Choosing and Validating Density Functionals for Quantum Materials: Strong Correlation, Spin–Orbit Coupling, Topology, and Two-Dimensional Magnetism

Abstract

Density functional theory (DFT) has become the principal computational instrument for predicting and interpreting the electronic, magnetic and topological properties of quantum materials, yet the reliability of any such prediction rests entirely on the exchange–correlation (XC) approximation chosen for the calculation. This review examines the practical problem facing a computational materials scientist: which functional, or combination of functional and correction scheme, is appropriate for a given class of quantum material, and how should that choice be validated. Four intersecting themes are addressed. First, the treatment of strong electron correlation in partially filled d- and f-shell compounds is discussed through the lens of the Hubbard-corrected DFT+U method, hybrid functionals and dynamical mean-field theory, with attention to the double-counting problem and the determination of the Hubbard interaction parameter. Second, the implementation of spin–orbit coupling and its consequences for the identification of topologically non-trivial band structures are reviewed, including high-throughput screening strategies and the special case of intrinsic magnetic topological insulators. Third, the emerging body of work on two-dimensional van der Waals magnets is surveyed, with emphasis on how functional choice, Hubbard corrections and dispersion corrections influence predicted exchange interactions, magnetic anisotropy and critical temperatures. Fourth, cross-cutting validation strategies are synthesised into a practical decision framework intended to guide functional selection according to the dominant physics of the material under study. Throughout, the meta-generalised-gradient approximation and its strongly constrained and appropriately normed variants are treated as a distinct and increasingly important middle ground between computationally economical semi-local functionals and costly many-body approaches. The review concludes that no single functional is universally reliable across strongly correlated, spin–orbit-coupled and low-dimensional magnetic systems, and that transparent, material-specific validation against experiment or higher-level theory remains indispensable.