MoS 2 nanosheet production by the direct exfoliation of molybdenite minerals from several type-localities †

Samples of the mineral molybdenite from three classic molybdenum mining localities were examined as a potential source of molybdenum disul ﬁ de (MoS 2 ) nanosheets. In all cases, ultrasonication-promoted exfoliation of these samples in N -methylpyrrolidone (NMP) was found to produce MoS 2 as dispersed nanosheets with lateral sizes in the range of 200 – 600 nm and thicknesses between 1 and 10 atomic trilayers. The MoS 2 nanosheets obtained were found to be highly crystalline and largely defect-free, but tend to contain small amounts of aggregates on their surfaces. The exfoliated MoS 2 dispersions were characterised by UV-Vis spectroscopy, Raman spectroscopy, scanning electron microscopy (SEM), (scanning) transmission electron microscopy ((S)TEM) and energy dispersive X-ray (EDX) spectroscopy. This work raises the possibility that mined, unre ﬁ ned minerals could be a source of low-dimensional MoS 2 .


Introduction
Atomically thin graphene exhibits exceptional electronic, optical and mechanical properties. 1 Since graphenes' initial discovery 2 there has been an explosion of interest 3 in other 2D materials, leading to the exploration of graphene analogues such as the family of transition metal dichalcogenides, or TMDCs (MX 2 ; e.g. M ¼ Mo, W and X ¼ S, Se, Te). TMDCs all display similar layered-graphitic superstructures. 4 These structures are comprised of a layer of transition metal atoms coordinated by two layers of close-packed chalcogen atoms, held together by strong covalent bonds. In the bulk material, these sheets are bound together by weak van der Waals forces, just as in graphite. 5 Of these 2D-TMDC materials, MoS 2 nanosheets are potentially a useful semiconducting component in a series of devices. 6,7 In addition, molybdenum disulde exhibits considerable anisotropy, giving rise to novel electronic, optical, structural and mechanical properties in its 2D form (2D-MoS 2 ). Due to these properties 2D-MoS 2 has been shown to be a suitable material for inclusion in a myriad of potential applications that includes photochemical sensors, 8,9 optoelectronics, 10,11 bioimaging, 12,13 catalysis, 14,15 solid-state lubricants, 16 and energy storage. 17,18 In many cases, laboratory-scale 2D-MoS 2 production is based on liquid exfoliation of molybdenum(IV) sulde powder. 19 The powder is rened from natural MoS 2 , molybdenite, which is mined in numerous pan-global locations and puried by comminution, otation and leaching. 20,21 In the years 2006-2011, demand has surged to 264,000 tonnes per year of molybdenum metal, an increase of 25%. 22,23 The demand of the mineral is driven by metallurgical, green-and now nanomaterial-based applications. 20 Herein we discuss the feasibility of MoS 2 nanosheet production by exfoliation of molybdenite ores, as exemplied by samples obtained from three locations (Queensland AUS [QL], New South Wales AUS [NSW] and Telemark NOR [TM]). The crystalline molybdenite samples were obtained untreated from the selected minerals (see Table S1 † for purity analysis). The potential to obtain MoS 2 nanosheet from these sources by employing the Coleman method of MoS 2 exfoliation is detailed in this report. 24,25 The nanomaterials produced have been subject to a comparison of exfoliation efficiency, as well as compositional and structural features, relative to the nanosheets produced from MoS 2 powder documented in the literature. 24 Results and discussion 50 mL solutions of N-methylpyrrolidone (NMP) containing akes of the selected molybdenite (375 mg, 7.5 mg mL À1 concentrations) were bath ultrasonicated for 48 hours, with aliquots obtained at various time intervals. Unexfoliated material was removed from the aliquots by centrifugation (45 minutes at 1500 rpm). UV-visible spectra of the supernatants showed two peaks with maxima at 628 and 674 nm, attributed to the characteristic A and B direct excitonic transitions of MoS 2 respectively (Fig. 1, S1a and S2a †). 24 The UV-Vis data was modied to remove the background noise associated with the scattering exponent of the MoS 2 nanosheets. 24 The intensity of the feature at ca. 674 nm was then used as a spectroscopic handle to estimate the concentrations of the MoS 2 dispersions produced. 