Chemical & Engineering News
September 29, 1997
Copyright © 1997 by the American Chemical Society
Detecting Illegal Substances
This Special Report looks at three aspects of the search for ways to detect illicit substances such as illegal drugs and explosives. The first part examines how specialists are trying to improve the already striking ability of dogs to zero in on specific "odor signatures." The second part explores the wishes and wants of law enforcement and aviation security officials with regard to technology for detecting explosives. The third part describes a novel way of capturing the boundary layer of air that surrounds every human as a means of sampling for residues of illegal substances.
Unlocking The Secrets Of Supersniffing Dogs
Jamie, an athletic, black Labrador retriever trained by the Australian Customs Service to search for narcotics, epitomizes the unrivaled ability of dogs to locate objects through their sense of smell. Yet, it is not fully understood how dogs do what they do and why they are so good at it. Basic studies of canine olfaction and breeding are helping researchers figure out the secrets of the dog's exquisite detection capabilities. They are also generating knowledge that may be used to enhance the technology of canine detection and to design detectors based on canine olfaction.
Australian detector dog Jamie and handler at work.
In the U.S., the rising frequency of criminal bombing incidents has increased the visibility of dogs as explosives detectors. Last June, for example, bomb-sniffing dogs were called in several times to assess a bomb threat at an office building just a few blocks from the White House. And as a result of the recommendations of the White House Commission on Aviation Safety & Security, chaired by Vice President Al Gore, the Federal Aviation Administration (FAA) has established a program that makes detector dogs available to all major U.S. airports in case of a security alert.
Those who have seen dogs in action swear they work. Dogs can find people, bombs, land mines, illicit drugs, illegal agricultural produce-anything that has a discrete odor that they have been trained to find. But skeptics question the wisdom of relying on a tool for which a scientific basis has not been fully established.
"The greatest disadvantage now to the use of dogs as detectors of illicit substances-terrorist bombs, illegal narcotics, and chemical and biological warfare agents-is that we don't know how they work," says James A. Petrousky, a chemical engineer and a program manager at the Department of Defense's Office of Special Technology, Fort Washington, Md. "Would you want your safety to depend on something not well understood?" he asks.
Petrousky also coordinates the Multiagency Canine Substance Detection Working Group, formed by federal agencies that use dogs-such as FAA, the Secret Service, the Customs Service, and the Federal Bureau of Investigation-to pursue the mutual goals of understanding and enhancing canine detection technology. The wide interest in this group's work is evident from its funding, which, Petrousky says, comes not only from member agencies but also from others such as DOD, the Technical Support Working Group of the National Security Council, and the Office of National Drug Control Policy, as well as several foreign governments.
Unlike machines, dogs do not come with precise specifications, notes Susan F. Hallowell, a research chemist and acting program manager of airport security technology integration at FAA. She explains: "If a vendor comes to me with an explosives detection system, the first thing I'll ask for is the data package: How does the device work? What does it detect? What is the detection limit? What is the probability of detection? What is the mean time between failure?"
The heightened interest in canine detection is keeping scientists very busy finding answers to such questions about dogs. Among them is James M. Johnston, a professor of psychology and director of behavioral research at the Institute of Biological Detection Systems, Auburn University, in Alabama. His federally funded research is coming up with very interesting results.
One of the nagging questions concerns the limit of detection: How small an amount of a material can dogs detect?
When tested with methyl benzoate, a degradation product of cocaine, the limits of detection of dogs range from 5 to 27 ppb, Johnston's studies indicate. Data for other compounds are being compiled, and similar levels of sensitivity are expected for most compounds.
This level of sensitivity is comparable with those of most analytical instruments and is not extraordinary, comments Petrousky. "The dog's supersensitivity is a myth," he says. "Rather it's the dog's accurate and discriminating detection ability that makes it invaluable."
Another question: What is it really that dogs respond to when they're being trained to recognize various substances through smell?
