Efficient Automatic Design of IGBT Structural Parameters Using Differential Evolution and Machine Learning Model

Analysis of the dynamic avalanche of punch through insulated gate bipolar transistor (PT-IGBT)

In this paper, a technology computer-aided design (TCAD) model of a silicon carbide (SiC) insulated-gate bipolar transistor (IGBT) has been calibrated against previously reported experimental data. The calibrated TCAD model has been used to predict the static performance of theoretical SiC IGBTs with ultra-high blocking voltage capabilities in the range of 20-50 kV. The simulation results of transfer characteristics, IC-VGE, forward characteristics, IC-VCE, and blocking voltage characteristics are studied. The threshold voltage is approximately 5 V, and the forward voltage drop is ranging from VF = 4.2-10.0 V at IC = 20 A, using a charge carrier lifetime of τA = 20 μs. Furthermore, the forward voltage drop impact for different process dependent parameters (i.e., carrier lifetimes, mobility/scattering and trap related defects) and junction temperature are investigated in a parametric sensitivity analysis. The wide-range simulation results may be used as an input to facilitate high power converter design and evaluation. In this case, the TCAD simulated static characteristics of SiC IGBTs is compared to silicon (Si) IGBTs in a modular multilevel converter in a general high-power application. The results indicate several benefits and lower conduction energy losses using ultra-high voltage SiC IGBTs compared to Si IGBTs.

Insulated-gate bipolar-transistors (IGBTs) are becoming the main switching devices in medium- and high-power applications, such as switched-mode power supplies, power switches, variable frequency driver circuits, traction motors, and inducting heating. Although the switching speed of IGBTs is lower than in the case of metal-oxide-semiconductor field-effect-transistors (MOSFETs), the forward voltage drop is significantly lower, particularly in higher blocking voltage rated devices, which provides major advantage for using IGBTs in many applications. In addition, IGBTs have higher breakdown voltage and currant handling capacity than traditional bipolar-junction-transistors (BJTs), which makes them attractive for a variety of applications related to electric power switches. In this presentation, we introduce two new analytical models for the breakdown voltage of the punch-through IGBTs. The models are based on the accurate analytical estimation of the ionization integral and the solution of the Poisson equation in the drift region, which lead to a nonlinear equation for the breakdown voltage as a function of the doping concentrations in the drift region, buffer layer, and gate and emitter junctions, as well as of the widths of these regions and junctions. In general, this equation does not have an analytical solution and needs to be solved numerically in order to compute the breakdown voltage. Analytical solutions can be found only in a number of particular cases that will be discussed at the conference. The numerical and analytical solutions of the breakdown voltages are compared to more accurate results obtained using Synopsys’s Sentaurus Device Simulator [1], which is a commercial-grade finite element semiconductor simulator. We observe a very good agreement between our models and the finite element simulations for a wide range doping concentration and geometric dimensions of the IGBTs. In addition, at the conference, we will also compare our analytical results with previous analytical results published in the literature [2]. The proposed analytical model provides an efficient way to design the geometry and doping concentration profiles in punch-through IGBTs and helps the optimization process of the breakdown voltage in these devices [3]. More details about the approximations used in our analytical calculations, the results of the comparison with the more precise finite element simulations, and the existing literature will be presented at the conference. [1] 2016 Sentaurus Device Simulator User Manual, www.synopsys.com[2] J. Baliga, Fundamentals of Power Semiconductor Devices: Springer, 2008. [3] Zhu, C., & Andrei, P. (2017). Sensitivity of Breakdown Voltage of Power Transistors to Dopant Impurities. ECS Transactions, 80(7), 171-179.

The dissipated turn-off losses of shorted drain non-punch-through and punch-through type IGBTs (insulated-gate bipolar transistors) are investigated for voltage resonant circuit application. These characteristics are analyzed experimentally and calculated by using a two-dimensional device simulator. It is shown that the shorted drain structure is not effective for decreasing the dissipated loss, whereas the optimized punch-through type IGBT is suitable for this circuit application.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

A practical and accurate parameter extraction method is presented for the Fourier-based-solution physics-based insulated gate bipolar transistor (IGBT) and power diode models. The goal is to obtain a model accurate enough to allow switching loss prediction under a variety of operating conditions. In the first step of the extraction procedure, only one simple clamped inductive load test is needed for the extraction of the six parameters required for the diode model and of the 12 and 15 parameters required for the nonpunch-through (NPT) and punch-through (PT) IGBT models, respectively. The second part of the extraction procedure is an automated formal optimization step that refines the parameter estimation. Validation with experimental results from various structures of IGBT demonstrates the accuracy of the proposed IGBT and diode models and the robustness of the parameter extraction method.

