False metals, real insulators, and degenerate gapped metals

Abstract

As science and technology of conductors have transitioned from studying simple elemental metals such as Al or Cu to compound conductors such as binary or ternary oxides and pnictides, a special class of degenerate but gapped metals has been noticed. Their presumed electronic configurations show the Fermi level inside the conduction band or valence band, yet there is an "internal band gap" between the principal band edges. The significance of this electronic configuration is that it might be unstable towards the formation of states inside the internal band gap when the formation of such states costs less energy than the energy gained by transferring carriers from the conduction band to these lower energy acceptor states, changing the original (false) metal to an insulator. The analogous process also exists for degenerate but gapped metals with Fermi energy inside the valence band, where the energy gain is defined by transfer of electrons from the donor level to the unoccupied part of the valence band. We focus here on the fact that numerous electronic structure methodologies have overlooked some physical factors that could stabilize the insulating alternative, predicting instead false metals that do not really exist (note, this is in general not a physical phase transition but a correction of a previous error in theory that led to a false prediction of a metal). Such errors include: (i) ignoring spin symmetry breaking, such as disallowing magnetic spin ordering in CuBi2O4, or disallowing the formation of polymorphous spin networks in paramagnetic LaTiO3 and YTiO3; (ii) ignoring structural symmetry breaking, e.g., not enabling energy-lowering bond disproportionation (Li-doped TiO2, SrBiO3, or rare-earth nickelates), or not exploring pseudo-Jahn-Teller-like distortions in LaMnO3 or disallowing spontaneous formation of ordered vacancy compounds in Ba4As3; (iii) ignoring spin-orbit coupling (SOC) forcing false metallic states in CaIrO3 and Sr2IrO4. Some of these modalities lead to intermediate bands inside the gap that might trap carriers. The distinction between false metals vs real insulators is important because, (a) predicting theoretically that a given compound is metal even though it is found to be an insulator often creates the temptation to invoke a high order novel physical effects (such as correlation in d electron Mott insulators) to explain what was in effect caused by a more mundane artifact in a lower-level mean-field band theory, (b) recent prediction of exotic physical effects such as topological semimetals were unfortunately based on the above compounds that were misconstrued by theory to be metal, but are now recognized to be stable insulators not hosting exotic effects, and (c) practical technological application based on stable degenerate but gapped metals such as transparent conductors or electrides for catalysis must rely on the systematically correct and reliable theoretical classification of metals vs insulators.

Semiclassics: The hidden theory behind the success of DFT

“…At the bond length at which that occurs, the semilocal approximations that most functionals use must fail, and in fact they do. The KS equations then yield a broken-symmetry solution which has lower energy [127,150,151].…”

Bipolar conduction asymmetries lead to ultra-high thermoelectric power factor

“…The effective masses of interest can be extracted from a dedicated code that we have developed, 24 and Eg can be extracted by DFT with a careful choice of pseudopotentials. [25][26][27][28] Terms 1c and 1g require deformation potentials , which can be extracted by ab initio calculations. 29,30…”

High-throughput inverse design and Bayesian optimization of functionalities: spin splitting in two-dimensional compounds

“…We propose a workflow based on the inverse design approach that automatically filters, selects, and designs compounds with specific functionalities 32 , i.e., a controllable material property with potential device applications. Remarkably, in addition to the simple material prediction, commonly performed for diverse functionalities in quantum materials 32 – 34 , we integrate the proposed workflow to the automatic optimization of the target functionality based on Bayesian statistics.…”