Answers are grounded in this paper and cite their sources.
Soft errors induced by ionizing radiation constitute a major reliability concern in contemporary semiconductor technologies, particularly in aerospace, automotive, and high-performance computing applications. As technology scaling continues, reduced operating voltages and node capacitances significantly increase the susceptibility of integrated circuits to radiation-induced charge collection. Although physics-based diffusion–collection models provide accurate insight into the underlying mechanisms of soft errors, their numerical solution leads to large-scale dynamical systems that are computationally expensive, especially when operating conditions such as temperature vary. In this work, we extend compact soft error rate (SER) modeling by incorporating temperature-dependent charge transport into a parametric model order reduction (MOR) framework. Starting from the diffusion–collection equation, temperature dependence is explicitly introduced through key physical parameters, resulting in a temperature-parameterized state-space formulation. To efficiently handle temperature variability, we propose an efficient parametric model order reduction (PMOR) methodology based on the extended Krylov subspace (EKS) combined with moment matching (MM), enabling the extraction of reduced-order models that remain accurate across a range of operating temperatures. The proposed parametric reduced models significantly reduce computational complexity while preserving the fidelity of temperature-dependent charge collection dynamics. Experimental evaluation on large three-dimensional benchmarks demonstrates that the reduced models achieve substantial speedups compared to full-order simulations, while maintaining high accuracy over the examined temperature range. These results indicate that the proposed approach enables fast and reliable temperature-aware SER estimation, making it suitable for large-scale reliability analysis under realistic operating conditions.
No takes yet. Share an insight, caveat, or question.
Stoikos et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: