No abstract

No abstract

The integration of novel logic and memory devices, fabricated from van der Waals materials, into CMOS process flows with a goal of improving system-level Energy-Delay-Product (EDP) for data abundant applications will be discussed. Focusing on materials growth and integration techniques that utilize non-equilibrium, kinetically restricted strategies, coupled with in-situ characterization, enables the realization of atomic configurations and materials that are challenging to make but once attained, display enhanced and unique properties. These strategies become necessary for most future technologies where thermal budgets are constrained and conformal growth over selective areas and 3-dimensional structures are required. In this work, we demonstrate the high-quality MBE heterostructure growth of various layered materials by van der Waals epitaxy (VDWE). The coupling of different types of van der Waals materials including transition metal dichalcogenide thin films (e.g., WSe2, WTe2, HfSe2), helical Te thin films, and topological insulators (e.g., Bi2Se3) allows for the fabrication of novel electronic devices that take advantage of unique quantum confinement and spin-based characteristics. We demonstrate how the van der Waals interactions allow for heteroepitaxy of significantly lattice-mismatched materials without strain or misfit dislocations. Yet, at the same time, the VDW interactions are strong enough to cause rotational alignment between the epi-layer and the substrate, which plays a key role in the formation of grain boundaries. We will discuss TMDs, Te, and TIs grown on atomic layer deposited (ALD) high-k oxides on a Si platform as well as flexible substrates and demonstrate field-effect transistors with back-end-of-line compatible fabrication temperatures (<450 °C). As an example, the room temperature I-V curves of WSe2 FETs grown by MBE on ALD-grown amorphous Al2O3 indicate strong ambipolar conduction with electron and hole ON currents reaching ~25 and 15 µA/µm, respectively and ON/OFF ratios exceeding 104. Hole mobilities of ~50 cm2/V-s are achieved. The subthreshold swing in the hole branch is ~230 mV/dec over three decades of ID , suggesting an immobile charge density of ~8 x 1012 cm-2. Esaki tunnel junctions in WSe2 FETs are also demonstrated with 40 nm channel length. A back-gate is employed which makes the negative differential resistance (NDR) gate-tunable at temperatures up to 140 K and a maximum peak-to-valley current ratio (PVCR) of 3.5 at T = 110 K. When compared to previous reports for WSe2, the fabricated Esaki junctions provide increased PVCRs, from ~2 to 3.5, 100x higher peak currents, from ~pA/µm to ~nA/µm levels, and a significant reduction in the peak voltage, VP , from 1.5 to 0.2 V, as should be expected for tunnel diodes. High performance transistors made from helical Te with field-effect mobilities as high as 700 cm2/V-s have also been demonstrated. These devices are fabricated from Te needle-like structures directly grown on thermal SiO2 on a back-gated structure at 120 °C. The achievement of high-mobility transistor channels at BEOL compatible processing temperatures shows the potential for integrating van der Waals materials into CMOS process flows. In summary, we have demonstrated the growth of a variety of van der Waals materials directly grown even on amorphous oxides at BEOL compatible temperatures. Using novel techniques to improve the crystal quality at these low temperatures has enabled the demonstration of devices that can be intergrated above high-performance circuits and metal levels. TFTs and tunnel devices in particular have been demonstrated with outstanding performance suggesting a path forward for heterogeneous integration and improved systems level performance. This work is supported in part by NEWLIMITS, a center in nCORE, a Semiconductor Research Corporation (SRC) program sponsored by NIST through award number 70NANB17H041. This work is also supported by the National Science Foundation under awards 1917025 and 1921818.

The incorporation of magnetic dopants is a way to realize magnetism in semiconductors. The 2D material WSe2 is a semiconductor with a bandgap of 1.3 eV, and Fe-doped WSe2 is predicted to have room temperature, long-range ferromagnetism. Moreover, WSe2 has been extensively studied for applications in electronic devices, and can be prepared by deposition methods that are compatible with semiconductor fabrication processes. 2D magnetic films are particularly intriguing in technologies that rely on exchange coupling since the efficiency of the exchange with the magnet is inversely proportional to the magnet thickness 1/tFM. Chalcogenide-based TMD magnets can also have ideal interfaces with chalcogenide-based topological insulators (like Bi2Se3), enabling efficient spin torque devices and potentially realizing the quantum Hall effect without an external magnetic field. In this work, we demonstrate the growth of Fe-doped WSe2 and report on its up to room temperature ferromagnetic properties. The evolution of Fe-doped WSe2 films under different doping levels indicates that strain from the impurity atom itself coupled with strain imparted by the substrate can drive phase separation at high Fe concentrations. We find that suppressing that film strain helps promote more Fe incorporation (and higher Curie temperatures). Cr2O3 has only 0.2% lattice mismatch with WSe2 and is a magnetoelectric that can couple with ferromagnetic films. By using seed-layer techniques, we successfully prepared layered Fe-WSe2 on Cr2O3 and demonstrate magnetic coupling between the two, a potentially promising route toward realizing electric field control of 2D magnets. Magnetic measurements show a clear hysteresis loop at 100 K in a 2 Tesla cooling field, from which a large negative bias field of 600 Oe is resolved. The bias field vs temperature mesurements, where a negative bias field emerges below 250 K, indicates the existence of ferromagnetic exchange coupling in this heterostructure. This work is supported in part by NEWLIMITS, a center in nCORE, a Semiconductor Research Corporation (SRC) program sponsored by NIST through award number 70NANB17H041. This work is also supported by the National Science Foundation under awards 1917025 and 1921818.