25 The nal concentrations of the exfoliates were found to be 0.57, 0.42 and 0.84 mg mL À1 , from QL, NSW and TM precursors, respectively. These values obtained are considerably lower than the concentration of 1.79 mg mL À1 produced from the exfoliation of bulk-MoS 2 powder, which was run alongside the molybdenite samples. In all cases, plots of concentration versus sonication times (insert in Fig. 1, S1b and S2b †) reveal at least two kinetic regimes of MoS 2 nanosheet production, as the exfoliation rates (k exf ) in the initial 12 hours of sonication are noticeably slower than those observed aer this period. Using the values of k exf obtained aer this initial 12 hour period, it was clear to see the selection of the concentrate results in different exfoliation rates (k exf between 9.4 and 20.7 mg mL À1 h À1 ; Table S2 †). The rates of exfoliation were also found to be signicantly lower than those observed for the exfoliation of bulk-MoS 2 powder (37.1 mg mL À1 h À1 ). Stability investigations show that there is a very small (<4%) reduction in the intensity of the A and B bands over a period of 4 weeks; indicating that the dispersions are highly stable (Fig. S3 †).
In all molybdenite sources, both large akes from the crystalline aggregates and the exfoliated nanosheets exhibit two Raman peaks that correspond to the A 1g and E 2g modes ( Fig. 2ac). These peaks are known to exhibit a well-dened dependence on layer z-thickness and thus can be used as a way of estimating nanosheet thickness. 26 All of our exfoliations gave few-layer MoS 2 nanosheets as the predominant product. Nanosheets from the TM dispersion showed the two modes separated by 19 cm À1 , similar to the peak separation observed for single-layered MoS 2 nanosheets exfoliated from MoS 2 powder. 24,27 It was expected that exfoliation of the QL and NSW concentrates also produced single-layered 2D-MoS 2 , however, these were not  detected in the Raman spectra obtained from either exfoliated sample, presumably due to the smaller lateral dimensions of the nanosheets. Statistical analysis (sample size ¼ 50, Fig. S4 †) of randomly selected akes indicates that more than 80% of the akes dispersed contain less that 6 stacked monolayers. The remaining areas were covered with thicker material whose thickness could not be determined by Raman spectropscopy. 26 Membrane-supported thin lms of MoS 2 nanosheets were prepared from the QL, NSW and TM dispersions by vacuum ltration 24,28 and imaged by scanning electron microscopy (SEM). All lms produced were found to have similar surface morphology (Fig. 2d-f and S5 †); the surface of the lms consist of partially-stacked 2D-nanosheets with lateral dimensions between 200-600 nm.
Nanosheets were deposited onto lacey carbon grids for (scanning) transmission electron microscope ((S)TEM) imaging. The resulting HAADF STEM images (Fig. 3) and TEM images (Fig. S6 †) show faceted morphologies and a range of irregular shapes. The lateral dimensions of the nanosheets ranged from 200 nm to greater than 1 mm. Selected area diffraction (SAED) patterns (Fig. S6 †) conrmed that in all samples, most akes consisted of single crystals and that all of the akes had the expected hexagonal crystal structure. Atomic resolution, highangle annular dark-eld (HAADF) STEM imaging ( Fig. 2 and S7 †) conrmed that all akes were highly crystalline and largely defect-free. The edge structure of akes observed in high magnication HAADF STEM images can be used to estimate ake thickness. 24 The majority of the akes were found to be 1-10 layers thick, although ake thicknesses varied for different areas within individual akes.
Energy-dispersive X-ray (EDX) spectroscopy measurements, performed in the SEM and STEM, were used to obtain compositional information from both the thin lms and individual akes. Quantitative analysis gave the expected Mo : S ratio of 1 : 2 in all cases. STEM-EDX mapping revealed calcium rich regions in all samples, believed to be nanometre scale calcium carbonate aggregates on the ake's surface (Fig. S8-11 †). Nanosheets prepared from the NSW source showed signicantly more calcium contamination than nanosheets made from the other sources. SEM-EDX analyses did not show any signicant quantities of calcium, but did detect other elements: all samples were found to contain non-negligible amounts of iron, in addition to bismuth found in QL and NSW thin lms

