Low Resistance Asymmetric III-Nitride Tunnel Junctions Designed by Machine Learning.

Area of Science:

• Materials Science
• Semiconductor Physics
• Device Engineering

Background:

• Tunnel junctions (TJs) are critical components in III-nitride electronic and optical devices.
• A key challenge in TJ design is minimizing resistance at high current densities.

Purpose of the Study:

• To propose and investigate a novel asymmetric p-AlGaN/i-InGaN/n-AlGaN tunnel junction structure.
• To develop a machine learning model for predicting TJ resistance.

Main Methods:

• Simulations of p-AlGaN/i-InGaN/n-AlGaN TJs using TCAD software with varying compositions and layer thicknesses.
• Development of a machine learning model trained on simulation data for real-time TJ resistance prediction.
• Prediction of resistances for 22,254 distinct TJ structures.

Main Results:

• An asymmetric TJ structure (p-Al_0.7Ga_0.3N/i-In_0.2Ga_0.8N/n-Al_0.3Ga_0.7N) demonstrated seven times lower resistance compared to a symmetric counterpart.
• The machine learning model enables efficient, real-time prediction of TJ resistance.
• Optimized Al composition in the p-layer significantly impacts TJ resistance.

Conclusions:

• The proposed asymmetric TJ structure offers a significant improvement in reducing resistance for III-nitride devices.
• This work introduces a new paradigm for designing efficient III-nitride tunnel junctions.
• The developed machine learning tool accelerates the design and optimization process for future devices.