Premium
Novel LIBS method for micro‐spatial chemical analysis of inorganic gunshot residues
Author(s) -
MenkingHoggatt Korina,
Arroyo Luis,
Curran James,
Trejos Tatiana
Publication year - 2021
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.3208
Subject(s) - computer science , sample (material) , laser induced breakdown spectroscopy , sampling (signal processing) , pattern recognition (psychology) , laser ablation , artificial intelligence , data mining , biological system , laser , chemistry , chromatography , optics , physics , computer vision , filter (signal processing) , biology
This study developed a reliable laser‐induced breakdown spectroscopy (LIBS) screening approach capable of detecting GSR in just a few minutes with minimal damage to the sample, high specificity, and sensitivity. Moreover, a novel micro‐sampling method was developed to gather three‐dimensional data of the simultaneous occurrence of IGSR markers from a discrete space. The method is capable of micro‐spatial chemical analysis from just two laser shots fired at an area of 100‐μm diameter. The performance of the micro‐spot method is compared with our previously published bulk‐line method. Superior accuracy, spatial information of IGSR distribution in the sample, and a less invasive sampling are some of the advantages of the newly proposed method. A benefit afforded by this approach is the use of the universal hand's collection method currently used by practitioners, while leaving over 99% of the stub left unaltered for further analysis. Machine learning algorithms were used for the classification of samples derived from shooters' hands versus nonshooters hands, based on their LIBS spectrochemical data. Four different approaches—critical threshold, logistic regression, naïve Bayes, and neural networks—were applied to examine the performance and accuracy of two different ablation patterns (micro‐spot and bulk‐line mode). A validation set of 326 samples originated from 51 nonshooters and 56 known shooters resulted in an overall accuracy between 87% and 100%, depending on the ablation pattern and the type of prediction model applied. The incorporation of this rapid screening and statistical decision‐making approach could offer more efficient case management in firearm‐related investigations.