arXiv subjects

Tianshi Wang

Publications and source records attributed to Tianshi Wang.

33 records · Page 2Linked to original sources

arXiv2019

Velocity Saturation in La-doped BaSnO3 Thin Films

BaSnO_{3}, a high mobility perovskite oxide, is an attractive material for oxide-based electronic devices. However, in addition to low-field mobility, high-field transport properties such as the saturation velocity of carriers play a major role in determining device performance. We report on the experimental measurement of electron saturation velocity in La-doped BaSnO_{3} thin films for a range of doping densities. Predicted saturation velocities based on a simple LO-phonon emission model using an effective LO phonon energy of 120 meV show good agreement with measurements of velocity saturation in La-doped BaSnO_{3} films.. Density-dependent saturation velocity in the range of 1.6x10^{7} cm/s reducing to 2x10^{6} cm/s is predicted for δ-doped BaSnO3 channels with carrier densities ranging from 10^{13} cm^{-2} to 2x10^{14} cm^{-2} respectively. These results are expected to aid the informed design of BaSnO3 as the active material for high-charge density electronic transistors.

cond-mat.mtrl-sci↗

arXiv2019

Late Breaking Results: New Computational Results and Hardware Prototypes for Oscillator-based Ising Machines

In this paper, we report new results on a novel Ising machine technology for solving combinatorial optimization problems using networks of coupled self-sustaining oscillators. Specifically, we present several working hardware prototypes using CMOS electronic oscillators, built on breadboards/perfboards and PCBs, implementing Ising machines consisting of up to 240 spins with programmable couplings. We also report that, just by simulating the differential equations of such Ising machines of larger sizes, good solutions can be achieved easily on benchmark optimization problems, demonstrating the effectiveness of oscillator-based Ising machines.

cs.ET↗

arXiv2019

OIM: Oscillator-based Ising Machines for Solving Combinatorial Optimisation Problems

We present a new way to make Ising machines, i.e., using networks of coupled self-sustaining nonlinear oscillators. Our scheme is theoretically rooted in a novel result that establishes that the phase dynamics of coupled oscillator systems, under the influence of sub-harmonic injection locking, are governed by a Lyapunov function that is closely related to the Ising Hamiltonian of the coupling graph. As a result, the dynamics of such oscillator networks evolve naturally to local minima of the Lyapunov function. Two simple additional steps (i.e., adding noise, and turning sub-harmonic locking on and off smoothly) enable the network to find excellent solutions of Ising problems. We demonstrate our method on Ising versions of the MAX-CUT and graph colouring problems, showing that it improves on previously published results on several problems in the G benchmark set. Our scheme, which is amenable to realisation using many kinds of oscillators from different physical domains, is particularly well suited for CMOS IC implementation, offering significant practical advantages over previous techniques for making Ising machines. We present working hardware prototypes using CMOS electronic oscillators.

cs.ET↗

arXiv2019

STFNets: Learning Sensing Signals from the Time-Frequency Perspective with Short-Time Fourier Neural Networks

Recent advances in deep learning motivate the use of deep neural networks in Internet-of-Things (IoT) applications. These networks are modelled after signal processing in the human brain, thereby leading to significant advantages at perceptual tasks such as vision and speech recognition. IoT applications, however, often measure physical phenomena, where the underlying physics (such as inertia, wireless signal propagation, or the natural frequency of oscillation) are fundamentally a function of signal frequencies, offering better features in the frequency domain. This observation leads to a fundamental question: For IoT applications, can one develop a new brand of neural network structures that synthesize features inspired not only by the biology of human perception but also by the fundamental nature of physics? Hence, in this paper, instead of using conventional building blocks (e.g., convolutional and recurrent layers), we propose a new foundational neural network building block, the Short-Time Fourier Neural Network (STFNet). It integrates a widely-used time-frequency analysis method, the Short-Time Fourier Transform, into data processing to learn features directly in the frequency domain, where the physics of underlying phenomena leave better foot-prints. STFNets bring additional flexibility to time-frequency analysis by offering novel nonlinear learnable operations that are spectral-compatible. Moreover, STFNets show that transforming signals to a domain that is more connected to the underlying physics greatly simplifies the learning process. We demonstrate the effectiveness of STFNets with extensive experiments. STFNets significantly outperform the state-of-the-art deep learning models in all experiments. A STFNet, therefore, demonstrates superior capability as the fundamental building block of deep neural networks for IoT applications for various sensor inputs.

