The MoS2 Nanotubes with Defect-Controlled Electric Properties

We describe a two-step synthesis of pure multiwall MoS2 nanotubes with a high degree of homogeneity in size. The Mo6S4I6 nanowires grown directly from elements under temperature gradient conditions in hedgehog-like assemblies were used as precursor material. Transformation in argon-H2S/H2 mixture leads to the MoS2 nanotubes still grouped in hedgehog-like morphology. The described method enables a large-scale production of MoS2 nanotubes and their size control. X-ray diffraction, optical absorption and Raman spectroscopy, scanning electron microscopy with wave dispersive analysis, and transmission electron microscopy were used to characterize the starting Mo6S4I6 nanowires and the MoS2 nanotubes. The unit cell parameters of the Mo6S4I6 phase are proposed. Blue shift in optical absorbance and metallic behavior of MoS2 nanotubes in two-probe measurement are explained by a high defect concentration.


Introduction
One-dimensional nanostructures such as nanorods, nanobelts, nanowires and nanotubes have attracted intensive attention due to their unique applications in mesoscopic physics and nanoscale devices. In analogy to graphite, nanoparticles of many two-dimensional inorganic compounds are unstable against folding and can form closed cage structures which are referred to as inorganic fullerene-like particles and inorganic nanotubes [1]. Recent discovery of MoS 2 nanopods called "mama"-tubes [2] with MoS 2 fullerene-like particles in-situ grown in a confined geometry of MoS 2 nanotubes and coaxial MoS 2 nanotube hybrids [3] have opened a new way of synthesis of MoS 2 nanotubes from Mo x S y I z ternary compounds, which allows for the production of mass quantities of nanomaterials. Weak van der Waals interactions among MoS 2 molecular layers enable low-strength shearing and several possible stackings [2,4].
Molybdenum disulfide nanostructures are receiving considerable attention because of their potential applications as heterogeneous catalysts for desulfurization processes [5], hydrogen evolution [6,7], and as materials for thermoelectric applications [8]. MoS 2 microplatelets have been used as a solid lubricant or as an additive in oil or grease for more than 60 years. Cage-like nanostructures, e.g. cylindrical MoS 2 nanotubes, represent a new generation of lubricants with extremely low friction resulting from the size, small enough to turn microvoids and nanovoids of the objects in mechanical contact into lubricant reservoirs, and by the curved geometry of the nanoparticles, which put them into constantly parallel orientation with the counterpart surfaces. The orientation relationship has been proposed to explain the ultralow friction measured on thin films composed of MoS 2 fullerene-like particles even at high humidity [9]. Lubrication is strongly related to electronic properties, more precisely to the filling of Mo d z 2 molecular orbitals [10]. Control of nanotube dimensions and the density of defects that influence transport properties are of great importance in the construction of nanoscale electronic devices and multifunctional materials.
The significantly lower molecular mass of MoS 2 in comparison with WS 2 is an advantage for many applications, although MoS 2 nanotubes are found to be more difficult to fabricate. Several different growth techniques are used for the synthesis of multiwall MoS 2 inorganic nanotubes, like sulfurization of molybdenium oxides [11,12] and chlorides [13] in a stream of H 2 S gas, thermal decomposition of (NH 4 ) 2 MoS 4 [14] and by the template method [15], hydrothermal synthesis [16] and chemical transport reaction using iodine as a transport agent [17]. Currently, most of the described techniques are not suitable for large-scale production of pure multiwall MoS 2 nanotubes, which would possess a relatively uniform size.
In the present paper, we report on a synthesis that can be scaled up for bulk production of pure multiwall MoS 2 nanotubes of lengths up to several tens micrometers and diameters up to 100 nm using groups of Mo 6 S 4 I 6 nanowires as precursor crystals. The structural data are combined with optical absorbance and Raman scattering. In addition, two-probe current-voltage measurements were performed on a single nanotube.

