Layer-Controlled Chemical Vapor Deposition Growth of MoS2 Vertical Heterostructures via van der Waals Epitaxy Leith Samad,† Sage M. Bladow,† Qi Ding,† Junqiao Zhuo,† Robert M. Jacobberger,‡ Michael S. Arnold,‡ and Song Jin*,† †
Department of Chemistry, University of WisconsinMadison, 1101 University Avenue, Madison, Wisconsin 53706, United States Department of Materials Science and Engineering, University of WisconsinMadison, 1509 University Avenue, Madison, Wisconsin 53706, United States
‡
S Supporting Information *
ABSTRACT: The fascinating semiconducting and optical properties of monolayer and few-layer transition metal dichalcogenides, as exemplified by MoS2, have made them promising candidates for optoelectronic applications. Controllable growth of heterostructures based on these layered materials is critical for their successful device applications. Here, we report a direct low temperature chemical vapor deposition (CVD) synthesis of MoS2 monolayer/multilayer vertical heterostructures with layer-controlled growth on a variety of layered materials (SnS2, TaS2, and graphene) via van der Waals epitaxy. Through precise control of the partial pressures of the MoCl5 and elemental sulfur precursors, reaction temperatures, and careful tracking of the ambient humidity, we have successfully and reproducibly grown MoS2 vertical heterostructures from 1 to 6 layers over a large area. The monolayer MoS2 heterostructure was verified using cross-sectional high resolution transmission electron microscopy (HRTEM) while Raman and photoluminescence spectroscopy confirmed the layer-controlled MoS2 growth and heterostructure electronic interactions. Raman, photoluminescence, and energy dispersive X-ray spectroscopy (EDS) mappings verified the uniform coverage of the MoS2 layers. This reaction provides an ideal method for the scalable layercontrolled growth of transition metal dichalcogenide heterostructures via van der Waals epitaxy for a variety of optoelectronic applications. KEYWORDS: heterostructures, van der Waals epitaxy, molybdenum disulfide, graphene, TaS2, SnS2, chemical vapor deposition
L
applications critically depends on the direct and controlled growth of 2D heterostructures, many of which have been prototyped as mechanically exfoliated van der Waals heterostructures with improved device performances over their bulk counterparts.15,33,34 Heterostructures based on 2D van der Waals materials are potentially more adaptable than heterostructures based on conventional semiconductors as van der Waals epitaxy enables the creation of tailored heterojunctions without the limitation of traditional lattice matching.35 The additional capacity for the MX2 materials and heterostructures to have variable or alternate bandgaps36−38 and work functions therefore provides a nearly limitless ability for heterostructure design and manipulation.
ayered metal dichalcogenides, MX2 (M = Mo, W, Sn, Ta, Nb, etc.; X = S, Se, Te), encompass a wide range of atomically thin two-dimensional (2D) materials with interlayer van der Waals interactions.1,2 Bulk MX2 materials have been previously investigated for a variety of applications such as fundamental charge density wave (CDW) studies, solar energy conversion,3−5 catalysis,6−9 and Li-ion batteries10 due to the wide range of properties varying from semiconductors with bandgaps from 1.0−2.2 eV to superconductors. Moreover, the fascinating semiconducting and optical properties of monolayer and few-layer MX2,11−14 commonly fabricated from mechanically exfoliated samples15 or synthesized directly via chemical vapor deposition (CVD),16−18 have made them promising candidates for nanoelectronic applications such as transistors,18−23 spintronics and valleytronics,24−26 as well as optoelectronic applications such as photodetectors,27 lightemitting diodes (LEDs),28−31 and solar energy conversion.32 The realization of the MX2 materials for potential device © 2016 American Chemical Society
Received: May 10, 2016 Accepted: July 3, 2016 Published: July 3, 2016 7039
DOI: 10.1021/acsnano.6b03112 ACS Nano 2016, 10, 7039−7046
Article
www.acsnano.org
Article
ACS Nano
pressure at low temperatures (650 °C), which elucidated in part the complexity of heterostructure growth in the limit of high temperature.44 Taking a different approach, we have previously developed a low-temperature CVD growth process using metal halide precursors to grow bulk (>5 layer) MX2−SnS2 heterostructures.46 This heterostructure growth process demonstrated van der Waals epitaxy with oriented MX2−SnS2 layers as observed from the Moiré patterns taken in high resolution transmission electron microscopy (HRTEM), which indicates that low temperatures may enhance growth orientation for potentially superior electronic interaction across a MX2−MX2 heterojunction. However, no report to date has achieved the controllable growth of MX2 from monolayers to bulk phase thicknesses (≥6 layers) in a MX2 heterostructure. Note that both monolayer and bulk-like multilayer MX2 heterostructures have potential device applications, as the former maximize the strong interlayer electronic interactions unique to 2D materials while the latter allow sufficient variability of both electronic and optical properties (e.g., absorbance, bandgap engineering, band alignments). Although less