Using nitroglycerine-based smokeless powder, Johnston finds that what humans expect the dog to be smelling-nitroglycerine, in this case-is not in fact what the dog is responding to. Indeed, when a type of nitroglycerine-based smokeless powder is presented to a dog, the odor the dog associates with the material, the odor signature, is composed of acetone, toluene, and limonene.
Johnston with research dog.
The finding that explosives have odor signatures unrelated to the energetic component has serious implications in training. "In the dog world, some people think the dogs should be alerting to the explosive component," says Johnston. "If you force the dog to recognize the component of your choice, you may be making the dog's job much harder." That's because most explosives have very low vapor pressures.
Even though nitroglycerine comprises the bulk of the smokeless powder tested, because of its very low vapor pressure (0.00026 mm at 20 °C), very few molecules are in the air at any time. Components with higher vapor pressures will have more molecules in the air even if they are present only in minute amounts in the material. If the canine nose has the appropriate receptors, these components would be far easier for the dog to smell.
Knowing the odor signature of target materials will help in the design of training aids. FAA, for example, is supporting an effort to study whether dogs can reliably find explosives after training on reconstituted odors. Such aids would not have the safety problems associated with using actual explosives to train dogs nor the security problems of illegal drugs.
Odor signatures will also help explain false-positive responses. People have reported bomb-sniffing dogs alerting-that is, signaling they have found something-on materials unrelated to explosives, says FAA's Hallowell. In such a situation, more dogs are brought in to sniff the material. "If all the other dogs also [alert], we conclude that it might be something that, to the dogs, smells like explosives," she explains.
"It doesn't mean the dog is wrong," explains Johnston. The dog may have detected the odor it was trained to smell, but the scent may be coming from some other material that has the same odor signature as the target substance. For example, a dog trained to look for explosives based on ammonium nitrate may alert at the scent of ammonium nitrate in garden fertilizer.
Johnston's group is identifying the odor signatures of other substances such as illicit cocaine and additional explosives. "As we build a library of signatures, we will understand [false alarms] better and even predict the kinds of false-positive responses we can expect," he says.
Johnston is now beginning to explore other issues that need to be addressed. One is the dog's real-time sampling capability, a major advantage of dogs over machines.
The real advantage of dogs is that they "go to source." That is, dogs find things, whereas machines now can only verify. As Petrousky puts it, "Machines are very good in answering the question, Is this particular sample cocaine? But only a dog can answer in a timely manner the question, Is there cocaine in this room?"
As a sampling device, the dog is remarkable, says Johnston. It takes little sniffs while it moves briskly back and forth, using the tiniest whiffs of an odor to decide whether to keep sniffing at a particular area or move on. If it senses an odor of interest, it will follow that odor and quickly locate the point of highest concentration. And then it alerts, as trained.
"The dog is obviously very good in localizing a source," which current state-of-the-art systems cannot do, says Johnston. "If we understand how the dog does that, it might tell us how to program or develop instruments that can."
Another issue that needs to be addressed is calibration. Many factors, internal and external, can affect canine olfaction, including heat stress, toxins, injury, and even disease at presymptomatic stages. Right now, with dogs, there's no "press-to-test" button to check if the system is working effectively-a major disadvantage, says Petrousky.
What's needed is some means of checking the dog's nose before the dog is deployed, "just like highway troopers test their radar guns before they go on patrol," says Johnston. "If you're the handler of a dog that searches for land mines, you definitely want to know if your dog's nose is working as well as it usually does," he adds, "because sniffing is just a behavior, and dogs will keep sniffing when they can't smell anything."
But it's not all about the nose. Although most dogs have a keen sense of smell, only a few can be trained to be detector dogs like Jamie, the Australian Labrador retriever. Even at an early age, Jamie exhibited a trait that Australian researchers now have found to be a key determinant of good detector dogs.