Transition metal dichalcogenides (TMDs) are two-dimensional (2D) layered materials covalently bonded within the layers, but with only weak van-der-Waals (vdW) interactions between individual monolayers [1]. Substantial progress has been made in better understanding of these materials, from the nature of their defects [2] to the achievement of large area epitaxially grown films [3-5]. These materials have attracted great attention for applications such as next-generation electronics - including sub-60 mV sub-threshold slope transistors, flexible electronics, and optoelectronics, novel applications in spintronic devices [6] and the use of heterostructures of TMDs for tunnel field effect transistors (TFETs) [7]. However, issues including control of channel and source/drain doping have impeded their implementation into device. Similar to three-dimensional semiconductors, doping of the TMDs is required to modulate carrier concentration, to achieve Ohmic contacts, and to generate n-type and p-type materials which are required for complementary metal-oxide-semiconductor (CMOS) technology and TFET applications. One of the most studied TMDs is MoS2. The transition metals Nb and Re are two candidate dopants for MoS2 with theoretical results showing their suitability as p- and n-type dopants, respectively [8]. Experimental results have confirmed that Nb substitutes at the Mo-site and acts as a p-type dopant in MoS2 [9-11]. While Re has been confirmed as an n-type dopant, with the additional benefit of reducing sulfur vacancies and defect-related gap states [12-13]. This study reports on the band structure and electrical characteristics of doped and unintentionally doped chemical vapor transport (CVT) grown MoS2 bulk crystals. We present a direct determination of the valence band structure of the MoS2 and the impact of transition metal doping (Nb and Re) using high-resolution angle-resolved photoemission spectroscopy (ARPES). Structural defects in the form of vacancies are widely known to strongly alter the MoS2 electronic structure. Therefore, we have performed highly-efficient density functional theory (DFT) based simulations to provide insight into the impact of vacancies in addition to the incorporation of transition metal dopants on the MoS2 band structure. Unfolded band structures obtained through our simulations [14], in comparison with the experimentally obtained occupied band structure of doped and un-doped MoS2 have shown excellent agreement and revealed that there has been significant distortion to the band structure due to the presence of vacancies, as well as the introduction of degenerate Nb-doping. Scanning tunneling microscopy (STM) studies revealed high quality crystals with point defects established by our first-principle calculations to be mainly Mo vacancies. We also report our Hall effect analysis to obtain the electrical metrics for the crystals. Secondary ion mass spectrometry (SIMS) shows the impurities present in the crystals, which is then used to explain the difference in transport behaviour and dopant types between similar crystals from different material sources. Figure 1. (a) Unfolded band structure of un-doped MoS2 with Mo vacancy obtained by DFT calculations shown using contour plot of total weight intensity. Black curve: primitive-cell of pristine MoS2 band structure. Mo vacancy induced localized states are located close to the valence band edge - shown by white rectangles. (b) ARPES spectra for a non-intentionally-doped MoS2 crystal acquired along the high symmetry Γ-K direction overlapped with DFT results showing excellent agreement considering the effect of a Mo vacancy. (c) DFT-obtained STM images of a pristine MoS2 (left), and Mo vacancy in MoS2 (right). (d) Charge density difference between pristine MoS2 and MoS2 with Mo vacancy. Red and blue indicate charge accumulation and depletion, respectively, at Mo vacancy sites. Figure 2. In situ STM, XPS and LEED measurements of MoS2. (a) Large-scale STM image with bright and dark defects (300×300 nm). (b) STM image (100×100 nm) with line profiles over the defects 1, 2, and 3. (c) STM image illustrates atomic resolution with interatomic distance 0.32 nm. (d) & (e) corresponds to the binding energies of Mo 3d and S 2s, and S 2p core levels, respectively. (f) LEED showing highly ordered structure. References: [1] Applied Materials Today, 9, 504, 2017. [2] ACS Nano, 8, 2880, 2014. [3] 2D Materials, 4, 045019, 2017. [4] 2D Materials, 4, 025044, 2017. [5] ACS Nano, 9, 474-80, 2015. [6] Nature Nanotechnology, 7, 699-712, 2012. [7] Applied Physics Letters, 103, 053513, 2013. [8] Physical Review B, 88, 075420, 2013. [9] AIP Advances, 6, 025323, 2016. [10] Nano Lett, 14, 6976-6982, 2014. [11] Applied Physics Letters, 104, 092104, 2014. [12] Applied Physics Letters, 111, 203101, 2017. [13] Advanced Functional Materials, 28, 1706950, 2018. [14] npj 2D Materials and Applications, 3, 33, 2019. Figure 1