cs.LG↗

arXiv2018

Band gap and band offset of Ga$_2$O$_3$ and (Al$_x$Ga$_{1-x}$)$_2$O$_3$ alloys

Ga$_2$O$_3$ and (Al$_x$Ga$_{1-x}$)$_2$O$_3$ alloys are promising materials for solar-blind UV photodetectors and high-power transistors. Basic key parameters in the device design, such as band gap variation with alloy composition and band offset between Ga$_2$O$_3$ and (Al$_x$Ga$_{1-x}$)$_2$O$_3$, are yet to be established. Using density functional theory with the HSE hybrid functional, we compute formation enthalpies, band gaps, and band edge positions of (Al$_x$Ga$_{1-x}$)$_2$O$_3$ alloys in the monoclinic ($β$) and corundum ($α$) phases. We find the formation enthlapies of (Al$_x$Ga$_{1-x}$)$_2$O$_3$ alloys are significantly lower than of (In$_x$Ga$_{1-x}$)$_2$O$_3$, and that (Al$_x$Ga$_{1-x}$)$_2$O$_3$ with $x$=0.5 can be considered as an ordered compound AlGaO$_3$ in the monoclinic phase, with Al occupying the octahedral sites and Ga occupying the tetrahedral sites. The direct band gaps of the alloys range from 4.69 to 7.03 eV for $β$-(Al$_x$Ga$_{1-x}$)$_2$O$_3$ and from 5.26 to 8.56 eV for $α$-(Al$_x$Ga$_{1-x}$)$_2$O$_3$. Most of the band offset of the (Al$_x$Ga$_{1-x}$)$_2$O$_3$ alloy arises from the discontinuity in the conduction band. Our results are used to explain the available experimental data, and consequences for designing modulation-doped field effect transistors (MODFETs) based on (Al$_x$Ga$_{1-x}$)$_2$O$_3$/Ga$_2$O$_3$ are discussed.

cond-mat.mtrl-sci↗

arXiv2018

Large Disparity Between Optical and Fundamental Band Gaps in Layered In2Se3

In$_2$Se$_3$ is a semiconductor material that can be stabilized in different crystal structures (at least one 3D and several 2D layered structures have been reported) with diverse electrical and optical properties. This feature has plagued its characterization over the years, with reported band gaps varying in an unacceptable range of 1 eV. Using first-principles calculations based on density functional theory and the HSE06 hybrid functional, we investigated the structural and electronic properties of four layered phases of In$_2$Se$_3$, addressing their relative stability and the nature of their fundamental band gaps, i.e., direct {\em versus} indirect. Our results show large disparities between fundamental and optical gaps. The absorption coefficients are found to be as high as that in direct-gap III-V semiconductors. The band alignment with respect to conventional semiconductors indicate a tendency to $n$-type conductivity, explaining recent experimental observations.

cond-mat.mtrl-sci↗

arXiv2017

Oscillator-based Ising Machine

Many combinatorial optimization problems can be mapped to finding the ground states of the corresponding Ising Hamiltonians. The physical systems that can solve optimization problems in this way, namely Ising machines, have been attracting more and more attention recently. Our work shows that Ising machines can be realized using almost any nonlinear self-sustaining oscillators with logic values encoded in their phases. Many types of such oscillators are readily available for large-scale integration, with potentials in high-speed and low-power operation. In this paper, we describe the operation and mechanism of oscillator-based Ising machines. The feasibility of our scheme is demonstrated through several examples in simulation and hardware, among which a simulation study reports average solutions exceeding those from state-of-art Ising machines on a benchmark combinatorial optimization problem of size 2000.

cs.ET↗

arXiv2017

Rigorous Q Factor Formulation and Characterization for Nonlinear Oscillators

In this paper, we discuss the definition of Q factor for nonlinear oscillators. While available definitions of Q are often limited to linear resonators or oscillators with specific topologies, our definition is applicable to any oscillator as a figure of merit for its amplitude stability. It can be formulated rigorously and computed numerically from oscillator equations. With this definition, we calculate and analyze the Q factors of several oscillators of different types. The results confirm that the proposed Q formulation is a useful addition to the characterization techniques for oscillators.

eess.SP↗

arXiv2017

Achieving Phase-based Logic Bit Storage in Mechanical Metronomes

Recently, oscillator-based Boolean computation has been proposed for its potentials in noise immunity and energy efficiency. In such a system, logic bits are encoded in the relative phases of oscillating signals and stored in injection-locked oscillators. To show that the scheme is very general and not specific to electronic oscillators, in this paper, we report our work on storing a phase-based logic bit in the relative phase between two mechanical metronomes. While the synchronization of metronomes is a classic example showing the effects of injection locking, our work takes it one step further by demonstrating the bistable phase in sub-harmonically injection-locked metronomes --- a key mechanism for oscillator-based Boolean computation. Although we do not expect to make computers with metronomes, our study demonstrates the generality of this new computation paradigm and may inspire its practical implementations in various fields, eg, MEMS, silicon photonics, spintronics, synthetic biology, etc.

cs.ET↗

arXiv2017

Analyzing Oscillators using Describing Functions

In this manuscript, we discuss the use of describing functions as a systematic approach to the analysis and design of oscillators. Describing functions are traditionally used to study the stability of nonlinear control systems, and have been adapted for analyzing LC oscillators. We show that they can be applied to other categories of oscillators too, including relaxation and ring oscillators. With the help of several examples of oscillators from various physical domains, we illustrate the techniques involved, and also demonstrate the effectiveness and limitations of describing functions for oscillator analysis.

