Modeling of III-Nitride Devices for Device-Circuit Co-optimization under Cryogenic Environments

Modeling of III-Nitride Devices for Device-Circuit Co-optimization under Cryogenic Environments Abstract: Electronic systems used in quantum computing, space missions, superconducting equipment, and high-performance computing must operate reliably at temperatures far below those encountered by ordinary electronics. Electronic devices made from gallium nitride are well suited for high-power and high-frequency operation. Certain gallium nitride transistor structures are also promising at very low temperatures because their current-carrying electron channels remain robust and electron mobility can increase as temperature decreases. However, engineers lack reliable and efficient tools for predicting how cooling changes electron motion, heat generation and flow, and sensitivity to manufacturing variations. Measurements under these conditions are difficult and expensive, while existing simulations may require extensive data and long computing times. This project will create an integrated framework that connects the physical behavior inside gallium nitride devices with the performance of complete electronic circuits. The framework will help identify device structures and material choices that improve performance while limiting fabrication complexity, cost, and reliability risk. By reducing dependence on repeated low-temperature experiments and making design decisions more systematic, the research may accelerate the development of reliable electronics for extreme environments. The project will publicly share software artifacts and educational materials and engage students from K-12 through graduate study to advance knowledge in microelectronics, device modeling, and circuit simulation. The research will develop a hierarchical device-to-circuit co-optimization framework for gallium nitride devices operating from room temperature to cryogenic (or extremely low-temperature) conditions. First, the research team will develop a cryogenic physical compact model by extending an existing quasi-ballistic model for gallium nitride high-electron-mobility transistors. The model will then be adapted to fin-shaped high-electron-mobility transistors and nanowire devices and used for circuit simulation. Across this broad temperature range, the model will describe the transition from scattering-dominated transport to quasi-ballistic and ballistic transport, in which electrons undergo few or no scattering events, as the average distance an electron travels between scattering events approaches the device channel length. It will also account for cryogenic changes in electron mobility, dominant scattering mechanisms, high carrier concentration, interface roughness, charge trapping, self-heating, and the nonlinear variation of thermal conductivity with temperature. Second, a full-band Monte Carlo solver, a simulation method that statistically tracks electron motion and scattering using a detailed representation of the material’s allowed electron energies, will be coupled with calculations connecting electrical transport, heat generation, and heat flow. Quantum-mechanical and electrostatic solvers will evaluate electrical charge created by material polarization, the restriction of electron motion by very small device dimensions, and electron mobility in structures that confine motion in one or two spatial directions. Experimental measurements and numerical device simulations will calibrate and validate current-voltage behavior, charge-voltage behavior, small-signal response, and high-frequency performance. Third, physics-informed machine learning, in which training is constrained by semiconductor equations, will create fast mathematical substitutes for costly simulations from the combined datasets. Bayesian optimization, a statistical method for selecting promising designs with relatively few costly evaluations, will identify device parameters that balance circuit performance, manufacturing complexity, process variation, reliability, and cost. Candidate designs will be reevaluated with coupled electrical and thermal simulation, incorporated into the compact model, and tested through circuit-level simulation. The central intellectual contribution will be a predictive and data-efficient connection among microscopic transport physics, practical circuit models, and automated device design for extreme-temperature electronics. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.