AVS2016 Session 2D+NS-FrM: 2D Materials: Device Physics and Applications

Ultra-thin two-dimensional (2D) semiconducting materials possess a combination of large, tunable electronic bandgaps, optical transparency, and mechanical flexibility, and will likely revolutionize electronic devices such as wearable sensors and flexible displays. A primary step in the development of such devices with integrated 2D materials is the development of scalable, transfer-free synthesis over large areas at low temperatures. Electrically insulating amorphous transition metal dichalcogenide (TMD) films can be deposited via physical vapor deposition on large area flexible substrates at room temperature, and crystallized with subsequent illumination with light. Focused laser light with a power density of ~1 kW cm²is suitable for writing micron scale features in semiconducting transition metal dichalcogenides on polymer substrates. Broad band illumination from a xenon lamp can also be used over the large substrate areas (> 100 cm²), or passed through a physical mask to print features only in desired locations. The semiconducting properties of 2D MoS₂ and WS₂ materials synthesized in this way have been characterized using conductive atomic force microscopy, and other techniques to observe the expected temperature dependence on electrical conductivity. Structure and composition of the materials can be controlled by altering the incident fluence as well as by controlling the ambient environment during illumination, as verified by Raman spectroscopy, X-ray photoelectron spectroscopy, cross-sectional and plan view transmission electron spectroscopy, and other techniques. Multiple layers of 2D materials can also be treated in this way. For example, both layers in a MoS₂/WS₂ heterostructure of 10 nm total thickness on a polymer (PDMS) substrate were crystallized upon laser illumination. Diverse 2D architectures and devices built from illumination-based crystallization techniques will be highlighted.

Black phosphorus (P) has emerged as a layered semiconductor with unique crystal structure featuring corrugated atomic layers and strong in-plane anisotropy in its physical properties. In particular, it is predicted to exhibit strong in-plane mechanical anisotropy, which shall lead to previously inaccessible dynamic responses in resonant 2D nanostructures [1], and new opportunities for studying carrier-lattice interaction in atomic layers. It is therefore of both practical and fundamental importance to systematically investigate the mechanical anisotropy in black P crystal and NEMS devices.

Enabled by the first demonstration of black P resonant nanostructures [2] with multimode responses, we show that the spatial mapping of the multimode resonance mode shapes [3] creates a new means for precise determination of black P crystal orientation (i.e., the anisotropic zigzag and armchair axes) [4] . Further, the multimode technique enables simultaneous quantification of the anisotropic mechanical properties (i.e., elastic moduli along both major crystal axes): combined with finite element method (FEM) modeling, we determine the Young’s moduli of multilayer black P to be 116.1 GPa and 46.5 GPa in zigzag and armchair directions, respectively. In addition, we demonstrate that electrostatic gating induced straining can continuously tune the mechanical anisotropic effects on multimode resonances in black P electromechanical devices. Our results show that multimode resonant response manifests the unique mechanical anisotropy effect in black P nanodevices, and provides a new method for determine the material’s crystal orientation and elastic properties in situ, independent from conventional optical, electrical, and nanoindentation calibration techniques.

Graphene, a hallmark of two-dimensional (2D) materials, has been employed as an atomically thin building block for highly miniaturized nanoelectromechanical system (NEMS) and shown attractive potential for nanoscale actuators and sensors. Thanks to its exceptional elastic modulus (E_Y~1TPa), ultralow mass density (r~2200kg/m³), and superior strain limits (e_limit~25%) [1], high performance and frequency tunable graphene resonators have been demonstrated using photothermal [2] and electrostatic actuation [3] schemes. In addition to excellent mechanical properties, graphene possesses high temperature stability [4] and negative thermal expansion coefficient [5], hence graphene resonators may inherently exhibit better performance under high temperature. In existing reports, graphene resonators are exposed in high temperature using annealing or Joule heating (e.g,, applying up to 1.8V from drain to source) only for thermal annealing [3], high temperature operation of graphene resonators has not been demonstrated yet.

