by Tony Cafe

This paper was presented by the author at the 2nd Australian Arson Fraud Seminar, 18-21 October 1990 at the Gazebo Hotel, Sydney Australia.


One of the many objectives of a fire cause and origin investigation is to determine whether flammable liquids were deliberately used to accelerate the spread of the fire. These liquids are termed accelerants and if their use is suspected at the fire scene then debris samples from various areas should be submitted to the laboratory to determine the presence, distribution and identity of the accelerant. This information will support the investigator's own understanding of the ignition and propagation of the fire.

The efforts of the forensic laboratory are entirely dependent on the quality of the samples provided and therefore a major objective in forensic fire cause determination is to successfully locate and sample fire debris for subsequent laboratory analysis for residual accelerants. To achieve this aim the investigator needs to apply the proper sampling techniques and have a basic understanding of the chemical and physical nature of some of the common accelerants and their behaviour during and after a fire. The aim of this paper is to briefly explain some of these considerations.

The Nature of Accelerants

The most commonly used accelerants are petrol, kerosene, diesel and mineral turpentine. They are all derived from crude oil which is a very complex mixture of hydrocarbons whose components with similar physical and chemical properties are collected to give the various fuels and solvents. Because the common accelerants are themselves complex mixtures it is best to examine the graphs produced from the analysis by capillary Gs Liquid Chromatography (GLC) to understand their properties and behaviour.

Gas Liquid Chromatography is the most widely used laboratory instrument for analysing accelerants because of its ability to detect and identify trace amounts. A headspace or liquid sample of the volatiles extracted from the fire debris is taken and introduced into the instrument where it is volatilised and swept by a gas stream through a long tubular column towards a detector. As the sample moves through the column the various components will separate so that the compound with the lowest boiling point will emerge from the column first to be detected followed by the other components I order of their boiling points. By measuring the time from injection at which the individual components emerge from the column it is possible to positively identify each component. The entire analysis is recorded on a chart called a chromatogram where each component of the sample is represented by a peak and the overall pattern is essentially a fingerprint for each accelerant.

Figure 1. shows the chromatograms produced from the analysis of fresh, kerosene, fresh diesel and evaporated kerosene exposed to the atmosphere for seven days. The major peaks in the chromatogram are labelled according to the chain length of the molecule producing the peak.

By comparing the chromatogram produced from fresh kerosene and diesel it can be seen they have similar component and are complex mixtures being produced from the fractional distillation of crude oil. Diesel is however composed of components that have a higher boiling point and so is termed a heavier fraction. As kerosene weathers the more volatile components tend to evaporate and its chromatogram begins to resemble that of diesel. For this reason it is difficult for an analyst to conclusively identify kerosene in fire debris samples if sampling is made some time after the fire and weathering of the accelerant has occurred.

Petrol is a more volatile mixture than kerosene and therefore more readily forms explosive mixtures in air which upon ignition can cause considerable damage to the surrounding environment. The chromatograms produced from the analysis of fresh petrol and evaporated petrol are shown in figure 2. and it can be seen there are fewer peaks present in the chromatogram of evaporated petrol. The analyst when presented with this chromatogram being essentially a partial fingerprint of the accelerant, has less information with which a conclusion can be made regarding the identification of the accelerant.

The common accelerants are all insoluble in water and as such are not readily washed away during the extinguishing of the fire. They tended to become sealed into porous surfaces which prevents their evaporation and have been found to remain at the fire scene for periods of up to three months. Water soluble accelerants such as methylated spirits, acetone and some of the industrial solvents tend to be washed away from the fire scene. Traces do remain but will readily evaporate because they are not sealed into the surfaces by water. If a water soluble accelerant is suspected in a debris sample the analyst should be notified in case the analytical procedures require some modifications to successfully analyse water soluble compounds.

Because of the reasons illustrated above it is important to sample debris as soon as possible after the fire so the laboratory analysis will yield as much data as possible on which to base a conclusion regarding the presence and identity of an accelerant.

Where to sample

At the fire scene various indicators are used to predict the presence and location of an accelerant. These may be eyewitness reports of a raid and intense fire in its initial stages or the presence of heavy localised burning to the flooring material and overhead damage which is inconsistent with the combustion of the naturally available fuel below. Accelerants are normally found at the area of origin of the fire, in doorways where an arsonist would attempt to leave a building and in large spaces such as in the centre of a room where the arsonist can move about freely when distributing the accelerant. It is often helpful when attempting to locate areas where accelerants may be present to visualise the scene before the fire and predict the movements and actions an arsonist would make whilst spreading an accelerant.

