Modeling and optimization of HS-IMPATT diode execution enriched with Si/SiC using ANN

This study offers a precise, expandable, and effective ANN (Artificial Neural Network) model to evaluate and calculate the important device parameters like breakdown voltage, efficiency, negative conductance, negative resistance, susceptance, and RF power of a heterostructure Si/3C-SiC-based IMPATT diode at an operating frequency of 94 GHz. The authors have compared the simulation and optimization of a Si/SiC-based heterostructure IMPATT diode with the neural network techniques for CW operation. The experimental data are almost 85 % to 90 % the same as computer simulation outcomes and provide numerically agreed results regarding breakdown voltage, efficiency, negative conductance, and power. Owing to several factors such as temperature, parasitic impacts, and appropriate hit sink arrangements, there is a 10 %–15 % discrepancy between the theoretical simulation result and the experimental output. This newly developed ANN technique, developed by the authors for the first time, was found to be in close agreement with the experimental findings available at 94.0 GHz. The simulation result gives the breakdown voltage of the IMPATT device as 188 V as compared with the experimental results of 185 V. Similarly, the neural network model shows approximately 183 V. The RF power of the simulated device is 2.5 W as compared to the experimental result of 2.0 W at 94 GHz, whereas the neural network model gives 2.2 W, which shows the validity of the model. The assessed outcomes clearly demonstrate the effectiveness of the device parameter estimations and optimizing IMPATT device design efficiently, and the findings will benefit missile and radar technology.