AVS2014 Session EM-WeA: High-K Dielectrics for 2D Semiconductor

Metal phthalocyanine (MPc) molecules are composed of a metal atom and a surrounding pthalocyanine ligand ring. The metal-ligand interaction and molecule-surface interaction are the two most important parameters, which controls the various physical and chemical properties of this adsorption system of MPc molecules onto the substrate. MPc molecules can potentially be employed for electrostatic doping of non-reactive substrates (for example 2D semiconductor materials), will lead to various applications such as sensors and electronics, and hence MPc become a model system for surface chemistry and nanotechnology. Tytanyl pthalocyanine (TiOPc) is another interesting molecule of pthalocyanine family having anisotropic intermolecular interactions. In this study, TiOPc electrostatic interactions with two nonreactive 2D semiconductor substrates, graphene and MoS₂, were studied. We have carried out first principle density functional theory (DFT) calculations and theoretical analysis to explore the structural and electronic properties of mono- and bi-layer of TiOPc molecule on both graphene and MoS₂. The adsorption of mono- and bi-layer films of TiOPc on graphene shows that there is a net 0.047 electrons and 0.016 electrons per TiOPc molecule charge transfer takes place from the graphene surface, respectively. Conversely, a net amount of 0.058 electrons and 0.029 electrons per TiOPc molecule charge transfer take place to the MoS₂ surface in case of mono- and bi-layer of TiOPc, respectively. Moreover, we find that the bandgaps of graphene/TiOPc(mono-and bi-layer) and MoS₂/TiOPc(mono-and bi-layer) heterostructures decrease with increasing number of TiOPc layers. Our results suggest that the band gap of 2D semiconductors, MoS₂/TiOPc and graphene/TiOPc heterostuctures, could be engineered with atomic layer precision by controlling the number of TiOPc layer deposited on the 2D substrate, which could serve as a potential candidate for both sensor and advanced electronics applications.

The growth of high-quality, layered HfSe₂ thin films by molecular beam epitaxy (MBE) on a variety of substrates is demonstrated for the first time. The cross-section of HfSe₂ thin films on highly ordered pyrolytic graphite (HOPG) shows the layered structure of crystalline HfSe₂ with an atomically sharp interface between the HfSe₂ and HOPG, verifying the van der Waals epitaxy. Crystalline HfSe₂ thin films with preferred orientation and hexagonal top surface symmetry are characterized by RHEED, STM, and XRD with a measured lattice constant consistent with the theoretical prediction of the 1-T phase. The stoichiometry (Hf: Se) of crystalline HfSe₂ on HOPG, measured by XPS, is very close to 1:2 and the in-plane and out-of-plane vibration mode peaks of the 1-T structure of HfSe₂ is confirmed by Raman spectroscopy. It is also noted that the crystal quality of the HfSe₂ changes as a function of substrate with the best growth results obtained on inert, hexagonal symmetry surfaces. These results indicate that the growth of novel TMDs by MBE is achievable and opens the possibility for exciting new 2D heterostructures and devices.

Basic ideas from pseudopotentials and semiclassical-transport will be used to discuss the properties of some novel 2D materials considered for post-Si-CMOS applications. First, it will be shown that in order to scale devices to 5 nm, simple electrostatic scaling laws demand the use of these two-dimensional materials, despite the daunting processing challenges they pose. Graphene will be considered discussing how its outstanding electronic properties become much less interesting when used as a component of some non-ideal structure, such when supported and gated and/or in nanoribbon form. A couple of very interesting ideas will be discussed next: 1. The Bose-Einstein condensation in bilayer systems (motivating UT-Austin’s BiSFETs) as an example of how issues of practical implementation may regrettably transform an excellent idea into a pure academic exercise; and 2. Monolayer tin (“stannanane”) as a 2D topological insulator with potential applications in spintronics and low-power high-performance devices. Besides discussing its potential electronic properties, the likelihood of actually fabricating such a material will be discussed on the basis of ab initio thermodynamics.

