Explosives, chemical warfare agents, and toxic industrial chemicals present the greatest threats for soft targets, including stadiums, malls, theme parks, airports, subways, train stations, and high-rise buildings. Most screening today uses a "portal" system that generally includes individual walk-through stations. These systems restrict movement, create traffic-flow bottlenecks, are physically large and costly, and are labor-intensive to operate. By definition, portal sensors are also fixed in place and cannot be deployed in a distributed format for continuous sensing over wide areas such as airport terminals. More complete security screens would survey waiting rooms, bathrooms, and shops, using distributed monitors that could detect hazardous materials carried by people moving through these areas.
Although some hazardous items are not volatile (e.g., viruses and some toxins), many explosives, toxic industrial chemicals, and chemical warfare agents fall into this category. If inexpensive, sensitive, and rapid detectors were deployed at multiple locations, networks of sensors and radionuclide detectors could be set up to monitor and report in identified areas. The addition of cameras would help pinpoint the threats.
 |
The Olfactory Technique
One of the most highly developed devices for detecting volatile airborne compounds is the "vertebrate olfactory system," as exemplified by the use of dogs for detecting vapor signatures. The nose can detect and discriminate among many different odors by using broadly responsive sensors that provide a pattern of responses across an array, rather than relying on individual, odor-specific sensors to capture information. The brain then interprets this response pattern to recognize odors. Dogs trained to sniff out and identify specific chemical substances (mostly explosives and narcotics) are often used in military and civilian settings. Yet dogs have a limited duty cycle, can function only with trained handlers, and cannot be used in the presence of chemical threats.
An inexpensive device that can mimic some features of vertebrate olfactory systems (broad sensitivity, pattern recognition), and provide continuous surveillance would be highly desirable. The challenges of creating a biomimetic olfactory device lie in mimicking the broadly responsive sensor array and in finding sensors similar to the olfactory receptor proteins in the vertebrate nose that would allow the device to identify specific odorants of interest and discriminate between targets and interferents.
Arrays vs. Monospecific Sensors
The broadly responsive sensor array has advantages over systems based on "monospecific" sensors. True monospecific sensors are difficult to produce, whereas broadly responsive sensors are readily made. Even if perfect monospecificity could be achieved, detecting and identifying multiple compounds of interest would require a separate sensor for each one. Conversely, a relatively small array of combinatorial, broadly responsive sensors can discriminate among a large number of different compounds.
Solving the Interferent Problem
The challenge of odor discrimination in vapor phase detection can be overcome in a number of different ways. First, a device can be trained to detect a target analyte under a number of background conditions, extremely beneficial when the environmental baseline has been determined. It was used when field-testing the electronic nose's ability to detect landmines by establishing training patterns that represented a number of different environmental contexts (e.g., bare earth vs. long grass).
Second, because sensors can be selected from large banks of potential detecting materials for certain tests, they can be chosen to have intrinsic sensitivities that exclude responses to the interferents defined for those conditions. For example, in developing sensors to detect toxic industrial chemicals and chemical warfare agents for first responders, many of the defined interferents consisted of hydrocarbons (e.g., diesel, gasoline, and toluene). The sensors we have elected as appropriate for the defined target vapors generated negligible responses to these interferents.
Third, the challenges of more common interferents, such as water vapor (polymer-based sensing devices are notoriously sensitive to water vapor), can be overcome by choosing sensors with different water vapor responses as well as different responses to the designated targets. This allows the assessment of the nature and extent to which water vapor contaminates the authentic target signature. Algorithms are then developed to treat the target signature in the presence of water vapor as a mixture of "target volatiles" plus "water volatiles," and to then separate these signatures from one another to achieve authentic identification.
 Figure 1. ScenTraK uses optical-electronic circuits and sensor arrays to detect and identify a large number of volatile chemicals |
ScenTraK
ScenTraK (Figure 1) is an inexpensive electronic nose platform that meets the requirements necessary for distributed sensor networks with regard to rapid identification, cost, size, weight, ease of use, and other factors. Sensors are arrayed in a sniffing chamber and fluorescence changes of the dye-polymers in response to volatile chemicals are detected by optical-electronic methods (Figure 2). A patented sensor technology significantly increases the number of potential sensors available and reduces the economics of sensor discovery through rapid screening of large numbers of candidates for detecting particular volatile compounds of interest. Whereas polymer/dye, metal oxide, or conducting polymer sensors are usually identified one by one through specific synthesis and screening, ScenTraK's odor detectors use either dye-polymers or short-chain, single-strand dried DNA oligomers coupled to fluorescent dye. Specific nucleic acid sequences yield sensors that interact differentially with volatile chemicals to give fluorescent signals that vary in amplitude, time course, and sign, resulting in characteristic patterns for each odor (Figure 3). The tremendous combinatorial complexity of DNA oligomers—yielding 421 sensors from a 21-mer oligonucleotide composed of standard DNA bases—and the ease of generating large numbers of specific sequences, generate huge arrays from which to select sensors with desirable traits and specificities.
 Figure 2. Diagram of ScenTrak sensing chamber, showing sniff pump and organization of 16 sensors deployed on substrates in the air chamber, along with LEDs, appropriate excitation, and emission filters for each sensor, and photodiodes to detect the fluorescence signal |
To draw on the power of these DNA libraries called for developing a way to rapidly and simultaneously screen thousands of potential sensors to find those that respond to particular volatiles of interest. For example, recent experiments have identified DNA sensors that respond to nitro-aromatic compounds such as DNT, at sensitivities, even without optimization, of ~2 x 10-2 , or ~6 ppb.
 Figure 3. Several examples of patterns of fluorescence changes (Y axis) during 3 s sniffs of several specific odors in a 16-sensor array (sensors along the X axis) as a function of time (Z axis) |
Ubiquitous Applications
The availability of such libraries of broadly responsive sensors allows devices to be customized. This in turn creates an opportunity for deploying sensing technologies in situations not currently covered except by human observation. For example, units can be permanently installed in strategic locations, configured as distributed network arrays that continuously monitor transit halls, checkpoint queues, ventilation ducts, and unsupervised areas such as restrooms. Handheld units can also be transported quickly through crowds and brought into proximity with persons and items of interest, such as knapsacks or vehicle trunks. The military can have one-person sensors that detect threats and accommodate interferents unique to the battlefield, and first responders and law enforcement can have devices tailored to their specific requirements as well.
One example is the development of a handheld chemical detection device for first responders using specialized DNA and other dye-polymer sensors for detecting toxic industrial chemicals and chemical warfare agents. These sensors can detect and identify targets in the presence of a variety of real-world scenarios, including varying humidity and temperature in the presence of common interferents. The sensing platform will incorporate features specifically designed for the application including a networking and locating capability in order to map a HAZMAT event in real time.
Clearing the Air
Rapid access to a large number of broadly responsive sensors that can be custom engineered to specific threat paradigms, in conjunction with a hardware platform that incorporates other elements of the mammalian olfactory system, yields an inexpensive, robust sensing capability. Devices based on these technologies allow monitoring equipment to be used in areas not currently regarded as capable of being monitored. Security can go beyond the portal and into environments currently underprotected.
John S. Kauer, PhD, and Barbara R. Talamo, PhD , can be reached at CogniScent Inc., North Grafton, MA; 508-839-7973, [email protected], [email protected], www.cogniscentinc.com.