Predictive device modeling for radiation effects through machine learning

Alphacore will evaluate and develop a new approach to multi-scale modeling of radiation effects in electronic device technologies based on novel material systems. Our approach aims to directly correlate nano-scale properties of novel materials systems with macro-scale electrical properties of devices constructed with those materials, and their radiation response. Radiation hardness assurance (RHA) involves several activities to ensure that the microelectronic systems perform to their design specifications under specific radiation requirements. An important but costly step is testing technologies to the mission established radiation hazards; the introduction of new technologies makes this an ongoing challenge, with fewer shortcuts to understanding basic device radiation response, and with increasingly heavier reliance on modeling and simulation. Alphacore’s proposed modeling approach will demonstrate the feasibility of using machine learning (ML) algorithms to develop models for predicting trends in the radiation response of devices. This method will use experimental data from high resolution microscopy and defect spectroscopy to develop a set of “nano-scale features.” Combined with device technological aspects, details of the operating radiation environment and electrical characteristics from device testing will be inputs to ML algorithms. Model parameters will be obtained using optimization routines to achieve a desirable prediction accuracy of device electrical properties. ML algorithms will also be evaluated for the classification of radiation hardness, and to support hardening techniques. This effort is of importance to the DoD microelectronics activities, since many new material systems are being considered for different electronic applications, and their radiation susceptibilities are unknown. The proposed material system for the Phase I feasibility study is β-Ga2O3 (beta-phase gallium oxide), a wide bandgap material attracting increasing interest for power electronics and truly solar-blind ultraviolet (UV) photodetectors. Beside its superior thermal and chemical stabilities, the effects of radiation damage on Ga2O3 are of fundamental interest in space-based and some terrestrial applications. Ga2O3 is of significant interest for high-power rad-hard electronics because of its large bandgap (4.8 eV), high-breakdown electric field (8 MV/cm), and its low-cost, mass-producible, high-quality, single-crystal growth methods. The types of devices that will be used in this study are β-Ga2O3 Schottky-barrier diodes (SBD).