Stralingsgeïnduceerde transiënte effecten in CMOS-circuits

Technology scaling has significantly increased the susceptibility of nanoscale CMOS circuits to radiation-induced transient effects. Reduced node capacitances, lower supply voltages, and stronger device nonlinearity amplify the impact of single-event transients (SETs), particularly in time-dependent circuits such as oscillators. While compact SET models exist at the logic level, they often rely on fixed-shape current sources that neglect voltage dependence and extended charge collection. At the circuit level, classical Impulse Sensitivity Function (ISF)-based phase models assume small-signal perturbations and fail to capture large nonlinear disturbances. A unified modeling framework linking device-level radiation physics to nonlinear phase-domain circuit behavior is therefore required for predictive radiation-aware design. This dissertation develops and experimentally validates a multi-level modeling methodology for SET analysis in 65\,nm CMOS technology, spanning calibrated TCAD simulations, voltage-aware compact modeling, and nonlinear phase--amplitude oscillator analysis. At the device level, detailed TCAD models of 65\,nm CMOS nMOS and pMOS transistors were calibrated against foundry SPICE data, including DC characteristics, capacitances, and FO4 delay benchmarks. Heavy-ion simulations were performed to capture drift- and diffusion-dominated charge collection. The extracted SET waveforms exhibit strong voltage dependence and pulse broadening effects that are not captured by conventional double-exponential models. Based on these observations, a voltage-aware analytical abstraction was developed in which the SET current is decomposed into drift and diffusion components with bias-dependent parameters. The resulting closed-form expressions were implemented in Verilog-A, enabling dynamic adaptation of the SET waveform to the instantaneous terminal voltage during circuit simulation. Experimental validation using heavy-ion and laser measurements in inverter and oscillator structures confirmed the predictive accuracy of the model. Building upon the validated current model, radiation-induced timing errors in ring oscillators were investigated. Classical ISF-based phase perturbation theory assumes linear superposition and negligible amplitude coupling, limiting its validity under large transient disturbances. To overcome this limitation, a Nonlinear Phase-Coupled (NPC) ISF framework was developed. The model propagates phase deviation recursively by dynamically updating the oscillation time origin, thereby capturing nonlinear phase feedback beyond the small-signal regime. To account for envelope perturbations and amplitude-to-phase conversion, the framework was extended to a Nonlinear Phase--Amplitude Coupled (NPAC) model. Building on Hajimiri's original oscillator phase-noise framework, the model employs both the Phase Impulse Sensitivity Function (P-ISF) and the Amplitude Impulse Sensitivity Function (A-ISF), while introducing an amplitude state with memory and exponential decay. The evolving amplitude modulates the effective phase sensitivity through a calibrated conversion factor obtained from unit-impulse characterization. This coupled formulation captures both direct phase shifts and secondary AM-to-PM effects, showing strong agreement with full transient SPICE simulations under worst-case PVT conditions. Finally, the proposed framework was applied to analyze radiation sensitivity in frequency-tuned ring oscillators using stage count variation, gate length scaling, and load capacitance adjustment. The results reveal clear trade-offs between oscillation frequency, power consumption, and SET-induced timing error. By incorporating power into the evaluation, a quantitative figure of merit was proposed to compare resilience-efficiency trade-offs across tuning techniques. In summary, this dissertation presents a physics-aware and nonlinear phase-domain modeling framework for radiation-induced transient analysis in nanoscale CMOS oscillators. By bridging calibrated TCAD device simulations, voltage-dependent compact modeling, and coupled phase--amplitude dynamics, the work enables predictive simulation of large-signal transient effects and provides systematic design insights for radiation-resilient oscillator architectures.