Talanta 67 (2005) 313–327 Identification of dominant odor chemicals emanating from explosives for use in developing optimal training aid combinations and mimics for canine detection Ross J. Harper, Jos´ e R. Almirall, Kenneth G. Furton ∗ International Forensic Research Institute, Department of Chemistry and Biochemistry, Florida International University, University Park, Miami, FL 33199, USA Available online 1 July 2005 Abstract Despite the recent surge in the publication of novel instrumental sensors for explosives detection, canines are still widely regarded as one of the most effective real-time field method of explosives detection. In the work presented, headspace analysis is performed by solid phase microextraction (SPME)/gas chromatography–mass spectrometry (GC–MS), and gas chromatography–electron capture detection (GC-ECD), and used to identify dominant explosive odor chemicals seen at room temperature. The activity of the odor chemicals detected was determined through field trials using certified law enforcement explosives detection canines. A chemical is considered an active explosive odor when a trained and certified explosives detection canine alerts to a sample containing that target chemical (with the required controls in place). A sample to which the canine does not alert may be considered an inactive odor, but it should be noted that an inactive odor might still have the potential to enhance an active odor’s effect. The results presented indicate that TNT and cast explosives share a common odor signature, and the same may be said for plasticized explosives such as Composition 4 (C-4) and Detasheet. Conversely, smokeless powders may be demonstrated not to share common odors. The implications of these results on the optimal selection of canine training aids are discussed. © 2005 Elsevier B.V. All rights reserved. Keywords: Solid phase microextraction; Explosives; Canine detection 1. Introduction The use of canines as a method of detection of explo- sives is well established worldwide and those applying this technology range from police forces and military to humani- tarian agencies in the developing world. Until recently, most data regarding optimal training protocols and the reliabil- ity of canine detection has been anecdotal, leading to suc- cessful challenges regarding the admissibility of evidence obtained with the assistance of canines and hampering the improvement of performance of canines as biological explo- sive detectors [1]. Challenges facing the field of canine detec- tion include the limited ability to evaluate their performance with standardized calibration standards. Unlike instrumental methods, it is currently difficult to determine detection levels, ∗ Corresponding author. Tel.: +1 305 348 2292; fax: +1 305 348 3772. E-mail address: furtonk@fiu.edu (K.G. Furton). perform a calibration of the canines’ ability or produce sci- entifically valid quality control checks. In addition, there are increasingly strict requirements being applied to the admissi- bility of the application of detector dogs in locating items of forensic interest, highlighting the need for better a scientific understanding of the process of canine detection. This cur- rent research is targeted towards the identification of active odors for canine detection of items of forensic interest and the development of what we are calling odor mimics, or train- ing aids that contain the odor chemicals that mimic the real substances. There are presently several theories about what is responsible for the canines’ high selectivity and specificity to explosives including (i) that canines alert to the parent explosives regardless of their volatility; (ii) that canines alert to more volatile, non-explosive chemicals that are present in explosives, and which are characteristic to explosives; or (iii) both parent explosives as well as characteristic volatiles are used to accurately locate explosives. To date, there are no 0039-9140/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.talanta.2005.05.019