Boston University
Principal component analysis of gasoline DART-MS data in the low-mass region
Abstract
dc:description.abstractGasoline is the most commonly used accelerant in arson, partly because it is widely accessible and inexpensive. The gasoline distributed to stations is not pure; it contains additives and blending agents in low percentages (≤1%). Gasoline mixing is a phenomenon in which new shipments of gasoline are added to residual amounts of previous batches in varying percentages. Each batch has a slightly different composition of additives. As mixing continues over time, gasoline develops a chemical attribution signature (CAS) due to the blending of slightly different batches at varying percentages. It is predicted that CASs can be used for source attribution by analyzing the mass spectral data of the additives obtained using Direct Analysis in Real Time–Mass Spectrometry (DART-MS) and Principal Component Analysis (PCA). The mechanism of DART used in this research generates protonated intact molecular cations through proton transfer, and is much more efficient at detecting additives than electron impact with Gas Chromatography-Mass Spectrometry (GC-MS). The ability to differentiate gasoline samples of the same brand using the mass spectral data of sample replicates has shown promise when analyzing the data with principal component analysis (PCA) in the range of 50-1000 mass-to-charge ratio (m/z). This study aimed to examine the mass spectral data solely in the low-mass range (0-400 m/z) to determine whether PCA analysis and Leave-One-Out Cross-Validation would indicate that the low-mass region is sufficient for differentiating between gasoline samples of the same brand. The low-mass region contains the more volatile analytes, and because of this, this region is more amenable to current American Society for Testing and Materials (ASTM) extraction methods and subsequent DART-MS analysis. Smaller molecules are more likely to enter the headspace and adsorb to a charcoal strip and be readily desorbed and ionized by the DART ionization source. A re-analysis of the same data collected from across five counties in Massachusetts using PCA analysis on the Mass Mountaineer™ © software supported the theory that source attribution using data collected in the low-mass region can accurately classify unknown samples. Problems arose during the blind studies, where the best results achieved a 100% precision, meaning that all replicates of the same sample were classified in the same class; however, the accuracy of the results was only 20%. However, this does not mean that source attribution is not possible using data from the low-mass region, as other factors are predicted to be at play. Further research will be necessary to thoroughly assess these issues.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- masters
- Discipline thesis:degree_discipline
- Biomedical Forensic Sciences
- Grantor
- Boston University
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Giambusso, Briana
- Advisor dc:contributor.advisor
-
- Hall, Adam B.
- ORCID dc:identifier.orcid
Subjects
dc:subject × 9Rights
dc:rights- Statement dc:rights
-
- Attribution 4.0 International
- Licence dc:rights.uri
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/53163