We have presented a stochastic Monte Carlo algorithm parameterized with ab initio calculations to investigate the thermal transport across lateral interfaces between two-dimensional materials, with a focus on hexagonal boron nitride and Graphene heterostructures. Ab initio parameters are incorporated into the Monte Carlo simulation, enhancing the accuracy of thermal properties. The interface thermal conductance, temperature profiles, response times, and modal contributions of phonon modes in all transport regimes are theoretically studied. This research significantly contributes to our understanding of thermal properties at the nanoscale, providing essential insights for electronics and energy management applications.
We propose a compact model that utilizes physical-based artificial neural networks (ANNs) to model the effect of temperature on n- and p-type gate-all-around nanosheet FETs. Our compact model comprises two independent ANNs, where the first ANN is designed to output parameters related to temperature and the second ANN is utilized for the device physical parameters. All outputs of ANNs are integrated into a physical equation of drain current to form the entire compact model. Compared with the BSIM-CMG model in circuit simulations, our results are highly consistent in transfer characteristics and timing dynamics.
Graphene is a promising material for gas sensing because of its unique electrochemical properties. Previous studies have revealed that the adsorption and desorption of gas molecules on graphene were controlled by mechanical strain. In this study, first-principles calculation was applied to the investigation of the effect of tensile and compressive strains on the gas adsorption properties of graphene. It was found that the adsorption energy varied almost linearly with the applied strain due to the change of its electronic band structure and, consequently, it caused the change of the magnitude of the charge transfer between them. Among them, NO 2 exhibited greater strain sensitivity, specifically, its adsorption energy increased by about 70% under compressive strain of 10%. The non-equilibrium green function (NEGF) method was also performed to confirm that the I-V characteristic was changed by the adsorption of gas molecules and strain. Notably, it was found that the change of current was about 14 times larger than that observed in 10%-tensile-strained graphene adsorbed with H 2 O system when no strain was applied. Overall, our results validated that tuning the gas binding behavior of graphene by application of strain is a promising approach for developing multi-gas sensors with high selectivity.
We explore GAA transistor design elements for performance scaling beyond 2nm technology node. We examine the various resistance components of a 1 st generation GAA FEOL transistor with typical 2nm node design rules. We propose different schemes to reduce each resistance component of the GAA transistor to improve its drive-current and circuit performance. Using our calibrated process, device and ring-oscillator modeling platform, we evaluate the device and circuit performance impact of each of these schemes. We show that the drive-current and circuit performance of GAA transistors can be improved by over 50% and 18%, respectively, which aligns with the performance scaling requirements for logic nodes beyond 2nm.
The atomistic investigation of modern nanoscale devices for RF applications requires time-resolved simulations, which often come with a high computational burden. Thus, the effective mass approximation is usually employed to reduce the complexity and computation times. Here, a novel approach combining a tight-binding like ansatz with a Quantum Liouville-type Equation is presented, that can effectively take into account any arbitrary energy dispersion relation. It is applied to analyze the charge carrier transport in a DGFET in a single non-parabolic valley and compared to the parabolic case for the stationary and transient cases. As it is shown, the application of the parabolic approximation leads to a severe underestimation of the current densities and does not capture distortion effects when RF amplifier operation is considered.
