Validation of thermoelectrics measurements. II

This paper reports the joint effort of Marlow Industries (MI) and members of the Naval Research Laboratory (NRL) to test the usefulness of a carefully constructed conventional apparatus for measurement of the thermal conductivity of thermoelectrics. The design of the NRL apparatus takes into account the listed error sources, but is especially tailored to eliminate radiative exchange. This choice is driven by the fact that the instrument is most often used to measure the thermal conductivity of metals at high temperatures. We have found that as a consequence of using insulation to prevent radiation, the apparatus is not well suited for measurement of low thermal conductivity samples, such as standard thermoelectrics. The samples used as standard materials were taken from ingots grown at MI by a commercial method. The p-type ingot was grown from a melt of Sb/sub 150/Bi/sub 50/Te/sub 291/Se/sub 9/ plus a few percent of excess Te. The n-type ingot was grown from a doped melt of Bi/sub 20/Te/sub 27/Se/sub 3/. With slight variations, these are the compositions used throughout the thermoelectrics industry.