Gauging Semiconductor Temperature

As any good cook knows, a recipe's cooking temperature is critical. For physicists constructing novel semiconductors, knowing the temperature has been a problem. But thanks to a new light-based technique developed at the University of Arkansas, researchers can know precisely at what temperature they're cooking, which could lead to improved and less expensive communication and other devices.

A schematic shows the University of Arkansas' optical temperature device used in a molecular beam epitaxy machine. The technique shines a time-chopped white light on the back of a substrate, then measures the wavelength where the light coming through the substrate drops in intensity.

Compound semiconductors offer electrical and optical properties quite different from standard silicon. Molecular beam epitaxy is often used in research to lay down compound semiconductors on a substrate, but such manufacturing can be tricky.

"Temperature is by far the most critical parameter, because everything is exponentially dependent on that value," said Paul Thibado, assistant professor of physics.

Thibado and physics professor Greg Salamo developed the measurement technique in conjunction with CIU Systems Ltd. of Migdal Ha'emeq, Israel, and Riber Inc. of Rueil-Malmaison, France.

Dangling a thermocouple near the surface is one way to measure substrate temperature during molecular beam epitaxy. This approach, however, can result in readings that are as much as 80 °C in error, according to Thibado. Another measurement method involves the color of a glowing substrate. Unfortunately, that requires substrate temperatures of more than 450 °C. It's also thrown off by any other hot sources in the research chamber, which are common in molecular beam epitaxy.

The technique shines a time-chopped white light on the back of a substrate, then measures the wavelength where the light coming through the substrate drops in intensity. This very accurately determines the energy gap between the valence and conduction bands of the substrate. The energy gap depends upon the substrate material and its temperature.

The light-dependent thermometer's accuracy is ±2 °C from 0 to 700 °C, Thibado said, noting that measurements could be updated every second.