EARLY TRANSISTOR
AND DIODE HISTORY AT BELL LABS Art Uhlir Jr. |
||
Oral History – Art Uhlir Jr. (Continued) |
|
|
The X-Y recorder shown in photographs arrived late in the
examination of point-contact transistors but was convenient for recording
characteristics of electrochemical cells. We were asked to test the life of point-contact transistors as a
function of process parameters and stress. The Model Shop made point-contact
transistors as requested. Not more than 100 samples were used, including a
few P-Type. This test was the first of its kind. My wife-to-be, Ingeborg Williams, started work as a technical
assistant months before receiving her 1953 chemistry degree from Douglass
College of Rutgers University. I had interviewed her a month earlier, but
didn't then know she would my assistant. The Model Shop prepared prototype
point-contact transistors with varied process parameters, such as resistivity
and various strengths of “forming” electrical pulses applied to the
collector. I asked Inge to distribute them in randomized fashion but balanced
between the two "identical" aging racks. The racks used vacuum-tube
regulated DC power supplies to subject the transistors to steady DC bias. After three months and a couple of
moves, we found that some of the transistors were little changed and others
suffered degradation of collector resistance or current gain. But most of the
degradation occurred on just one of the test racks. (Naturally we blamed that
on the tubes.) At least we didn't learn any false information about process
variables and power levels. |
Please describe your work with these first germanium junction transistors. We started experimenting with germanium grown-junction
transistors even before the point-contact life tests were turned on. Not on
circuit development, but on a specific reliability issue. I was given the assignment to cure the "floating
potential" problem of NPN germanium grown-junction transistors. The open-circuit emitter-base voltage
sometimes agreed reasonably well with the Shockley relation to the current
gain. But all too often it drifted up in the direction of the positive bias
on the collector in a matter of seconds, minutes or hours. It seemed
plausible that an N-type channel developed over the exposed edges of the base
region. It didn't take much investigation on bare NPN bars mounted on
headers to find that humidity was a cause. Plastic encapsulation only slowed
the effect. In general, the channels were favored by reducing agents and
opposed by oxidizing agents. This drift was intolerable in some circuit
applications, embarrassing in any event. Along with the assignment, Jack Morton gave me the imprecation.
“Do not come up with putting them in a vacuum envelope.” Go
To Uhlir Oral History, Page 5 |
|
COPYRIGHT
© 2004 by Jack Ward. All Rights
Reserved. http://www.transistormuseum.com/ PAGE
4 |