The thermoelectric properties of 28 sintered Si0.8 Ge0.2 alloys, heavily doped with either boron or phosphorus and prepared from powders with median particle sizes ranging from about 1 μm to over 100 μm, have been determined from 300 to 1300 K. The thermal conductivity decreases with decreasing particle size, however, the figure of merit is not significantly increased due to a compensating reduction in the electrical conductivity. The thermoelectric figure of merit is in good agreement with results of Dismukes et al. [J. Appl. Phys. 10, 2899 (1964)] on similarly doped alloys prepared by zone-leveling techniques. The electrical and thermal conductivity are found to be sensitive to preparation procedure while the Seebeck coefficient and figure of merit are much less sensitive. The high-temperature electrical properties are consistent with charge carrier scattering by acoustic or optical phonons.

Four conductively coupled thermoelectric cells, developed under the SP 100 program, have been life tested. These cells, referred to as TOC (Task Order Contract) cells, were fourth generation cells, and incorporated design improvements to extend operating life. GS526 glass had been added to suppress the loss of Ge from the MoGe bond between the SiGe and the barrier graphite. The previous generation was life limited by the degradation of this electrical bond at the SiGe to graphite hot side interface due to the Ge loss. This led to abnormal internal resistance trends. The TOC cell test data and post test diagnostics have confirmed the effectiveness of the perimeter glass. c _ Two cells were tested at the Lockheed Martin Missiles and Space facility in Valley Forge and two at JPL in Pasadena. The last cell (No. 150) was taken off test on 23 June 1997 at Valley Forge after having reached 23,065 hours (2.6 years) of operation. Cell 156, also tested at Lockheed Martin, was shut down after 14,270 hours (1.6 years) of operation. The two cells at JPL (139 and 148) were tested for 15,100 (1.7 years) and 18,720 hours (2.1 years), respectively. Three of the four cells demonstrated normal electrical performance trends. The fourth cell (No. 139) tested at JPL showed an abnormal increase in internal resistance and a shift in temperature levels at 12,400 hours following a facility shutdown and restart. When the cell was removed from the test fixture, separation occurred between the hot side compliant pad facesheet and the niobium filament bundles. No degradation of the bond between the SiGe and the barrier graphite was found and the change in slope of the internal resistance was attributed to changes in the rate of dopant precipitation caused by the shift in temperature levels. Cell 156 tested at LMMS showed low power output, initially and throughout the life test. Post test diagnostics showed an approximately 50 % unbonded region on both the hot and cold side compliant pad facesheet to porous tungsten braze. No failures were detected in the thermoelectric section of the cell. The problems in both cells 139 and 156 were related to the compliant pad facesheet. These compliant pad problems were not entirely unexpected. It was known that all of the identified bonding process improvements had not been incorporated into the manufacture of the available compliant pads.