An open-source CD-SAXS library framework for parallelized calculations to facilitate the modeling and uncertainty calculations of complex material stacks

Critical Dimension Small-Angle X-ray Scattering (CD-SAXS) has emerged as a powerful metrology for characterizing nanoscale structures in semiconductor devices, offering sub-nanometer precision and non-destructive probing capabilities. However, the complexity of CD-SAXS data analysis demands robust computational tools to accurately translate scattering data into structural information. Current open-source solutions lack the necessary efficiency, modularity, and accessibility to keep pace with evolving metrology needs. This paper presents cdsaxs, a new open-source Python package designed to address these challenges. cdsaxs features a modular architecture that decouples simulation and fitting components, enabling flexible model integration and optimized performance. The package implements key functionalities for CD-SAXS analysis, including CMA-ES-based optimization, MCMC-driven uncertainty quantification, and correlation analysis. We demonstrate the capabilities of cdsaxs through the implementation and fitting of representative nanostructure models, validating its accuracy against both synthetic and experimental datasets. Performance benchmarks showcase significant improvements in computational speed compared to existing solutions, particularly with GPU acceleration. cdsaxs provides a valuable resource for the CD-SAXS community, offering an efficient, extensible, and accessible platform for advancing nanoscale metrology.