Two-temperature preparation method for PDMS-based canine training aids for explosives

Abstract

Canine training aids based on vapor capture-and-release into a flexible polymer, polydimethylsiloxane (PDMS), have been used for in canine detection of explosives that have volatile or semi-volatile odorants. To enhance the rate of odor capture for less volatile targets, two temperatures are used for aid preparation. By using an elevated temperature for the target explosive, the amount of vapor is enhanced, increasing the production of the characteristic odor profile. The polymeric adsorbent is maintained at a cool temperature, favoring vapor capture. The success of this two-temperature approach is demonstrated for training aids targeting the low volatility explosive TNT using SPME (solid-phase microextraction) headspace analysis. In addition, the effect of using two temperatures on preparing training aids based on TNT and its more volatile impurities 2,4-DNT and 2,6-DNT are evaluated in canine trials. A thermal pretreatment to minimize the non-target odors in the PDMS polymer is presented.

1. Introduction

Canines are a sensitive and mobile system for highly effective detection of explosive threats [1–3]. In an age where terrorist attacks have become increasingly common, the importance of canine detection has grown, both for civilian and military use. Secondary explosives such as TNT (trinitrotoluene), composition C-4, and Semtex, which require blasting caps or other primary explosives to initiate detonation, are comparatively safe to handle for canine training. However, some international agreements prohibit the use of any type of real explosives in certain theaters of U.S. military operations. In addition, the use of all explosives has requirements for record-keeping and secure storage magazines. The development of safe, alternative training aid materials that provide the characteristic odor profile of explosives would be of great value, particularly for canine maintenance training.

To provide the odor of real explosives, a “capture-and-release” training aid technology has been developed based on the flexible polymer, polydimethylsiloxane (PDMS) [4–8]. Odorants are transferred into the PDMS in the vapor phase by locating the polymer close to the target explosive for vapor transfer of the characteristic odor and is colloquially called ‘charging.’

Transport of target odors into the PDMS and the subsequent release of the vapors as a training aid is facilitated by its flexible, semi-solid properties relative to other solid adsorbents. PDMS provides favorable target odorant transport within the polymer with small molecule diffusion coefficients determined using PDMS SPME to be in the 10⁻⁷⁻ cm²/s range [6]. This polymer also has the ability to notably swell [9], absorbing substantial amounts of target odorants to > 2% mass fraction [6], thus increasing the useful lifetime of the training aid. PDMS has good affinity for the incorporation of non-polar to somewhat polar molecules [9,10]. Incorporation of any vapor-captured active explosive agents into the PDMS also renders them inert by dilution. These properties, as well as the thermal stability of PDMS, have made it useful for headspace preconcentration of vapor-phase target analytes in solid-phase microextraction (SPME) [11]. The vapor capture-and-release properties of PDMS have been studied for the volatile odorants in C-4 and TATP [4–6,8] and used successfully for canine training for detection of these two targets in unpublished work.

One of the limitations of this approach to training aid fabrication is the time required to completely absorb the vapors of the target (to charge the PDMS in the aid) at room temperature - particularly for targets of low volatility. For easily captured, volatile targets like TATP and the characteristic odorants in C-4 (cyclohexanone and 2-ethylhexanol), we have typically used between a week and a month for successful training aid charging [6]. Longer times may be required for less volatile targets such as TNT and HMTD. In addition, using the capture-and-release technology would be of great value for the rapid preparation of training aids for newly emerging improvised explosives, where supplies of training aids are limited and/or dangerous to use directly.

The concept of using two temperatures to enhance the absorption process has been demonstrated in SPME [12,13] via heating the target material and cooling the adsorptive phase of the fiber. Heating of the target sample increases the vapor pressure of all components of the sample, generating more vapor needed for capture. However, without cooling of the PDMS phase, the ability of the absorptive phase to capture target vapor would be decreased. The magnitude of this latter effect was clearly demonstrated in a study of temperature on the preconcentration of a compound by a PDMS-phase SPME fiber [14]. The analyte calibration factor obtained from a constant, isothermal analyte vapor concentration was found to decline logarithmically with increasing PDMS SPME fiber temperature over the range of 15 °C to 35 °C. Thus, it is advantageous cool the training aid PDMS phase while simultaneously heating the target explosive during the training aid charging process. Another beneficial effect of the heated/cooled vapor generation system is generation of convective air circulation within the closed two-temperature system, enhancing vapor transport.

We provide a preliminary evaluation of the ‘two-temperature’ idea for preparing the PDMS-based canine training aids, addressing the need for more rapid charging of aids, particularly for targets with characteristic odorants of low volatility. Using carefully designed canine trials, successful aids for the detection of TNT were created based on charging with the previously identified canine odorant 2,4-dinitrotoluene (DNT) [1,3] using this two-temperature approach. Since training aids require a low non-target odor signature, a ‘bake out’ protocol to minimize the odor of PDMS is also developed.