math.CA↗

arXiv2017

Sub-harmonic Injection Locking in Metronomes

In this paper, we demonstrate sub-harmonic injection locking (SHIL) in mechanical metronomes. To do so, we first formulate metronome's physical compact model, focusing on its nonlinear terms for friction and the escapement mechanism. Then we analyze metronomes using phase-macromodel-based techniques and show that the phase of their oscillation is in fact very immune to periodic perturbation at twice its natural frequency, making SHIL difficult. Guided by the phase-macromodel-based analysis, we are able to modify the escapement mechanism of metronomes such that SHIL can happen more easily. Then we verify the occurrence of SHIL in experiments. To our knowledge, this is the first demonstration of SHIL in metronomes; As such, it provides many valuable insights into the modelling, simulation, analysis and design of nonlinear oscillators. The demonstration is also suitable to use for teaching the subject of injection locking and SHIL.

nlin.CD↗

arXiv2017

Thermal Transport Across Metal Silicide-Silicon Interfaces: An Experimental Comparison between Epitaxial and Non-epitaxial Interfaces

Silicides are used extensively in nano- and microdevices due to their low electrical resistivity, low contact resistance to silicon, and their process compatibility. In this work, the thermal interface conductance of TiSi$_2$, CoSi$_2$, NiSi and PtSi are studied using time-domain thermoreflectance. Exploiting the fact that most silicides formed on Si(111) substrates grow epitaxially, while most silicides on Si(100) do not, we study the effect of epitaxy, and show that for a wide variety of interfaces there is no difference in the thermal interface conductance of epitaxial and non-epitaxial silicide/silicon interfaces. The effect of substrate carrier concentration is also investigated over a wide range of p- and n-type doping, and is found to be independent of carrier concentration, regardless of whether the interface is epitaxial and regardless of silicide type. In the case of epitaxial CoSi$_2$, a comparison of temperature dependant experimental data is made with two detailed computational models using (1) full-dispersion diffuse mismatch modeling (DMM) including the effect of near-interfacial strain and (2) an atomistic Green' function (AGF) approach that integrates near-interface changes in the interatomic force constants obtained through density functional perturbation theory. At temperatures above 100K, the AGF approach greatly underpredicts the CoSi$_2$ data, while the DMM prediction matches the data well. The full-dispersion DMM is also found to closely predict the experimentally observed temperature-dependent interface conductance for epitaxial NiSi/Si and non-epitaxial TiSi$_2$/Si interfaces. In the case of epitaxial PtSi/Si interfaces, full dispersion DMM significantly overpredicts the experimental data.

cond-mat.mes-hall↗

arXiv2016

Microstructural Characteristics of Reaction-Bonded B4C/SiC Composite

A detailed microstructural investigation was performed to understand structural characteristics of a reaction-bonded B$_4$C/SiC ceramic composite. The state-of-the-art focused ion beam & scanning electron microscopy (FIB/SEM) and transmission electron microscopy (TEM) revealed that the as-fabricated product consisted of core-rim structures with α-SiC and $B_4$C cores surrounded by β-SiC and $B_4$C, respectively. In addition, plate-like β-SiC was detected within the $B_4$C rim. A phase formation mechanism was proposed and the analytical elucidation is anticipated to shed light on potential fabrication optimization and the property improvement of ceramic composites.

cond-mat.mtrl-sci↗

arXiv2016

A Functional Package for Automatic Solution of Ordinary Differential Equations with Spectral Methods

We present a Python module named PyCheb, to solve the ordinary differential equations by using spectral collocation method. PyCheb incorporates discretization using Chebyshev points, barycentric interpolation and iterate methods. With this Python module, users can initialize the ODEsolver class by passing attributes, including the both sides of a given differential equation, boundary conditions, and the number of Chebyshev points, which can also be generated automatically by the ideal precision, to the constructor of ODEsolver class. Then, the instance of the ODEsolver class can be used to automatically determine the resolution of the differential equation as well as generate the graph of the high-precision approximate solution. (If you have any questions, please send me an email and I will reply ASAP. e-mail:shaohui_liu@qq.com/2013141482143@stu.scu.edu.cn)

cs.MS↗

arXiv2016

Well-Posed Models of Memristive Devices

Existing compact models for memristive devices (including RRAM and CBRAM) all suffer from issues related to mathematical ill-posedness and/or improper implementation. This limits their value for simulation and design and in some cases, results in qualitatively unphysical predictions. We identify the causes of ill-posedness in these models. We then show how memristive devices in general can be modelled using only continuous/smooth primitives in such a way that they always respect physical bounds for filament length and also feature well-defined and correct DC behaviour. We show how to express these models properly in languages like Verilog-A and ModSpec (MATLAB). We apply these methods to correct previously published RRAM and memristor models and make them well posed. The result is a collection of memristor models that may be dubbed "simulation-ready", i.e., that feature the right physical characteristics and are suitable for robust and consistent simulation in DC, AC, transient, etc., analyses. We provide implementations of these models in both ModSpec/MATLAB and Verilog-A.

cs.ET↗