Glass fragments are frequently recovered during criminal investigations as they can provide links between the crime scene, suspects, or victims. The elemental composition of glass is often used to determine if there is a source commonality between glass samples. Micro-X-ray fluorescence spectrometry (μXRF) is a standard technique for elemental glass comparisons due to its high informing power, low cost, rapid analysis time, and non-destructive nature. In recent years, advancements in μXRF technology, such as silicon drift detectors (SDD), have improved precision and analysis time and increased adoption in crime laboratories. Furthermore, the superior precision afforded by SDD technology has been hypothesized to allow for accurate analysis of much thinner glass fragments, expanding its applicability to casework scenarios where smaller fragments are encountered. This study compares results for the μXRF analysis of full-thickness (≈2 mm) and thin glass fragments (10 to 50 μm) for different types of float glass. The proposed modified 3s comparison criterion results in a false exclusion rate of less than 2.5% and a false inclusion rate of less than 1.5% for full-thickness fragments. Thin fragments yielded false exclusion and false inclusion rates of less than 12% and 7.5%, respectively. A spectral similarity metric, spectral contrast angle ratio (SCAR), was tested to quantitatively evaluate spectral similarity, achieving accuracies of greater than 98% and 91% for full-thickness fragments and thin fragments, respectively. These findings show that while full-thickness fragments produce more precise data, μXRF-SDD is suitable for fragments as thin as 10 μm.
(Publisher abstract provided.)