2. Experimental¹

2.1. Reagents

Military-grade TNT was provided by the Bureau of Alcohol, Tobacco, Firearms, and Explosives (Ammendale, MD) and was used for vapor release studies of aids prepared by two-temperature charging and for the preparation of canine training aids. The externally-sampled internal standard (ESIS) for the SPME-ESIS measurements [8] of TNT was one of the geometric isomers of DNT, 3,4-DNT (ChemService, Westchester, PA). 2,4-Dinitrotoluene (2,4-DNT) (Tokyo Chemical Industry Co., Ltd.) has previously been identified as the primary volatile impurity associated with TNT used by canines for explosives detection, owing to its higher vapor pressure, and was also used for training aid preparation. An additional impurity found in TNT, 2,6-DNT (Picatinny Arsenal, Atlantic City, NJ) was also tested as a canine odorant for training aid fabrication. This material was highly purified and the purity determined to be > 99% by the provider. All of these explosive materials were stored in a locked storage magazine, logged in and out, and handled by trained staff.

2.2. Training aid preparation

The training aid was fabricated from ‘2-oz’ shallow round cans (Paper Mart, Orange, CA). A 12-hole can was prepared by drilling 3.2 mm (1/8″) holes. Six holes were symmetrically drilled hexagonally around the sides, midway to the upper lid and 6 were drilled hexagonally in the lid centered on a 3.5 cm circle, as seen in Fig. 1. The 6 holes in the lid of the can provide direct access for odor sniffing and are spaced apart approximately the same distance as the 2 naris inlets of a Labrador retriever’s nose. The 6 holes in the circumference of the can provide outlets for the heavier-than-air target molecules to spread easily in a 360° fashion, achieving well-dispersed training aid odor. The potentially adsorptive protective coating of these commercial metal cans was removed by alternately soaking for about ≈ 1 h in ethyl acetate and then acetone. Each solvent treatment was followed by abrasive wiping with a clean Kimwipe^™ until the visible coating had been removed. To remove any residual stripping solvents, the cans were soaked in pentane for 30 min and then baked for 2 h at 150° C in a pre-warmed oven. An absorptive layer of PDMS, either Space Grade 93–500 or Sylgard 184 PDMS (Ellsworth, Germantown, WI), was added to the cans after thorough mixing of the 10:1 monomer and catalyst. 5.50 g of the uncured polymer was weighed into the can producing a ≈ 2 mm layer of PDMS. For curing, the can was set on a perfectly level surface, covered by inversion of an outer ‘4-oz’ storage can (Paper mart) and allowed to cure overnight. For storage, the aid was placed in a pentane-cleaned 4-oz can, subsequently baked at 150°C for 2 h, and the outer rim of the metal can lid was tightly wrapped with Teflon^™ tape to minimize exchange of odors. The headspace compositions of Space Grade and Sylgard 184 PDMS were compared without a ‘bake out’ thermal treatment and after the above baking protocol. All TNT training aids were fabricated with a nominal ≈ 100 mg of TNT or DNTs, limited by the available amount of materials.

Fig. 1.

Vapor-charging setup for explosives training aids and a completed 12-hole training aid can. For this study, ≈ 100 mg of TNT or two different DNTs (C-4 is pictured) was used to provide charging vapor into 5.5 g of PDMS in the aid.

For one aid, the TNT was incorporated in the PDMS by a ‘mixing’ method with dissolution of the TNT directly into the PDMS. 110 mg of TNT was blended into 10 g of the PDMS monomer and mixed vigorously several times over a 3 h period. This added amount of TNT far exceeded the short-term solubility limit in the PDMS monomer. However, some TNT did dissolve as evidenced by visual inspection of the light yellow color. A 5.00 g portion of the saturated monomer was decanted into the receiving training aid can, followed by the addition and mixing of the 0.50 g catalyst with thorough stirring, providing 5.50 g of PDMS. After curing overnight, this aid was baked at 150 °C for 2 h.

Training aids prepared by the two-temperature method used ≈ 100 mg of target material placed in a 3 cm diameter glass crystallizing dish inside the outer 4 oz charging container. The training aid container was affixed to the lid of the charging container (as in Fig. 1) via a neodymium magnet embedded in an aluminum block. The Al block had a hole for the temperature probe controlling a cooling plate (STIR-KOOL Model SK-12D-AW, Thermoelectrics Unlimited, Inc., Midland, ON, Canada). The charging container and target material were placed on a temperature-controlled hot plate (Super-Nuova, Thermo Scientific, Waltham, MA) to raise the temperature of the target material to as high as 82 °C. The cooling plate was placed upside down on the aluminum heat-transfer block. The configuration is shown in Fig. 2. For all two-temperature experiments, the cooling plate was maintained at 20 °C. Colder temperatures resulted in atmospheric water condensation on the setup. Given the proximity of the heated sample zone to the PDMS, it seems likely that the temperature of the outermost surface of the PDMS may rise somewhat above the 20 °C setpoint, especially for higher vapor source temperatures.