TFETs (Tunnel Field-Effects Transistors) are assumed to be ultra-low-power devices because they have steep subthreshold swing and a more favourable scaling characteristic. Their electrical characteristics, however, cannot be adequately simulated using extensive Technology Computer-Aided Design (TCAD) simulations, which are computationally intensive and time-consuming to analyze. In an attempt to overcome this bottleneck, a proposal for a machine learning (ML)-based surrogate modeling framework is provided where the prediction of the critical TFET properties of transfer curves, Ion, Ioff, threshold voltage (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$V_{\text{th}}$</tex>), and subthreshold swing (SS) is endeavored with the help of the Random Forest Regression (RFR) model. The data collected based on TCAD simulations and literature reports was processed, trained, and tested on various ML models such as the Artificial Neural Networks (ANN), Support Vector Regression (SVR) and RFR. The calculated R2 of the proposed RFR model was better than ANN and SVR, with a much smaller MSE (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{3. 2 1} \times \mathbf{1 0}^{-\mathbf{7}}$</tex>) and an RMSE (<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{1. 7 9} \times \mathbf{1 0}^{-\mathbf{4}}$</tex>), showing that the model is robust and exhibits better generalization within the conditions of the devices used. Also, the ranking of intrinsic feature importance offered by RFR has a better interpretation rate and complies well with the previous findings on TCAD. These findings place RFR as a sound and efficient substitute for TFET modelling, with lower computational expense and intact physical stability, and thus add another facet and a practical approach to future device design optimization, assisted by a novel and powerful ML

The performances of various 1200 V insulated gate bipolar transistor (IGBT) structures for hard- and soft-switching (both the zero-voltage and the zero-current switching) at high temperature and under various test conditions are investigated in detail. A comparison is made between a conventional planar punchthrough IGBT (P-IGBT) using global lifetime control, a trench punchthrough IGBT (T-IGBT) using local lifetime control and a enhanced planar punchthrough IGBT (N-IGBT) also using the local lifetime control process. A simple and effective specific test circuit allows hard- and soft-switching operating modes, and provides also an independent control on the test parameters (load current, clamping voltage, gate resistance, temperature, dead times). It is highlighted that the local lifetime control is very efficient at high temperature reducing especially the turn-off losses. Furthermore, the T-IGBT exhibits very good performances during hard- and soft-switching turn-off, but this is not the case during hard-switching and zero-voltage switching turn-on. Due to its high input capacitance, the hard-switching turn-on losses are relatively high. Regarding the N-IGBT, it presents good performances under hard-switching and also zero-voltage switching. However, like the T-IGBT, its structure has a strong inductance behavior during zero-voltage turn-on increasing the conductivity modulation lag and also the voltage spike leading to high losses. All these data allow to provide useful information for the devices modeling, design, and optimization.

The switching characteristics (turn-on and turn-off) and forward conduction drop of trench-gate punch-through insulated gate bipolar transistors (IGBTs) are examined over a temperature range of from -50/spl deg/C to 125/spl deg/C. An analytical description of the forward conduction voltage drop is presented based on temperature dependencies of the appropriate physical parameters and mechanisms. A physics-based PSpice model, incorporating much of the device behavior, is also described. Results from the model are compared to experimental waveforms.

Significance of activation functions in developing an online classifier for semiconductor defect detection

