The metal-insulator transition in disordered solids: How theoretical prejudices influence its characterization. A critical review of analyses of experimental data

In a recent experiment, Siegrist et al. [Nature Materials 10, 202 (2011)] investigated the metal-insulator transition (MIT) of GeSb_2Te_4 on increasing annealing temperature. The authors conclude that this material exhibits a discontinuous MIT with a finite minimum metallic conductivity. The striking contrast to reports on other disordered substances motivates the present in-depth study of the influence of the MIT criterion used on the characterization of the MIT. First, we discuss in detail the inherent biases of the various available approaches to locating the MIT. Second, reanalyzing the GeSb_2Te_4 measurements, we show that this material resembles other disordered solids to a large extent: according to a widely-used approach, these data may also be interpreted in terms of a continuous MIT. Checking the justification of the respective fits, however, uncovers inconsistencies in the experimental data. Third, comparing with previous experimental studies of crystalline Si:As, Si:P, Si:B, Ge:Ga, CdSe:In, n-Cd_{0.95}Mn$_{0.05}Se, Cd_{0.95}Mn_{0.05}Te_{0.97}Se_{0.03}:In, disordered Gd, and nano-granular Pt-C, we show that such an inconclusive behavior occurs frequently: the analysis of the logarithmic temperature derivative of the conductivity highlights serious inconsistencies in the original interpretations in terms of a continuous MIT. In part, they are common to all these studies and seem to be generic, in part, they vary from experiment to experiment and may arise from measurement problems. Thus, the question for the character of the MIT of these materials has to be considered as yet open. The challenges now lie in improving the measurement precision and in developing a microscopic theory capable of explaining the seemingly generic features.