EARLY TRANSISTOR HISTORY AT

TEXAS INSTRUMENTS 

Mary Anne Potter

 

Oral History – Mary Anne Potter

(Continued)

 

The temperature at which this was accomplished was 1200 ° C.  The globar furnaces were capable of only about 6 inches of usable space for accomplishing the oxidation, deposition, and diffusion processes. Even 6 inches took much technician time to adjust globar settings and re-profile. We profiled every furnace once a day and frequently had to spend time tweaking. When the usable area of the furnace decreased and was no longer “tweakable”, we would often just reduce the number of slices in a load to use the space that was ok because, when the globars had to be changed, we would lose at least a day and possibly days in getting the furnace back up.   The highest temperature we used for the quad-diffused structures was 1250°C, and the first collector diffusion (the deep N-type one) took about 15 hours of diffusion time. Much of my first weeks at TI were spent developing and graphing diffusion coefficients for various times and temperatures for each of the diffusion sources we used so that we could make better educated decisions as to how to build these complicated ICs.

 

We used the diffusion depth determined by the lap and stain operation along with the sheet resistance measured on pilots with each lot to determine the time for the next diffusion step. Once we got to the emitter processes, we would run one slice from the lot through the calculated times and strip part of it to actually measure the gain of the resulting transistors.

 

    

 

Oral History – Mary Anne Potter

(Continued)

 

Based on this measurement, the remainder of the lot would be committed to the emitter process with an even more fine-tuned (tweaked) diffusion time. Delmo, one of the technicians who was outstanding at doing these measurements and tweaking calculations, was still doing this in the low-power Schottky production area in the 1980s. He had it down to a science then and actually learned to do statistical process control charts so that he could tell when the emitter process was going out of control.

 

The “cleans” required for the slice processing were all interesting. Most of them used a 10% hydrofluoric (HF) acid mixture to remove any residual oxides from the surfaces to be diffused. A typical cleanup required immersing the slices in 180° sulfuric to remove all organics, a dip in cold sulfuric so that the immersion into flowing water to remove the sulfuric would not splatter acid and water all over everything, a dry, then a dip into the HF solution, a flowing water rinse, then dips into various solvents for quick-drying. We eliminated acetone from those solvents after some fires. Most of the slice-cleaning carriers in the diffusion area were made of quartz. So were the diffusion carriers, called boats, which went into the diffusion furnaces.

 

 

  Potter Oral History, Page 6

COPYRIGHT © 2002 by Jack Ward.  All Rights Reserved.  http://www.transistormuseum.com/

Page 5