Petro Yavorskyi Ab initio Method for the Certification of Semiconductor Materials for Dosimetric Research. – A Qualification Scientific Work in Manuscript Form. Dissertation Submitted for the Degree of Doctor of Philosophy in Speciality 104 “Physics and Astronomy” – Institute of Electron Physics of the National Academy of Sciences of Ukraine, Uzhhorod, 2025.
The dissertation is devoted to the development of a theoretical framework and methodology of the ab initio certification of semiconductor materials for dosimetric studies, integrating contemporary methods from the theory of dosimetric and luminescent phenomena in semiconductors, numerical modelling, and the software implementation of the Lumini computational package.
The first chapter analyzes theoretical foundations and practical achievements of contemporary dosimetry based on thermoluminescence in irradiated wide–band gap semiconductors. It outlines the interaction of ionizing radiation with semiconductors, mechanisms of energy storage in solids, and criteria for selecting dosimetric indicators. Particular attention is given to energy–level schemes responsible for electron–hole recombination and thermoluminescence, the OTOR (One Trap–One Recombination) model, and existing theoretical methods. Requirements for semiconductors as dosimetric materials are discussed, including chemical composition, band gap, and effective atomic number. It is shown that traditional approaches to thermoluminescent analysis are insufficient for reliable material certification, thus requiring ab initio techniques.
The second chapter introduces the scaling method, proposed for the first time for ab initio solutions of differential equations describing thermoluminescent characteristics of irradiated semiconductors. Scaling transformations reduce OTOR model parameters to a dimensionless form, enabling effective use of numerical algorithms such as Runge–Kutta. The chapter also presents a novel application of the Monte Carlo method to account for the stochastic nature of microprocesses in transitions between the conduction band and trapping/recombination levels. This approach allowed realistic modeling of dosimetric processes. The scaling procedure proved essential for reformatting OTOR equations into a dimensionless representation, simplifying the fitting of theoretical and experimental data, and reducing dependence on computational techniques while ensuring universality of the simulation code.
The third chapter presents the implementation of scaling procedures into the Lumini computational package, developed for comprehensive ab initio modeling of thermoluminescence in irradiated semiconductors, optimization of dosimetric indicators, and material certification. The architecture and functional capabilities of Lumini are described, with examples ranging from purely theoretical simulations to processing of experimental data. The package is shown to be a necessary platform for integrating experiment and theory in modern dosimetric studies and for employing artificial intelligence methods in data analysis.
The fourth chapter addresses the principles of certification of dosimetric materials to ensure accurate dose evaluation under diverse irradiation conditions. The creation of reliable databases of energy and kinetic parameters of dosimeters is emphasized, achievable through ab initio approaches. Attention is given to optimization criteria and AI algorithms capable of identifying parameter sets in best agreement with experimental data. Applications of machine learning, luminescence curve recognition, and multiparametric data analysis are considered. The use of neural networks, deep learning models, decision trees, and ensemble methods is shown to enhance both accuracy and efficiency of certification, while automating routine calculations.
The dissertation presents the first certification results for irradiated dosimetric materials LiF, Al₂O₃, and Li₂B₄O₇, demonstrating the effectiveness of the developed ab initio methodology. The research establishes a theoretical and computational framework that integrates physical modeling, numerical methods, and artificial intelligence, providing a reliable basis for the development of standardized dosimetric systems.