commonly studied than other MX2 materials, SnS2 has demonstrated potential for a variety of nanoelectronic devices such as monolayer transistors50 and monolayer through bulk photodetectors.27,51 Bulk SnS2 has demonstrated conductivity of 10−5−100 Ω−1·cm−1 and carrier mobility of 15−51.5 cm2·V−1·s−1,52 whereas monolayer SnS2 has demonstrated photodetector response times of ∼5 μs27 and carrier mobility of 230 cm2·V−1·s−1.50 That MoS2 forms a Type II heterojunction with SnS2 therefore makes the pairing appealing for photovoltaics and photodetectors as well as for potentially new applications. In this work, we present a low-temperature, direct, and layercontrolled CVD growth of MoS2 monolayer/multilayer vertical heterostructures with complete and uniform coverage on a variety of layered materials via van der Waals epitaxy. The use of the more volatile metal halide precursors instead of the more traditional metal oxide precursors that have negligible partial
RESULTS AND DISCUSSION Growth of Single-Crystal SnS2 Plates as Heterostructure Substrates. We first discuss the growth of the large single-crystal SnS2 plates used for heterostructure growth, as the use of single-crystalline substrates for heterostructure formation is a crucial requirement to ensure high quality heterojunctions for electronic and optical applications. Typical chemical vapor transport (CVT) single-crystal growth process requires long reaction times (typically >1 week);52 however, we have developed a facile atmospheric-pressure CVD process for growing large single-crystal SnS2 plates using SnO2 and elemental sulfur precursors at 800 °C (Figure 1a) in 20−30 min as the first step to realizing large-scale heterostructures
Figure 1. (a) Illustration of the CVD reaction for SnS2 plate growth; (b) optical microscopy image of a typical SnS2 plate after a 20−30 min reaction at 800 °C; (c) Raman spectrum demonstrating the primary SnS2 Raman shift at 318 cm−1; (d) photograph of centimeter-scale SnS2 plates grown at 850 °C for 1 h; (e) illustration of the CVD reaction for MoS2 heterostructure growth. 7040
DOI: 10.1021/acsnano.6b03112 ACS Nano 2016, 10, 7039−7046
Article
ACS Nano
cm−1 between the E12g and A1g Raman shifts is indicative of a monolayer (1L) MoS2 growth forming a well interfaced heterostructure with the underlying layered material (SnS2). The photoluminescence spectrum (Figure 2b) demonstrates strong PL response of the MoS2 film and the clear presence of both the A and B exciton peaks at 680 and 630 nm, respectively. This further suggests the interaction between the monolayer MoS2 and the underlying SnS2 plate (an exfoliated MoS2 monolayer demonstrates preferential emission from the lower energy exciton at ∼680 nm whereas the second exciton peak at ∼630 nm is clearly visible in an exfoliated MoS2 bilayer). Raman mapping of the main SnS2 Raman shift at 318 cm−1 and photoluminescence mapping of the MoS2 A exciton at 680 nm were also used to confirm the complete and uniform coverage of the MoS2 monolayer growth on SnS2 (Figure 2d,e), which was further verified via qualitative EDS mapping (Figure 2g,h). These mappings also reveal that monolayer MoS2 preferentially grew on a layered material (SnS2) with only minor growth observed on the SiO2/Si wafer, which suggests preferential MoS2 growth occurs at temperatures lower than those typically required for monolayer MoS2 growth on SiO2/Si substrates. To further confirm that the heterostructure was comprised of one monolayer of MoS2, we also performed HRTEM analysis of a 1L MoS2−SnS2 heterostructure cross-section that was prepared by polymer-suspension microtome (Figure 3a). We
(See Methods for experimental details). The resulting 0.1−1 mm wide single-crystal SnS2 plates (Figure 1b) with welldefined shapes and typical thicknesses of 0.1−1 μm (Figure S1 in Supporting Information) were confirmed to be 2T-SnS2 by Raman spectroscopy (Figure 1c). Further increasing the reaction temperature to 850 °C and time to 1 h resulted in an increase in plate growth up to 0.5−1 cm (Figure 1d), which demonstrates the potential for growing large single-crystals for large-scale applications. Monolayer MoS2 Heterostructure Growth. Monolayer MoS2 heterostructure growth was achieved by reacting MoCl5 and elemental sulfur precursors heated upstream and reacted over 1−2 min at temperatures of 420−450 °C (Figure 1e) with variations in reaction parameters dependent on relative atmospheric humidity (See Methods for experimental details). We employed both Raman (Figure 2a) and photoluminescence
Figure 2. (a) Raman spectrum of a 1L MoS2−SnS2 heterostructure with the inset highlighting the E12g and A1g Raman peak shifts corresponding to 1L MoS2 with a heterostructure interface; (b) PL spectrum of the 1L MoS2−SnS2 heterostructure compared with a bulk (>6L) MoS2−SnS2 heterostructure; (c) optical image of a 1L MoS2−SnS2 heterostructure; (d) MoS2 PL mapping at 680 nm demonstrating uniform coverage of MoS2 with preferential growth on SnS2; (e) Raman mapping of the primary SnS2 peak at 318 cm−1 (all white scale bars: 20 μm); (f) SEM image of a 1L MoS2−SnS2 heterostructure; (g, h) SEM-EDS mapping of Mo and Sn K lines demonstrating the even distribution and growth enhancement of MoS2 on SnS2 relative to the SiO2/Si substrate (all black scale bars: 40 μm).
