Automated Data Analysis of Complex Arson Samples

Scott J Campbell1 and Mary Blackburn2

1SpectralWorks Ltd, The Heath Business & Technical Park, Runcorn, UK; 2Thermo Electron, 355 River Oaks, San Jose, CA, USA

First Published: BMSS 2004


The investigation of suspected arson cases is both a time and labor-intensive operation; not only is the work highly repetitive in nature, but very large volumes of data are generated which need to be reviewed. The characterization of pure flammable liquids is a relatively simple task. However, real samples are complex mixtures containing both artifacts and pyrolysis products in addition to the masked accelerant; collectively complicating the identification process. In the process of analyzing these complex samples the analyst is required to engage in exhaustive reviews of complex GC/MS chromatograms (A typical Total Ion Chromatogram (TIC) is shown in Figure 1). A skilled arson investigator can often identify an accelerant hidden in complex chromatographic data by recognizing patterns formed by the relative abundance of key compounds. The specialized skills of an experienced analyst can be significantly augmented by the use of a sophisticated data collection and analysis system, relieving many of the more tedious aspects of the work and delivering rapid identification. Historically, arson analysis has consisted of long GC/MS run times followed by the review of the mass chromatographic data. The application of GC/MS using fast GC techniques can greatly reduce the data acquisition period and when coupled with suitable data processing software, it can greatly reduce the total analysis time. The data presented here describes the rapid analysis of arson samples using GC/MS and a rapid data processing software package.

Materials and Methods

Arson samples were analyzed as follows; arson debris is incubated at 70°C, in a stainless steel can containing a 15×3 mm carbon strip suspended from the lid. The strip is then placed in a standard 1.8 ml vial and 1 mL of carbon disulfide is added to desorb the entrained material. Injections of 1 µL were made directly from the vial using an autosampler. The GC conditions were as follows; injector temperature 200°C, oven 50°C, ramped 50°C / minute to 300°C, hold 1 minute, split injection, split flow 100 mL/min, constant column flow of 2.0mL/min. The column used for analysis was a 30m DB-5, 0.18mm ID, 0.18 µm film. Data was acquired at a rate of 20 spectra per second over the mass range 40-440 amu with a solvent delay of 0.5 minutes, total run time 6 minutes. Data was processed and results generated using AnalyzerPro™ and MatrixAnalyzer™ from SpectralWorks.

Figure 1. Total Ion Chromatogram (TIC) of a Typical Arson Sample

Figure 2. Target Analysis Results for the Typical Arson Sample Depicted in Figure 1

Figure 3. Expansion of 7.9 to 8.3 minutes from Figure 2

Figure 4. XICs for the Component Detected at 8.01 mins

Figure 5. Extracted Spectrum for Component at RT 8.01 mins

Figure 6. NIST Library Search Result of the Extracted Spectrum from Figure 5

Figure 7. Blanks

Figure 8. Kerosene

Figure 9. Gas

Figure 10. Lighter Fluid

Figure 11. Unknown

Results and Discussion

Real arson samples are often quite complex in nature. Figure 1 depicts a complex chromatogram representative of a typical arson sample. To aid in the analysis of this complex chromatogram and suggest the identity of a possible accelerant the software tools AnalyzerPro™ and MatrixAnalyzer™ were employed. Figure 2 illustrates the detection of target components by labeling the TIC with green markers. In addition target components that were not detected are indicated by a red marker at the appropriate retention time.

Once a target component has been detected the ions that were specified in the detection routine are displayed to aid in confirmation of the detected target component (Figure 4).

The next step in the process is the extraction of the spectrum for the detected component from the full scan data (Figure 5) and subsequent searching of the extracted spectrum against the NIST library for final confirmation of a detected target component (Figure 6).

To assist the analyst in making a final determination as to the identity of a suspected accelerant the software displays only the detected target component for a sample (Figure 7 through 10). Using this graphical representation it is relatively simple to make the assignment of kerosene to the arson sample analysed here (Figures 1 through 3). This assignment is based on the similarity of the graphical representations of detected components for the kerosene standard (Figure 8) and the unknown arson sample (Figure 11). Recall that each detected componentFs was first confirmed by a co-elution of target ions followed by the extraction of the spectrum and finally searching the extracted spectrum against the NIST library. Thus the analyst has a high level of confidence in their identification of kerosene as the accelerant in this example.


Currently the analysis of arson samples is laborious and time consuming. This fact is due to both the actual data acquisition and the subsequent manual data analysis. We have shown that through the use of rapid GC/MS analysis and rapid data processing (AnalyzerPro / MatrixAnalyzer) that arson cases can be rapidly and accurately analyzed. Actual data acquisition time is reduced by the use of fast GC techniques without loss of resolution or sensitivity. The application of automated data reduction software resulted in accurate and rapid identification of actual arson samples. The reduction of both runtime and analysis time increases sample throughput and productivity. Finally, a repetitive and arduous task is removed from the analyst while the analyst’s confidence is increased in the ultimate identification of the accelerant.


John Lucey – Michigan State Police, 714 S. Harrison Road, East Lansing, Michigan, USA