Thin WS2 nanotubes from W18O49 nanowires

ABSTRACT Single- and double-walled WS2 nanotubes of small diameter (<10 nm) and few layers (<4) have been synthesized by heating ultrathin W18O49 nanowires and S powders in an H2/Ar atmosphere at 840°C. In the process of formation and growth of a WS2 nanotube, the W18O49 phase transformed first into an amorphous WS3 structure via the absorption and diffusion of the sulfur atoms to substitute the oxygen atoms, and followed by a phase transition from amorphous WS3 to 2H-WS2 by losing sulfur atoms to form the shell of the WS2 nanotube. GRAPHICAL ABSTRACT IMPACT STATEMENT Synthesis of thin WS2 nanotubes with single- and doubled-walled structure from ultrathin W18O49 nanowires is developed and their structure and growth are studied using electron diffraction and microscopy.


Introduction
Following the discovery of fullerenes [1] and carbon nanotubes [2], it was soon realized that other inorganic compounds, having layered structures similar to that of graphite, could also form fullerene-like and nanotube structures. Metal dichalcogenides (MX 2 , M = Mo, W, Nb, Hf; X = S, Se) are the typical example materials. Fullerene-like WS 2 nanoparticles and WS 2 nanotubes were first reported by Tenne et al. in 1992 [3]. Attention has also been drawn to other inorganic nanotubes since then owing to their interesting properties. Substantial progress has been made and demonstrated that MoS 2 and WS 2 nanotubes have great potential for applications in scanning probe microscopy [4,5], lubrication [6], catalysis [7,8], field-effect transistors [9], nanocomposites [10], and energy storage [11,12].
Several experimental methods have been explored to synthesize WS 2 nanotubes, including gas-solid reactions CONTACT Lu-Chang Qin lcqin@unc.edu Department of Physics and Astronomy, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA [3,[13][14][15][16][17][18][19][20], chemical transport [21,22], irradiation activation [23][24][25], sulphurization of tungsten film [26], and the solution route [27]. Most of the synthetic methods are based on high-temperature gas-solid reactions, where hydrogen and hydrogen sulfide or sulfur and metal oxide precursors were used as the reactants [20]. However, a common feature of these WS 2 nanotubes is that they all have relatively large diameters and always have more than five layers, resulting in a low-specific surface area. Theoretical calculations using the density functional tight-binding theory have shown that MoS 2 nanotubes with an outer diameter larger than about 6 nm are more stable than nanostrips of comparable width [28]. Therefore, the threshold for the outer diameter of single-to four-walled nanotubes is 6.45, 7.13, 7.61, and 8.59 nm, respectively. WS 2 nanotubes are structurally analogous to MoS 2 nanotubes. Hence, the synthesis of WS 2 nanotubes of small diameter (<10 nm) and few shells (<4) has been considered a great experimental challenge. Whitby et al. [29] reported use of multi-walled carbon nanotubes as templates for synthesis of single-walled WS 2 nanotubes. As for template-free methods, it was concluded that the generation of such thin nanotubes requires highly exergonic conditions to drive the reactions into windows of stability far enough from equilibrium, such as using laser ablation [25]. However, it was also suggested that these thin nanotubes were not obtainable by the conventional thermally driven process (<1000°C), which is not sufficiently exergonic.
In this work, we use ultrathin W 18 O 49 nanowires as the precursor to grow thin WS 2 nanotubes. The major advantage of our method is that the ultrathin W 18 O 49 nanowires would serve directly as the templates for the formation and growth of WS 2 nanotubes. Since the W 18 O 49 phase is non-volatile, the produced WS 2 nanotubes would retain the morphology of their template and a diameter below 10 nm would be reachable. As a result, both single-and few-walled WS 2 nanotubes have been obtained successfully using this method, where the W 18 O 49 nanowires were reduced by sulfur powders in the reducing H 2 /Ar atmosphere at about 840°C. It is therefore demonstrated that thin WS 2 nanotubes are obtainable by the conventional thermally driven process of synthesis. The structure of and growth mechanism for the thin WS 2 thin nanotubes are also studied.

