Identifying strong correlation in metals by a DFT calculation

Strongly correlated systems—materials configurations in which electron–electron interactions play a decisive role—have long been a central topic in condensed matter physics. In metals, these strong interactions can arise from a high density of single-particle states near (just above and just below) the Fermi level. These configurations are described using correction schemes like DFT+U or DFT+DMFT. However, symmetry breaking (SB) itself may serve as a mechanism that qualitatively captures the energy of strong correlation within standard DFT. One hypothesis is that SB transforms a strongly correlated metal, represented by a symmetric configuration with many states near the Fermi level, into a normally correlated configuration with fewer states near the Fermi level — a regime that can then be well described by DFT. This idea naturally connects the nonmagnetic (NM) configuration, which is symmetric and strongly correlated, to the magnetic configurations, which are symmetry-broken and normally correlated. In this work, we explore this hypothesis by applying it to strongly correlated systems such as Ni, Fe, NiO, VO₂ (rutile and monoclinic), EuB₆ and SmB₆, and to normally correlated metals, such as Cu and Ag. Upon SB of a strongly correlated configuration, we observe a pronounced reduction in the density of states near the Fermi level and a significant lowering of the total energy. To identify strong correlation by a DFT calculation, we introduce a correlation parameter (α) based on comparison between the density of states near the Fermi level in the NM configuration and that of the uniform electron gas, which serves as a benchmark for a normally correlated system. We find that our method clearly distinguishes strongly correlated from normally correlated systems, such as Cu and Ag, in excellent qualitative agreement with physical expectations.