Over the past two years, transition metal dichalcogenide (TMD) materials, as 2-D crystals, have attracted much interest for a wide range of electronic and optoelectronic device applications. Yet, uniform deposition of sub-10 nm dielectrics on 2-D materials remains challenging.There is a great demand for the visualization and analysis of interfaces and defects in 2-D crystals and gate oxides, which play an important role in the growth and properties of heterostructures. Site-specific cross-sectional TEM imaging and associated techniques have become essential in providing detailed atomic scale information. High angle angular dark field (HAADF)-scanning transmission electron microscopy (STEM) can be used to investigate the atomic structure and chemistry of TMD-oxide interfaces and defects. In-situ TEM techniques such as heating and electrical/mechanical probing can also provide information regarding the dynamic behavior of the materials/devices of interest.

Herein, we report our results on HfO₂ and Al₂O₃ thin films deposited on MoS₂ by atomic layer deposition and SnSe₂ FETs with an Al₂O₃ gate oxide. Our in-situ TEM work on 2-D materials will also be presented. Based on our study, a possible growth mechanism of oxides on TMD with different surface treatments has been proposed [1,2]. These unique and effective site-specific and in-situ analysis tools can be widely applied to other gate oxides of interest.

This work was supported in part by the Center for Low Energy Systems Technology (LEAST), one of six centers supported by the STARnet phase of the Focus Center Research Program (FCRP), a Semiconductor Research Corporation program sponsored by MARCO and DARPA.

[1] ACS Nano, 7, 10354–10361, 2013.

[2] Applied Physics Letters, 104, 111601, 2014.

In order to investigate neutron-induced effects on HfO₂ resistive random-access memory (RRAM), HfO₂ films are subjected to irradiation by neutrons with energies above 1 MeV. Changes in the defect state concentrations of HfO₂ may lead to changes in the states of memory which has the potential to convert a “0” into a “1” in RRAM or vice versa.

Electron-spin resonance (ESR) is used to detect defect-state concentrations of the HfO₂ film deposited on high-resistivity substrates. Additionally, leakage currents of HfO₂ films are also measured to support the ESR data.¹

Neutron irradiation at low fluence decreases the Pb-type and E’ defect levels in ultra-thin HfO₂ films because electrons can fill existing states. These electrons come from electron-hole pairs generated by neutron interactions with silicon and oxygen. Thus, a low fluence of neutrons “anneals” the sample. However, when neutron fluence increases, more neutrons collide with oxygen atoms and cause them to leave the lattice or to transmute into different atoms. This causes the E’ state concentration to increase. The changes in the number of defect states lead to changes in leakage currents.

This work was supported by the Semiconductor Research Corporation under Contract 2012- KJ-2359 and by the National Science Foundation under Grant CBET-1066231.

One of the major obstacles impeding the advancement of III-V MOSFETs is the large density of interfacial trap states (D­_it) at the high-k/III-V interface. Poor nucleation of the gate oxide can lead to dangling bonds, strained bonds, and metallic bonds which contribute to D_it­; additionally, low nucleation density requires a thicker oxide to avoid pinholes which increase gate leakage. The nucleation of HfO₂ was studied using tetrakis(ethylmethylamino)hafnium (TEMAH) and H₂O on the InGaAs (001) and InGaAs (110) surfaces at low temperature, 120 ˚C, using atomic layer deposition (ALD). Low temperature ALD reduces subcutaneous oxidation of the channel when this is an activated process. The pulse and purge times of the oxidant and reductant were varied and their impact on the nucleation of HfO₂ were studied by fabricating MOSCAPs and extracting the D_it­ using the full interface surface state model. All samples had 10 cycles of an in-situ pre-ALD surface clean developed by Choptabanna and Carter which utilizes atomic H and TMA. The effectiveness of the ex-situ buffered oxide etch (BOE) was examined by fabricating samples with and without this clean. The BOE was more effective on the (001) samples since the non-BOE samples had a larger D_it­­ and a higher C_{max ­}indicating poor nucleation of the HfO₂. The dispersion in accumulation remained constant indicating the BOE had an immediate effect on the interface, rather than deep traps in the oxide. The BOE and non-BOE (110) samples had a nearly identical C_max­, D_it­, and dispersion in accumulation indicating the nucleation of HfO₂ is not as dependent on the BOE as the (001) surface. This result obviates the need for the BOE allowing an all dry process on InGaAs(110) and is likely a result of the inherent stability of the (110) surface compared to the (001) surface.