Fig. 2.

Two-temperature vapor-charging setup set for 35 °C/20 °C PDMS training aid. The lower hot plate (set to 35 °C) raised the temperature of the target material while the upper cooling plate kept the training aid cool (20 °C) through contact with an aluminum contact plate and temperature probe.

The temperatures chosen for heating the vapor-generating compounds for training aid charging using in the canine studies were selected based on the relative vapor pressures of TNT, 2,4-DNT, and 2,6-DNT. There is not great agreement on explosives vapor pressure data [16], which seems to depend on the technique used for determination. However, a reasonable comparison of the relative vapor pressures of these three compounds can be made using data generated at NBS (now NIST) [17]: TNT 0.00120 Pa; 2,4-DNT 0.0290 Pa; and 2,6-DNT 0.0756 Pa. Thus, in choosing temperatures for charging TNT aids, higher temperatures were used (50 °C, 60 °C, 82 °C) than for the order-of-magnitude more volatile 2,4- and 2,6-DNT (35 °C and 30 °C, respectively).

In our experience with analysis of solid samples of the military TNT used for charging, the amounts of DNT impurities are typically near the 1% level for 2.4-DNT; whereas the 2,6-DNT content is a few tenths of a percent. Despite its low concentration in the solid, 2,4-DNT dominates the headspace of military-grade TNT and has been previously identified as the primary odorant used by canines for detection of TNT [1,3].

The various compositions and temperatures for canine evaluations are listed in Table 1. The designation for the two temperatures, such as 50 °C/20 °C symbolize a heating temperature of the lower hot plate of 50 °C and an upper cooling plate temperature of 20 °C. All charging of training aids with TNT, 2,4-DNT, and 2,6-DNT used a nominal amount of 100 mg of odor source material. Charging times were nominally 6 to 7 days except for some aids charged with TNT. The purity of the 2,6-DNT source material (>99%) was of particular importance as the presence of notable 2,4-DNT might conflate the results of DNT aids in canine testing. In the cases of the TNT aids prepared at 60 °C/20 °C and 82 °C/20 °C, charging was interrupted when crystals of TNT had formed on the surface of the can and/or PDMS surface via sublimation and condensation. For the two 60 °C/20 °C aids, the excess crystals formed on the rim of the can and were allowed to evaporate in a fume hood for 1 h. For a single 82 °C/20 °C aid, crystals also formed both on the rim of the can and two on the surface of the PDMS. The TNT crystals were left in place and the aid stored in the absence of source TNT in the sealed outer 4 oz storage can maintained at 82 °C for 25 h. This was sufficient time for the crystals to volatilize and absorb into the PDMS. Canine testing occurred about 2 months after preparation of the aids allowing ample time for any unseen residual crystals to be fully absorbed into the PDMS.

Table 1.

Conditions for PDMS training aid preparation.

Designation	Mass of Target	Charging Method	Temperature/Condition	Days Charged	Notes
Saturated TNT	< 50 mg	TNT Mixing	Room Temperature	N/A	decanted from saturated solution
TNT 50°/20 °C #1	99.96 mg TNT	TNT vapor	Two-temperature 50 °C/20 °C	6
TNT 50°/20 °C #2	101.46 mg TNT	TNT vapor	Two-temperature 50 °C/20 °C	7
TNT 60°/20 °C #1	100.48 mg TNT	TNT vapor	Two-temperature 60 °C/20 °C	3.9	crystals formed, wiped off
TNT 60°/20 °C #2	102.72 mg TNT	TNT vapor	Two-temperature 60 °C/20 °C	3.6	crystals formed, wiped off
TNT 82°/20 °C	100.03 mg TNT	TNT vapor	Two-temperature 82 °C/20 °C	2.6	Crystals formed, absorbed when stored for 25 h at 82 °C
2,6-DNT 30°/20 °C	99.40 mg 2,6-DNT	DNT vapor	Two-temperature 30 °C/20 °C	7
2,4-DNT 35°/20 °C	100.78 mg 2,4-DNT	DNT vapor	Two-temperature 35 °C/20 °C	6
2,4 + 2,6 DNT 30°/20 °C	100.42 mg 2,4-DNT; 1.20 mg 2,6-DNT	DNT vapor	Two-temperature 30 °C/20 °C	6

