Akademik Veri Yönetim Sistemi
Radiation Measurements, cilt.189, 2025 (SCI-Expanded)
Kinetic models based on deterministic ordinary differential equations (ODEs) are effective for macroscopic systems, but a breakdown of their foundational assumptions is observed for single-grain and nanodosimetric applications where particle numbers are small. In this study, the metrological consequences of this stochastic-deterministic divergence are quantitatively investigated. The study's primary contribution is to deconvolve the divergence into two distinct components: a robust, inherent imprecision and a model-dependent, systematic inaccuracy. Intrinsic physical stochasticity is confirmed to be a dominant source of imprecision, generating an irreducible dose uncertainty of over 20 % that accounts for a significant fraction of single-grain overdispersion. Conversely, a systematic inaccuracy (e.g., a >50 % dose bias), initially observed in a simplified model, is demonstrated to be a methodological artifact, not a universal consequence of discreteness. It is shown that this systematic bias can be reduced to negligible levels (<1 %) by using either a more physically realistic multi-trap model or a correctly specified simple model. Based on this deconvolution, analytical protocols are assessed. An “Average-Dose-First” protocol is identified as the superior method, as it provides an accurate final dose estimate while correctly propagating measurement uncertainty. A general framework for understanding and partitioning variance in luminescence data is thereby established. Practical recommendations are provided for improving the accuracy of modern luminescence science by selecting appropriate models and using correct statistical protocols, with a strong emphasis on the critical need for model validation.