Experimental
We obtained WS 2 nanotubes in a two-step process. In the first step, ultrathin W 18 O 49 nanowires were prepared. In a typical procedure [30], 25 mg of WCl 6 was dissolved in 20 ml of ethanol and subsequently transferred to a Teflonlined stainless steel autoclave heated to 180°C for 24 h. The obtained solution was then centrifuged and purified with ethanol for five times before the precipitates were collected. In the second step, 2 mg of W 18 O 49 nanowires were first dispersed in 10 ml of ethanol. A quartz burette was then dipped into the ethanol solution to collect the W 18 O 49 nanowires for reactions with sulfur powders at 840°C in a furnace for 3 h to allow the formation and growth of WS 2 nanotubes.
Transmission electron microscopy (TEM) images and electron diffraction patterns (EDPs) were obtained with Tecnai F20 operated at an accelerating voltage of 200 kV. Nano-beam electron diffraction (NBED) patterns were obtained by using a nearly-parallel nano-probe, which was generated under the microprobe mode with the usage of a smallest 30 μm condenser aperture. The diameter of a typical nano-beam electron probe is ∼ 50 nm. Figure 1(a) shows a powder X-ray diffraction (XRD) pattern of the tungsten oxide nanowires which are in the monoclinic W 18 O 49 structure (PDF#05-0392). The broadened reflections are attributed to the small crystallites. The low-magnification TEM image (Figure 1(b)) revealed that the nanowires had large aspect ratios and lengths of up to several microns. A low precursor concentration (C WCl6 ≤ 5 mg/ml) resulted in ultrathin and welldispersed nanowires [30]. Selected-area electron diffraction (SAED) patterns showed the (010) and (020) Figure 2(a) shows a HRTEM image of a single-walled WS 2 nanotube with a diameter of about 12 nm, which is larger than the threshold diameter of 6.45 nm. The additional wall (indicated with black arrow) is due to the folded edge of a nano-platelet, which was confirmed by comparing the power spectra of the two different regions (I and II). Careful examinations of the power spectra of the image reveal that this single-walled nanotube has a structure close to the zigzag configuration, which renders smaller folding energy in comparison with chiral nanotubes. Double-walled WS 2 nanotubes were also often observed in the samples. Figure 2(b) shows an HRTEM image of two double-walled WS 2 nanotubes. The outer and inner diameters of the smaller nanotube are 7.29 and 5.89 nm, respectively. The outer diameter is also beyond the threshold value of 7.13 nm for double-walled nanotube. It should be noted that the interlayer separation of this nanotube is different at the two sides with one side being 0.64 nm and the other 0.76 nm, suggesting that this nanotube has two nonconcentric shells. The interlayer separation of bulk 2H-WS 2 crystals is 0.62 nm and the interlayer separation multi-walled WS 2 nanotubes varied between 0.63 and 0.65 nm [25]. Since the outer diameter is 1.4 nm larger than the inner diameter, the closer contact of one side would lead to a larger separation of the other side. The NBED pattern (Figure 2(c)) and the power spectrum (inset of Figure 2(b)) obtained from this small nanotube reveal that both shells exhibited an armchair structure. The thicker double-walled nanotube has outer and inner diameter of 11.04 and 9.67 nm, respectively. The interlayer distance of one side is 0.64 nm, while the other is 0.73 nm. Figure 3(a) shows an HRTEM image of a triple-walled WS 2 nanotube and the inset is an EDP revealing three   the NBED pattern [31][32][33][34]. Figure 3(b) displays a magnified portion of the diffraction layer lines where three groups of layer-lines are classified, from which the helical angles for these three shells are 19.9°(inner shell), 13.2°(middle shell), and 7.4°(outer shell), respectively, indicating that this nanotube consisted of three shells of different helicity. Since the electron scattering intensities due to each shell are proportional to the diameter of nanotube, the integrated intensities of each layer line can also be used for the identification of the contributing shell. As displayed in Figure 3(b), the layer lines in Group A have the highest reflection intensities and group C has the lowest. Hence, the final assignment of helical angle for outer, middle, and inner shell is 7.4°, 13.2°, and 19.9°, respectively. It is also noted that the helical angles have a wide distribution, indicating that there was no strong correlation in the growth of respective shells.

Results and discussion
The sulphurization of thicker W 18 O 49 nanowires would usually result in multi-walled WS 2 nanotubes of more than five shells. Figure 3(c) shows an HRTEM image and NBED pattern (inset) of a quintuple-walled WS 2 nanotube. The outermost and innermost diameter is 13.72 and 8.34 nm, respectively. The helical angle of this nanotube is in 4.1°-5.5°. Figure 3(d) shows an HRTEM image and NBED pattern of another thick WS 2 nanotube (>10 shells), which has the outermost and innermost diameter of 20.39 and 7.93 nm, respectively, with an average inter-shell spacing of 0.62 nm. The helical angle is in a range of 15.9°and 20.5°. Similar features have also been observed in other multi-walled WS 2 nanotubes in this study. The inter-shell spacing varied between 0.62 and 0.65 nm and the smallest innermost diameter was even below 5 nm. Since the inter-shell spacing is close to the corresponding lattice spacing of the bulk 2H-WS 2 phase and the chiral angle has a relatively narrow range, it seems natural that the synthesized shell formed earlier would serve as a template for the formation and growth of fresh shells [35].
To gain insight into the growth mechanism for the WS 2 nanotubes, the morphology and structure of one incipient nanotube was studied. Figure 4 shows the TEM images of a nanotube with diameter of 40 nm, where the oxide was encapsulated by sulfide shells. The TEM images (Figure 4(a,c)) revealed that this nanotube had one end open and the other closed and it was covered with two to three layers of WS 2 with d-spacing of 0.62 nm, in conformation with the (002) spacing of the 2H-WS 2 structure. The sulphurization should have started from outside and from the ends to the middle of the nanowhisker, which is different from the observations of a previous report [20]. The space between the grown WS 2 shells provided a diffusion channel for S atoms for further sulphurization. The SAED pattern (Figure 4(b)) is a superposition of the Bragg reflections from both the sulfide shells and the oxide core. The (010) and (004) reflections (indexed yellow) are due to the W 18 O 49 phase, while the (100) and (010) reflections (indexed white) pertain to the WS 2 phase. The diffraction pattern indicates that the oxide was single-crystalline. It should also be noted that electron diffraction analysis indicated that all shells have the same helical angle and are of an armchair structure. Figure 4(c) shows an HRTEM image of region I. The thin area shows only the WS 2 shells as indicated by the NBED pattern (Figure 4(d)). This result is also supported by an analysis of the Fourierfiltered images (Figure 4(e,f)). The Fourier-filtered image in Figure 4(f) reveals that the nanotube exhibited an armchair structure. The axial direction of the nanotube is also indicated in Figure 4 The ultrathin W 18 O 49 nanowires usually contained many oxygen vacancies-they provided reductive sites for the conversion of carbon dioxide into methane and isopropyl alcohol into propylene and also exhibited a 'trapping' effect to enhance the capture and adsorption of the substrate material [30,36]. In our study, the oxygen vacancies contained in the W 18 O 49 nanowires also served as reductive sites and traps for catching S atoms from the atmosphere. Likewise, oxygen vacancies were not rich in the one-dimensional W 18 O 49 nanostructures with relatively larger diameters [30]. These phenomena give the explanations for why the reaction temperature required in our experiments was even below 840°C, while the needed temperature was above 1000°C in [15]. The energy for breaking the W-O bonds was also likely greater in the case when long and crystalline whiskers were used compared to the use of short semi-crystalline needle-like particles [16]. Based on these earlier studies, we believe that the reactions started from the oxygen vacancy sites and expanded into the remaining parts via the diffusion of S atoms. Since hydrogen does not reduce W 18 O 49 further [20], its role is to prevent the further oxidation of W 18 O 49 nanowires at the beginning and enhance the subsequent sulphurization.
In addition, the densities of W 18 O 49 and 2H-WS 2 are 7.71 and 7.73 g/cm 3 , respectively, and the corresponding densities of W atoms in W 18 O 49 and 2H-WS 2 are 6.23 and 5.73 g/cm 3 , respectively. If the mass of tungsten were to suffer no loss in the process of sulphurization, it cannot result in the formation of a hollow tubule with W 18 O 49 converted directly into the 2H-WS 2 structure. Since the WS 2 nanotube always contained a hollow core, it suggests that a volume expansion must have occurred in the reactions. Furthermore, two-dimensional 2H-WS 2 platelets were also observed when the W 18 O 49 nanowires were placed side by side on the substrate. Hence, we suggest that conversion of the W-O bonds into the W-S bonds would undergo initially the formation of amorphous WS 3 with inflated volume and followed by a phase transition to 2H-WS 2 with a loss of S atoms at a temperature close to 850°C according to the W-S phase diagram [37]. For ultrathin W 18 O 49 nanowires, phase transition took place after the complete sulphurization. But for thicker W 18 O 49 nanowires, sulphurization and phase transition could take place simultaneously. These are the reasons why the inner diameter of thinner nanotubes was always larger than that of thicker nanotubes. Figure 5 shows a schematic diagram illustrating the formation and growth of WS 2 nanotubes from W 18 O 49 nanowires. It is remarkable that a good dispersion of W 18 O 49 nanowires was essential to avoid tangling with each other in order to produce thin WS 2 nanotubes. Since the formation of thin WS 2 structure occurred locally, nanotubes consisting of single-to triple-walled would require ultrathin W 18 O 49 nanowires as precursors. On the other hand, a precursor subject to starting temperature above 850°C would result in the formation of fullerene-like structures.

Conclusions
A route for synthesis of few-layer WS 2 nanotubes has been developed by heating ultrathin W 18 O 49 nanowires together with S powders in an H 2 /Ar atmosphere at 840°C to demonstrate that WS 2 nanotubes with fewer than three shells could be formed by conventional thermally driven synthesis below 1000°C. Single-and doublewalled WS 2 nanotubes exhibited close to either zigzag or armchair structure, while multiple-walled WS 2 nanotubes showed a chiral structure with dispersed helical angles. The study of one incipient nanotube indicates that the conversion of a W 18 O 49 nanowire into a WS 2 nanotube undergoes the following major steps: (i) S atoms are attracted preferentially to the O vacancies at the surface of the W 18 O 49 nanowire, and subsequently diffuse to other sites to substitute O atoms; (ii) The W 18 O 49 phase transforms gradually to amorphous WS 3 , accompanied by a volume expansion; and (iii) Phase transition from amorphous WS 3 to crystalline 2H-WS 2 by losing S atoms, which forms the shells of the WS 2 nanotube.

Disclosure statement
No potential conflict of interest was reported by the authors.