|
|
Oral History – Bill Brower (Continued)
|
|
The
crucibles were inside a sealed quartz tube filled with inert gas – nitrogen
or helium. The top of this enclosure contained a rotating shaft that could
be raised and lowered and had a chuck on the end to hold a small diameter
single crystal silicon seed. Once the RF power was applied and the silicon
material was molten, the seed was lowered until it touched the surface of
the melt. Raising the melt temperature would cause the end of the seed to
melt, while lowering the melt temperature would cause a single crystal
growth to commence at the seed interface. The RF power could thus be
programmed to change the diameter of the resultant growing crystal from the
small seed diameters of around 1/8 inch to the full size crystal of 1 to
1½ inch diameter.
Starting with pure silicon, the
initial melt was doped N-type using arsenic or antimony. This produced the
N-type collector region. After about ½ inch of vertical growth, a doping
pellet of phosphorous or gallium was introduced to form the P-type base
layer, followed quickly by a much larger amount of N-dope to make the low
resistivity emitter area. The emitter region was further grown another
3/8 inch or so to complete the transistor crystal, then raised from the
melt.
|
|
|
Oral History – Bill Brower (Continued)
|
|
By controlling the doping
levels and thus the resistivities of the three regions, plus the growth
width of the P base region, the desired transistor parameters were
obtained.
The next step was to cut off
the upper and lower N regions to yield a round NPN wafer about ¼ inch
thick, with the P base in the center of the sandwich. Using a multibladed
diamond saw, the wafer was then sliced into narrow rectangular NPN bars.
These bars were chemically etched to remove surface damage and mounted in a
header by the collector and emitter ends. To contact the base, a thin
P-doped gold wire was pulse bonded to the edge of the P base layer, then
welded to the base header terminal. The diameter of the bar and thus the
area of the base layer had to be small to yield a high frequency
transistor, but small bars severely limited the power handling
capabilities. To increase the power properties, the transistor cans were
filled with a heat transferring media we called “wimp”. It consisted of a
silicone liquid loaded with a thermal conducting ceramic powder to make a
thick liquid filler. Power transistors had shorter and thicker bars for
lower internal resistance, and the collector end was mounted directly to
the header for better heat conduction.
Brower
Oral History, Page 3
|
|