DNA-based sensors know what the nose knows

Can a machine mimic the human—or better yet, the canine—sense of smell? To do so, it would have to not only determine whether a chemical vapor is present in small amounts but also figure out, at least partially, what chemical it is.

Carbon nanotubes and other nanomaterials do a good job on the first front. Their small size means that the presence of just a few gas molecules is enough to measurably change their electrical properties. But to discriminate among many different molecules, an “electronic nose” must contain an array of sensors, each with different response characteristics.

In 2005 A. T. Charlie Johnson (University of Pennsylvania), Alan Gelperin (Monell Chemical Senses Center in Philadelphia), and their colleagues began to investigate whether a nanotube sensor decorated with single-stranded DNA, as shown in figure 1, might provide the needed specificity to serve as an array element. ¹ They found that sensors made with different DNA strands did indeed show different responses, as measured by the nanotube’s conductivity, to the same odorant chemicals. The sensors responded to the odorants within seconds, recovered their equilibrium conductivity when the odorant was removed, and maintained a reproducible response for dozens of cycles.

Now, the same researchers have turned their attention to the problem of telling the difference between very similar molecules. They’ve found that with suitably chosen DNA sequences, they can create sensors that discriminate between organic molecules that differ by a single carbon atom, and even between molecules that are enantiomers, or mirror images, of each other. ² Human noses can do that, but not many electronic sensors can.

Figure 2 shows one pair of enantiomers the researchers looked at, (+)-limonene and (−)-limonene. To us, one smells like lemon–orange; the other smells like sour orange and turpentine. In DNA–nanotube sensors made with one particular DNA sequence, conductivity through the nanotube increased—by up to 40%—in the presence of (+)-limonene and decreased just as much in the presence of (−)-limonene. The same sensor could also distinguish, though less strongly, between the two enantiomers of carvone, one of which smells like spearmint and the other like caraway.

The researchers tested their sensors in the lab under carefully controlled conditions, with just one odorant in a stream of argon gas. But to be useful components of an electronic nose, sensors would have to operate in air under a range of atmospheric conditions—humidity, for example—and in the presence of background odors.

It’s still not understood exactly how the DNA–nanotube sensors work. “But that’s the case for essentially all chemical detection schemes based on nanostructure transistors,” says Johnson. “It would be terrific if we could develop that understanding in the coming years, ideally to the point where we could model the responses quantitatively.”