“The sound we were listening to was created by arbitrary thermal fluctuations generated elsewhere in the sample, such as an electron hitting a lattice imperfection or an air molecule striking the surface,” Weaver said. “While no one had really doubted that these tiny fluctuations existed, no one had ever measured them before.”
Weaver and Lobkis not only proved that the vibrations were indeed measurable, they also showed that by correlating what appeared to be random noise, considerable information could be gleaned about an object’s interior. First, they listened to the noise, then they used mathematical operations that looked for patterns and repetitions – a process called autocorrelation.
“Like BBs rattling inside a box, phonons will bounce off the walls of the aluminum, ricochet off some internal structure, and bounce off the walls again, corresponding to the round-trip travel time of an echo,” Weaver said. “We looked for correlations within the echoes.”
Weaver and Lobkis validated their technique by autocorrelating the noise from a passive piezoelectric transducer mounted to the sample and then comparing that result with an active measurement they obtained using conventional ultrasonics.
“The waveforms were almost identical,” Weaver said. “When you autocorrelate the ambient noise, you see nearly the same signal as when you pulse the transducer and listen to the echoes.” This surprising result is something scientists have been overlooking for decades, Weaver said. “We’ve been throwing away this noise – not realizing that it’s full of useful information.” In principle, the passive technique could work on nearly any object, but would be most helpful in applications where conventional sound sources are scarce. At very low frequencies, for example, seismologists could pick up the random vibrations from distant earthquakes to obtain local stratigraphic information without setting off directed explosives. At extremely high frequencies, the technique could be used to noninvasively probe micron-sized features and material properties in microchips.
“The technique also might be useful for monitoring building vibrations to anticipate potential collapse,” Weaver said. “By measuring the natural frequencies of the building as it responds to random vibrations in the neighborhood, even subtle changes in structural rigidity could be detected.”
The National Science Foundation funded the research.
Physical Review Letters