Chris Regan’s group has been investigating nanoscale thermodynamics in hot wires. This year they demonstrated a thermometry technique with record-setting spatial resolution.
The physics underlying their technique is the same as that employed by Fahrenheit's glass-bulb thermometer, which uses the calibrated thermal expansion of mercury to indicate temperature. Such a device is too clumsy to map temperature on sub-micron length scales, but thermal expansion is a well-characterized, near-universal property of solids. By measuring thermal expansion, one can turn almost any solid into a thermometer. By measuring thermal expansion precisely and locally, one can turn a solid into many thermometers, so that temperature variations across the solid can be detected. With this approach, scanning the thermometer across the device-under-test becomes unnecessary. The device takes its own temperature, functioning as a collection of localized thermometers.
Regan’s group implemented this reasoning in an electron microscope, using the electron beam to measure the density of a heated, aluminum wire at fifty-thousand separate locations. The microscale aluminum wire thus became fifty-thousand nanoscale thermometers, each of which was queried individually. Capturing a temperature reading from each, they generated maps, thereby solving a well-known open problem: thermometry with nanoscale spatial resolution.
This technique, described in the February 6, 2015 issue of Science, promises to open new avenues of fundamental research. The ultimate, limiting resolution of their technique is only a few nanometers, which is smaller than the electron and phonon mean-free-paths in many materials. Since electrons and phonons are the main carriers of heat in a solid, this technique should be able to probe the crossover between normal, diffusive thermal transport and ballistic thermal transport. Moreover, in very tiny system it may be able to see the breakdown of the usual, large-number, statistical assumptions of thermodynamics that allow the classical definition of temperature.
The technique has practical applications as well. Modern microprocessors have transistors with characteristic feature sizes of only 14 nm, and their performance is deliberately limited to prevent overheating. Currently thermal transport within individual transistors can only be studied numerically. Circuit designers must assume, despite evidence to the contrary, that heat moves at the nanoscale much as it does at the macroscale. In other words, the $300 billion semiconductor industry is designing its flagship devices using physical assumptions that have not been tested at the relevant length scales. With an improved understanding of nanoscale thermal transport, new methods for heat management may unlock better performing and more efficient microelectronic devices. Regan’s group is currently verifying their technique in silicon, and hopes to apply it soon to functioning transistors.