 1 in Figure 3) compared to the experimental distance of 1.0 nm (Figure 2c). However, the DFT calculations partly neglect interwire dispersion forces, which leads to exaggerated distances in the bulk compound.18 The other low-energy trimer structures selected from a large number of model structures do not interlock to the same extent as structure 1, and accordingly they exhibit even larger equilibrium spacings of up to 1.4 nm (see Figure 3). Also, for the dimers and tetramers, the calculated nanowire spacings are slightly exaggerated compared to the experimental distances of 0.65-0.69 nm for the dimer and of 1.8-2.0 nm for the tetramer models. From the interplay of atomic resolution STM images and theoretical DFT calculations, we reveal the exact atomicscale structure of the Mo6S6 trimer nanowires. The basic unit of the nanowire bundle observed by STM is the single Mo6S6 nanowire depicted in Figure 1e. The unit cell consists of an octahedral Mo6 unit dressed by an exterior cap of six S atoms each coordinated to three Mo atoms (Figure 1c). The Mo6S6 nanowire is created by a face-sharing repetition of these units, forming a backbone of Mo3 triangles arranged with the base up or down in an alternating fashion as depicted in Figure 1d. The distance between Mo6 units measured along a single 3930
wire is calculated by DFT to be 0.44 nm, which is in perfect agreement with the periodicity determined from the STM line scans in Figure 2b. Across the nanowire trimer, the arrangement of protrusions on adjacent wires are shifted 0.22 nm along the wire, which suggests that adjacent wires are rotated 60° with respect to each other and that the two outermost wires have the same orientation. The structural model displayed in Figure 2d illustrates the calculated equilibrium configuration of the wire corresponding to the trimer structure 1 in Figure 3. The top part of the image shows an STM simulation superimposed on the structural model. For the simulation, the width of the trimer was fixed to match the experimental geometry with a spacing of 1.0 nm between the two outermost wires. The simulated STM image perfectly reproduces the shifted geometry observed on the nanowire in the experimental STM image. We have also simulated STM images (not shown) for other trimer configurations (2-4 in Figure 3), obtained as stable local minima from DFT calculations. These simulations do not, however, in any way resemble the experimental STM images. The STM simulation reveals that the bright protrusions separated by 0.44 nm in the STM image reflect the position of the sulfur atoms exposed at the top of the nanowires. In the STM line scan across the wire (Figure 2c), slightly weaker protrusions with an interatomic distance of 0.26 nm are observed, and from the comparison with the simulated line scan in Figure 2f, these features are associated with lower lying sulfur atoms in the middle region of the trimer nanowire, thus confirming the orientation of the wires. To assess the potential of Mo6S6 nanowires for nanoelectronic applications, we explored the conductivity properties of the nanowires by scanning tunneling spectroscopy (STS) studies because STS allows us to probe the conductivity of the nanowires directly at the atomic scale by measuring the tunneling current (I) as a function of bias voltage (V) at a well-defined position of the nanowires. The measurement was performed at room temperature by disabling the scanning and the current feedback loop and subsequently measuring the current (I) as a function of the bias voltage (V) applied to the sample at a specific position on the nanowire (Figure 4). We investigated the uniformity of the tip states by recording STS spectra on the plain HOPG surface prior to the STS measurements on the Mo6S6 wires and concluded that tip states only had a minor influence on the recorded STS spectra, i.e., only changed the intensity of the observed features but not their position. The final output of a STS measurement is a normalized conductance curve, (dI/dV)/ (I/V), which can be shown to reflect the local density of states (LDOS) distribution around the Fermi level in a first-order (tip-independent) approximation.19,20 Using STS, we explored the electronic structure of the Mo6S6 nanowires. The normalized STS conductance curve measured on the Mo6S6 nanowires displays a number of sharp peaks as seen in Figure 4a. Such distinct peaks in the DOS spectra are due to van Hove singularities (VHS) in the density of states as expected for a 1D nanowire system and have previously been observed for, e.g., carbon nanotubes.21 The fine structure in the experimental STS curves is well Nano Lett., Vol. 8, No. 11, 2008
wires is a matter of adding additional Mo6S6 units in a repeated manner, we expect that creation of longer wires is possible by modifying the synthesis parameters or the dispersion of the defect-based nucleation sites. Furthermore, the sulfur termination of the wires may facilitate a strong and well-defined atomic bonding to metallic electrodes,25 which has previously proven to be crucial for a well-defined transport behavior in nanoscale electronic contacts. Acknowledgment. We gratefully acknowledge fruitful discussions with Henrik Topsøe, Bjerne Clausen, and Philip Hoffman. The experimental work was supported by a grant from the Danish Strategic Research Council to iNANO and financial support from Haldor Topsøe A/S. J.V.L. acknowledges a grant from the Carlsberg Foundation.
