Testing whether flat bands in the calculated electronic density of states are good predictors of superconducting materials

If one could systematically search the calculated electronic structures of materials, already available on online materials databases, for electronic characteristics that are associated with superconductivity, then the rate of new superconductor discovery could be accelerated. Here we test whether one such proposed characteristic, flat bands in the calculated electronic density of states, can broadly distinguish known superconductors from known non-superconductors. Our data mining exercise, performed on the materials for which the relevant data can currently be downloaded, does not support flat bands as strong predictors of superconductivity in materials. Beyond the specific test, our study's limitations underscore the importance of standardizing digital resources for achieving the accelerated materials discovery paradigm promised by the Materials Genome Initiative.