Materials for sampling

After locating the area where it is felt an accelerant may be present a sample of debris which will have the highest probability of retaining traces of accelerant is required. As a general rule if the area to be sampled is wet traces of petroleum derive accelerant would be expected to remain. Therefore the best materials to sample are wet porous materials such as soil, paper, cardboard, bagging, carpet, cloth and to a lesser extent concrete. Readily combustible materials such as rubber and timber are generally not good materials to sample because their combustion supports the depletion of the accelerant.

When sampling a material that is difficult to remove such as concrete, an absorbent may be sprinkled onto the surface to absorb water and in turn traces of an accelerant if present and the absorbent recovered and analysed. Absorbents that can be used are diatomaceous earth which is commonly used for swimming pool filtration, fullers earth, inorganic carbonates and some industrial absorbants. Flour has been used but is unsatisfactory because its subsequent fermentation in the container will produce ethanol which is the major component of methylated spirits. Household absorbents such as sanitary napkins and disposable nappies can also be used to sample from concrete.

To assist in the selection of a sample the investigator normally uses their own sense of smell to detect any odours of accelerants. The debris can be smelt directly or warmed in one's hands to release vapours. The use of portable gas detectors (Sniffers) at the fire scene to detect traces of accelerants has also been used.

Use of a sniffer

Various types of sniffers are manufactured and are best classified according to the principle of operation of their detector. They may employ a flame ionisation detector or a catalytic oxidation probe. The latter is the most commonly used because of its low cost and robust design. The major problem when using nearly all types of sniffers regardless of their principle of operation and their price is they cannot discern between accelerant vapours and pyrolysis products and because of this their use at a fire scene remains a continual source of controversy amongst investigators.

The main advantage when using a sniffer rather than relying on one's sense of smell are:

Table 1 is a summary of a sniffer response to 112 fire debris samples submitted to the laboratory over a period of six months. The samples were classified according to their basic composition and the sniffer response was classified as being either positive or negative. When testing with the sniffer the debris was not disturbed for fear of losing accelerant vapour. The figure shown in parenthesis is the number of samples that gave a positive analytical result for accalerants when tested using the more conclusive technique Gas Liquid Chromatography.

Table 1.
Material Positive sniffer result Negative sniffer result
Ash & charcoal18(5) 10(0)
Carpet22(18) 9(1)
Cardboard & paper10(10) 7(2)
Concrete2(0) 4(0)
Soil1(1) 12(8)
Felt & cloth7(7) 2(0)
Plastic1(1) 5(0)
Timber2(1) 0(0)
TOTAL63(43) 49(11)
Sniffer response for 112 samples
(number in parenthesis indicate +ve GLC result).

It can be seen from table 1 that the material which gave the highest ratio of true positive readings were carpet, cardboard and paper, felt and cloth. The investigator when using a sniffer should sample these materials if an option exist. The soil sample gave the highest ratio of false negative readings (8 out of 12) indicating that when testing on site the soil must be freshly disturbed so that accelerant vapours are released to be detected. The high overall number of false positive readings obtained (20 out of 63) using the sniffer indicates the lack of specificity of the instrument.

For testing materials such as rubber backed carpet or polystyrene which can both produce liberal amounts of pyrolysis products as indicated by their odours, the sniffer will give random positive readings that could confuse the operator. Also a poorly tuned instrument, a low battery or an instrument malfunction may give the investigator a false impression that accelerants are absent from the fire scene.

Sniffers can be a valuable aid at the fire scene however the operator must be aware of the principle of operation of the instrument so that it may be tuned correctly before use and its results are interpreted correctly. It must be stressed they are only to be used as an aid for the collection of samples for laboratory submission and their results are not conclusive regarding the presence or absence of an accelerant. The best option for selecting debris samples is to use a combination of relying on one's sense of smell and having a sniffer on hand to use as the circumstance requires.

Sampling containers to use

Various containers have been used for sampling fire debris however unlined metal paint cans are regarded as the most suitable because of their excellent sealing capabilities, their robust design and are harmonious with most analytical techniques. Plastic bags are easily pierced and are prone to diffusion of vapours both into and out from the bag and glass jars are fragile.

Metal paint cans come in a variety of sizes and types being unlined or lined with an epoxy coating designed for the storage of water based paints. Figure 3. is a chromatogram produced from analysing the volatiles extracted from a lined can together with a chromatogram of petrol. It can be seen an industrial solvent, similar to petrol has been used in the manufacture of the lining and for this reason lined cans should never be used for sampling fire debris.

Control samples

Control samples or blanks generally form part of the scientific process to ensure that background materials do not contribute to the result.


Figure 1.
Fresh kerosene, diesel and evaporated kerosene.

Figure 2.
Evaporated petrol, fresh petrol.

Figure 3.
Petrol, lined can.

Figure 4.
Kerosene standard, burnt polyethylene.

Figure 5.
Petrol standard, varnished wood.

Figure 6.
Soil (ex motor yard), petrol.