A Good Nose for Business
In a windowless underground laboratory, a Caltech grad student squirts water into the brain of an anesthetized locust.
He keeps his eyes on the dropper as measured moisture falls into the bug’s brain. “He’s getting dry,” Michael Wehr explains.
The bug says nothing.
That is, the electron conductors poking at two specific neurons in the olfactory portion of the locust’s brain say nothing. The bug is not giving any readings--yet. But Wehr has 78 to 98 more neurons to test, and he will keep the bug’s brain wet until every last neuron has been examined.
This intricate work with locusts is one of several experiments in a five-year project to produce the world’s first electronic nose. It combines biology, chemistry, computer science and silicon technology to come up with a silicon chip the size of a dime that is endowed with a sense of smell.
So far it can detect only three odors--beer, wine and hard alcohol--but electrical engineering professor Rod Goodman said he expects that Caltech’s whiffer will attract a variety of industry buyers.
The chips eventually might work as smelling silicon slaves, sniffing without rest and replacing dogs and hypersensitive human olfactories that detect everything from drugs to perfumes.
“There are people who sit at the end of a factory line and smell if the cheese is bad,” said chemistry professor Nate Lewis. “We can have it do that.”
Goodman said he also hopes to gain backing from the automotive industry to build a nose that can “sit underneath your car hood and smell for things such as oil leaks.”
Most of the funding for the project comes from an $11-million grant from the National Science Foundation. The foundation awarded the money to Caltech’s Center for Neuromorphic Engineering, which is dedicated to electronically imitating the body’s nervous system, said Goodman, the center’s director.
Private industry thus far has adopted a wait-and-see approach. Earlier this year, General Motors said the technology needed more work before receiving private industry support, according to GM senior researcher K.P. Unnikrishnan.
GM has shown an interest in the neuromorphic center, working on a joint project in its optics research department. But Unnikrishnan said the smelling chip has a long way to go before it noses its way under the hood of cars.
“Will it be in GM cars in 10 years? Maybe. In two years? I don’t know,” he said. “We don’t have a joint project at this point in olfaction.”
Hoping to convince motor companies to engage in a joint project, Caltech researchers are working toward a nose that can “smell the interiors of cars to test for consistency in the leather,” Lewis said.
An even more discriminating nose could follow--one that could sniff out truly expensive wines and look down . . . itself . . . at the cheap stuff.
“We can have it assign human value judgments,” Lewis said. “We could teach it what perfumes or wines are considered good and what are bad smells.”
Although it cannot make value judgments just yet, the nose knows how to differentiate among some chemicals.
It does so through a series of minute polymer sponges that lie in a row inside Pinocchio, Lewis’s pet name for a large, glass-encased system of tubes, pumps, hoses and wires. Like tissues in our own noses, the 17 sponges react to the vapors by swelling. But each sponge is made from a different polymer and swells differently in response to specific scents.
Once vapors are sent racing inside Pinocchio and the polymer sponges start to swell, a computer snaps to life: It measures the amount of swelling and graphs the results to diagnose a smell.
Other groups throughout Europe and the United States are doing similar nose work, but Caltech is the only project using polymers that can pick up several scents at once, instead of just one. Because the sponges react to all odors, they should--once etched into silicon--give a chip a broad-based sense of smell.
Industries, such as car manufacturers, that deal with a variety of vapors “couldn’t afford specific sensors for each smell. It costs too much,” Lewis said. “But if [the chip] used a broad-based smell sense, then it becomes more economic.”
It also becomes more natural. Humans and animals most likely lack specific sensors for every scent in the world, and the economic value of having one chip recognize a variety of odors is “a good reason to do it the way biology does it,” Lewis said.
The only problem, according to Caltech associate biology professor Gilles Laurent, is that scientists do not really know how biology does it.
That is where the locusts come in, said Laurent, who, along with associate biology professor Jim Bower, is involved in the biology department’s side of the project.
Laurent and his graduate students plan to poke tiny glass micropipettes, or electron conductors, at each of the locusts’ 80 to 100 olfactory neuron receptors. It is a tedious task, as grad student Wehr demonstrated on a recent day, dressed in a Los Angeles City Jail T-shirt that he swears he got at a thrift shop. But the researchers hope it will show them a pattern to the way the locust’s brain smells--a pattern they could imitate electronically.
For now, Wehr and his colleagues will remain stuck in the windowless research lab poking conductors at bug brains.
“Outta one jail and into another,” said Wehr, presumably commenting on his shirt as he stares at his locust friend. And from beneath the hot lights and microscope of the neuron tester, the bug stares back at him in silence.
