Semiconductor-to-Metal Transition in the Blue Potassium Molybdenum Bronze, K 0.30 Mo O 3 ; Example of a Possible Excitonic Insulator

The electrical conductivity of the blue-phase potassium molybdenum bronze ${\mathrm{K}}_{0.30}$Mo${\mathrm{O}}_{3}$ has been extended down to 4 \ifmmode^\circ\else\textdegree\fi{}K. Below 10 \ifmmode^\circ\else\textdegree\fi{}K, $\ensuremath{\sigma}$ is weakly metallic. Above 10 \ifmmode^\circ\else\textdegree\fi{}K, there is an activation energy for conduction of 0. 025 eV. Above 180 \ifmmode^\circ\else\textdegree\fi{}K, $\ensuremath{\sigma}$ becomes metallic again with a discontinuity only in the slope. Several models are suggested to account for the semiconductor-to-metal transition in the absence of an antiferromagnetic N\'eel point or heat effect at 180 \ifmmode^\circ\else\textdegree\fi{}K. Included are the Mott-Hubbard model with a short-range electron-hole attraction proposed by Ramirez, Falicov and Kimball, and the excitonic-insulator model on the semimetallic side. The latter, although semiquantitatively consistent with the electrical conductivity, the Seebeck coefficient, and the weak temperature dependence of the magnetic susceptibility, cannot be distinguished from a correlation-free model involving only a simple lattice distortion on the semimetallic side. The exciton binding energy is calculated to be 0.021 eV in satisfactory agreement with the low-temperature-conductivity activation energy (0.025 eV).