ALBUQUERQUE, N.M. Seven years of cooperative development recently resulted in a new suite of electronics reliability tests that squeezes a 10-year test period into just a few weeks.
The new tests can validate even inexpensive commercial parts at as high a reliability level as expensive radiation-hardened devices. The project was a cooperative effort between Sandia National Laboratories here, the Department of Energy (DOE) national security research lab, and a consortium headed by the Defense Special Weapons Agency (DSWA).
The bipolar electronics reliability tests were released by Sandia National Labs' Dan Fleetwood, a physicist in the Radiation Technology and Assurance Department. "The challenge was to come up with tests that would predict a 10-year response in just a few weeks of testing. Of course, failing the test often takes much less time," said Fleetwood. Fellow researcher Peter Winokur also participated in developing and releasing the test-suite specs.
For instance, one satellite maker recently designed a bipolar voltage regulator that the new tests quickly identified as prone to failure in less than 10 years. Parts that pass the test validate not only themselves, but any other parts in their lot as well as subsequent lots in which the process has not changed. Bipolar tests were needed because satellites use bipolar chips for analog functions. The test suite was recently adopted as the official long-term radiation-exposure test for the American Society for Testing of Materials (ASTM), an electronics industry standards organization.
Winokur pointed out that for space vehicles, thorough testing needs to be performed long before any spacecraft is ever launched. "When a satellite fails in space, it's hard to send a repair crew to see what broke. You need to put reliable parts in from the beginning to prevent future problems," he said.
In 1991, researcher Ed Enlow of Mission Research Corp. (Albuquerque, N.M.) invalidated the previous standard radiation test suite, which gave a quick, high dose of radiation on the assumption that the longer times delivered in space would only improve things because of self-healing effects.
Fleetwood said, "Defects from high-energy protons and electrons mainly trap charge, temporarily changing, say, the gain of an amp somewhat, and when the charge finally escapes, the gain goes back to normal."
Enlow, however, discovered that bipolar electronics prematurely failed over the long term at considerably lower radiation levels than the old tests indicated. "By 1993 we had responded in cooperation with the Defense Department to find a new method of testing the long-term response of chips to radiation in space," Fleetwood said.
The main problem ignored by the old test was that the base of the bipolar transistor typically is covered by a layer of insulation that can not rebound from the lower dose rates encountered in space. Instead, the active base of the transistor slowly degrades. The effect was described in papers presented at the annual Institute of Electrical and Electronic Engineers (IEEE) Nuclear and Space Radiation Effects conference by Fleetwood, Winokur, and colleagues from Vanderbilt, The Aerospace Corporation and other members of the DSWA consortium.
Two long-term failures were identified. One type preys on defects that can trap charge inside insulators. For instance, if an impurity on a chip has only one oxygen atom instead of two, then the oxygen vacancy can capture a permanent positive charge to compensate.
A second defect type releases a trapped hydrogen atom, which interacts at the surface of the emitter-base junction, causing a permanent interface trap that can also reduce gain.
"Our first breakthrough was when we discovered that elevating the temperature at which the test is performed can simulate the effects of long-term radiation. After that it was just a lot of hard engineering work to validate an effective procedure," said Fleetwood.
The final procedure divides the test chips into three groups. Group one is the control and gets the same quick, high dose of radiation as the old test. Group two receives a lower dose of radiation over a few weeks' time. And group three receives a high dose of radiation at a high temperature.
A comparison of the response of groups one and two reveals any emerging trend toward failure. When those results are compared with the third group, any non-linear degradation over the long term is revealed.
"Group three with the elevated temperature reveals defects that suddenly occur in step-function fashion," said Fleetwood.
Though designed with bipolar in mind, Fleetwood claimed that the new tests can validate both bipolar and CMOS parts even when they are not radiation-hardened. It takes four to six weeks to validate a part with the new test suite.
Sandia's work was funded primarily by the DSWA, Mission Research Corp., Naval Surface Warfare Center, Jet Propulsion Laboratory, The Aerospace Corp. and Vanderbilt University. Sandia is a multiprogram DOE laboratory, operated by a subsidiary of Lockheed Martin Corp.