Amid acres of gently waving corn, just north of Missouri's Ozark Mountains, neuroscientist John Kauer strides through a grassy field studded with land mines. He swings a foot-long brushed-aluminum box by its black handle, sweeping it back and forth over the ground. There is no sound save the rustling of the cornstalks. Then the box abruptly bellows "Land mine!" in a voice recorded by a former drama student working in Kauer's lab at Tufts University in Boston.
The defused mines at this 22-acre government test site are used to help Kauer find a better way to locate the roughly 110 million live ones buried in Afghanistan, Bosnia, Cambodia, and other past and present combat zones. "Because they are so difficult to find, people are losing limbs and lives, and countries with economies based on farming can't rebuild after war," he says. In most of the world, land-mine clearing is a slow, risky job done by workers in Kevlar suits crawling around and feeling the ground with an 18-inch fiberglass or wooden stick. Metal detectors cannot locate the plastic casings of modern land mines; trained dogs are both expensive and difficult to maintain. So Kauer is working on a machine that can sniff them out— an electronic nose.
Companies have been selling chemical sensors to monitor carbon monoxide and other airborne gases for nearly a decade, but land-mine detection requires a higher level of sophistication. Trace amounts of TNT and its chemical precursor, DNT, vaporize from a land mine and hover closely above the ground in concentrations of no more than a few parts per trillion. Finding those scattered TNT and DNT molecules requires a daunting mix of sensitivity and selectivity. Kauer isn't quite there yet— his test run requires near-optimal weather, and it misses many potentially deadly explosives— but he and other electronic-nose pioneers are getting close. And the technology may soon find a host of uses: diagnosing disease, improving airport security, monitoring food safety, even enhancing your own sense of smell.
Four years ago, the Defense Advanced Research Projects Agency (DARPA) boosted smell-recognition research with its $25 million Dog's Nose Program. "The idea was to bring the sensing ability of a dog's nose to humans," says Regina Dugan, the program's former manager. Two distinct solutions emerged. Kauer and his team at Tufts sought to create a true electronic nose, capable of distinguishing a wide variety of smells. Chemist Timothy Swager of the Massachusetts Institute of Technology and his partners at Nomadics Inc. of Stillwater, Oklahoma, single-mindedly focused on sensitivity to TNT and DNT— a less generic, more practical approach. As a result, the Nomadics device, called Fido, was the first artificial nose capable of sniffing out a land mine in the real world.
Fido, an electronic nose, can locate land mines by sensing their chemical vapors. Modern plastic-encased mines are undetectable to WWII-era metal detectors.Photograph courtesy of C. Reding/NOMADICS
Fido draws air through tiny glass tubes coated with a fluorescent polymer, developed by Swager, that binds to DNT, TNT, and related molecules. Normally the polymer glows under ultraviolet illumination, but the fluorescence diminishes where a particle of the explosive adheres. Photomultiplier tubes detect the light and convert it into an electric signal, which is amplified and sent to a computer. This technique is touchy enough to detect a land mine from just 30 million molecules of airborne explosive and practical enough to use outside the lab. "At the end of the year, we hope to have something ready to ship," says Colin Cumming, president of Nomadics, who expects the device will cost less than $5,000. But the company's electronic nose has an Achilles' heel. Although Fido is exquisitely sensitive to the DNT family of compounds, it also responds to innocuous molecules whose chemical components resemble those of DNT. That leads to false alarms.
Kauer, whose work grows out of his theories about how the brain encodes smell, looks to biology to solve the problem of look-alike molecules. In place of Fido's fine-tuned detector, his artificial nose relies on an array of 32 broadly reactive fluorescent sensors to mimic the human olfactory system. We don't have specific receptors for oranges or for coffee. Rather, we use a wide variety of chemically promiscuous cells, each of which responds to many different cues. The vapors emanating from a cup of coffee or a peeled orange trigger a unique pattern of electrical activity in those cells, which the brain then interprets as a particular odor. Rather than zeroing in on DNT, Kauer built a system that similarly senses the overall smell of its environment: "We wanted it to be as sloppy as possible."
That sloppiness comes at a price. Last fall, when the Dog's Nose Program arranged a contest, Fido proved 100 times more sensitive to DNT than the Tufts sensor. Since then, Kauer's team has upgraded the optics, improved the airflow, and rebuilt the electronics. Recently, the Tufts nose went snout to snout with land-mine-detecting dogs in trials at Auburn University in Alabama. Researchers pumped an aromatic compound into a set of cages. Dogs trained to press a lever if they could detect the odor competed against their electronic doppelgänger. "We were probably about a factor of 10 times less sensitive than the best dogs but about par with the worst dogs," says Joel White, one of Kauer's collaborators.
Nomadics, too, is working on further refinements, most notably an attachment to vacuum up dust particles containing TNT by-products in the soil. During the coming year the company plans to begin trials with real land-mine sweepers. "The key is to work with people in the field to meet their needs so that they adopt it," says Cumming.
Meanwhile, other researchers are bringing artificial noses to bear on less dire problems. With support from the Dog's Nose Program, chemist Nathan Lewis of the California Institute of Technology developed another polymer-based system, one in which certain odors make the sensor expand slightly. Although this approach lacks the sensitivity to find land mines, it is easily miniaturized. Lewis licensed his technique to Cyrano Sciences of Pasadena, California. Last September the company started selling $8,000 portable sniffers to monitor food, air, and packaging quality. Now Cyrano is sponsoring research into smell-based diagnosis of upper-respiratory infections. Likewise, Osmetech of Crewe, England, is awaiting Food and Drug Administration approval of a multisensor gas-sensing array to screen patients for bacteria that cause infections of the urinary tract.
Regina Dugan, of DARPA, foresees military uses for an electronic nose with interchangeable cartridges, each specialized for a different task: measuring contamination in a tank's diesel fuel, detecting drugs in airports, or sounding an alert if traces of chemical-warfare agents waft by. "Consumer applications are theoretically possible," she adds— perhaps a food-safety monitor in the refrigerator or a sensor to check for indoor pollutants.
Someday a bionic nose might even replace our own. White imagines a day when a supersensitive electronic nose could be wired directly into the human brain to allow us to explore the vivid world of smells currently experienced by our pets. After all, scientists have developed cochlear implants that tap into hearing and are making astounding progress with silicon-vision prosthetics. But Kauer, who has spent a career trying to decode the process of smell, will believe it when he smells it. "Noses are amazingly complicated," he says.
The Defense Advanced Research Projects Agency put together a nice overview of the Dog's Nose Program when it began in 1997. The site is a bit dated now but gives a good sense of the research goals and various teams' approaches: web-ext2.darpa.mil/ato/programs/UXO/html/making.html.
Several companies and research groups have helpful information available online. Nomadics (www.nomadics.com) features a whole section on land-mine detection. Cyrano Sciences (cyranosciences.com) and Osmetech (www.aromascan.com) discuss how existing electronic noses are being used.
Check out Discover's archives (www.discover.com) to learn more about Nathan Lewis, Cyrano Sciences, and the current and potential uses for their Cyranose: "Future Foods," Brad Lemley, December 2000; and "The Electronic Nose," Gary Taubes, September 1996.
The International Campaign to Ban Landmines (www.icbl.org) has a slow but extensive site of photos and information about the land-mine problem. This organization won the 1997 Nobel Peace Prize for its efforts.