Anomalous high-temperature ionic conductivity in the silver halides

Abstract

The ionic conductivity of pure AgCl and AgBr crystals has been measured carefully for a range of temperatures of 300 \ifmmode^\circ\else\textdegree\fi{}C below the melting point. The usual cation-Frenkel-defect model is used, including contributions from vacancy and from both collinear and noncollinear interstitialcy jumps. Coulomb interactions are treated by the first-order Debye-H\"uckel-Lidiard (DHL) corrections, and further constraints are also introduced to keep the calculated results consistent with data from doped conductivity and tracer-diffusion measurements. Parameters for formation and mobility of defects are obtained from a least-squares fit of the conductivity in an intermediate temperature range up to 150 \ifmmode^\circ\else\textdegree\fi{}C below the melting point. When these parameters are used to extrapolate to higher temperatures, the observed conductivity shows a large excess over the extrapolated values, amounting to nearly 100% at the melting point. The size of the anomaly is expressed in terms of a correction -$\ensuremath{\Delta}g$ to the free energy of formation needed to bring observed and calculated values of the conductivity into agreement. In both halides it is found that $\ensuremath{\Delta}g$ increases rapidly and exceeds the values provided by the DHL corrections (as well as those from the extended-interaction theory of Sevenich and Kliewer for AgCl) by more than a factor of 2 at the highest temperatures. It is concluded that long-range Coulomb interactions are not sufficient to account entirely for the observed anomaly, and it is surmised that some other physical process, such as a general softening of the lattice, must also be present.