This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Heyd–Scuseria–Ernzerhof | |
|---|---|
| Name | Heyd–Scuseria–Ernzerhof |
| Acronym | HSE |
| Type | hybrid density functional |
| Developers | Jochen Heyd, Gustavo E. Scuseria, Matthias Ernzerhof |
| First appeared | 2003 |
Heyd–Scuseria–Ernzerhof is a screened hybrid density functional used in computational Kohn–Sham electronic structure calculations for molecules and solids. It combines short-range exact exchange with a generalized gradient approximation exchange–correlation functional to improve band gaps and reaction barriers in comparisons with pure Perdew–Burke–Ernzerhof, B3LYP, and PBE0 functionals. The approach has become widely adopted in implementations within software packages developed by groups associated with Gaussian (software), Quantum ESPRESSO, VASP, CP2K, and CASTEP.
Heyd–Scuseria–Ernzerhof was introduced by researchers affiliated with institutions connected to Rice University and Duke University and published in 2003, with a later revision in 2006. The functional’s construction built on developments from Walter Kohn, Lu Jeu Sham, early generalized gradient approximations such as John Perdew’s work, and hybrid functional concepts advanced by Axel Becke and collaborators. By blending screened Hartree–Fock exchange with a Perdew–Burke–Ernzerhof-type correlation, HSE addresses shortcomings documented in comparisons involving Hartree–Fock, Møller–Plesset perturbation theory, and various coupled cluster studies.
The theoretical foundation uses the Kohn–Sham formalism and partitions the Coulomb operator into short-range and long-range components via an error function kernel parameterized by a screening parameter. The short-range component mixes a fraction of nonlocal Hartree–Fock exchange with a semilocal exchange functional derived from Perdew–Burke–Ernzerhof, while correlation is treated at the semilocal level. The formulation references mathematical tools and concepts associated with Fourier transform techniques, Ewald summation ideas used in periodic electrostatics, and operator-splitting strategies familiar from implementations in plane-wave and Gaussian basis frameworks. The construction explicitly targets improvements over functionals tested in benchmark studies involving systems previously examined by researchers such as Martin Karplus, Walter Kohn, and groups working on band gap problem analysis.
Two widely cited parameterizations are often denoted by their publication years: the original 2003 parameter set and a 2006 revision that adjusted the screening parameter to improve agreement with experimental data for solids and molecules. The 2003 form adopted a specific fraction of exact exchange and a screening length tested against datasets influenced by benchmarking efforts from groups including David R. Reichman and Gonzalo Álvarez. The 2006 revision altered the screening parameter to better reproduce optical properties measured in experiments performed by groups at institutions like Bell Labs, IBM Research, and national laboratories such as Argonne National Laboratory. Users commonly refer to these parameter sets when comparing results with those obtained using B3LYP, PBE0, SCAN, and other modern functionals.
Implementations exploit efficient evaluation of Hartree–Fock exchange integrals in periodic boundary conditions, combining techniques from Fast Fourier Transform, localized basis optimizations seen in Gaussian (software), and projector-augmented wave approaches popularized in software from the Vienna Ab initio Simulation Package group. Algorithms incorporate k-point sampling strategies associated with methods developed at Uniform Electron Gas studies and accelerate exchange evaluation using real-space truncation schemes and density-fitting approaches related to research from Frank Neese and teams at Max Planck Institute for Coal Research. Parallelization strategies follow high-performance computing paradigms applied at centers such as Oak Ridge National Laboratory and Lawrence Berkeley National Laboratory to scale calculations to supercomputers like Summit and Perlmutter.
Benchmark studies compare HSE variants to experiment and to high-level quantum chemistry methods including coupled cluster with single, double, and perturbative triple excitations and quantum Monte Carlo results across datasets curated by consortia involving National Institute of Standards and Technology and academic groups such as those at Harvard University and MIT. HSE typically reduces systematic underestimation of semiconductor band gaps observed with Local Density Approximation and Generalized Gradient Approximation functionals, and provides improved reaction kinetics relative to pure semilocal functionals in datasets used by GMTKN55-style benchmarking efforts. Performance assessments also reference spectroscopic comparisons from experiments at facilities including European Synchrotron Radiation Facility and Advanced Photon Source.
HSE has been applied extensively to predict electronic structure, optical spectra, defect energetics, and surface chemistry in materials investigated by research groups at Stanford University, University of California, Berkeley, and California Institute of Technology. Representative applications include studies of photovoltaic absorber materials examined alongside work from National Renewable Energy Laboratory, investigations of oxide interfaces connected to research at Los Alamos National Laboratory, and catalysis research related to efforts at ETH Zurich and University of Cambridge. Molecular applications span organic photovoltaics and charge-transfer complexes studied in collaborations involving Max Planck Institute for Solid State Research and industrial labs such as BASF.
Limitations include residual self-interaction errors and challenges in describing van der Waals interactions unless augmented by dispersion corrections developed by researchers like Axel Grimme. Extensions and hybrid schemes combine HSE with many-body perturbation approaches such as GW approximation and embedding frameworks inspired by dynamical mean-field theory to treat strongly correlated systems studied by communities at Princeton University and Columbia University. Research continues to refine screening strategies and to integrate machine-learning potentials developed by groups at Google DeepMind and MIT-IBM Watson AI Lab to accelerate HSE-quality predictions.