Sample Preparation
The Mo 6 S 4 I 6 nanowires were fabricated in evacuated (10 -4 Pa) quartz ampoules directly from molybdenum and sulfur powder (Aldrich, 99.98 %), and iodine flakes (99.999%, Aldrich) in a molar ratio of 6: 3: 9. The iodine was used as the transport agent in the chemical transport reaction, which took place in a two-zone horizontal furnace for 48 h under a temperature gradient of 5.5 K/cm. A fraction (5-10 wt.%) of the total synthesized material was transported to the low-temperature zone (1010 K) of the ampoule and grew in the form of a hairy foil composed of Mo 6 S x I y nanowires with some traces of MoI 2 at the interface with the quartz, while the material remaining at the hot zone (1123 K) appeared as a dark-brown powder. The stoichiometry of this remaining powder in the form of Mo 6 S 4 I 6 nanowires was determined by wave dispersive analysis using a scanning electron microscope, Jeol JSM 6500F. These nanowires were used as the precursor material for transformation into MoS 2 nanotubes by annealing at 1073 K in the reactive gas composed of 98 vol% of Ar, 1 vol% of H 2 S, and 1 vol% of H 2 for 1 h. In a typical experiment, around 600 mg of the starting material was sulfurized and transformed into MoS 2 nanotubes. The total mass of the starting material during the transformation was decreased for 40% due to the complete removal of iodine and its substitution by the lighter sulfur. X-ray powder diffraction and X-ray energy dispersive analysis of the end product reveal the iodine-free MoS 2 compound.

Characterization
The Mo 6 S 4 I 6 precursor crystals and MoS 2 nanotubes have been studied by high-resolution 200 keV Jeol 2010 F field-emission transmission electron microscopes (HRTEM) and scanning electron microscope FE-SEM, Supra 35 VP, Carl Zeiss. X-ray diffraction (XRD), optical absorption, Raman spectroscopy, and wave dispersive analysis (WDS) were used to characterize the obtained materials. X-ray diffraction (XRD) was performed at room temperature with a D4 Endeavor diffractometer (Bruker AXS) using quartz monochromator Cu Kα1 radiation source (l = 0.1541 nm) and Sol-X energy dispersive detector. Angular range 2 θ was chosen from 6°t o 73°with a step size of 0.04°and a collection time of 4 s. The samples were rotated during measurements at 6 rpm. Raman spectra were recorded in a micro-Raman 180°backscattering configuration on a Labram HR spectrometer with a spectral resolution of 1.5 cm -1 determined by the width of 3 CCD-pixels. For excitation, a frequency-doubled Nd:YAG 532 nm laser operated with 100 μW power on the sample was used. Under these conditions, heating or degradation effects were excluded. Transport properties were measured using an Agilent 4155 semiconductor parameter analyzer using on-wafer probing of two-terminal test structures.

