Identification of Factors Affecting SEM/EDS Analysis for Discrimination and Classification among Common Items of Evidence using Particle Combination Profiles

Abstract This article presents key factors affecting the analysis of particle profiles as used for discrimination and classification among items commonly collected at crime scenes. Identification of these factors is a necessary step to enable systematic improvement, optimization, and transition to practice. Prior research, employing reasonable initial choices of analytical and statistical parameters, has (1) demonstrated the presence of highly discriminating sets of very small particles (VSP) on the surfaces of items commonly collected at crime scenes, (2) developed statistically rigorous measurements of correspondence between VSP profiles, and (3) produced objective measures for the resulting probative value. In the present work the analytical and statistical parameters were examined more critically, identifying key factors affecting method performance. Experiments were conducted using scanning electron microscopy (SEM) with energy dispersive x-ray elemental analysis (EDS) to characterize the elemental composition of thousands of individual particles within each specimen. The experiments studied: Reproducibility of VSP analyses at given parameters Effects of the SEM/EDS parameters used for the detection of each particle Effects of SEM/EDS x-ray analysis parameters used for elemental analysis of each particle Effects of the number and choice of elements used in the elemental analysis Effects of particle size on the strength of correspondence between particle sets Effects of data filtration parameters on the strength of correspondence between particle sets The experiments confirmed the presence of abundant, highly discriminating VSP on items commonly collected at crime scenes. The numbers of particles available for analysis was not a limiting factor: many more particles (usually greater than 50 times more) were present than were used for the analysis. A very high level of reproducibility was observed. Many of the parameters tested had no measurable effect on particle combination analysis performance and others had minor or interactive effects. Four factors were identified as having significant impact on the strength of correspondence between particle profiles, three factors were identified as having a significant impact on the numbers of particles detected and nine factors were identified as having a significant impact on analytical time and costs. The approach in its current state of development offers crime laboratories an additional capability suitable for high priority cases. The identification of key factors affecting performance of the VSP analytical protocol allows existing methods to be further developed and systematically improved to facilitate transition to routine practice.