2.3. Vapor-Time measurements

To characterize the vapor release of TNT-charged training aids fabricated at various temperatures, solid-phase microextraction with an externally-sampled internal standard (SPME-ESIS) [4,6,8] was used in conjunction with gas chromatography-mass spectrometry (GC–MS). A 20 mL autosampler vial with vapor-saturated 3, 4-dinitrotoluene was used as the ESIS. The various TNT materials were placed in a 3.9 L (‘1 gallon’) can with a small hole in the center of the lid covered by an affixed septum. Charged 5.5 g PDMS aids were placed in the can and the time course of vapor release for each aid was compared to a can with ≈ 100 mg of neat TNT in an open 3 cm glass crystallizing dish. In these ‘static’ SPME-ESIS experiments [6] the time for the vapors to approach equilibrium within the sealed 1 gallon can is monitored by insertion of the SPME sheath needle ≈ 1 cm into the top of the lid and using a 1 cm fiber extension for sampling. The accumulation of the ESIS used the ‘fiber retracted’ method where the SPME fiber remains retracted in the outer sheath needle, held at the zero position of extension [8]. The ESIS sampling time was 30 s. For collection of the TNT headspace vapor of the test samples, the fiber was extended 1 cm for a sampling time of 1 h. GC separations used a non-polar J & W Scientific DB-5 column (15 m, I.D. 0.25 mm, film 0.25 μm, Folsom, CA) with a He flow rate of 1.0 mL/min. For SPME analysis, splitless injections were performed with an inlet temperature of 180 °C. The temperature program started with an initial temperature of 50 °C (held for 3 min), then advanced to 180 °C at 10 °C/min, followed immediately by an increase to 280 °C at 20 °C/min (hold 1 min at 280 °C). The single quadrupole mass spectrometer was operated in scan mode from 40 m/z to 400 m/z with an electron multiplier voltage of 2082 V. The source and analyzer temperatures were set at 230 °C and 150 °C respectively. The MS transfer line was maintained at 280 °C. The mass spectrometer was autotuned daily using perfluorotributylamine (PFTB). The Enhanced Data Analysis feature was used in conjunction with the NIST Mass Spectral Search Program (NIST/EPA/NIH Mass Spectral Library, Version 2.0 d) to analyze all spectra collected and identify unknown volatile components.

2.4. Canine detection studies

Six healthy adult male Labrador Retrievers were trained to detect explosives. The dogs were purpose-bred for detection work from a colony of detection dogs developed at the Auburn University Canine Performance Sciences Breeding Program. Each dog had one to three years of experience as an explosive detection canine. The dogs were trained in an olfaction laboratory during the training period and typically received up to 15 to 30 trials per day, four to five days per week for over two months. All activities for this project were approved by the Institutional Animal Care and Use Committees of Auburn University and NIST.

The target odor learning source was 20 g of military-grade flake TNT in a small glass Ball^™-type jar. Distractors, or ‘non-target odors’, are used to provide unrelated odors to ensure that the dog is truly indicating on the target and are similar to the target or of potential canine interest. Distractors used in this study included a large variety of common materials such as duct tape, WD40 Spray, glue, grass clippings, paper clips, window cleaner, plastic bags, cotton, and pieces of the canine’s reward toy. For every PDMS testing trial, PDMS-blanks occupied at least one position of the wheel as well as a novel distractor that the canines had not previously encountered.

A 3.66 m × 3.66 m (12 ft × 12 ft), climate controlled, indoor, olfaction laboratory was used for odor testing. In the center of the room was a presentation wheel with six arms of equal length. For each trial, one TNT-related target odor and five distracting odors (or six distracting odors for distractor-blank trials) were placed on each arm of the presentation wheel and shielded from physical contact of the dog using a wire mesh basket. The various PDMS TNT and DNT target aid formulations to be tested were removed from their closed storage containers and then “aired out” for 30 min before use. The test aids remained open throughout testing and being interrogated by multiple dogs. All target odors were randomly assigned a position (1 through 6) on the presentation wheel. Dogs were brought into the room and allowed to search, starting at position 1 and working to position 6. The dogs were always free to repel from the source of the odor and never manually encouraged to sample the vapor from any target source. The duration of a search of all six presentation wheel positions was typically 3 to 4 s. For data collection, test training aids, PDMS-blanks, distractors, and a sample of military TNT were evaluated 3 times by the 6 canines for a total of 18 contacts per individual test sample.

Extensive efforts were made to reduce confounding factors that could lead to false positive results or inflated measures of detection performance unrelated to detection/null detection of the target odor. All target and distractor holding baskets on the presentation wheel were changed after each trial. Baskets, basket holders, presentation wheel apparatus, and jars were only handled using nitrile gloves and metal forceps to eliminate human scent. Baskets were sanitized daily in a commercial dishwasher on high heat without the use of detergent. The handler was not informed as to the target location and stood out of sight of the dog to avoid influencing the dog. Dogs were monitored for behavior changes related to detection of a target odor at a specific position. Positive alerts included providing its trained final response of sitting at a target and also any “change of behavior” (COB) revealed by pausing and abrupt head turning. All dogs were operated off lead by the handler. When the dog alerted to the military TNT target odor, it was rewarded with a toy using a variable reward schedule. Rewards for the positive alert were provided randomly about 2/3 of the time. The duration of exposure to target and non-target odors was very short, typically < 0.5 s. This brief exposure provided sufficient time to sniff a basket and for the observer to determine individual dog search behavior.

