Influence of annealing on the spatial distribution of thermoelectric properties in bismuth-telluride specimens

Abstract

A p-type (Bi_0.25Sb_0.75)₂Te₃ specimen doped with 8 wt. % excess Te alone and an n-type Bi₂(Te_0.94Se_0.06)₃ specimen codoped with 0.017 wt. % Te and 0.068 wt. % I were grown by the Bridgeman method and were cut into a parallelepiped of 5×5×15 mm³, where the length of 15 mm is parallel to the freezing direction. Their local Seebeck coefficient α_ℓ and local electrical resistivity ϱ_ℓ were measured before and after annealing at a scan step of 1 mm along the freezing direction of the specimen, where annealing was done at 673 K for 2 h in vacuum. The specimen was mounted on an X–Y stage and the temperature difference between two probes set at an interval of 1 mm apart was approximately 2.6 K and measured with an accuracy of 0.1 K. The ϱ_ℓ of the as-grown and annealed specimens changed more significantly from place to place than their α_ℓ, so that the effect of annealing on the local power factor P_ℓ (=α_ℓ ²/ϱ_ℓ) was not uniform throughout the specimen surface. Such a spatial variation in P_ℓ is considered to result mainly from the spatial variation in ϱ_ℓ which originates from only a slight deviation of the melt composition from the stoichiometry inside a specimen, and also from a difference in the degree of alignment of the c planes of grains inside a specimen. The maximum P_ℓ (=α_ℓ ²/ϱ_ℓ) of the annealed p- and n-type specimens reached surprisingly high values of 7.78 and 8.93 mW/K² m, respectively, which correspond to being about twice as large as the macroscopic P values obtained by conventional methods.