Abstract
Reduction of thermal conductivity is one of the most effective strategies for increasing the figure of merit (ZT) of thin-film thermoelectric (TE) devices. Recently, thin-film structures using Bi-Sb-Te TE materials attracted significant attention because of their low thermal conductivity and anisotropic thermal properties. The four-point-probe three omega (3-omega) method is the most widely used technique to measure the thermal conductivity of various dimensional materials, including 1D nanostructures, 2D-thin films, and 3D-bulk materials, because it provides a simple measurement setup and high accuracy (less than similar to 10%) for the measurement. In addition, it was confirmed that both cross-plane and in-plane thermal conductivities of thin films can be measured by altering the width of the heater on the films and by using a proper substrate underneath the films in the 3-omega method. Here, we first report an uncertainty analysis of both the cross-plane and in-plane thermal conductivities of 500-nm-thick p-Bi0.5Sb1.5Te3 (p-BST) thin films, which were prepared on a SiO2/Si substrate and measured by the four-point-probe 3-omega method at room temperature, by varying the width of the heater on the substrates. From the uncertainty calculations, reasonable in-and cross-plane thermal conductivities of p-BST thin films in the 3-omega method were estimated with lower measurement error. The results suggest that the heater widths are strongly related to both in- and cross-plane thermal conductivities in the measurement. The corrected cross-plane and in-plane thermal conductivities of the 500-nm-thick thin films were similar to 0.43 +/- 0.02 W m(-1) K-1 and similar to 0.61 +/- 0.05 W m(-1) K-1, respectively, at room temperature.