Review: Hierarchical Cluster Analysis of Ignitable Liquids Based on the Total Ion Spectrum
Emily C. Lennert
ignitable liquid residue, fire debris, arson, ASTM E1618, standards, gas chromatography, mass spectrometry, GC-MS, aromatic products, gasoline, petroleum distillates, iso-paraffinic products, napthenic paraffinic products, normal alkane products, oxygenated solvents, hierarchical cluster analysis, HCA
Waddell, E. E.; Frisch-Daiello, J. L; Williams, M. R.; Sigman, M. E. Hierarchical cluster analysis of ignitable liquids based on the total ion spectrum. Journal of Forensic Sciences. 2014, 59 (5), 1198-1204.
ASTM. E1618-14 standard test method for ignitable liquid residues in extracts from fire debris samples by gas chromatography – mass spectrometry. West Conshohocken, PA: ASTM International, 2014.
The opinions expressed in this review are an interpretation of the research presented in the article. These opinions are those of the summation author and do not necessarily represent the position of the University of Central Florida or of the authors of the original article.
Ignitable liquid residues present in fire debris are typically analyzed by gas chromatography – mass spectrometry (GC-MS). Ignitable liquid residues are then assigned to one of the seven classes outlined by the ASTM E1618 standard, or to a miscellaneous class. The authors of this study used ASTM standard test method E1618-11; however, an updated version has since been published and the most recent version of the standard is ASTM E1618-14.2 The seven major classes outlined in the standard are aromatic products, gasoline, petroleum distillates, iso-paraffinic products, napthenic paraffinic products, normal alkane products, and oxygenated solvents. Samples that do not meet criteria for one of these seven classes are assigned to a miscellaneous category. The major classes, excluding gasoline, are subcategorized based on the carbon range observed into light, medium, and heavy subclasses. Samples are assigned to classes based on visual pattern recognition of the total ion chromatogram (TIC) from the GC chromatogram, with the interpretation based on the ASTM E1618 standard. This data interpretation is inherently subject to the skill and expertise of the analyst, due to the reliance on visual pattern recognition. The authors of this study present the use of the total ion spectra (TIS), i.e. average mass spectra across the chromatographic profile, as a means of objectively identifying ignitable liquid class. Using the TIS, samples were placed into chemically distinct groups corresponding to ASTM E1618 classes through hierarchical cluster analysis (HCA). HCA is a statistical technique that groups samples with a similar composition into the same group and samples with different chemical compositions into a separate group.
The dataset for this study was obtained from the National Center for Forensic Science Ignitable Liquids Reference Collection and Substrate databases. The dataset consisted of 436 unweathered ignitable liquid samples, 9 weathered gasoline samples, and 88 substrate samples. Weathered samples consist of ignitable liquids that evaporated under normal atmospheric conditions, which allow for the light chemical compounds to evaporate resulting in a weathered sample that primarily contains heavier chemicals. TIS were calculated for each sample. Cluster dendrograms from HCA and heat maps were generated from the data, in which samples with similar chemical composition clustered together.
As seen in the heat map in Figure 2 within the study, samples along the diagonal with the dark shading reflect samples that are highly similar. This demonstrates that a high degree of similarity is observed between samples within the same ASTM class. Samples in the miscellaneous and oxygenated classes, as well as the substrates, showed less similarity among samples within the same groups. The authors stated that the variation was expected due to the criteria for assignment to these classes. Miscellaneous samples are assigned to the category when the sample does not meet criteria for any other class or meets criteria for several classes, resulting in higher variation among miscellaneous samples compared to a more well defined class such as gasoline. Oxygenated solvent samples must contain at least one significant oxygenated component, but may also contain other compounds, which would be a source of variation within the class. The remaining classes have characteristics that allow for more defined classification, with greater similarity among samples within the group. The classes with the lowest variation, i.e. high similarlity, were considered for subsequent analysis: i.e. iso-paraffinic products, normal alkane products, napthenic-paraffinic products, petroleum distillates, gasoline, and aromatic products. These classes accounted for 282 samples, which were re-plotted in a heat map, seen in Figure 5 within the study. Two distinct groups were observed in this heat map: aliphatic and aromatic. Aliphatic compounds are compounds where the carbon atoms form open chains and there are no double bonds, i.e. non-aromatic. Aromatic compounds have rings and double bonds. Refer to Figure 1 (below) for an example of an aliphatic and aromatic compound.
Figure 1: Aliphatic and aromatic compound examples
Using this dataset, HCA clustering was performed again. Once again, the heat map showed distinct aliphatic and aromatic clusters. The cluster dendrogram corresponded with the heat map. When examining the portion of the dendrogram corresponding to the aliphatic compounds, the dendrogram could be further divided into 4 clusters. A table summarizing these clusters can be found in Table 1 within the study. For example, cluster 1 contained normal alkanes, cycloalkanes, and isoalkanes in the C5-C9 region. Some sub-clusters were also assigned ASTM classes. For example, samples in cluster 2 corresponded with iso-paraffinic products and normal alkane products.
Examination of the portion of the dendrogram corresponding to aromatics allowed for further division into 5 clusters. A table summarizing the clusters of the aromatic region of the dendrogram can be found in Table 2 within the study. For example, cluster 1 contained samples corresponding to the gasoline ASTM class. Clusters 2-5 contained samples corresponding to the aromatic product class. The specific compounds present in the samples separated these clusters. For example, cluster 2 consisted of samples containing a carbon range of C8-C12 as well as C2- and C3- alkylbenzenes.
- The TIS may be used for hierarchical cluster analysis of ignitable liquid residues.
- The HCA resulted in clusters based on the chemical composition of the ignitable liquids.
- Clusters identified through HCA were consistent with ASTM E1618 classes.
Ignitable liquid residue class assignment is currently a subjective classification based on the analyst’s interpretation of the evidence and application of ASTM E1618 standard. This method provides an objective method of determining ignitable liquid class; thereby, reducing subjective analysis.
The use of the TIS and multivariate statistical methods may provide support for the assignment of classes.