🔬 materials science

Dynamical Hubbard approach to correlated materials: the case of transition-metal monoxides

This paper extends the dynamical Hubbard functional formulation to spin-polarized and multi-site cases, demonstrating that this efficient and transparent approach accurately captures electronic correlations in transition-metal monoxides (MnO, FeO, CoO, and NiO) with results comparable to or better than state-of-the-art theories.

Original authors: Mario Caserta, Tommaso Chiarotti, Marco Vanzini, Nicola Marzari

Published 2026-10-08

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Mario Caserta, Tommaso Chiarotti, Marco Vanzini, Nicola Marzari

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the world of materials science, some substances are deceptively simple to look at but incredibly difficult to understand. Take the transition-metal oxides, a family of compounds formed when metals like manganese, iron, cobalt, or nickel bond with oxygen. These materials are often insulators, meaning they do not conduct electricity, yet their behavior is driven by electrons that are stubbornly stuck in place, refusing to flow freely. This stubbornness arises from a phenomenon known as electronic correlation, where the movement of one electron is deeply tied to the presence of its neighbors, creating a complex, crowded dance of repulsion and attraction. Standard theories used to predict how these materials behave often fail here, treating electrons as if they move independently when they are actually locked in a tight, collective struggle. Understanding these materials is crucial because they are the building blocks for future technologies in energy storage and electronics, but to design better devices, scientists first need a way to accurately describe this crowded, chaotic environment.

A team of researchers has now developed a new, streamlined approach to solve this problem, offering a clearer window into the behavior of these complex oxides. In a recent study, the authors applied a method called the dynamical Hubbard functional to four specific compounds: manganese oxide, iron oxide, cobalt oxide, and nickel oxide. These materials serve as the perfect test cases because they sit on a spectrum of difficulty, ranging from those where electrons are moderately correlated to those where they are heavily entangled. The researchers wanted to see if their new method could capture the subtle, time-dependent interactions between electrons without getting bogged down in the massive computational costs that usually plague such calculations. By treating the electrons' interactions as a dynamic process that changes over time, rather than a static snapshot, they were able to generate detailed maps of the materials' electronic energy levels.

The results were strikingly accurate. When the team compared their simulations to real-world experimental data, the match was excellent across the board. For manganese oxide, the predicted energy levels aligned closely with what has been measured in laboratories. For iron oxide, the method correctly identified a small energy gap of about 2 electron-volts, a detail that other advanced theories had missed, often predicting a much larger gap that did not fit the experimental reality. In the case of cobalt oxide and nickel oxide, the new approach successfully reproduced the complex patterns of electron energy, including how the strength of the material's insulating properties shifts as you move across the series. Perhaps most notably, the method revealed specific electronic bands in nickel oxide that had previously disappeared in other high-level simulations but were clearly visible in experimental measurements. This suggests that the new approach captures a level of physical detail that was previously lost in the noise of more complex, expensive calculations.

What makes this achievement particularly significant is the balance it strikes between accuracy and efficiency. Traditional methods that try to solve these problems often require immense computing power and time, scaling exponentially with the size of the system, which limits their use for larger, more realistic materials. The dynamical Hubbard functional, by contrast, offers a transparent and computationally efficient path forward. It manages to include the essential physics of electron correlation—such as the shifting of energy weights and the renormalization of electron bands—without needing the heavy machinery of exact solutions. The researchers found that by using a specific type of starting calculation and applying their dynamic correction, they could predict the properties of these materials with a precision that rivals the most sophisticated theories available today.

The study confirms that it is possible to move beyond static approximations and capture the living, breathing nature of electron interactions in these materials. The authors demonstrated that their method not only reproduces known experimental facts but also resolves discrepancies found in other state-of-the-art theories, particularly regarding the size of energy gaps and the presence of specific electronic states. By providing a tool that is both physically transparent and numerically efficient, this work opens the door to studying even more complex correlated materials that were previously out of reach. It suggests that the key to unlocking the secrets of these stubborn electrons lies not in making calculations infinitely more complex, but in finding the right dynamic formulation that respects the time-dependent nature of the interactions.

Technical Summary: Dynamical Hubbard Approach to Correlated Materials

Problem Statement
Transition-metal monoxides (MnO, FeO, CoO, NiO) serve as prototypical systems for studying complex electronic correlations, categorized as Mott and charge-transfer insulators. While standard Kohn-Sham density-functional theory (DFT) with semi-local functionals fails to predict the insulating nature of CoO and FeO and underestimates band gaps in MnO and NiO, more advanced methods face significant trade-offs.

Static Mean-Field Approaches: Methods like DFT+U and hybrid functionals can recover insulating states and gap magnitudes but are inherently static. They fail to capture frequency-dependent self-energies necessary for describing band renormalization, spectral weight transfer, and particle lifetimes—key signatures of electronic correlations.
Many-Body Perturbation Theory (MBPT): Approaches like $G_0W_0$ and $GW_0$ often struggle when starting from metallic DFT ground states. While extensions like QSG $\hat{W}$ improve gap descriptions, they frequently discard full frequency dependence due to computational cost, limiting their ability to describe $k$ -resolved spectral weight transfer.
DFT+DMFT: Dynamical Mean-Field Theory (DFT+DMFT) successfully models local dynamical correlations and fluctuating magnetic moments by mapping the correlated subspace to an Anderson impurity model. However, this comes with high computational costs and numerical complexity, with exact solutions scaling exponentially with system size. Furthermore, recent DFT+DMFT results have predicted wide gaps for all four monoxides and failed to reproduce certain spectral features observed in Angle-Resolved Photoemission Spectroscopy (ARPES), such as specific dispersive bands in NiO.