In this ‘double blind’ experiment [15], the handler instructed the dogs from a waiting room to go search in the presentation wheel room. The handler remained in the waiting room while the dog entered presentation room by itself. After the dog provided a null, COB, or alert response, it was called out of presentation room, back to the waiting room. The test proctor was in a separate observation room with a one-way glass mirror to view the presentation room. If the dog alerted on a basket, the test proctor would tell the handler to reward or withhold the reward. PDMS-blank aid trials were utilized to ensure that the dogs did not provide false alerts when they searched the presentation wheel. The test proctor recorded null responses, COB, and alerts to the targets and distractors.

3. Results

3.1. Minimizing the Non-target odors from the PDMS phase

Although siloxanes that comprise PDMS are comparatively odorless to the human nose, the reagents used for polymerization contain volatile organic carbon (VOC) compounds. The ‘Space Grade’ PDMS is advertised as particularly low in VOCs for specialized applications with a stated value < 71 mg/g. Based on our headspace evaluations, and the Safety Data Sheet, a notable VOC in the Space Grade is ethylbenzene. Sylgard 184 has a stated VOC value of 86 mg/g. It was determined to have ethylbenzene, p-and m-xylene in the headspace. When freshly cured, both of these polymer formulations have a noticeable odor to humans. One of the useful properties of PDMS is its thermal stability, enabling its use in coatings for SPME fibers for temperatures up to 300 °C. It is possible to remove most of the VOCs in the cured polymer by heat treatment. For this study, PDMS aids were baked at 150 °C for 2 h. This temperature is above the boiling point of the non-target odorants ethylbenzene, and p- and m-xylene found in the headspace of unbaked PDMS. A comparison of the headspace of unbaked and baked Space Grade and Sylgard 184 PDMS as evaluated by SPME GCMS is presented in Fig. 3. Most of the VOCs are removed by this heat treatment, particularly for the Space Grade. Besides the aromatic compounds, a number of volatile siloxanes such as hexamethylcyclotrisiloxane and octamethylcyclotetrasiloxane are largely removed. However, a semi-volatile component bis(trimethylsilyl)silicic acid ester remains after this heat treatment which may be related to its high affinity for the PDMS phase. For the entirety of this canine training aid study, the Space Grade PDMS was used. However, other canine studies have successfully used the heat-treated Sylgard 184 PDMS as the odor absorptive matrix.

Fig. 3.

Studying the effect of the PDMS baking protocol on headspace vapors of two grades of PDMS. Left: the headspace of cured Space Grade PDMS (upper chromatogram) and cured Sylgard 184 PDMS (lower chromatogram) PDMS. Right: the headspace of Space Grade PDMS (upper chromatogram) and Sylgard 184 PDMS (lower chromatogram) after bake-out protocol at 150 °C for 2 h.

3.2. Results for TNT training aids

In unpublished work, detection of TNT based on the PDMS aids vapor charged for one month over ≈ 100 mg of pure TNT at ≈23 °C provided no alerts by canines trained on 20 g of neat TNT. As an alternative to vapor capture, a PDMS aid was prepared by direct dissolution of the TNT into the PDMS monomer using the mixing method. It was obvious these conditions far exceeded the solubility of the TNT crystals in the PDMS, although the PDMS did take on some light-yellow color of TNT. In a subsequent dissolution experiment, heating of the TNT to its melting point (82 °C) while mixing with the PDMS monomer did not seem to notably increase the solubility of liquid TNT and resulted in two immiscible layers.

For the two-temperature experiments in preparing TNT aids, a PDMS cooling temperature of 20 °C was investigated in conjunction with three different heating temperatures of 50 °C, 60 °C, and 82 °C. Whereas at the 50 °C and 60 °C temperatures, the vapor source TNT crystals were a solid, charging at the 82 °C melting point temperature potentially allowed more rapid production of vapor from the TNT liquid. As noted in the Experimental section, TNT crystal formation was noted for some aids prepared by this two-temperature method and these were removed as the presence of pure TNT crystals on the training aid would likely be a source of detectable vapor for subsequent canine studies.

3.3. Headspace studies

To evaluate the relative success of these various TNT training aid preparation methods at producing TNT vapor, the headspace was monitored with SPME-ESIS GCMS as they came to equilibrium in a closed 1 gallon can. The resulting vapor-time profiles were compared to the profile of ≈ 100 mg of pure TNT crystals, shown in Fig. 4. On the y-axes are plotted the A/E ratios (integrated analyte signal divided by the integrated externally-sampled internal standard signal) which are directly proportional to concentration [5]. The pure TNT (right y-axis) provided the most vapor. The two-temperature 60 °C/20 °C TNT training aid (left y-axis) provided the next most vapor, achieving approximately 1/8th as much vapor at equilibrium as 100 mg of pure TNT. The 50 °C/20 °C TNT aid (left y-axis) provided approximately 1/50th the pure TNT vapor. The mixing method TNT aid (left y-axis) provided barely detectable TNT vapor and only at the longer equilibration times. The headspace of the 82 °C/20 °C TNT aid was not evaluated.

