Explicit Function Model of Electromagnetic Reliability for CMOS Inverters Under HPM Coupling Based on Physical Mechanism Analysis and Neural Network Algorithm

Currently, commonly used compact device models for CMOS inverters lack an explicit functional model that can efficiently describe the specific physical processes of HPM-coupled voltage excitation damaging the inverter. This paper proposes an explicit function model that describes the failure process of CMOS inverters under HPM irradiation conditions using neural network algorithms and physical mechanism classification. First, the coupling paths and mechanisms were simulated and verified, proposing a parameterized description of HPM-coupled voltage excitation. Then, based on the source of burnout power, the burnout mechanisms were studied and classified in detail into two categories: direct burnout and latch-up burnout, serving as the physical mechanism logic basis for the model. A two-stage cascade neural network model was established based on the physical logic. The first stage model predicts the working status and burnout type of the CMOS inverter, with a classification accuracy of 99.2%. The second stage model branches predictions based on the burnout type, with prediction indicators being burnout time and peak power supply current. The prediction accuracies for the two types of burnout situations were 98.5%, 99.2%, and 97.8%, 99.5%, respectively. Finally, the model performance was evaluated, with detailed discussions on the advantages of the proposed model in terms of different performance.