Evaluation of odor compounds sensed by explosives-detecting canines

Kelley M Kitts, Purdue University


Canines are regularly utilized by law enforcement agencies to detect explosives. However, the mechanism by which canines respond to explosive vapors is not well understood, leading to difficulties in canine training and testing. It is known that the amount of vapor generated from explosive compounds is dependent upon several factors including sample amount, vapor pressure, and the degree of confinement. Underlying these factors is the basic process of evaporation of an unconfined explosive, which is crucial to understanding how explosive vapors behave in other, more confined, systems. In Stage One of this study, evaporation rates were determined for several explosive liquids using an analytical balance. These rates were compared to one another as well as to theoretical models for the evaporation of liquids. In general and as expected, mass decreased linearly with time and evaporation rates decreased logarithmically as boiling point increased. Several examples of solvent "pinning" on a metal surface were also observed. While an empirical model for the evaporation of unconfined explosive liquids was developed, a comprehensive model for the escape of explosive vapors from sealed containers (i.e., a suitcase, knapsack, or IED container itself) is needed. The second part of Stage One of this study was to determine that the flow rate of explosive vapors escaping from relatively large orifices does not conform to Fick's Law of Diffusion. Fick's model states that the flow rate is linearly dependent upon the cross sectional area of the orifice and the material's diffusion coefficient. Instead, the flow rate was found to be linearly dependent upon the diameter of the orifice due to the tendency of the flow to diffuse outwards from its circular edge. A clear relationship between flow rate and diffusion coefficient was seen, however. Additional uncertainty arises concerning the complexity of the odor generated from explosive compounds. Because explosive vapors are often complex (they consist of multiple chemical compounds), confusion exists regarding the cause of canine alert; that is the "odor compound" that allows for canine detection of various explosives. Although 2, 4- dinitrotoluene (DNT) has been explored as a potential odor compound, the possibility of a nitrated explosive inherently producing nitrated gas upon decomposition has not. Stage Two of this study focused on evaluating nitrate as a potential cause of canine alerts. An LC/MS method for the detection of nitrate ions in Composition C-4 and flake trinitrotoluene (TNT) was developed and tested. Instrumental analysis was not successful in detecting nitrate ions in any of the explosives tested. The lack of nitrate was confirmed using a diphenylamine color test for nitrates, thus eliminating nitrate as an odor compound and cause of canine alert to nitroaromatic compounds. 2, 4-DNT has been introduced as a potential odor compound of TNT, however, the mechanisms behind its vapor emission have not been thoroughly explored. More specifically, due to the "sticky" nature of the 2, 4-DNT isomer, the effects of surface adhesion to container walls are of concern. In particular, whether the amount of material lost to surface adhesion is significant enough to effect canine detection of TNT. A second focus of Stage Two explored this concern. A GC/MS method for the detection and separation of TNT and DNT isomers in liquid extracts was developed and the amount of 2, 4-DNT residues adhering to container walls was quantified. These values, compared to the amount 2,4-DNT expected to saturate each container (determined by the Ideal Gas Law), showed a significant preference of 2,4-DNT in the solid phase as opposed to in the gas phase. The amount of residue adhering to the walls of a gallon can differed from expected values by nearly 70%. The amount of material extracted from a quart can exceeded expected values by 137%. The apparent sticky nature of 2, 4-DNT resulted in a significant loss of material needed to fully saturate a container and thus canine detection success may be affected. In the final stage of this study, theories regarding odor compounds and odor availability of nitromethane, TNT, and Composition C-4 were tested using certified explosives-detecting canines. These trials included thirty-three canine-handler teams from eight government agencies. The odor availability of nitromethane was tested by placing varying volumes of nitromethane in containers with differing degrees of confinement and studying the effects on canine detection success. The odor availability trial showed no significant effect of sample amount or degree of confinement on canine detection so long as the sample volume was sufficient to saturate its container. In this study that volume was determined to be < 1 mL. Detection of 2, 4-DNT, TNT-NESST (Non-Hazardous Explosives for Security Training and Testing), and flake TNT were also studied using certified canines. The purpose of this was to identify the odorant responsible for canine alert to the explosive TNT. These trials showed a significant response to 2, 4-DNT compared to TNT and its training aid; this suggests that 2, 4-DNT is the primary cause of canine alerts to TNT. Additionally, Composition C-4 and RDX-NESTT were tested along with potential odor compounds that included the manufacturing solvent, cyclohexanone, the energetic "taggant" 2, 3-dimethyl-2.3-dinitrobutane (DMNB), the plasticizer dioctyladipate (DOA) and its degradation product 2-ethyl-1-hexanol. While some response to DMNB and cyclohexanone was seen, the most significant response was to the actual Composition C-4. This suggests that the cause of canine alert to Composition C-4 is the explosive mixture as a whole and not a single chemical component of the mixture.




Goodpaster, Purdue University.

Subject Area

Analytical chemistry

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server