Fig. 4.

Vapor-time profiles of TNT PDMS training aids prepared under different conditions. SPME-ESIS A/E measurements were made with 3,4-DNT as ESIS for ≈ 100 mg of military-grade TNT (orange squares, right axis); TNT mixed into the PDMS (blue diamonds, left axis); TNT PDMS aid using two temperature preparation at 50 °C/20 °C (red squares, left axis); TNT PDMS aid using two temperature preparation at 60 °C/20 °C (gray triangles, left axis). Data are single point measurements of unknown uncertainty to establish trends in the results. Lines are drawn to simply help visualize general trends and are not mathematically computed.

3.4. Canine detection of PDMS training aids

Six Labrador Retrievers previously trained to alert 100% of the time to 20 g of military TNT were used to evaluate the detectability of the training materials. Canine evaluations provided 3 detection passes by the 6 dogs for all materials, providing 18 effective measurements.

The temperatures chosen for heating the vapor-generating target compounds used for canine evaluations were based on the relative vapor pressures of the various compounds related to TNT detection. There is not great agreement on explosives vapor pressure data [16] which seems to depend on the technique used for determination. However, a reasonable comparison of the relative vapor pressures of these three compounds can be made using data generated at NBS (now NIST) [17]: TNT 0.00120 Pa; 2,4-DNT 0.0290 Pa; and 2,6-DNT 0.0756 Pa. Thus, in choosing temperatures for charging TNT-related aids, higher temperatures were used (50 °C, 60 °C, 82 °C) than for the order-of-magnitude more volatile 2,4- and 2,6-DNT (35 °C and 30 °C, respectively).

In our experience with analysis of solid samples of the military TNT used in this study, the amounts of DNT impurities are typically near the 1% level for 2,4-DNT; whereas the 2,6-DNT content is a few tenths of a percent. Despite its low concentration in the solid, 2,4-DNT dominates the headspace and has been previously identified as the primary canine odorant for detection of TNT [1,3].

The nature of the training aids tested is summarized in Table 1. Two training aids of each of the following types were used, each with separate canine evaluations: the baked and unbaked PDMS-blanks, TNT prepared by the mixing method, and the 50 °C/20 °C and 60 °C/20 °C two-temperature TNT training aids. Single aids were also prepared and evaluated by the canines: TNT 82 °C/20 °C, 2,6-DNT 30 °C/20 °C, 2,4-DNT 35 °C/20 °C, mixed 2,4- and 2,6-DNT 30 °C/20 °C aids, and the 20 g sample of military TNT. In the canine evaluations, null, COB, and alert responses were noted by the proctor in these double-blind evaluations. Although not scored as an alert, we have found it instructive to record COB in experiments where the odor delivery may be close to the canine detectability limit. The canine detection results for these training aids are presented in Fig. 5. The percent alerts are calculated for the 18 passes over each individual aid and do not include the COB data.

Fig. 5.

Results of trained canine evaluations of various training aid materials for detection of TNT. Three evaluations by six canines were made for each aid. Blue bars are positive alerts, orange bars are change of behavior. The percent positive alerts are noted for each aid.

The two baked PDMS-blank aids provided the fewest (2) false positive responses [15]. The use of the unbaked PDMS aids provided a higher number of false positives (5) with 1 COB for the 36 total passes over the two aids. Mixing the TNT directly into the PDMS to create a saturated solution of monomer prior to polymerization for two aids only produced 3 positive alerts. The two-temperature charging method of 50 °C/20 °C for the 2 TNT aids also only provide 3 positive alerts with 1 COB. The canine results for the two 60 °C/20 °C aids were slightly better with 6 alerts/36 passes and 2 COBs. Raising the charging temperature to the melting point of TNT possibly helped provide more odor for the 82 °C/20 °C aid that was tested with 6 positive alerts for the 18 passes for the single aid tested.

For the single aids based on the characteristic DNT odorants in TNT, the 30 °C/20 °C 2,6-DNT aid produced 13 alerts and 1 COB for the 18 passes. To our knowledge, this is the first time that 2,6-DNT alone has been identified as a potential odorant recognized by canines as TNT. 100% alerts were provided by an aid charged at 35 °C/20 °C for pure 2,4-DNT for all 18 trials. To simulate a more realistic ‘bouquet’ of TNT that typically contains both 2,4- and 2,6-DNT, a mixed DNT aid charged under the 30 °C/20 °C condition also produced 100% alerts for the 18 passes. Finally, all 6 dogs were able to provide 100% alerts for the 20 g military TNT sample for the entire duration of the study.

