Page 263 - Chalcogenide Glasses for Infrared Optics

238 Cha pte r Ni ne

AMI accepted an order from McDonald Douglas in Kansas City,
on a best-effort basis, to provide four GaAs plates 4 in × 10 in × 0.5 in
thick. The BN boat was used for all four plates. The two undoped
plates were prepared without any difficulty, ground, and polished,
and infrared transmission measured 2.5 to 25 µm at six points down
the plates with no appreciable variation point to point. The other two
plates were to be doped to a resistivity of 1 to 10 Ω·cm to be tested
relative to EMI protection. The doping was accomplished by adding
to the melt silicon heavily doped Hall samples that had been evalu-
ated. The measured resistivity of the samples determined the Si atoms
per gram in each sample. Thus the amount of Si required for a con-
centration corresponding to the desired plate resistivity was found
using Fig. 9.17. The doped samples were weighed and added to the
melt at the start. The concentration corresponding to 1 to 10 Ω·cm was
14
13
low, 10 to 10 per cubic centimeter. The resistivity of both plates
after grinding and polishing was measured using the ASTM conduc-
tivity method applied by AMI in evaluating germanium bars for the
GSA purchases, during the government stockpiling effort and the
evaluation of float zoned silicon crystals. Changes in resistivity along
the bars were evident on both plates. Silicon was not evenly distributed.
Unfortunately, the plates were polycrystalline with grain boundaries
running in the direction of growth. Contacts tended to be nonohmic.
The readings were different with current in the direction of growth
compared to current across the plates. Values taken for the entire
plate were unreliable. Silicon was a bad choice as a dopant because of
its distribution coefficient, Se or Te would have been better. A discrep-
ancy was found between the measured resistivity and the absorption
in the infrared and RF measurements made by McDonald Douglas.
To better understand the discrepancy, two other doped plates were
made. Strips 2 in wide were cut from the plates, and indium contacts
were hot soldered to all four polished edges using an ultrasonic
soldering iron. Conductivity was measured at various currents in
both directions with and across the grains until they all agreed. The ir
transmission at 10.6 µm was measured in several locations in each
plate, and the absorption coefficients were calculated and compared
to experimental values for different carrier concentrations found in
the literature. Figure 9.17 was used to convert the carrier concentra-
21
tions to resistivity. The results are shown in Fig. 9.20 in a log-log plot
that covers 5 orders of magnitude in absorption and resistivity. In the
figure, the crosses are the literature values while the circle points are
the AMI experimental values from the plates. Figure 9.20 may be used
to measure the dopant uniformity in GaAs plates provided the resis-
tivity is 10 Ω or less so that the 10.6-µm absorption is appreciable.
The EMI measurements made on delivered plates to McDonald
Douglas indicated the AMI resistivity values based on conductivity
were 4 times too high. The AMI results indicated a resistivity rise front
to back of 1 to 2 Ω·cm. The new AMI ir measurement method indicated