Methodology
The authors propose a "Dynamical Hubbard" (dynH) functional, a generalization of their previously introduced formulation to handle spin-polarized, multi-site magnetic systems. The approach is designed to be numerically efficient and physically transparent while capturing dynamical correlations.

Formalism: The method extends the Dudarev rotationally invariant DFT+U formulation by introducing a local, frequency-dependent, screened interaction potential. The self-energy is derived from the functional derivative of the dynamical Hubbard functional with respect to the local Green's function:
$\Sigma^{\sigma, I}_{\text{dynH}}(\omega) = -\int d\omega' U_I(\omega') G^{\sigma}_I(\omega + \omega') + \frac{U^{\infty}_I}{2}$
Here, $U_I(\omega)$ represents the local average of the screened Coulomb interaction (calculated via RPA spin-DFT polarization), and $U^{\infty}_I$ is the unscreened bare Coulomb potential.
Implementation:
- The local self-energy is upfolded to $k$ -space to solve the Dyson equation ( $G^{-1} = G^{-1}_0 - \Sigma$ ).
- The solution utilizes the Algorithmic Inversion Method on Sum Over Poles (AIM-SOP). This technique represents both the local screened interaction and the Green's function as sums over first-order poles.
- Crucially, AIM-SOP operates directly on the real frequency axis, avoiding the use of imaginary Matsubara frequencies and the subsequent analytic continuation required in standard DMFT impurity solvers. This enhances calculation accuracy.
- The ground-state calculations employ the r2SCAN functional, which correctly predicts insulating phases for all four compounds on an equal footing.
- The effective screened interaction is calculated at the first-order in perturbation theory, with a zero-frequency limit of $U(0) = 5.3$ eV.

Key Results
The framework was applied to the ground states of MnO, FeO, CoO, and NiO, comparing results against experimental Photoemission (PES) and Inverse Photoemission (IPES) data, as well as state-of-the-art DFT+DMFT and QSG $\hat{W}$ calculations.

MnO: The occupied Density of States (DOS) aligns with DFT+DMFT results, showing a slight contraction in overall bandwidth compared to experiment. The LUMO excitation at 5 eV matches IPES data in position, width, and intensity. High-energy satellites remain elusive, a limitation shared with DFT+DMFT and QSG $\hat{W}$ .
FeO: The method predicts a small band gap of approximately 2 eV, consistent with IPES data. This contrasts with DFT+DMFT and QSG $\hat{W}$ , which predict wide gaps for all four monoxides. The intensity decrease after the first excitations is well-described.
CoO: The DOS compares well with both DFT+DMFT and PES data. A notable difference is the distribution of empty states: DFT+dynH shows comparable weights for $t_{2g}$ and $e_g$ symmetry states, whereas DFT+DMFT transfers weight primarily to the higher energy state.
NiO: The approach achieves excellent agreement with experimental ARPES data and PES/IPES spectra.
- The overall bandwidth (6 eV) is consistent with ARPES.
- The method successfully reproduces the Zhang-Rice doublet bound state and flat Ni(3d)/O(2p) bands.
- Critical Finding: Unlike DFT+DMFT, which predicts vanishing spectral weight for specific bands, DFT+dynH reproduces the dispersing band at -8 eV and the antiferromagnetic (AFM) flat band around -4 eV. These features are present in experimental ARPES data but were previously missing in DMFT impurity solutions. The authors attribute the slight underestimation of the AFM bands to the r2SCAN starting point, lack of self-consistency, or the approximate nature of the double-counting term.

Significance and Claims
The paper claims that the DFT+dynH approach offers a "simple and transparent" alternative to computationally expensive methods like DFT+DMFT. Its primary contributions are:

Generalization: Successfully extending the dynamical Hubbard functional to multi-site, spin-polarized magnetic systems.
Accuracy vs. Cost: Achieving results comparable to or improved upon state-of-the-art theories (DFT+DMFT, QSG $\hat{W}$ ) for spectral properties, including band renormalization and spectral weight transfer, but with significantly lower computational cost and numerical complexity.
Resolution of Discrepancies: Correcting specific failures of previous theories, such as the prediction of a small gap for FeO and the recovery of spectral features in NiO that were absent in DMFT solutions.
Physical Insight: Providing access to fundamental signatures of electron correlations (particle lifetimes, satellites, spectral weight transfer) without the prohibitive scaling of exact impurity solvers.

The authors position DFT+dynH as a promising tool for simulating realistic, complex correlated materials, particularly those with symmetry-broken orbital orderings and hybridizations, where methods employing exact solutions may be prohibitive.

Technical Summary: Dynamical Hubbard Approach to Correlated Materials

More like this