Results and Discussion
The Mo 6 S 4 I 6 Nanowires Mo 6 S 4 I 6 nanowires grew as hedgehog-like selfassemblies ( Figure 1a) composed of nanowires of very homogeneous size, up to 100 nm in diameter and up to 20 μm in length. Considering that little information is available about this phase with no unit cell determined [18], we describe the direction of growth and assignment of the diffraction pattern in accordance with the similar Mo 6 S 2 I 8 phase [19]. We find close similarities of electron and X-ray diffraction patterns of both phases, which generalize the report [20] on the stability of the Mo 6 S 9-x I x nanowires (4.5 < x < 6) with different S and I stoichiometries, to the Mo 6 S 4 I 6 phase. Nanowires of different stoichiometries grow in skeletal structures composed of one-dimensional polymer chains of Mo 6chalcogen-halogen clusters, which differ only in the site occupation by sulfur and iodine. This makes difficulties in the determination of a particular phase, especially based on X-ray results. In our studies, we used electron diffraction obtained on a single nanowire for the elucidation of the symmetry rules, X-ray diffraction for the determination of interlayer distance with sufficient accuracy, and wave dispersive analysis for the determination of the stoichiometry on a single nanowire. Due to a mixed range of selective area diffraction, one cannot exclude the presence of other Mo 6 S 9-x I x and Mo 6 S 10-x I x nanowires in the starting materials, like Mo 6 S 3 I 6 or Mo 6 S 2 I 8 [20]. Nevertheless, most of the starting materials can be attributed to one phase, i.e. Mo 6 S 4 I 6 , while the others incorporate impurities that cause broadening of the X-ray peaks.
The Mo 6 S 4 I 6 nanowires possess a high aspect ratio and grow in a longitudinal direction along the [001]. The needles are rigid and well crystallized (Figure 1b). One-dimensional chains are mutually ordered and in contrast with reported Mo 6 S 3 I 6 nanowires do not exhibit a tendency for easy splitting. A stacking fault marked in Figure 1c with the component of the Burger's vector perpendicular to the nanowires axis can contribute to the resistance of the needles against longitudinal cleavage and decreases a strong anisotropy of these quasi one-dimensional cluster compounds.
The electron diffraction pattern of a single Mo 6 S 4 I 6 nanowire (Figure 1d) is assigned in accordance with the proposed space group P6 3 /m and estimated lattice parameters of a hexagonal structure with: a = 1.88(5) nm and c = 1.18 nm. The nanowires grow with the [001] axis along their longitudinal direction.
Besides Mo 6 S 4 I 6 nanowires, X-ray investigation of the starting material (Figure 2a-A) reveals the presence of the Mo 6 S 2 I 8 and traces of MoS 2 . The (002) MoS 2 peak is shown by an asterisk in spectrum (a), while other MoS 2 peaks cannot be resolved. Due to nearly identical skeletal structures, most of the diffraction peaks of Mo 6 S 4 I 6 and Mo 6 S 2 I 8 nearly match, leading to a broadening of the peaks in addition to the size effect broadening. As an example, the peak at~.817 nm is composed of two peaks situated at 0.816(1) nm and at 0.825(8) nm. The last one can be associated with the Mo 6 S 2 I 8 (110) planes, while the first one was used for the approximation of the unknown unit cell of Mo 6 S 4 I 6 nanowires. The symmetry of the TED pattern (Figure 1d) was utilized for assignation of the peak to the (200) planes of the Mo 6 S 4 I 6 . The peak positioned at 0.197 nm was associated with the (006) planes. Complete determination of the Mo 6 S 4 I 6 structure needs further studies.

Transformation of the Mo 6 S 4 I 6 Nanowires into MoS 2 Nanotubes
The X-ray powder diffraction of the product after the transformation (Figure 2a ( Figure 2b-A). The only exception is the relative intensity of the peak at 117 cm -1 , which is higher in the Mo 6 S 4 I 6 compared to the Mo 6 S 2 I 8 . Some traces of MoS 2 could also be observed in some spectra, marked by arrows. The spectroscopic similarity is a further support for the nearly identical crystal and local structure of Mo 6 S 4 I 6 and Mo 6 S 2 I 8 . The Raman spectrum of the final product-MoS 2 nanotubes (Figure 2b-C) contains the usual signature of MoS 2 , A 1g mode at 409 cm -1 and E 2g mode at 384 cm -1 [21]. No traces of precursor Mo 6 S 4 I 6 nanowires are observed. A small peak at the 287 cm -1 is attributed to the MoS 2 , E 1g mode that is forbidden in the backscattering experiments on the basal (001) plane. The reason for E 1g occurrence is the cylindrical geometry with a great part of the basal planes oriented in parallel with the axis of illumination. The line widths of the MoS 2 nanotubes are larger than that of MoS 2 single crystals. The FWHM of the A 1g line increases from 2.0 to 3.3 cm -1 and that of the E 2g from 1.6 to 3.8 cm -1 . The line broadening is attributed to a smaller crystallite size and to a larger amount of defects in multiwall MoS 2 nanotubes compared to MoS 2 single crystals [22,23].
The electronic band structure calculations show that the bulk MoS 2 is an indirect gap semiconductor with two exciton absorption bands at the absorption edge [24]. The absorption associated with the direct band gap is located in the visible spectrum around 700 nm and results from a direct transition at the K point. Two peaks assigned as A1 (690 nm) and B1 (620 nm) are attributed to two excitons of the Rydberg series. The band at around 500 nm is associated with the direct transitions from the valence band to the conduction band [25].
UV-Vis absorption spectroscopy was performed at room temperature. The MoS 2 powder, Mo 6 S 4 I 6 nanowires, and MoS 2 nanotubes were ultra-sonicated in ethanol. MoS 2 powder (Aldrich) was used as a reference material. The UV-Vis absorption spectrum (Figure 2c) of the Mo 6 S 4 I 6 nanowires (A) reveals two broad peaks at 748 nm (1.66 eV) and 487 nm (2.55 eV). The absorption spectrum of the MoS 2 nanotubes (B) reveals the A1 peak occurring at 681 nm and B1 at 631 nm with energy separation of 0.14 eV. The third peak dominates the spectrum, and it is blue-shifted with respect to MoS 2 . It is composed of a peak at 472 nm (2.63 eV) and its shoulder at 416 nm (2.99 eV). Comparison with the spectrum obtained by dispersed MoS 2 powder in the form of platelets (C) reveals two main differences in the MoS 2 nanotube sample: (i) a decrease in energy separation between A1 and B1 associated with the spin-orbit splitting of the top of the valence gap at the K point [20] and (ii) a blue shift, relative intensity and the shape of the absorption peak centered at the 472 nm, revealing changes in the direct transitions from the valence band to the conduction band. Besides these differences, the spectrum of MoS 2 nanotubes matches well with the spectrum belonging to the dispersed MoS 2 polycrystalline sample.

