Giampiero Ottaviani

Opportunities and challenges in the use of heavily doped polycrystalline silicon as a thermoelectric material. An experimental study

Large-volume deployment of Si-based Seebeck generators can be foreseen only if polycrystalline rather than single crystalline materials can be actually used. The aim of this study was therefore to verify whether polycrystalline Si films deposited on top of a SiO$_2$ insulating layer can develop interesting thermoelectric power factors. We prepared 450-nm thick heavily boron doped polysilicon layers, setting the initial boron content in the film to be in excess of the boron solubility in polycrystalline silicon at 1000 °C. Isochronal thermal annealings were then used to modify the B$_{Si}$ content by precipitation. Quite unexpectedly, a concurrent increase of the thermoelectric power and of the conductivity was observed for heat treatments at temperatures above 800 °C. Upon annealing at 1000 °C we found a power factor $P$ of 13 mW K$^{-2}$ m$^{-1}$, more than three times higher than previously reported $P$ for Si nanowires. These findings could be explained observing that degenerate polysilicon displays a remarkable enhancement of its Seebeck coefficient as an effect of the large amount of boron it can dissolve. Band gap narrowing and band tailing modify the density of states around the Fermi energy leading to a dramatic improvement of its log-derivative in the Mott equation. These results apparently point out an interesting direction for the development of Seebeck and Peltier devices sharing low cost and relatively high efficiency.