This is an archived article.

September 19, 1996

UW engineer helping Washington company develop instrument enabling semiconductor manufacturers to make faster, cheaper computer chips

An innovative temperature-measuring instrument developed with the assistance of a University of Washington engineering professor has yielded improvements in processing semiconductors that may lead to faster, cheaper computer chips.
The DRS (diffuse reflectance spectroscopy) 1000™ Temperature Measurement System is being produced and marketed by Thermionics Northwest, based in Port Townsend, Wash. Prototypes already are being tested by major semiconductor manufacturers and national research laboratories. The National Science Foundation this month announced it will award Thermionics a $300,000 grant to further refine the DRS 1000™ and develop design specifications for large-scale production of the instrument.

“If this instrument is adopted as standard processing equipment for manufacturing semiconductors, it could represent a new multi-million dollar market,” said Thomas Pearsall, professor of electrical and materials science engineering and Boeing-Johnson chair of semiconductor electronics at the UW. Pearsall is working on the DRS 1000™ project through a research partnership funded by Thermionics, the National Science Foundation and the Washington Technology Center, a state-funded enterprise linking the technology needs of Washington companies with the expertise and resources of Washington research universities. “This project is a great example of how the university and government can work together to help a small company in this state expand its business.”

Functioning much like the human eye, the DRS 1000™ picks up diffuse reflections of light to detect the color of an object. From the color reading, the instrument can gauge the temperature of the object to within a quarter of a degree Celsius in less than a second. Pearsall compares it to being able to instantaneously measure the temperature of an oven element by simply looking at the color of the element as it heats up.

This precise temperature measurement capability is a boon to manufacturers of computer chips, which now run everything from computers to car engines and microwaves. To make computer chips, manufacturers heat up wafers of silicon or other semiconductor materials and, at various temperatures, either apply coatings or etch grooves which form the parts of the chip. To ensure the correct level of deposition or etching, the processing steps must occur at precise temperature intervals. If they’re off by even a few degrees, the entire batch of wafers may need to be scrapped at a cost of hundreds of thousands of dollars.

Currently, manufacturers measure processing temperatures by placing thermocouples near the wafers; nothing can be attached to the wafers themselves. Unfortunately, the thermocouples disrupt temperature measurement by bleeding off heat, and they only measure in one spot so the temperatures of other parts of the wafer are unknown. In an effort to guard against temperature variations, manufacturers rotate the wafers during processing.

The DRS 1000™ measures temperature over the whole wafer and has revealed that rotating some wafers may actually introduce temperature variation rather than uniformity. It also has shown that the temperature of a batch of wafers changes during processing and from one batch of wafers to the next– though the heating element’s temperature remains constant.

“Nobody knew any of this until we developed our instrument; I think a lot of manufacturers will be surprised,” Pearsall said. “The DRS 1000™ not only measures the temperature more precisely, it can monitor and regulate temperature throughout a process run. If we can control temperature uniformity, we can improve chip performance by using more aggressive designs.”

The speed of a transistor, for example, is directly related to the thickness of the controlling gate, which is determined in part by the wafer processing temperature. If the processing temperature is uncertain, designers have to make the controlling gate thicker than may be necessary to compensate for possible temperature variation, and performance suffers.

The potential of the DRS 1000™ to improve semiconductor performance and production cost has captured the interest of some of the leading semiconductor manufacturers and research laboratories in the world who have already purchased the instrument.

“It’s a unique instrument that does what nothing else in the market can do,” said Thermionics President Jim Worthington, who predicts sales of the DRS 1000™ could climb from about 10 units a year to several hundred units a year. “It has proved it is an extremely valuable research and development tool. What remains to be seen is whether it will break into the much larger market in the production end of the industry.”

For more information, contact Pearsall at (206) 543-3063, pearsall@maxwell.ee.washington.edu or Worthington at (360) 385-7707, TNW@thermionics.com