That trait is called independent possession. It relates to how well a dog remains motivated to play a game that's used in training, explains Kathryn A. Champness, breeding manager at the Royal Guide Dogs Associations of Australia (RGDAA), Kew, Victoria. In collaboration with RGDAA and the Australian Customs Service, Champness completed a Ph.D. degree at the University of Melbourne on the development of a breeding and rearing program for drug detector dogs.
Training of Australian drug detector dogs is based on hunting for a toy that is scented with a narcotic odor. When a dog finds the toy, the handler rewards it with lots of praise and a game: The handler plays tug-of-war with the dog, and when she gets the toy off the dog, she throws the toy, and the dog chases it and brings it back.
TO SIDEBAR - Dog talk: Extracting unambiguous answers to questions
By observing techniques used by the Australian Customs Service, Champness arrived at several traits that are important for work as detector dogs; for example, high chase/retrieve drive, possessiveness of toy, strong hunt drive, good stamina, stable temperament, and positive response to praise.
These traits are moderately heritable, Champness found. Therefore, a program that breeds selectively for these traits should be successful. But because other factors such as the environment or gene interactions also play a significant role in the expression of these traits, "progress by selective breeding will not be as rapid as it would be if the traits were highly heritable," explains Champness.
The most highly heritable character is related to the dog's possessiveness of its toy. Champness says the trait was assessed in three ways. One is mental possession, or how well the dog stays focused on watching its toy all the time. Another is physical possession, or how strongly the dog holds on to its toy when it has retrieved it. But the most highly heritable is independent possession, which is how well the dog remains focused on playing regardless of input from its handler.
This finding "ultimately was very nice for us," says Champness. "Because we're really after dogs that are totally focused on playing this game and don't get distracted by other things."
Champness also studied the effect of environment and found that dogs need to be exposed to the training games as juveniles in order for the characteristic to be expressed in adulthood.
Guided by these findings, the Australian Customs Service now has a very successful breeding program for detector dogs. Before the breeding program was in place, the service recruited dogs from pounds, shelters, and private breeders, says John Vandeloo, manager of the service's National Breeding & Development Centre. Not only was the supply insufficient, the quality of the dogs was unreliable. Less than 50% graduated as detector dogs, and there was no guarantee of the dog's working life. The cost was about $17,000 per detector dog per eight-year cycle (eight years being the average working life of a Labrador retriever).
Now the service has a steady supply of dogs suitable for training as detector dogs. "Each generation is getting better and better," says Vandeloo. As a result of better selection of dogs for training, the cost per detector dog per eight-year cycle has dropped to only $3,600.
The Australian Custom Service's use of detector dogs, such as Jamie, must contend with a point widely regarded as a disadvantage of dogs as detectors: They can't be used for routine screening because their duty cycles are short.
"Australian customs has to use the dog as a broad screening tool because of the millions of people and articles moving across borders," says Vandeloo. "We like to put them out front so they're constantly working."
There's a lot of diversity in what they do on any given day, as Jamie's average schedule shows. Such diversity is good, because canines, like humans, need variety. "If you just keep throwing them bag after bag on a conveyor belt-humans lose interest and so would dogs," says Vandeloo.
Optimizing rest time is also very important. For example, the travel time from one job to another should be a rest period. "Some dogs never shut down," says Vandeloo. "In between jobs, they may go stir crazy in the back of the van, barking, carrying on. By the time they reach the next job, they're tired."
Thus, part of the training is teaching the dogs to respond to a "trigger," for example a harness. "Once we put the harness on, it's like turning on the power, and the dogs are very enthusiastic," Vandeloo explains. "Once we take the harness off, the dogs know to cool off, relax, and conserve energy."
By late afternoon on Jamie's average day, the local police may have called, needing the assistance of a dog. And Jamie finishes the day by searching two or three premises.
Champness first tested Jamie when he was about six months old, during the very early stages of her research. "I had the gut feeling right then that he was very special," she recalls. He's been working in the field very successfully and also works as a stud. His offspring are also performing very well. "Jamie's success is living confirmation that I was doing the right thing," she says.
