Overview of emerging semiconductor device model methodologies: From device physics to machine learning engines

Li, Xufan; Wu, Zhenhua; Rzepa, Gerhard; Karner, Markus; Xu, Haoqing; Wu, Zhicheng; Wang, Wei; Yang, Guanhua; Luo, Qing; Wang, Lingfei; Li, Ling

doi:10.1016/j.fmre.2024.01.010

Cited by 3 publications

(4 citation statements)

References 74 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…N eff is the effective doping concentration of the N-drift region. Solving (11) with constraints ( 8)- (10) gives the potential distributions in the N-pillar as…”

Section: Electric Field Of Sipos Modulated Drift Regionmentioning

confidence: 99%

See 1 more Smart Citation

Machine Learning-Based Figure of Merit Model of SIPOS Modulated Drift Region for U-MOSFET

Cao,

Sun,

et al. 2024

Micromachines

View full text Add to dashboard Cite

This paper presents a machine learning-based figure of merit model for superjunction (SJ) U-MOSFET (SSJ-UMOS) with a modulated drift region utilizing semi-insulating poly-crystalline silicon (SIPOS) pillars. This SJ drift region modulation is achieved through SIPOS pillars beneath the trench gate, focusing on optimizing the tradeoff between breakdown voltage (BV) and specific ON-resistance (RON,sp). This analytical model considers the effects of electric field modulation, charge-coupling, and majority carrier accumulation due to additional SIPOS pillars. Gaussian process regression is employed for the figure of merit (FOM = BV2/RON,sp) prediction and hyperparameter optimization, ensuring a reasonable and accurate model. A methodology is devised to determine the optimal BV-RON,sp tradeoff, surpassing the SJ silicon limit. The paper also delves into a discussion of optimal structural parameters for drift region, oxide thickness, and electric field modulation coefficients within the analytical model. The validity of the proposed model is robustly confirmed through comprehensive verification against TCAD simulation results.

show abstract

“…N eff is the effective doping concentration of the N-drift region. Solving (11) with constraints ( 8)- (10) gives the potential distributions in the N-pillar as…”

Section: Electric Field Of Sipos Modulated Drift Regionmentioning

confidence: 99%

“…Recent research has seen a surge in innovative approaches using machine learning techniques for device modeling and optimization [ 10 , 11 , 12 , 13 , 14 , 15 , 16 ]. For example, Klemme [ 12 ] developed a machine learning method for accurately predicting the transfer characteristics of negative-capacitance FinFET devices.…”

Section: Introductionmentioning

confidence: 99%

Machine Learning-Based Figure of Merit Model of SIPOS Modulated Drift Region for U-MOSFET

Cao,

Sun,

et al. 2024

Micromachines

View full text Add to dashboard Cite

show abstract

“…However, decades of research have provided electric and thermal models for the devices, which yield accurate predictions of the device characteristics in a much denser form by analytical equations governed by a few fundamental parameters. A historical overview of the development of device modeling methodologies from the original Shockley equation to artificial-intelligence-based black-box models can be found in [1].…”

Section: Introductionmentioning

confidence: 99%

“…Although the traditional physics-driven compact models are still in the mainstream nowadays, due to the complexity of the devices and the availability of automated extraction methods, AI-assisted compact model extraction from measured and/or TCAD-simulated data is gaining popularity [1]. The fundamentals of physics-based compact electrical modeling for classic devices are summarized for example in [2], and extensions for modeling wide-bandgap power devices can be found in [3,4].…”

Section: Introductionmentioning

confidence: 99%

Analytical Prediction of the Thermal Behavior of Semiconductor Power Devices from Room-Temperature I–V Measurements

Ress,

Farkas,

Rencz

2024

Energies

View full text Add to dashboard Cite

The thermal characterization of power devices is an inevitable task in the industry. Thermal transient testing is one of the major tools for this characterization, as it is not only capable of giving information about the actual thermal parameters but may also reveal the root cause of potential device failures. The testing may occur on single packages or modules on a dedicated standard test bench, or “in situ”, in an actual assembly. The testing process itself is very fast in both cases, on the order of seconds, but the transient measurement needs to be preceded by a calibration step to determine the temperature dependence of a temperature-sensitive parameter (TSP) of the semiconductor device. This may require a long time, as the device has to be measured at many stabilized temperatures, which in the case of power devices may take hours. It also has to be considered that, especially in “in situ” measurements, reaching the highest device temperatures of power devices may even damage other parts of the surrounding electronics. Moreover, the temperature distribution inside a module will be different at calibration time than during operation with the same junction temperature, as bond wires and copper traces do not reach the chip temperature in the latter case. This paper presents a methodology that can be used to replace the lengthy measurement of one parameter at many temperatures with a fast single I–V characteristics measurement at room temperature. Physics-based calculations assign a unique temperature-sensitive parameter to each item of interest in the system. After the presentation of the theoretical background, the usability of the method is demonstrated by verifying measurements on silicon power devices.

show abstract

Explainable and generalizable AI-driven multiscale informatics for dynamic system modelling

Luo,

Li,

Xiao

et al. 2024

Sci Rep

View full text Add to dashboard Cite

Ultra-precision machining requires system modelling that both satisfies explainability and conforms to data fidelity. Existing modelling approaches, whether based on data-driven methods in present artificial intelligence (AI) or on first-principle knowledge, fall short of these qualities in high-demanding industrial applications. Therefore, this paper develops an explainable and generalizable ‘grey-box’ AI informatics method for real-world dynamic system modelling. Such a grey-box model serves as a multiscale ‘world model’ by integrating the first principles of the system in a white-box architecture with data-fitting black boxes for varying hyperparameters of the white box. The physical principles serve as an explainable global meta-structure of the real-world system driven by physical knowledge, while the black boxes enhance local fitting accuracy driven by training data. The grey-box model thus encapsulates implicit variables and relationships that a standalone white-box model or black-box model fails to capture. Case study on an industrial cleanroom high-precision temperature regulation system verifies that the grey-box method outperforms existing modelling methods and is suitable for varying operating conditions.

show abstract

Overview of emerging semiconductor device model methodologies: From device physics to machine learning engines

Cited by 3 publications

References 74 publications

Machine Learning-Based Figure of Merit Model of SIPOS Modulated Drift Region for U-MOSFET

Machine Learning-Based Figure of Merit Model of SIPOS Modulated Drift Region for U-MOSFET

Analytical Prediction of the Thermal Behavior of Semiconductor Power Devices from Room-Temperature I–V Measurements

Explainable and generalizable AI-driven multiscale informatics for dynamic system modelling

Contact Info

Product

Resources

About