4. Discussion

4.1. Headspace study of the vapor-release by different TNT aid formulations

In Fig. 4 the vapor-time profiles of various aid formulations for the detection of TNT were studied by the SPME-ESIS GC–MS method. This 1 gallon can experimental configuration was chosen to simulate the NORT (National Odor Recognition Test) used by ATF (Bureau of Alcohol, Tobacco, Explosives and Firearms). It is notable that it takes about 24 h for the vapor from ≈ 100 mg of neat TNT to come to equilibrium saturation within a closed 1 gallon can. This delay is determined by a number of factors including the rate of vapor release by the test material, the size and configuration of the 1 gallon odor receiving container, and the often-overlooked factors of the effective density and diffusivity of the target gas relative to air. To understand the significance of this last factor, consider how a training aid works. The source aid is put in a large room and is likely never to reach vapor equilibrium as the room has openings and ventilation. Thus, the use of a training aid in practice is an ‘infinite dilution’ experiment. This implies that the ability to provide canine-detectable odor is dependent on the rate of target odor release and the behavior of that odor as it moves in the room. Many studies of training aid odorants (including our own) have used concentration as a figure-of-merit for understanding vapor release. SPME-ESIS can provide estimates of concentration provided the relative vapor-pressure of internal standard and analyte are known [3]. However, for this study, we rely only on the proportionality of the SPME-ESIS to concentration to evaluate the behavior of test training aid materials for TNT compared to the headspace behavior of the neat TNT target material.

The situation controlling this vapor-time curve is more complicated than the simple rate of release of target odorants by either the target material or the fabricated training aid. In the experiment in Fig. 4 – a closed aka ‘static’ [4] 1 gallon can experiment – the released vapor behavior also depends on the density of the target gas and its diffusivity. As density of a gas is directly proportional to molecular weight, the TNT gas is 7.85 times as dense as air (227.1/28.9 – the average molecular weight of air). In addition, the diffusivity of a gas relative to air is inversely proportional to the molecular weight of the compound through Graham’s law. Thus TNT diffuses at ≈ 1/3 the rate of air (√(28.9/227.1). These factors contribute to the slow equilibration as measured near the top of the 1 gallon can. In addition, consideration of the gas density and diffusivity helps account for the typical behavior of the canine sniffing for the target odor ‘plume filaments’ [18] moving along the floor for these heavier-than-air targets that disperse slowly.

Thus, the generally parabolic shape of the vapor-time curves noted in Fig. 4 can be explained by a combination of the rate of release of the TNT from the aid and the effect of its relatively high density and slow diffusivity to compared to air. The slope of such ‘vapor-time’ curves – revealing the rate of vapor progression - may be of more interest than absolute concentrations in simulating training aid behavior in the field.

When compared to ≈ 100 mg of neat TNT, all of the fabricated TNT training aid materials – prepared by charging with ≈ 100 mg of TNT material - provided relatively small amounts of vapor. In no case, was a notable amount of the ≈ 100 mg source material consumed by the evaporative process during charging. Thus, the rate of vapor release of the aids relative to neat ≈ 100 mg TNT was smaller relative to the TNT in the PDMS phase. This may be a result of several factors including dilution of the TNT in the 5.5 g of PDMS, the affinity of the TNT to partition into the PDMS phase, rate limiting transport of TNT to the surface of the PDMS for vapor release, and the available surface area of PDMS.

Of particular note is the complete failure of the mixing method to prepare a TNT training aid that provided notable TNT vapor relative to other formulations. The absolute solubility of the TNT in the monomer and cured polymer are probably not all that different. So limited dissolution solubility alone does not account for the difference in mixing versus vapor charging. An additional factor must limit the transport of TNT toward release into the vapor phase. We hypothesize that when polymerization is performed after incorporation of the TNT into the monomer, this tends to entrap the TNT molecules in the web of the polymer. Whereas, vapor-phase incorporation into the cured, flexible polymer tends to allow the TNT molecules to absorb on the surface of the polymer. This facilitates diffusion of the target molecules within the polymer for transport to the polymer/air interface. Thus, the aid prepared by the mixing method provided very limited TNT vapor compared to vapor capture.

However, using the two-temperature approach with progressively higher source temperatures is demonstrated to increase the amount of available TNT vapor for PDMS charging over room temperature. In none of these charging experiments was the source material notably consumed. It should be noted that the progress of vapor incorporation was not measured and was sometimes stopped after a shorter period of charging time when TNT crystals visually appeared. Sublimation is an unintended consequence of this two-temperature approach. A particularly disturbing conundrum was the appearance of crystals on the face of the PDMS. In that condition, the rate of vapor deposition is exceeding the rate of PDMS absorbing the TNT (or alternatively the PDMS has reached TNT saturation). In the case of the 82 °C/20 °C TNT material, the solubility in the PDMS was not exceeded. This was evidenced by allowing the aid to re-equilibrate at 82 °C in the closed charging container in the absence of source TNT, permitted the polymer to absorb the sublimed TNT.