Better Detectors Needed For Varied Threats
Upon the recommendation last year of the White House Commission on Aviation Safety & Security and with funds provided by Congress, FAA is buying and deploying 54 units of the CTX 5000. This commercial detection system, made by InVision Technologies, Foster City, Calif., is the only one that has passed the agency's strict performance criteria. The move reflects "a change in national philosophy as to who is responsible for purchasing bomb-detecting equipment," says Lyle O. Malotky, FAA's scientific adviser for civil aviation security.
In the U.S., this responsibility used to rest squarely with the airlines. In response to the loss of TWA flight 800, which initially was suspected to have been caused by a bomb, "Congress and the Administration basically said [bomb detectors in airports are] important enough that perhaps the federal government should have a role in at least priming the pump to get the technology out there," Malotky says.
The CTX 5000, which is based on computer tomography and costs about $900,000 per unit, is doing "an adequate job" in screening checked-in luggage for explosives, says Malotky. More important, he says, it's giving FAA and U.S. policymakers a "baseline of equipment performance" under operational conditions.
"We would like to have equipment that is faster, has lower nuisance alarm rates, and is less expensive" than the CTX 5000, says Malotky. So far, a better system does not exist. "It may well be that we have looked at all the ways to detect the presence of explosives," he notes. "If indeed there are technologies left to exploit, we could use them to make the next revolutionary jump in capabilities."
An airline crash is sure to capture headlines whatever the cause, but especially if it's due to a bomb explosion. Yet more bomb-related incidents occur outside of aviation, and technology to deal with everyday threats-whether in U.S. towns and cities or in troubled areas of the world-is practically unavailable.
The latest statistics from the Federal Bureau of Investigation show numbers of criminal bombing incidents in the U.S. rising from 1990 to 1995. Except for the bombings of New York City's World Trade Center in 1993 and of the Alfred P. Murrah Federal Building in Oklahoma City in 1995, most are relatively small events that involve small homemade devices and don't do serious damage. However, the number of significant incidents is increasing, and the devices are becoming more sophisticated, according to David G. Boyd, director of the Office of Science & Technology of the National Institute of Justice, Washington, D.C.
"It's important to remember that [in the U.S.] bombings of big targets such as in Oklahoma City are very rare," says Boyd. "But at the local level, bombings are not infrequent and bomb squads are called out regularly."
In case of a bomb threat, the first response usually has to come from local authorities, such as the local police department, explains Boyd. However, these local units are the least trained, least funded, and least equipped of all law enforcement forces when it comes to dealing with bombs, he says.
The general rule is that local law enforcement doesn't "have anything at all in terms of bomb detection equipment," says Boyd. Although bomb detectors are available commercially, they are not practical for local bomb squads. "Most are expensive," he says, explaining that most police departments cannot afford any equipment that exceeds the cost of a laptop computer. In addition, most commercial equipment requires trained operators. "Police departments have a tough time sending people for training," says Boyd, because staff sizes are very small.
The ideal equipment for local bomb squads and first responders should cost no more than $10 and be so small it can be worn like a badge or on a belt or stuffed in a pocket, says Boyd. "It should be so reliable that it never misses the real thing and only false alarms once or twice a year."
That's a tall order, especially for Boyd, whose office engages in R&D of technologies to meet the needs of local law enforcement. And he acknowledges: "We're not close to where we would like to be."
Although criminal bombings are a tough problem, what U.S. local law enforcers have to deal with is probably a fraction of what their counterparts in Israel have to face day by day.
Jonathan Shoham, head of Israel's Branch for Counterterrorism & Detection of Explosives, explains the magnitude of the job: "We have to identify explosives on a person getting into a bus or a cinema or an airport check-in hall. We have to find it in a journalist's camera pointing at the president, in a bag or carry-on luggage of someone getting on an aircraft, in a cargo container, in an envelope sent to your house, or in a car traveling toward a government building. And all that as quickly, as precisely, and as cheaply as possible."
