Researchers introduce non-empirical Δf_xc(q,ω) to TDDFT for nonlocal XC; dramatic spectral improvement; aligns with experiments.

Nonempirical Time-Dependent Density Functional Theory Framework for Nonlocal Exchange–Correlation Potentials

Advanced, orbital-dependent exchange–correlation (XC) functionals can significantly improve the description of electronic structural properties, but they substantially worsen spectral properties that are computed within standard linear-response time-dependent density functional theory (TDDFT) frameworks. This is not a failure of the underlying Kohn-Sham states, but due to a formal inconsistency in the treatment of the dynamic density response when the non-locality of the XC potential is not taken into account consistently on the level of the full off-diagonal density matrix. To avoid these complexities, we present a non-empirical additive correction Δ f xc ( 𝐪 , ω ) \Delta f_{\textnormal{xc}}(\mathbf{q},\omega) to the dynamic XC kernel that re-enforces the exact f-sum rule of the non-local KS Hamiltonian within TDDFT. Comparing our new results against a representative set of accurate experimental measurements (ambient aluminum, silicon and carbon, as well as heated and compressed aluminum) reveals a dramatic improvement in all cases without any additional computational cost. The corresponding extension to the open-source GPAW code is made freely available online. We further investigate the implications of non-local pseudopotentials and show that the non-locality has an important, physically motivated effect that is indispensable to capture the correct plasmon dispersion in lieu of full all-electron simulations. In addition to being important for the estimation of a plethora of dynamic and spectral properties, our work constitutes an important step towards a universal XC functional that can be used to estimate all kinds of observables with high accuracy. Finally, we outline the potential utility of our framework for the development of advanced non-local XC functionals, and suggest a new way to rigorously verify non-local pseudopotentials against experimental measurements of collective excitations.

Over the last decades, density functional theory (DFT) has become established as the most successful tool for electronic-structure simulations of real materials in a great variety of contexts [ 56 , 53 , 73 , 51 ] . Its characteristic balance between computational efficiency and accuracy is based on the formally exact mapping of the complicated quantum many-electron system of interest onto a single-electron problem governed by the effective Kohn-Sham (KS) Hamiltonian H ^ KS \hat{H}_{\textnormal{KS}} [ 48 , 61 ] . The crucial ingredient is given by the exchange–correlation (XC) functional, which has to be approximated in practice, and which decisively determines the accuracy of a given DFT simulation [ 36 ] . Starting from relatively simple local density approximation (LDA) [ 93 , 118 ] and generalized gradient approximation (GGA) [ 94 ] type functionals, there has been remarkable recent progress on the development of more advanced XC functionals on higher rungs of Jacob’s ladder of functional approximations [ 95 ] . For example, meta-GGA and hybrid XC functionals that explicitly depend on the KS orbitals have been shown to give a vastly improved description of some important material properties such as band gaps in semiconductors and insulators [ 46 , 74 , 122 , 126 ] .

Recently, there has also emerged an increased interest in the properties of matter at extreme densities, temperatures and pressures. Such warm dense matter (WDM) [ 117 ] naturally occurs in a plethora of astrophysical objects such as giant planet interiors and brown dwarfs, and is also of key importance for cutting-edge technological applications such as inertial fusion energy (IFE) [ 12 , 13 , 1 ] and the discovery and synthesis of novel materials [ 68 , 62 ] . In practice, the accurate description of WDM systems using thermal DFT methods [ 75 ] requires the development of corresponding thermal XC functionals, that explicitly take into account the dependence on the temperature and which must include an additional entropic XC contribution [ 108 , 59 , 39 , 57 , 101 , 87 ] . Here, too, recent seminal works by Karasiev and co-workers [ 77 , 23 , 58 ] and others [ 123 , 81 ] have highlighted the importance of advanced, non-local XC functionals.

A particularly important class of observables that can be computed using time-dependent DFT (TDDFT) methods [ 113 ] is given by spectral properties such as electrical and thermal conductivity and opacity, which are of key importance for the modeling of both IFE experiments and astrophysical objects [ 117 , 38 ] . In addition, the dynamic density response function can be obtained using TDDFT [ 105 , 7 , 100 , 10 , 82 ] . The latter is directly probed in x-ray Thomson scattering (XRTS) experiments [ 35 ] via the electronic dynamic structure factor (DSF) S ⁡ ( 𝐪 , ω ) S(\mathbf{q},\omega) at finite momentum transfer q = | 𝐪 | q=|\mathbf{q}| . Due to this excellent point of contact between experiment and theory, XRTS has emerged as a standard method of diagnostics for extreme states of matter [ 20 , 19 , 21 , 63 ] , creating a high demand for accurate ab initio simulations. Spectral measurements such as electron energy loss spectroscopy (EELS) [ 110 ] and XRTS constitute an important way to probe the electronic structure of a given system at ambient conditions as well, allowing for more detailed comparisons between theory and experiment compared with integrated thermodynamic properties or lattice constants.