Morphology of MoS 2 Nanotubes
The MoS 2 multiwall nanotubes gained by the sulfurization after complete iodine removal keep the shape of the precursor Mo 6 S 4 I 6 nanowires (Figure 3a). The nanotubes are still organized in hedgehog-like groups (Figure 3b). The tube walls with a typical thickness below 10 nm are strongly defected (Figure 3c). The lattice defects condensate forming faceting of the dome closure or disorder areas near stacking faults of MoS 2 layers. Some inner cylinders can be terminated by curved players of outer cylinders (Figure 3d) revealing a dominant tendency of self-termination of surface molecular layers, while the inner ones did not gain sufficient energy for a closure.

Transport Measurement on a Single MoS 2 Nanotube
Two-terminal device was fabricated from MoS 2 nanotubes to measure transport properties (Figure 4a). The tubes were placed on a thermally-oxidized, heavilydoped p-Si substrate with a 90-nm-thick SiO 2 overlayer. Titanium/gold contacts 20/220 nm in respective thickness were formed by electron beam evaporation using a photolithographic lift-off process with separations between the electrodes of 5 μm. Two-terminal measurements on a MoS 2 nanotube, 105 nm in diameter, show a metallic (approximately ohmic) conduction. No photoconductivity was observed between measurements taken in the dark and under strong microscope illumination using a halogen lamp (Figure 4b). In addition, the wire conductivity could not be modulated independently on light illumination when backgated through the p + Si substrate in the range from -50 to +50 V with a tube bias ranging from 0 to 5 V, consistent with the metallic character of the nanotube. The nanotube conductivity is estimated to about 2 mS/cm by assuming that the wall thickness is 10 nm, which is the most frequently Figure 3 The MoS 2 nanotubes: a A typical TEM image of a bunch of nanotubes grown from sacrificed Mo 6 S 4 I 6 nanowires keeping their outer geometry; and b self-organization in hedgehock-like groups revealed by SEM; c HRTEM image of a nanotube's wall with resolved MoS 2 molecular layers, stacking faults and atomically rough surface; d TEM image of a single nanotube cap of a nanotube with inner cylinders terminated by curved layers of outer cylinders. observed value in these nanotubes. The metallic conductivity could indicate a high density of defects, which form energy states near the Fermi level and influence also the optical absorbance spectra.

Conclusion
In conclusion, we have reported an easy and straightforward way of fabricating pure multiwall MoS 2 nanotubes using Mo 6 S 4 I 6 nanowires as the precursor crystals. We synthesized the nanowires from elements in a furnace gradient and proposed the unit cell of the Mo 6 S 4 I 6 compound, which was first reported in 1985. The nanotubes produced by the sulfurization process keep the outer geometry and self-assembly of the precursor nanowires. They are of relatively homogeneous size in diameter and length. The lattice structure is strongly defected causing Raman line broadening, a blue shift in visible light absorption and metallic conductivity at room temperature of this otherwise semiconducting compound. The synthesis, which can be easily scaled up, and peculiar energy level distribution in these MoS 2 nanotubes could find application in nanoelectronics and tribology.