Development of a spectral X-ray fluorescence database to strengthen the scientific foundations for the forensic analysis and interpretation of modern soda-lime glass

This study offers scientific support for the updated standard for micro X-ray fluorescence (μXRF, ASTM E2926) used in forensic glass analysis, reflecting recent changes in modern soda-lime glass formulations and newer μXRF systems with silicon drift detectors (SDDs). To develop this standard, an extensive database has been developed, showing changes in the elemental profiles of modern glass and the forensic implications of the improved analytical capabilities of SDDs. The database comprises over 2000 fragments from 112 known sources of vehicle and architectural windows, including glasses treated with laminating, tempering, and e-coating techniques. The study reveals that different sampling and comparison criteria are required due to the improved precision and sensitivity of SDD and variations in spatial heterogeneity of glass. The comparison interval recommended for older systems with SiLi detectors produces higher false exclusion rates (up to 15 %). Therefore, a modified 3 s comparison criterion is recommended for SDDs to minimize error rates. The modified 3 s criterion accounts for data with precision lower than 3 % RSD by defining the range of the known sample as three times the standard deviation of the measurements or three times 3 % of the mean, whichever is greater. This criterion yielded false exclusion and inclusion rates less than 3 % and 0.7 %, respectively for various sets of full-thickness glass fragments when using at least ten fragments to characterize the heterogeneity of the known sample. The findings raise awareness of the detection of distinctive elements in the nanometer-thick e-coated layer and lead to recommendations for addressing spatial variations. • This study provided the framework for updates to a test method for μXRF glass analysis ASTM E2926–25 • A database of 112 glass sources was analyzed by μXRF with modern SDD • Error rates for soda-lime glass comparisons using silicon drift detectors are assessed • A modified 3 s comparison interval for element ratios produced the best results • A similarity metric includes within and between source variation to support comparisons