Shoham admits it would be very difficult to use technology to find a suicide bomber or to stop an attack in a crowded open market. "It's a complicated area," he says, and he hopes the explosives detection research community will share knowledge "to create new ideas to help deal with a threat that influences the lives of all of us."
Bomb detection technology would be very useful at strategic sites-at border crossings, for example. Shoham says it would help to have technology that could alert border agents to people carrying explosives as the individuals pass through without anyone stopping them. Another place where technology may be applied is in shopping malls. Something like a handheld metal detector to check people for explosives can diminish the threat at such targets, he says.
In the end though, says Shoham, the citizens' perception of the threat is what really matters. He says Israelis know they live in a threatened area. And they're willing to pay for security, for example by arriving very early at airports to be questioned or by accepting the presence of security agents almost everywhere.
By contrast, in the U.S. for example, "TWA flight 800 was in the news for a while," Shoham notes. "But now," he wonders, "how many people are still thinking of security?"
Sampling Body Heat, Skin Flakes
How these portals take a sample and bring it to the detector-the so-called front end of an integrated detection system-is an important issue. Some portals use a barrier that comes in contact with a person's body and clothing. Others blow currents of air to dislodge explosive traces from the person.
These sampling methods are inefficient and may even be counterproductive. Sampling by contact is not only intrusive but useless unless the barrier touches the contaminated area. Blowing air currents to dislodge explosive traces further dilutes the sample and could lead to its loss.
So researchers are looking to natural processes for a better way.
In the comic strip "Peanuts," the character named Pig-Pen is stubbornly dirty, invariably followed by a trail of dust. It appears Pig-Pen's dusty envelope is not unique, just an exaggeration of an air boundary layer that naturally surrounds any human body. That phenomenon now is being applied to sampling of trace explosives by Huban A. Gowadia, a mechanical engineering graduate student, and Gary S. Settles, a professor of mechanical engineering and director of the Gas Dynamics Laboratory at Pennsylvania State University, University Park.
Skin is usually warmer than ambient air and heat flows from the body to the surroundings, creating a freely moving boundary layer, explains Gowadia. For a person standing still, the layer begins at the ankles and moves up the torso, becoming thicker and moving faster and with more turbulence as it travels upward. Finally, swirling at the rate of about 40 L per second, it takes off from the shoulders and the head in the form of a thermal plume.
At the same time, humans constantly shed their outer skin layer. So much is shed that up to 90% of environmental dust in homes and offices is composed of human skin flakes, says Gowadia. Because the flakes are very small, they move freely through clothing and into the motion of the boundary layer. Thus the moving layer is a particle-laden flow, transporting skin flakes at about a third of a milligram per second for an average body surface area of 1.8 sq meters. Carried along are any materials lodged in the flakes, such as body fluids or traces of cosmetics, perfume, or explosives.
Air flow around the human body images by a color flow visualization system.
Gowadia believes the sheer numbers of skin flakes in the human plume provide a large cross-section to which trace explosives will attach. "In fact," she says, "explosive traces on skin and clothing will likely be shed continuously along with skin flakes and textile fibers without the need for external agitation."
The skin-flake-laden human plume provides "a simple and elegant approach" to sampling for explosives detection portals, says Gowadia. In research supported by the Federal Aviation Administration (FAA), she is developing a sampling system that will collect a person's natural thermal plume and filter the airborne particulates, made up mostly of skin flakes. Her preliminary tests show that explosive traces from samples concealed under clothing can be found easily in the thermal plume.
Susan F. Hallowell, a research chemist and acting program manager of airport security technology integration at FAA, says the Penn State research is "really quite novel." And she is not surprised that imaginative approaches are coming from areas other than chemistry.
"Chemists have been so fixed on detector development, and that's exactly what we got: very well developed detectors that have no front ends," says Hallowell. " We're going to have to reach out to other disciplines to develop novel sampling systems."