Forensic glass analysis by ICP-MS: a multi-element assessment of discriminating power via analysis of variance and pairwise comparisons

Abstract

Glass fragments from 81 automobile side windows were collected and analyzed by the FBI Laboratory using ICP-AES in 1991. The FBI selected 9 elements (Al, Ba, Ca, Fe, Mg, Mn, Na, Sr and Ti) to use for discrimination among the glass samples. This multi-element discrimination showed a significant improvement in the discrimination statistics over using only refractive index (RI) measurements. Oak Ridge National Laboratory (ORNL) recently analyzed fragments from 76 of the original side window fragments using inductively coupled plasma mass spectrometry (ICP-MS). The ICP-MS analyses measured 45 elements using a hierarchical sampling scheme to estimate variances due to sampled population (V_P), variance due to sample dissolution and within sample heterogeneity (V_D), and variance due to replicate measurements (V_M). The between-to-within ratio [B/W = V_P/(V_D + V_M)] afforded a measure of the variance within the population to that in the analytical measurement, providing a first approximation of the discriminating power of each element. Florida International University updated the RI measurements on 72 available glass fragments. These RI measurements along with ICP-AES and ICP-MS elemental analyses were used for pairwise comparisons of all possible pairs of the 72 glasses that had a complete set of measurements. The pairwise comparisons used Tukey's HSD method to compare RI and element-by-element discrimination potential of ICP-AES and ICP-MS for analyzing glass in forensic casework.