On data quality and optimal exposure time in powder diffraction experiments

by Dr Vladislav Kochetov (FRMII TUM)

Monday 3 Nov 2025, 14:30 → 15:30 Europe/Berlin

PH HS 3 (Physics Department)

Determining the optimal exposure time in powder diffraction experiments is essential for balancing data quality and experimental efficiency. While longer exposures improve signal-to-noise ratio and the reliability of Rietveld refinement, they can also lead to impractical measurement times and potential sample degradation, particularly relevant for synchrotron and neutron sources. Despite the importance of this parameter, few systematic studies directly link exposure time with refinement quality.

In this talk, I will address that gap by investigating how exposure time influences key data quality indicators and refinement outcomes in both real and simulated powder diffraction datasets. By combining analytical and statistical methods with machine learning approaches, we propose a strategy for predicting optimal exposure times for a given instrument. In particular, we explore semi-supervised learning strategies that leverage large volumes of unlabeled theoretical data to enhance model generalization. The resulting methodology aims to improve experimental throughput and reliability while maintaining the precision required for robust Rietveld analysis.