Figure 3. (a) Cross-sectional HRTEM image of a 1L MoS2−SnS2 heterostructure; (b) intensity histogram of the first 5 layers of the sample with the top layer spacing (0.64 nm) representative of a monolayer of MoS2 and the spacing of subsequent layers (0.58 nm) representative of SnS2; (c) fast Fourier transform of the HRTEM in panel a demonstrating alignment of the sample along the SnS2 [110] zone axis.
observed layer stacking in the cross-sectional HRTEM image, but it is difficult to visually differentiate the MoS2 layer from the SnS2 layers. However, using an image intensity histogram to measure the layer spacing (Figure 3b) revealed that all layers except the top layer were separated by ∼0.58 nm, indicating that the bulk of the film was comprised of SnS2 (with a lattice constant of c = 5.899 Å), while the top monolayer of the scanned region had a thickness of ∼0.64 nm that is characteristic of the spacing of one monolayer of MoS2 (with a lattice constant of c = 12.299 Å). The fast Fourier transform (FFT) confirmed the cross-sectional sample was aligned along the SnS2 [110] zone axis (Figure 3c) to ensure an accurate measurement of the layer spacing. Similar HRTEM previously performed on thicker MoS2−SnS2 heterostructures demonstrated consistent spacing between individual layers of MoS2 and the spacing at the MoS2−SnS2 interface was consistent with a layer of MoS2 rather than an average of the spacing of SnS2.46
(Figure 2b) spectroscopy to confirm monolayer growth. The Raman and photoluminescence behaviors of 1−6 layers of MoS2 has been well-established in previous literature and is a convenient tool for identifying the exact layer thickness.11,12 Specifically, the wavenumber separation between the in-plane E12g and out-plane A1g Raman shifts is directly related to the number of layers. More recently, a report demonstrated changes in the separation of the Raman shifts based on the formation of a heterostructure interface, which effectively increased the Raman shifts of n layers of MoS2 to demonstrate peak separations consistent with n + 1 layers.53 On the basis of these previous reports, we conclude that the separation of 21.5 7041
DOI: 10.1021/acsnano.6b03112 ACS Nano 2016, 10, 7039−7046
Article
ACS Nano
reaction time based on the relative atmospheric humidity during setup (see Methods for experimental details). Raman spectroscopy (Figure 5a) revealed the increasing separation of the E12g and A1g Raman shifts from 21.5 to 26 cm−1 that confirmed the progressive increase of the layer thicknesses from 1 to 6 layers. By comparison, the observed separation of the E12g and A1g peaks in exfoliated 1−6 layers MoS2 is 18.6−25.6 cm−1.11 The same systematic n + 1 increase of the E12g and A1g peak separation (Figure 5b) described previously similarly attests to the long-range heterostructure interaction of the layer growth and provides further confirmation of the monolayer growth by the complete identification of each layer thickness. The corresponding photoluminescence measurements taken on the same set of samples (Figure 5c) also demonstrate a unique layer-dependent response. Increased MoS2 thickness directly correlates with a decrease in PL intensity further reduced by the aforementioned heterostructure interaction. We found the consideration of the partial pressure of the MoCl5 precursor on the reaction rate to be crucial to the successful layer-controlled growth of MoS2. Importantly, because metal halides are notoriously hydroscopic, as they incorporate more water the effective partial pressure increases rapidly with temperature. Therefore, careful tracking of the ambient humidity was essential for reproducible layercontrolled growth. In all reactions, the MoCl5 precursor was exposed to air for ∼20 s with variable water incorporation dependent on the percent relative humidity during precursor loading. As the humidity increased, the effective partial pressure of the MoCl5 precursor increased. In order to compensate for this variability, it was necessary to reduce reaction times and precursor temperatures accordingly to achieve fully reproducible layer-controlled syntheses. Table S1 in the Supporting Information lists a representative set of experimental conditions that have been explored and verified multiple times to yield heterostructures with specific MoS2 layer numbers as examples for the requirements for the layer-controlled syntheses demonstrated. For the case of single monolayer MoS2 growth, when the ambient humidity decreased from 44% to
