Incorrect sample classification in "Electron localization induced by intrinsic anion disorder in a transition metal oxynitride"

Abstract

In the recent study of the metal-insulator transition (MIT) in the disordered crystalline solid SrNbO$_{3-x}$N$_x$ by Daichi Oka et al. [Commun. Phys. 4, 269 (2021)], the data evaluation relies on the Al'tshuler-Aronov theory of the interference of electron-electron interaction and elastic impurity scattering of electrons. The present comment shows that this evaluation approach is inappropriate. For that aim, we reconsider data for the samples with $x = 0.96$ and $x = 1.02$ from three different perspectives: (i) analysis of the logarithmic temperature derivative of the conductivity, (ii) study of the deviations of the measured conductivity data from the Al'tshuler-Aronov approximation of the temperature dependence, and (iii) comparison of the measured temperature data with the values obtained treating the sample as secondary thermometer in terms of that approximation. This way, for the sample with $x = 0.96$, classified as metallic by Daichi Oka et al., qualitative contradictions between the measurements and the zero-temperature extrapolation according to the Al'tshuler-Aronov theory are uncovered. Thus, this sample very likely exhibits activated instead of metallic conduction. In consequence, our findings question the continuity of the MIT resulting from the highly cited scaling theory of localization.