Through the Fire and Flames: Addressing Challenges in Wildfire Debris Analysis
Date
2024-04-19
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
ORCID
Type
Thesis
Degree Level
Doctoral
Abstract
Wildfires are a global concern with increasing occurrence and areas burned. Associated costs do not only apply to individual fires, such as the estimated 10 billion US dollars for the 2016 Fort McMurray fire, but also to fire prevention efforts as over 50% of wildfires are caused by human activity. Arson, which is the deliberate or reckless starting of a fire resulting in ecosystem and infrastructure damages or loss of life, is a major crime. Forensic analysis of fire debris plays a key role in determining the cause and origin of a fire, requiring meticulous analysis of evidence to determine if the fire was started intentionally. One vital aspect of these investigations involves the detection and identification of Ignitable Liquid Residue (ILR), the remnants of accelerants. Arson cases have one of the lowest conviction rates for major crimes with a large percentage of unsolved cases since the presence of trace levels of ILR in fire debris samples are frequently obscured by abundant matrix compounds with similar physico-chemical attributes. Three main challenges were identified in conventional ILR analysis in the context of wildfires: Interferences coupled with matrix composition and relative signal abundance of ILR target compounds, classification according to the American Society for Testing and Materials (ASTM) standard, and potential for cross-contamination stemming from long transport and storage times.
The application of design of experiment (DoE) principles to method development of flow-modulated comprehensive multidimensional gas chromatography coupled with time-of-flight mass spectrometry (GC×GC-ToF MS) for ILR analysis to address interference-related issues are discussed. Both, hardware (column selection and modulator settings) and GC run parameters (oven programming, inlet pressure, and modulation period) were optimized. For hardware, carbon loading potential, dilution effect, target peak amplitude and skewing effect were evaluated. GC run parameters optimization compared DoE approaches (Box-Behnken and Doehlert designs) to assess sensitivity, selectivity, peak capacity, and wraparound; alongside target peak retention, resolution, and shape evaluation. Sample alignment is addressed via the development of retention time indices (RI) to facilitate ASTM classification and resulting advantages for sample comparison across arson cases and laboratories. Using a combination of two well-established GC RI systems, non-isothermal Kovats index and Lee index, certified standards and simulated wildfire debris are used in the performance verification for ILR classification prior to performance validation on wildfire scene samples. Lastly, the thesis explores the importance of sample integrity by employing targeted and untargeted chemometric analysis to detect and characterize cross-contamination for various sample matrices in a controlled environment. It investigates the potential for associated false positives, with an outlook on potential quantitative analysis and source appointment for future development, before analysing the impacts of different packaging and storage methods.
Both DoE models operated in the optimal zone after hardware optimisation. The final method developed from this research separated all target compounds successfully from varied matrix compositions without wraparound for compounds with at least four aromatic rings. During validation, remaining co-elutions could be resolved with a deconvolution algorithm. For alignment, the developed RI system showed very good correlations to predicted values (r2 = 0.97 in first dimension, r2 = 0.99 in second dimension) and was valid for a wide range of analyte concentrations and operational settings (coefficient of variance (CV) < 1% in first dimension, < 10% in second dimension). It resulted in 86% coverage of total chromatogram area, leading to the successful development of an ILR contour map. The cross-contamination investigation showed a notable potential for false positive identification with gasoline compound transmission detectable after a 1-hour exposure, and a full profile transfer after 8-hour exposure. Matrix interaction effects were observed in the form of inherent native compound interference as well as adsorbate-adsorbate interaction during transmission and extraction. Chemometric analysis allowed for distinction between negative, positive, and contaminated samples with classification confidences of 88% for targeted and 93% for untargeted to 95% for diagnostic ratio analysis of three ratios deployed in tandem. Employing packaging reduced the extent of cross-contamination by varying degrees. Nylon-based packaging performed better than polyethylene-based packaging since the material itself emitted interfering compounds. Heat-sealing was the most reliable sealing mechanisms, and refrigerated storage offered several advantages. While double-packaging is a recommended practice, triple-packaging did not show significant benefits.
This thesis successfully developed several tools for ILR analysis, including a GC×GC-ToF MS method which fulfills and exceeds the current ASTM requirements, an ILR classification contour map utilising the roof-tiling effect of the developed method, and chemometric tools to evaluate cross-contamination. The implications for environmental and civil engineering performance are numerous, particularly as relating to the paramount safety of the public and environment in the conduct of engineering projects, particularly those in increasingly vulnerable wildfire and arson-associated regions.
Description
Keywords
GCxGC, ILR, chemometric
Citation
Degree
Doctor of Philosophy (Ph.D.)
Department
Civil and Geological Engineering
Program
Civil Engineering