In this work, we rigorously examine the implications of using non-local XC functionals and non-local pseudopotentials within standard TDDFT set-ups. For example, non-local XC functionals—commonly considered as a gold standard for the estimation of a wide range of electronic structure properties—lead to a drastically decreased quality in spectral observables compared to the simpler and computationally less expensive LDA or GGA type functionals without further corrections [ 91 ] . A representative example is depicted in Fig. 1 , where we compare a high-resolution XRTS spectrum of the plasmon of ambient aluminum at q = 0.93 q=0.93, Å -1 obtained at the European XFEL in Germany [ 30 ] (orange) with linear-response TDDFT results obtained using the GGA-level PBE [ 94 ] (solid black), meta-GGA level r2SCAN [ 27 ] (dotted black) and hybrid level HSE06 [ 47 ] (dashed black). Clearly, the comparably simple PBE calculation is in good agreement with the experimental spectrum, whereas r2SCAN and in particular HSE06 overestimate the true plasmon shift by more than 1.5 1.5, eV. This is not an inherent deficiency in the underlying KS orbitals, but due to the inconsistent treatment of the non-locality of the respective XC potentials in standard TDDFT implementations. Conversely, we show that, in lieu of computationally challenging all-electron calculations, a certain degree of non-locality is indispensable to estimate the correct plasmon excitation in the long wavelength limit. This insight opens up a new way for the experimental assessment of different non-local pseudo potentials. In addition, this finding points to limitations of time-dependent orbital-free DFT [ 18 , 8 , 55 , 120 ] , which commonly employs local pseudopotentials for the electron–ion interaction, when applied to the description of plasmon excitations. This has practical relevance for warm dense matter and IFE applications because orbital-free DFT provides an efficient framework for large-scale simulations of dynamic properties, including electronic stopping power [ 18 ] .

Throughout the remainder of this work, we methodically disentangle these complexities, leading to the following results:

• Non-local contributions to the effective KS Hamiltonian H ^ KS \hat{H}_{\textnormal{KS}} from both the pseudopotential and

or from the XC functional lead to a modified f-sum rule with direct and important implications for the estimated spectrum and for the plasmon dispersion.

Non-local contributions to the effective KS Hamiltonian H ^ KS \hat{H}_{\textnormal{KS}} from both the pseudopotential and

or from the XC functional lead to a modified f-sum rule with direct and important implications for the estimated spectrum and for the plasmon dispersion.

• Within a pseudized description of the electron–ion interaction, non-locality of the pseudopotential is essential to estimate the correct plasmon dispersion. Moreover, we show how any non-local pseudopotential gives rise to an effective electron–ion coupling energy Δ ion \Delta_{\textnormal{ion}} that is directly related to the plasmon shift and, thus, can be measured in XRTS [ 67 , 30 ] or EELS [ 22 ] experiments.

Within a pseudized description of the electron–ion interaction, non-locality of the pseudopotential is essential to estimate the correct plasmon dispersion. Moreover, we show how any non-local pseudopotential gives rise to an effective electron–ion coupling energy Δ ion \Delta_{\textnormal{ion}} that is directly related to the plasmon shift and, thus, can be measured in XRTS [ 67 , 30 ] or EELS [ 22 ] experiments.

• The combination of orbital-dependent non-local XC potentials (meta-GGA or hybrid level XC functionals) with standard TDDFT implementations leads to systematic inconsistencies, resulting in biased spectra and unphysical plasmon shifts.

The combination of orbital-dependent non-local XC potentials (meta-GGA or hybrid level XC functionals) with standard TDDFT implementations leads to systematic inconsistencies, resulting in biased spectra and unphysical plasmon shifts.

• Re-enforcing the exact f f -sum rule of the non-local KS Hamiltonian introduces a non-empirical dynamic correction Δ f xc ( 𝐪 , ω ) \Delta f_{\textnormal{xc}}(\mathbf{q},\omega) to the XC kernel f xc ( 𝐪 , ω ) f_{\textnormal{xc}}(\mathbf{q},\omega) that removes this inconsistency for both static (adiabatic) and dynamic (non-adiabatic) XC kernels without additional computational cost. A corresponding implementation for the open-source GPAW [ 89 ] code is being made freely available [ 84 ] .

Re-enforcing the exact f f -sum rule of the non-local KS Hamiltonian introduces a non-empirical dynamic correction Δ f xc ( 𝐪 , ω ) \Delta f_{\textnormal{xc}}(\mathbf{q},\omega) to the XC kernel f xc ( 𝐪 , ω ) f_{\textnormal{xc}}(\mathbf{q},\omega) that removes this inconsistency for both static (adiabatic) and dynamic (non-adiabatic) XC kernels without additional computational cost. A corresponding implementation for the open-source GPAW [ 89 ] code is being made freely available [ 84 ] .

• We demonstrate the value of the new framework by rigorous comparisons with high-precision XRTS and EELS measurements for a number of representative examples: ambient aluminum [ 30 ] and warm dense aluminum [ 10 ] , as well as ambient silicon [ 29 ] and carbon [ 119 ] , i.e., two metallic systems and two semiconductors. In all cases, applying Δ f xc ( 𝐪 , ω ) \Delta f_{\textnormal{xc}}(\mathbf{q},\omega) gives drastically improved spectra and good agreement with the experimental data for the considered meta-GGA and hybrid XC functionals.

We demonstrate the value of the new framework by rigorous comparisons with high-precision XRTS and EELS measurements for a number of representative examples: ambient aluminum [ 30 ] and warm dense aluminum [ 10 ] , as well as ambient silicon [ 29 ] and carbon [ 119 ] , i.e., two metallic systems and two semiconductors. In all cases, applying Δ f xc ( 𝐪 , ω ) \Delta f_{\textnormal{xc}}(\mathbf{q},\omega) gives drastically improved spectra and good agreement with the experimental data for the considered meta-GGA and hybrid XC functionals.

In this way, our work opens up a gamut of new possibilities. First, our findings constitute an important step towards a future DFT framework in which all observables