ALBUQUERQUE, N.M. The world's first micrometer-sized acoustic-wave sensors have been successfully integrated onto the Sandia Microelectronics Development Laboratory's "lab-on-a-chip." The handheld chemical-identification system is scheduled to be completed in the fourth quarter.
Acoustic-wave sensors use piezoelectricity to set up a high-frequency surface wave that is sensitive to molecules absorbed on the surface. A molecule causes the wave propagation to slow, and the specific dynamics are analyzed by intelligent pattern-recognition algorithms that can deduce the type of molecule.
"By adapting techniques developed for silicon micromachining, we can create acoustic waves in extremely thin membranes of gallium arsenide," said Sandia researcher Steve Casalnuovo.
"This piezoelectric effect in extremely small sensors is nevertheless 100 times more sensitive to chemicals than conventional sensors." Casalnuovo has led a team of engineers and scientists in the effort to develop the integrated sensor for the lab-on-a-chip for the past three years.
The project models its goal on the mythical "tricorder" used on the TV series "Star Trek," and accordingly the device was designed to display the chemical composition of a substances at a distance by simply pointing at them and "sniffing." The molecular samples enter a chip that measures 1 mm2, into which a 1-meter-long column micromachined from a 50-micron-wide trench is formed into a spiral. It takes about 1 minute for the sample to get through the one-chip spiraled gas chromatograph. The sample emerges separated into purified puffs, which are then blown across the acoustic-wave sensor chip.
"The two chips work together by providing our pattern-recognition algorithms with two ways of identifying substances and preventing false alarms," said Casalnuovo. "We can use neural networks to reliably learn the distinctive patterns in sensor activation that different substances cause."
The term lab-on-a-chip is somewhat of a misnomer, since the current incarnation of the device, including all its support circuitry, is about the size of a brick. The acoustic-wave sensor, however, packs not only an array of 2,000 x 500 x 500-micron sensors but also all their high-frequency support components onto a single, 6-mm2 GaAs die. That includes a 1-GHz signal source for the acoustic wave; a comparator, to detect propagation delays; and all the circuitry to convert the results into a dc signal that indicates the amount of substance detected.
"I wouldn't have believed it possible before this project, but we have fit all the sensor's support circuitry onto the single GaAs chip. All that goes onto the chip is a 3-V dc power supply, and coming out all we have is a dc signal between 1 V and 3 V," said Casalnuovo. "All of those troublesome high-frequency signals are confined to the chip, eliminating all the noise and interference problems that other acoustic-wave sensors have to deal with."
The actual mechanism used by an acoustic-wave sensor is to measure the propagation delay of a high-frequency signal across its surface. When there are no foreign substances present, the delay is at a minimum.
However, if the substance to which the sensor is sensitive is present, some of it will "stick" to the surface of the sensor, thereby changing its mass. The increase in mass, in turn, slows down the propagation of the acoustic wave in direct proportion to the amount of substance stuck to the surface.
Right now the acoustic-wave chip can hold up to six separate sensors, which share the support circuitry. But Casalnuovo already has denser versions of the chip on the drawing board. For different applications, distinct sensors will be included for sniffing application-specific substances.
Soldiers, for instance, will want models sensitive to chemical-warfare agents, whereas utility companies will want versions that can detect radon and natural-gas leaks.
There is even an effort to array a set of "artificial tongue" sensors that could distinguish among sweet, sour, salt and bitter flavors for commercial purposes. It may be a long time, however, before the universal detection capabilities of the "Star Trek" tricorder are realized.
"There are many different ways to deposit a chemically selective layer onto the surface of these acoustic-wave sensors. This chemically selective layer is traditionally a polymer of some kind deposited on quartz," said Casalnuovo.
Unfortunately, quartz cannot be integrated onto a chip. Casalnuovo therefore had to look to other substances that exhibit the piezoelectric effect.
Silicon is not piezoelectric, but GaAs can be made to vibrate like quartz. GaAs is also the best solution for high-frequency components, making a single-chip approach not only possible but highly desirable.
"I see our major accomplishment as fabricating the acoustic sensors alongside the high-frequency microelectronics without degrading the performance of either one," he said. "We had to engineer around many hidden incompatibilities by adapting advanced silicon fabrication techniques to GaAs."
Putting a 'lid' on it
Next for Casalnuovo's team is to reduce the size of test samples by harnessing nanotechnology-fabrication techniques to put a "lid" on the acoustic-wave sensor chip that permits minute quantities of the gases to flow over it.
The team plans to grow the lid directly on top of the GaAs substrate with selective deposition, etching and micromachining techniques, resulting in a sensor that would be able to detect substances in extremely small amounts.
"We want to be able to set off alarms about the presence of substances before they become a hazard and to do that you need to be able to detect them in extremely dilute samples," Casalnuovo said. "For instance, you need to be able to tell a soldier to put on his breathing filter before the substance has caused any damage to his lungs."