Scholarlink

<p>This paper investigates the design optimization of GaN-based Terahertz (THz) IMPATT sources using Artificial neural networks (ANNs). Traditional design optimization of IMPATT sources, relying on the self-consistent quantum drift–diffusion (SCQDD) model, is computationally intensive, manual and time-consuming. To overcome these challenges, we developed ANN models trained on simulation data from SCQDD-based DC and large-signal simulations. Forward ANN models predict critical performance parameters, including power output and efficiency, for GaN IMPATT diodes with flat and hi-lo doping profiles. These models demonstrated high accuracy when validated against experimental data. In addition, inverse ANN models were developed to optimize diode design parameters based on specific target performance metrics, significantly reducing the computational resources and time required for optimization. The proposed ANN-based approach allows for efficient design optimization of GaN IMPATT diodes across a wide frequency range, including the THz band, addressing the growing demand for high-performance THz radiation sources. This work introduces a scalable method for the rapid development and commercialization of GaN THz devices, advancing high-frequency applications.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design Optimization of GaN-Based Terahertz IMPATT Sources by Using Artificial Neural Networks

  • 期刊论文

  • 发表日期：

    2025年4月24日

查看全文
  • Santu Mondal,
  • Sneha Ray,
  • Biru Rajak,
  • Aritra Acharyya,
  • Gurudas Mandal,
  • Arindam Biswas,
  • Rudra Sankar Dhar

摘要

This paper investigates the design optimization of GaN-based Terahertz (THz) IMPATT sources using Artificial neural networks (ANNs). Traditional design optimization of IMPATT sources, relying on the self-consistent quantum drift–diffusion (SCQDD) model, is computationally intensive, manual and time-consuming. To overcome these challenges, we developed ANN models trained on simulation data from SCQDD-based DC and large-signal simulations. Forward ANN models predict critical performance parameters, including power output and efficiency, for GaN IMPATT diodes with flat and hi-lo doping profiles. These models demonstrated high accuracy when validated against experimental data. In addition, inverse ANN models were developed to optimize diode design parameters based on specific target performance metrics, significantly reducing the computational resources and time required for optimization. The proposed ANN-based approach allows for efficient design optimization of GaN IMPATT diodes across a wide frequency range, including the THz band, addressing the growing demand for high-performance THz radiation sources. This work introduces a scalable method for the rapid development and commercialization of GaN THz devices, advancing high-frequency applications.