In this work, we fabricate mono- and bi-layer (1L and 2L) graphene resonators and investigate their resonance characteristics at high temperature up to ~500 K using Joule heating. We conveniently use DC voltage to heat graphene resonators, and apply AC voltage to excite resonance motion. Then, we simultaneously measure temperature and resonance characteristic of graphene resonators using a home-built, integrated Raman spectroscopy and laser interferometry measurement system. We first test electrothermal frequency tuning and find that frequency of graphene resonators upshift from ~80MHz to ~86MHz as DC voltage increases from 0.5V to 2.5V. Unlike electrostatic force resonance tuning and excitation [3], we do not observe capacitive softening or loaded Q effects which may compromise performance of resonators. We then investigate mechanical nonlinearity of graphene resonators in high temperature by changing both DC and AC voltage. This study opens new capabilities for engineering tunable graphene NEMS resonators and oscillators for a number of emerging applications.

[1] C. Lee, et al., Science321, 385-388 (2008).

[2] J. S. Bunch, et al., Science.315, 490-493 (2007).

[3] C. Chen, et al., Nat. Nanotech.4, 861-867 (2009).

[4] K. Kim, et al., Phys. Status Solidi RRL4 302-304 (2010).

[5] M. Pozzo, et al., Phys. Rev.Lett.106 135501 (2011).

Two-dimensional semiconductors (2DSCs) such as MoS₂ have attracted intense interest as an alternative electronic material in the post-silicon era. However, the on-current density achieved in 2DSC transistors to date is considerably lower than that of silicon devices. It remains an open question whether 2DSC transistors can offer competitive performance. To achieve a high performance (high on-current) device requires (1) a pristine channel with high carrier mobility, (2) an optimized contact with low contact resistance and (3) a short channel length. The simultaneous optimization of these parameters is of considerable challenge for atomically thin 2DSCs since the typical low contact resistance approaches either degrade the electronic properties of the channel or are incompatible with the fabrication of short channel devices. Here I will first review different strategies that have been developed to optimize these factors, and discuss how we can combine these strategies together to achieve high performance 2DSC semiconductor transistors. In particular, we will discuss a unique approach towards high-performance MoS₂ transistors using a physically assembled nanowire as a lift-off mask for creating ultra-short channel devices with pristine MoS₂ channel and self-aligned low resistance metal/graphene hybrid contact. With the optimized contact in short channel devices, we demonstrate that a sub-100 nm MoS₂ transistor can deliver a record a high on-current density comparing well with that of silicon devices, demonstrating thte exciting potential of 2DSCs for future electronic applications.

2D semiconductors have attracted attention for future electronic devices due to their excellent electronic and optoelectronic properties. These devices require few nanometer thick and pin hole-free dielectric layers as gate insulators. However, due to the inert nature of 2D semiconductors such as graphene and Transition Metal Chalcogenides (TMDs), the dielectric layer selectively nucleates on defect sites or step edges. In the conventional atomic layer deposition (ALD) process on graphene or other 2D semiconductors, such non-uniform oxides result in large leakage currents in 2D semiconductor based device. Therefore, for successful integration into device, uniform and insulating gate oxides on 2D semiconductors should be prepared.

In this work, Al2O3 was directly deposited on HOPG and MoS2 surface by low temperature ALD with trimethylaluminum(TMA) and H2O or O3 without any seeding layer or surface treatments. Using short purge time between two precursor pulses at 50C, a CVD growth component was intentionally induced to provide more nucleation sites on the surface. The CVD growth component induces deposition of 1 nm Al2O3x particles on the surface which provide a uniform layer of nucleation centers. Before ALD, HOPG and MoS2 samples were cleaned by mechanical exfoliation method. For HOPG substrate, 50 cycles of ALD Al2O3 was deposited at 50C using 600ms of TMA and 50 ms of H2O pulse time with 500ms purge time between two pulses. In the case of MoS2, 300ms of O3 pulse was employed instead of H2O pulse. The same ALD recipes were performed on SiGe substrates in order to compared the quality of the oxide. After ALD process, MOSCAP devices were fabricated to measure the capacitance and leakage current of the oxide. Non-contact mode AFM was performed to check the topography of the oxide and the results showed that uniform and pin hole-free oxide layer was formed on the surface. The leakage current of the oxide on HOPG and MoS2 was as low as 10^-5 A/cm² which was comparable to the oxide on SiGe substrates.