Although the choices of source temperature were limited in this preliminary study, there are likely to be optimal temperature/charging time combinations for each target compound and these must be determined empirically to maximize the vapor output of the finished aid.

4.2. Dispersion of vapor from the training aid

It is clear from our vapor-time measurements in the 1 gallon can, and from observing canine searching behavior, the majority of odorants from heavier-than-air target explosives are not immediately well-mixed with the air, but rather sink to the floor in a more concentrated fashion creating lasting plume filaments. The 12-hole aid can design provides many facets for vapor dispersal, where heavier than aid targets can exit the annular side holes and lighter-than-air odorants such as ammonia [19,20] (a volatile odorant from ANFO, ammonium nitrate, and urea nitrate), with a density relative to air of 0.59, can exit out the top of the can. Most explosive odorants are heavier-than-air such as 2,4- and 2,6-DNT (6.3 times as heavy as air) and can easily exit through the side holes providing efficient 3-directional dispersion of odorants.

4.3. The two-temperature charging protocol

The primary compound used for the analytical study of the vapor-time behavior associated with aids fabricated by the two-temperature approach was TNT. Although this is not the primary odorant used by canines in detection of TNT, it was used to evaluate the viability of the approach shown in Fig. 4. From this study, it became apparent that using pure, crystalline, easily-sublimed target compounds was not the best system to provide a preliminary investigation of the two-temperature charging approach. Most explosives such as plastic explosives, ANFO, or dynamite have characteristic odorant ingredients that are often liquids at room temperature. Such ingredients will benefit from the two-temperature approach without the concomitant formation of sublimed crystals. Fortunately, for this study the crystals were easily and quantitatively removed by wiping and evaporation. The two-temperature charged TNT aids were determined to provide approximately 1/8th and 1/50th of the vapor of pure TNT at equilibrium depending on the source temperature used for charging.

4.4. Canine studies

Evidently, the vapor-charged TNT aids were not capturing and releasing sufficient amounts of the more important detectable impurity 2,4-DNT to provide reliable canine alerts (>90%) from the TNT aids. This may be related to the relatively small amounts of these impurities that would be available from the TNT charging quantity of only ≈ 100 mg.

As previously determined [1,3] this study confirms that the primary odorant in the detection of TNT is 2,4-DNT. By virtue of its higher vapor pressure than TNT, 2,4-DNT is likely to be more rapidly and effectively incorporated into the PDMS. Of additional note is the relative success of the canines detecting ‘TNT’ based on a 2,6-DNT training aid. The 2,6-DNT used for this study was reported to be high purity (99+ %), so contamination by 2,4-DNT is not likely to be a factor in providing the positive canine alerts. These two DNTs isomers are quite likely to have notably different odors to the canine. Although canines have the ability to ‘generalize’ similar odor profiles for identification, it would seem more likely that 2,6-DNT is a secondary odorant that contributes to the ‘bouquet’ for canine recognition of military TNT, at least for some dogs. Detection of this familiar potential odorant may account for the 72% alerts (but less than the required 90% alerts [15]) noted for the 2,6-DNT-based aids.

All of the aids tested were likely to provide less detectable odor than the 20 g of TNT upon which the canines were initially trained and also was used as a positive control in all presentation wheel experiments. It is well known that dogs may not reliably alert to amounts of target substances that are very different from their training materials. This is a conundrum for hazardous substances such as the drug fentanyl and the improvised explosive TATP, where small amounts of training materials are used (of necessity) but much larger amounts may be encountered in field work. There is a need for a series of training materials that provide a range of odor output, enhancing generalization by the dog.

In this preliminary investigation, none of the TNT/DNT PDMS “training aids’” were used for canine training. The dogs were trained on real TNT and the prototype materials were tested to determine the first level of utility for detection. Obviously, this is in opposition from their intended use. However, in other unpublished investigations, the PDMS capture-and-release technology has proven very useful as successful initial canine training materials for Composition C-4 and TATP.

5. Conclusion

Through this preliminary investigation, the concept of using the two-temperature approach to prepare more effective PDMS capture-and-release canine training aids for targets of low volatility was demonstrated. In carefully executed canine evaluations of this limited set of elevated source temperatures, masses of target material, and odor charging times, some of the resulting training aids associated with detection of TNT provided successful alerts. Aids providing more than random positive alerts were based on using DNTs for odor charging. The evaluation of 2,4-DNT charged PDMS aids further demonstrated that 2,4-DNT is the primary odorant that canines use to detect TNT and supported the possibility that 2,6-DNT is a secondary odorant for TNT canine detection.

In addition, a thermal treatment protocol was developed to minimize non-target odors in 2 different types of PDMS materials.