Accurate prediction of energy consumption in district heating systems plays an important role in supporting effective and clean energy production and distribution in dense urban areas. Predictive models are needed for flexible and cost-effective operation of energy production and usage, e.g., using peak shaving or load shifting to compensate for heat losses in the pipeline. This helps to avoid exceedance of power plant capacity. The purpose of this study is to automate the process of building machine learning (ML) models to solve a short-term power demand prediction problem. The dataset contains a district heating network’s measured hourly power consumption and ambient temperature for 415 days. In this paper, we propose a hybrid evolutionary-based algorithm, named GA-SHADE, for the simultaneous optimization of ML models and feature selection. The GA-SHADE algorithm is a hybrid algorithm consisting of a Genetic Algorithm (GA) and success-history-based parameter adaptation for differential evolution (SHADE). The results of the numerical experiments show that the proposed GA-SHADE algorithm allows the identification of simplified ML models with good prediction performance in terms of the optimized feature subset and model hyperparameters. The main contributions of the study are (1) using the proposed GA-SHADE, ML models with varying numbers of features and performance are obtained. (2) The proposed GA-SHADE algorithm self-adapts during operation and has only one control parameter. There is no fine-tuning required before execution. (3) Due to the evolutionary nature of the algorithm, it is not sensitive to the number of features and hyperparameters to be optimized in ML models. In conclusion, this study confirms that each optimized ML model uses a unique set and number of features. Out of the six ML models considered, SVR and NN are better candidates and have demonstrated the best performance across several metrics. All numerical experiments were compared against the measurements and proven by the standard statistical tests.

Gasification reaction rates of biomass-derived char were predicted using a machine learning model based on fuel properties. A comparison among a kinetic model, a machine learning model, and a hybrid model confirmed that the prediction formula based solely on machine learning demonstrated high accuracy in predicting the gasification reaction rates of biomass-derived char. While the kinetic model, originally developed for coal, requires manual parameter tuning to be applied to biomass fuels, the machine learning model showed potential for making accurate predictions without the need for manual tuning. Looking ahead, we aim to further improve the accuracy of the prediction formula using machine learning, thereby facilitating the exploration of diverse fuel applications and contributing to the realization of a carbon-neutral society.

There is a growing consensus that the physics-based model needs to be coupled with machine learning (ML) model relying on data or vice versa in order to fully exploit their combined strengths to address scientific or engineering problems that cannot be solved separately. We propose several methodologies of bridging technology computer-aided design (TCAD) simulation and artificial intelligence (AI) with its application to the tasks for which traditional TCAD faces challenges in terms of simulation runtime, coverage, and so on. AI-emulator that learns fine-grained information from rigorous TCAD enables simulation of process technologies and device in real-time as well as large-scale simulation such as full-pattern analysis of stress without high demand on computational resource. To accelerate atomistic molecular dynamics (MD) simulation, we have done a comparison study of descriptor-based and graph-based neural net potential, and also show their capability with large-scale and long-time simulation of silicon oxidation. Finally, we discuss the use of hybrid modeling of AI- and physics-based model for the case where physical equations are either fully or partially unknown.

Study of parasitic oscillation of a multi-chip SiC MOSFET circuit based on a signal flow graph model by TCAD simulation

Silicon trench-gate insulated gate bipolar transistors (IGBTs) were analyzed using technology CAD (TCAD). Excellent agreement was confirmed between the J <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">C</inf> -V <inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">CE</inf> characteristics obtained by 3D-TCAD simulations and experiments. The carrier lifetime requirement for scaled trench-gate IGBTs was determined by extraction of the on-resistance of the n-base layer derived from the electric potential profile.

We propose a physics-based analytical model that accurately captures the effects of buffer traps on dc characteristics of gallium nitride (GaN)-based high-electron-mobility transistors (HEMTs). The model is then semi-analytically extended to additionally include the transient behavior. Analytical formulations for the shift in the threshold voltage <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${(}{V}_{\text {OFF}}{)}$ </tex-math></inline-formula> and two-dimensional electron gas (2-DEG) density due to the presence of buffer traps in the steady state are presented. In pulsed operation, technology computer-aided design (TCAD) simulations indicate that a time-dependent negative potential (NP) is developed under the gate, resulting in a modified <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V}_{\text {OFF}}$ </tex-math></inline-formula> and current collapse (CC). An expression for the modified <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${V}_{\text {off}}$ </tex-math></inline-formula> helps capture the pulsed current–voltage characteristics. The model captures the dependence of bias, time, temperature, trap concentration, capture cross section area, and activation energy of traps on the steady-state and transient characteristics. The model is implemented in Verilog-A in an existing compact model framework using a diode and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">RC</i> sub-circuit and validated using measured data and TCAD simulations. The modeling results are in excellent agreement with the experimental data and TCAD simulations. Since the model is physics-based, it requires fewer number of parameters compared to that in the existing models.