Percorrer por autor "Serol, Miguel de Lima Caliço"
A mostrar 1 - 1 de 1
Resultados por página
Opções de ordenação
- Chemometric profile of gunpowderPublication . Serol, Miguel de Lima Caliço; Família, Carlos; Ahmad, Samir Marcos Esmail; Quintas, AlexandreThe forensic investigations regarding firearm-related crimes are of paramount importance, focusing mainly on visual analysis of marks in firearm-related elements. When the cartridge or the projectile, printed with specific marks from the firearm who fired the round, are not available, the analysis needs to find alternative approaches to continue with the forensic investigation. This often leads to the chemical analysis of the compounds created when the firearm is discharged. With the popular and political pressure on manufacturers, the most characteristics elements of gunshot residue, such as lead, antimony and barium, are quickly being removed from the formulations and replaced with others that grant gunshot residue chemical analysis less probatory value. This led to an increase in studies focused on the organic compound profiles of the ammunitions. These are mainly located in the propellant – smokeless gunpowder. The most common procedures for the profile analysis of chemical compounds in gunshot residues are spectroscopy and chromatography. In the present work, we propose the use of solid-phase microextraction in combination with Gas Chromatography-Flame Ionisation Detection and Attenuated Total Reflectance- Fourier Transform Infrared Spectroscopy to evaluate their discriminative capability for the determination of chemical profile of different gunpowder samples. The results showed visible differences among samples from different manufacturers and models. The obtained data was used in for training and validation of predictive models, with the objective of identifying the manufacturer and model of specific ammunitions. The results showed an overall accuracy of around 60% when classifying data not used for the predictive model. Problems in data acquisition may have harmed the predictor's accuracy, while overfitting of the models is also a possibility. Nevertheless, these results showed that the analytical approaches and predictive model herein proposed have great potential for identifying specific ammunition manufacturers and models.