Figure 4. Electronic structure of Mo6S6 nanowires. (a) Experimental normalized conductance curve (red) measured by STS at the position indicated in the STM image. (b) Density of states (DOS) plot (blue) of the trimer structure (insert), obtained from DFT calculations.
reproduced in the density of states (DOS) plot in Figure 4b, resulting from the DFT calculations of the trimer configuration (1) with the distance between the two outer wires fixed to the experimental value. With respect to the band structure of a single Mo6S6 wire,22 the interwire interaction in the trimer introduces additional splitting of the signals in the vicinity of the Fermi level due to a weak elliptic elongation of the wires by 0.01-0.02 nm parallel to the surface normal. These splittings considerably improve the match with the measured STS spectrum beyond a merely qualitative comparison. Thus, the present DFT calculations allow us to assign the origin of the different characteristic features observed in the experimentally observed STS spectra. The filled electronic bands just below the Fermi level are dominated by Mo d states with a small contribution from sulfur p states, whereas the conduction band above the Fermi level almost exclusively originates from Mo d states. Thus, the observed shift of the Fermi level in the experimental normalized conductance curve (dI/dV)/(I/V) compared to the theoretical DOS curve (∆EF ) 0.7 eV) is a result of charge transfer from the substrate to mainly the Mo d states. However, both the experimental normalized conductance results and the theoretical DOS curve exhibit nonzero values at and around the Fermi level (EF), proving that the Mo6S6 nanowire trimers are indeed one-dimensional electric conductors. We thus conclude that Mo6S6 nanowires may be very well suited as nanoscale metallic leads with a conductivity originating from Mo d states that create the metallic backbone of the Mo6S6 nanowires. The Mo6S6 nanowires thus possess uniquely welldefined structural and electronic properties and may thus have a distinct advantage as compared to the structurally much less well-defined 1D carbon nanostructures such as nanotubes21,23 or the more recently discovered graphene nanoribbons.24 The Mo6S6 wires presented in this study are, on the other hand, considerably shorter (10-20 nm) than, e.g., carbon nanotubes. However, because the formation of longer Nano Lett., Vol. 8, No. 11, 2008
Supporting Information Available: Methods. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Wilson, J. A.; Yoffe, A. D. AdV. Phys. 1969, 18, 193–335. (2) Feldman, Y.; Wasserman, E.; Srolovitz, D. J.; Tenne, R. Science 1995, 267, 222–225. (3) Yang, T.; Okano, S.; Berber, S.; Tomanek, D. Phys. ReV. Lett. 2006, 96, 125502. (4) Seifert, G.; Terrones, H.; Terrones, M.; Jungnickel, G.; Frauenheim, T. Phys. ReV. Lett. 2000, 85, 146–149. (5) Chianelli, R. R.; Prestbridge, E. B.; Pecoraro, T. A.; DeNeufville, J. P. Science 1979, 203, 1105–1107. (6) Remskar, M.; Mrzel, A.; Skraba, Z.; Jesih, A.; Ceh, M.; Demsar, J.; Stadelmann, P.; Levy, F.; Mihailovic, D. Science 2001, 292, 479– 481. (7) Popov, I.; Yang, T.; Berber, S.; Seifert, G.; Tomanek, D. Phys. ReV. Lett. 2007, 99, 085503. (8) Lauritsen, J. V.; Kibsgaard, J.; Helveg, S.; Topsoe, H.; Clausen, B. S.; Lægsgaard, E.; Besenbacher, F. Nat. Nanotechnol. 2007, 2, 53–58. (9) Wilcoxon, J. P.; Newcomer, P. P.; Samara, G. A. J. Appl. Phys. 1997, 81, 7934–7944. (10) Zhang, J.; Soon, J. M.; Loh, K. P.; Yin, J. H.; Ding, J.; Sullivian, M. B.; Wu, P. Nano Lett. 2007, 7, 2370–2376. (11) Meden, A.; Kodre, A.; Gomilsek, J. P.; Arcon, I.; Vilfan, I.; Vrbanic, D.; Mrzel, A.; Mihailovic, D. Nanotechnology 2005, 16, 1578–1583. (12) Vilfan, I. Eur. Phys. J. B 2006, 51, 277–284. (13) Murugan, P.; Kumar, V.; Kawazoe, Y.; Ota, N. Nano Lett. 2007, 7, 2214–2219. (14) Potel, M.; Chevrel, R.; Sergent, M. J. Solid State Chem. 1980, 35, 286–290. (15) Venkataraman, L.; Lieber, C. M. Phys. ReV. Lett. 1999, 83, 5334– 5337. (16) Helveg, S.; Lauritsen, J. V.; Lægsgaard, E.; Stensgaard, I.; Nørskov, J. K.; Clausen, B. S.; Topsøe, H.; Besenbacher, F. Phys. ReV. Lett. 2000, 84, 951–954. (17) Kibsgaard, J.; Lauritsen, J. V.; Lægsgaard, E.; Clausen, B. S.; Topsøe, H.; Besenbacher, F. J. Am. Chem. Soc. 2006, 128, 13950–13958. (18) Rydberg, H.; Dion, M.; Jacobson, N.; Schröder, E.; Hyldgaard, P.; Simak, S. I.; Langreth, D. C.; Lundqvist, B. I. Phys. ReV. Lett. 2003, 91, 126402. (19) Stroscio, J. A.; Feenstra, R. M.; Fein, A. P. Phys. ReV. Lett. 1986, 57, 2579–2582. (20) Lang, N. D. Phys. ReV. B 1986, 34, 5947–5950. (21) Wildoer, J. W. G.; Venema, L. C.; Rinzler, A. G.; Smalley, R. E.; Dekker, C. Nature 1998, 391, 59–62. (22) Gemming, S.; Seifert, G.; Vilfan, I. Phys. Status Solidi B 2006, 243, 3320–3324. (23) Odom, T. W.; Huang, J. L.; Kim, P.; Lieber, C. M. Nature 1998, 391, 62–64. (24) Li, X. L.; Wang, X. R.; Zhang, L.; Lee, S. W.; Dai, H. J. Science 2008, 319, 1229–1232. (25) Ploscaru, M. I.; Kokalj, S. J.; Uplaznik, M.; Vengust, D.; Turk, D.; Mrzel, A.; Mihailovic, D. Nano Lett. 2007, 7, 1445–1448.
NL802384N 3931