Conclusions
We have shown that untreated molybdenite mineral samples can be used as a precursor for exfoliation to produce few-layered MoS 2 nanosheets that are highly crystalline, and largely defectfree, with small amounts of aggregates on the surface of the sheets. The selection of molybdenite source from various locations worldwide appears to have little effect on the purity or crystallinity of the nanosheets produced. In addition thin lms comprised of stacked MoS 2 nanosheets have been produced using a simple ltration method. The relatively high purity of the nanosheets and thin lms produced by the methodology described herein gives hope that a cost-effective approach for the production of MoS 2 for components in electrical circuitry using relatively unrened material is possible. The electronic properties of the nanosheets produced are currently the subject of further work.

Background information
The molybdenite aggregates were obtained from selected molybdenite-containing mines: Toowong, Queensland (AUS); Kingsgate, New South Wales (AUS) and Siljord, Telemark (NOR). Compositional analysis of the aggregates were obtained by a Chemeca Camebax SX100 microsonde Electron Probe Micro-Analyser (EPMA), running a 20 nA beam at 20 keV with a 1.5 mm spot size on polished sections. The measurements were compared to elemental standards: Ni, Co, Fe, Re, W and Bi metals, MoS 2 , CuFeS 2 , ZnS and ZnSe. Molybdenum(IV) sulde powder and N-methylpyrrolidone (NMP) were purchased from Sigma-Aldrich, which were used as supplied.
The dispersions were prepared in a bench-top sonication bath (Elmasonic P 70H [HF power 820 W across four horns]).
Centrifugations were carried out using Centurion 822 series (up to 15 mL dispersions) and Jouan B4i (up to 50 mL dispersions) centrifuges. UV-Vis absorption spectroscopy was performed with an Agilent 8453 spectrometer using a 1 cm glass cuvette. Accurate dilutions of the MoS 2 dispersions in NMP were produced to obtain suitable UV-Vis spectra. The subsequent data was then analysed aer modication (implementing the Beer-Lambert law) to represent the absorption intensities at the concentrations when the aliquots were rst taken from the reaction.
Membrane-supported thin lms of MoS 2 nanosheets suitable for Scanning Electron Microscopy (SEM) were produced by the dilution of 30 mL suspensions in isopropanol (270 mL), and ltration through a polyvinylidene uoride membrane (PVDF, pore size 0.1 mm). The thin lms were washed with isopropanol (3 Â 50 mL) before drying in a vacuum chamber for 12 hours at ambient temperature. The thin lms are carbon-coated using a Gatan Model 682 Precision Etching coating system before carrying out SEM and EDX analyses by the use of Philips XL 30FEG scanning electron microscope and a DX4 detector.
Raman spectra were acquired on a Renishaw 1000 system, with a solid state 50 mW 514.5 nm laser. The laser beam was focused onto the MoS 2 sample by a 100Â objective lens. The scattered signal was detected by an air cooled CCD detector. Samples were prepared by dilution of the NMP dispersions with ethanol (1 : 40 v/v), before deposition onto a 300 nm SiO 2 /Si substrate and drying under ambient conditions over 48 hours.
Samples were prepared for transmission electron microscope (TEM) imaging by drop casting a dilute solution of the MoS 2 dispersions (1% solution in EtOH) onto holey carbon support lms which were then washed with ethanol and air dried. Flake size and shape was assessed using low magnication bright eld TEM in a Philips CM20 and a FEI Tecnai F20, both microscopes used a LaB 6 electron source and were operated at 200 kV. High angle annular dark eld (HAADF) scanning transmission electron microscope (STEM) imaging was performed using a probe side aberration corrected FEI Titan G2 operated at 200 kV. For HAADF imaging a convergence angle of 26 mrad, a HAADF inner angle of 52 mrad, and a probe current of $200 pA were employed. Energy dispersive X-ray (EDX) spectroscopy was performed in the Titan using a Super-X EDX detector with a solid angle of collection of 0.7 s rad. EDX quantication was performed using Brüker Esprit soware, where the bremsstrahlung background was subtracted, peaks deconvoluted, and quantication performed using the Cliff-Lorimer approach. Fe, Cu, Zr, and C counts were known to originate from the microscope (Fe and Zr) and support lm (Cu and C), so these elements were included in the deconvolution procedure but excluded from the nal quantication.

Purity of molybdenite aggregates
The molybdenite aggregates were analysed by Electron probe Microanalysis (EPMA) and found to be of high purities (>99.5%; Table S1 †). The trace element populations are low (below detection limits in many analysis spots), nor were there any major substitution. Bi was detected in the sample from QL where Bi 2 S 3 was an associated phase. The high standard deviations reveal the high variations. The Re and W do appear in several points in the analyses above detection and are probably real.

Synthesis of MoS 2 dispersions from the aggregates
In a 60 mL screw-cap glass jar, NMP (50 mL) was added to 375 mg of the chosen molybdenite aggregate (manually broken up into approx. 10 akes) before suspending the sealed jars in an operational ultrasonic bath for 48 hours. For concentration determination, 3 mL aliquots of the dispersions were removed at selected time intervals and diluted in NMP (3 mL) within a 15 mL centrifuge tube. Each aliquot was le to settle overnight before centrifugation at 1500 rpm for 45 minutes. Two-thirds of the resulting supernatant was pipetted to provide a suspension of MoS 2 nanosheets in NMP (note: at 50% of its initial concentration). Aer 48 hours the dispersions that remained in the reaction vessels (approx. 25 mL) were decanted into 50 mL centrifuge tubes containing 20 mL NMP and le to settle overnight. Centrifugation at 1500 rpm for 45 minutes followed by pipetting of 30 mL of the supernatant gave a suspension of MoS 2 nanosheets of suitable quantities for thin lm production.