interFIRE Home interFIRE Home interFIRE VR Support Training Calendar Training Center Resource Center Message Board Insurance Info

Part 3: Detection Limits: Is More Sensitive Necessarily Better?

Presented by:
Dr. John DeHaan

For most of the thirty years or so this author has been involved with fire debris analysis, investigators have been insisting that lab methods for volatile accelerants in fire debris are not sensitive enough and that results would be a lot more accurate (and presumably a lot more reflective of their conclusions about the fire) if only the lab techniques were more sensitive. Not all that many years ago, the human nose was the standard technique for identifying the presence of ignitable liquids in fire debris. It was pretty sensitive (20ppm) and could be trained to be quite discriminating, but it was subject to variables (person-to-person variations, colds, allergies, fatigue) and somewhat subjective (ask five different people what a sample of fire debris smells like and you are likely to get two or three different answers). It was not until 1962 that gas chromatography (GC) was first suggested for fire debris analysis. GC was pretty crude ­ 10 or 12 peaks was considered a good result ­ but you still had to get the volatile part of the sample from the can into the GC. This usually meant solvent extraction for heavy compounds, heated headspace for light, volatile ones, or steam distillation for a wide variety. Each technique had its own limit of detection. Headspace sampling was simple and non-invasive but it was not very sensitive (with 1­2cc of vapor sampling, a lower limit was on the order of 20­50ppm in the air sample) and then only for liquids volatile enough to evaporate significantly at the temperatures used. Steam distillation and solvent extraction had the potential to be more sensitive but were limited by the physical limits of sample handling. Adding a dollop of solvent could increase the volume of a tiny amount of distillate such that it could be recovered from a distillation condenser receiver but that added to the risk of contamination. Evaporating the solvent in a solvent extraction increases the sensitivity, but one runs the risk of driving off the sample itself with too much evaporation. All this meant that the same concentration of sample that could be isolated and identified by GC was about the same as a well-trained human nose could detect. Some investigators complained that if they could not smell anything, the lab never identified anything anyway, so why bother with the lab? There was always a demand for more sensitivity for arson debris.

In the late 1970s and early 1980s. GCs got better, as did extraction methods. Stealing the concept from the environmental sciences of concentration by trapping on activated charcoal , it was found that exposing a small quantity of activated carbon to a container of fire debris would allow trapping of the volatile species, leaving the water, ash, and debris behind. Drawing a volume of air from (or through) a container of debris would allow concentration of the volatiles into the trapping medium (adsorbent). The volatiles could then be eluted from carbon by a small quantity of solvent or driven off from a polymeric trapping medium like Tenax thermally. During the 1980s, these techniques were tested and refined, improving the sensitivity and the simplicity until very sensitive, very reproducible extraction methods were established. "Dynamic" extraction methods were more sensitive, particularly to heavier compounds but ran the risk of destroying samples in a "one-shot" analysis if the sample got too hot or the extraction carried out for too long. "Passive" methods were found to me most sensitive to light- and mid-range volatiles but did not interfere with the sample (other than opening the container) and could be repeated if the extraction went awry. These methods were shown to be 10 to 100 times more sensitive than the old "bulk" (daresay "bucket chemistry") methods.

Fortunately, GCs became more sensitive with the advent of capillary columns and more sensitive detectors. Older GC columns were 1/8 to 1/4 inch in diameter tubes, 6 to 10 feet long. In gas chromatography, length means better separation, so research grade columns were 100 to 200 feet long. They were capable of showing 150 or more compounds in gasoline but they took a long time for each analysis and took a lot of sample volume. Capillary columns today are typically 40 to 60 feet long, but only 0.01 inch in internal diameter. They have no packing, the active ingredient to achieve separation is coated on the inside of the column so they allow very rapid analysis (very short run times) with great resolution and very small sample volumes. These improvements meant that very small traces of volatiles could yield excellent GC results.

As the sensitivity increased, we started seeing some interesting things. Just like centuries ago when telescope optics improved, suddenly the sky was filled with faint stars no one ever expected to be there, more sensitive extraction and analysis methods meant we could see traces of volatiles in all sorts of fire debris samples. Most of these were pyrolysis products, the partially disintegrated molecules of the solid fuels in the debris, that could be sorted out by the lack of a "pattern" to the GC peaks and their retention times, which did not match those of hydrocarbons from petroleum-based fuels. But we also started seeing traces of volatiles in unburned materials like athletic shoes, newspapers, carpet tiles and even the packaging we used for fire debris evidence (like the "old" Kapak bags and paint cans lined with solvent-based varnishes). Many of these volatiles were petroleum products, residues of the manufacturing processes. So we learned to be more careful in interpretation and to check our containers for contaminants before use. But for the most part, we could recognize pyrolysis product or product contamination and distinguish it from "real" accelerants. Comparison samples of carpet were strongly suggested but most accelerants could be identified in their absence.

