Effect of Sn oxides on the thermal conductivity of polycrystalline SnSe

Abstract

SnSe is a promising thermoelectric material, with intrinsically low lattice thermal conductivity, κL. Surprisingly, in several reports, polycrystalline samples are found to have a higher thermal conductivity than single crystals. This disparity has been attributed to trace amounts of thermally conductive Sn oxides at the grain boundaries of polycrystalline samples. The same culprit was recently proposed to explain the reduction of κL in purified, oxide-free, SnSe polycrystals. Here, we test this hypothesis by: (i) tuning the type of oxide in SnSe by exploiting thermodynamic stability regions, since Sn-rich or Sn-poor compositions favour the formation of SnO or SnO2, respectively; and (ii) varying the quantity of SnO2 by intentionally oxidizing SnSe powder before consolidation, to obtain samples with quantifiable amounts - up to 15% - of SnO2. We find that the κL of SnSe is impervious to changes in the type or the amount of Sn oxide present in the samples. Our results show that a simple “rule of mixtures” cannot be used to estimate the effect of grain boundary oxides on the thermal conductivity of SnSe. These results call for an improved understanding of the intriguing thermal transport mechanisms in SnSe and numerous other systems where a two-phase transport is presumed.