Tailoring band alignment layer-by-layer using heterojunctions of two-dimensional (2D) semiconductors is an attractive prospect for producing next-generation electronic and optoelectronic devices that are ultra-thin, flexible, and efficient. 2D layers of transition metal dichalcogenides (TMDs) in laboratory devices have already demonstrated properties favorable for electronic and optoelectronic applications. Despite these strides, a systematic understanding of how band alignment evolves from monolayer to multilayer for MoS₂, a model TMD system, is still missing owing to the lack of a suitable experimental approach. Quantitative determination of the electronic band alignment necessitates that measurements be performed in a controlled environment (such as vacuum) using a substrate that interacts minimally with the overlying TMDs (preferably insulating) to suppress the electronic influence of supporting substrates and prevent chemical modification of TMDs due to adsorbates (primarily water).

Here we report on the local band alignment of monolayer, bilayer, and tri-layer MoS₂ on a 285-nm-thick SiO₂ substrate, measured using a new approach to probe the occupied electronic states based on photoemission electron microscopy with deep ultraviolet excitation. The spatially-resolved, simultaneous measurements of the vacuum level and the valence band edge at the Brillouin zone center show that the addition of layers to monolayer MoS₂ increases the relative work function, and pushes the valence band edge toward the vacuum level. We also find n-type doping of few-layer MoS₂ and type-I band alignment across monolayer-to-bilayer and bilayer-to-trilayer lateral junctions. Our results differ from some earlier reports based on Kelvin probe and scanning photocurrent microscopies [Sci. Reports, 5, 10990 (2015), Nano Lett., 15, 2278 (2015)], and highlight the strong influence of environmental effects on the band alignment in MoS₂ homojunctions. We are now applying this exciting new metrology to systematically examine the ionization energies of a series of TMDs. The results will provide fundamental information necessary to assess the band alignments of TMD heterojunction devices, and to validate or refine existing theoretical predictions [APL, 103, 053513 (2013), J. Phys. Chem. C, 119, 13169 (2015)].

This work was performed at CINT (DE-AC04-94AL85000), and is supported by Sandia LDRD, US DOE EERE SunShot Initiative BRIDGE (DE-FOA-0000654 CPS25859), and Army Research Office MURI (W911NF-11-1-0362). SNL is a multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin Co., for the US DOE NNSA (DE-AC04-94AL85000).

We investigate the light scattering properties of a silicon nanobeam waveguide through wavelength- and polarization-dependent scanning photocurrent measurements of a black phosphorus (BP) photodetector on top of the silicon waveguide. The measured photocurrent responses exhibit similar patterns as the light intensity distribution calculated by finite-difference time-domain simulations, suggesting that the light scattering properties of the waveguide can be detected as photocurrent signals by the BP photodetector. Interestingly, no photocurrent signals are observed when the incident photon energy is below the bandgap of silicon, indicating that the photocurrent response generated in the BP photodetector is mainly attributed to the photo-excited electron-hole pairs in the silicon waveguide which can be injected into the BP and dominate its photocurrent generation. Our experimental results suggest that two dimensional (2D) material based photodetectors can offer an effective approach to visualize the light scattering properties of photonic structures by photocurrent mapping, which not only opens up avenues for learning about light matter interaction of photonic structures but also provides a way of engineering future 2D material based optoelectronic devices with integrated photonic structures.