And then came the 1990s when petroleum chemists came up with a whole universe of new products that were no longer petroleum distillates and so no longer had the distinctive pattern of their GC peaks that made them readily identifiable. Refinements in instrumentation meant that mass spectrometry could be added to bench-top GC's for nearly anyone to use (and thus were no longer the complex, expensive machines that were solely the province of specialists). Mass spectrometry made it possible to identify individual components in a collection of peaks rather than just look at the retention times and peak patterns. GC/MS systems could be asked to look just for particular molecular types (just the aromatics, for instance) and plot those out separately from the alkanes. "Target" compounds were identified for gasoline and other common accelerants and they could be searched for amidst the forest of peaks of different species found in pyrolysis products. These advances meant that possible accelerants could be identified even at very low concentrations even when lots of volatile pyrolysis products were present. As petroleum products (at least in the U.S.) changed, GC/MS became the only way to identify the aromatic blends, isoparaffinic products, naphthenic/aliphatics, oxygenated compounds, and all the rest for which GC/FID was inadequate. As the minimum detection level dropped, we saw just what a petroleum-based world we live in. From the volatile residues in roofing paper and newspaper print, the solvents in glues and adhesives used in floor coverings and footwear, to residues of dry cleaning solvents, insecticides, and even cleaning agents (like the limonene used in those "green" grease removers), we can detect petroleum products in all manner of consumer goods today. Without specific identification of the product and comparison against a reference sample, the mere presence of a volatile petroleum product in fire debris may well be meaningless.

Much of the sensitivity issue came to light when canine detection was being debated. Yes, a properly trained canine nose can detect petroleum products at very low levels with a high degree of accuracy, sometimes at concentrations below even what a qualified lab could analyze. Yes, there may be samples where a canine correctly detects traces of gasoline that the lab cannot confirms if it adheres to the accepted standard practices. But the issue is not simply sensitivity, there are far too many other possible sources of petroleum products, it is the accuracy of the identification of the actual product present that is essential. A canine can only signal yes or no, it cannot make the identification as to whether that is a trace of the insecticide (with its aromatic blend solvent) the owner used the week before or gasoline that an arsonist used to set the fire. Only the lab with its verified and accepted methods and criteria can distinguish those products. The canine cannot discriminate between the residues of carpet glue used to repair a section of carpet from the residues of isoparaffinic solvent used as an accelerant. Only a properly equipped and qualified lab can make such calls. Another issue of extreme sensitivity is erroneous reconstruction. If someone steps in a puddle of gasoline leaking from a Bob Cat or a spilled from a fuel can outside the scene and then walks through the scene, traces might be found in several locations. A blind reliance on mere presence of an "accelerant" could lead to the conclusion that gasoline was used in several places throughout the scene. We must ask: " Why was this found in these concentrations, and how else might it have gotten there?"

Dr. John DeHaan has been a criminalist for some 32 years. He has worked at county, State, and Federal forensic labs. He is a native of Chicago and his Bachelor of Science degree in physics was from the University of Illinois at Chicago. He has been involved with fire and explosion investigations for over 30 years, and has authored dozens of papers on fires, explosions, and their investigation and analysis. He is probably best known as the author of the textbook Kirk's Fire Investigation (now in its Fourth Edition). His doctorate (in 1995), from the University of Strathclyde in Glasgow, Scotland, was on the Reconstruction of Fires Involving Flammable Liquids.

He is a member of NFPA, and served on its 921 Technical Committee from 1991-1999. He is a member of the IAAI and serves on its Forensic Science Committee. He holds a diploma in Fire Investigation from the Forensic Science Society (United Kingdom) and one from the Institution of Fire Engineers (U.K.). He is a Fellow of the American Board of Criminalistics in Fire Debris Analysis and a member of the Institution of Fire Engineers. He retired from the California Department of Justice in December 1998 and is now the president of his own consulting firm, Fire-Ex Forensics, Inc., Based in Vallejo, California, where he now serves as a consultant in fire and explosion cases all over the U.S., Canada and overseas.


Other articles in the Our Changing World Series

Home | interFIRE VR Support | Training Calendar | Training Center | Resource Center | Message Board | Insurance Info
Sponsorship Opportunities
Web Site Designed for 800 x 600 by Stonehouse Media Incorporated® Copyright © 2024 All Rights Reserved.