Session T46: Excited State V: real time TDDFT

Exact-factorization-based methods for coupled electrons, ions, and photons.

The description of excited state phenomena requires both an adequate description of the electronic structure and electron-nuclear correlation, while calling for computationally efficient methods to deal with the size and complexity of the systems of interest.  This talk focusses on the electron-nuclear correlation problem within the exact factorization (XF) framework, which provides exact potentials that drive different coupled quantum subsystems, and, more practically,  provides a rigorous starting point for approximate methods. We show that XF-derived modifications to the surface-hopping equations contain terms that correctly lead to population transfer and decoherence, while retaining a computationally-efficient independent trajectory picture, with particularly improved predictions of non-adiabatic dynamics when several electronic states become coupled. We further discuss the extensions of these methods to polaritonic systems where strong light-matter coupling through confinement modifies the chemical and physical properties of matter. Finally, the exact potential driving the photons is used to analyze the accuracy of Ehrenfest methods adapted to account for many photon modes.   

A novel Transition-Based Constrained DFT (TCDFT) for the Robust and Reliable Treatment of Pure and Mixed Excitations in Molecular Systems.

Despite the variety of available computational approaches, state-of-the-art methods for calculating excitation energies such as time-dependent density functional theory (TDDFT), are typically computationally demanding and thus limited to moderate system sizes. We present a new variation of constrained DFT (CDFT), implemented in the BigDFT code, wherein the constraint corresponds to a particular transition (T) between occupied and virtual orbitals, rather than a region of the simulation space as in traditional CDFT. We perform benchmark T-CDFT calculations for a set of gas phase acene molecules and OLED emitters considering both pure and mixed excitation states. For both classes of molecules, we find that T-CDFT based on semi-local functionals is comparable to hybrid functional results from ΔSCF and TDDFT. Furthermore, T-CDFT proves to be more robust than ΔSCF and does not suffer from the well-known problems encountered when applying TDDFT to charge-transfer (CT) states, and is therefore applicable to both local excitations and CT states. Finally, T-CDFT is designed for large systems  and it is ideally suited for exploring the effects of explicit environments on excitation energies, paving the way for future simulations of excited states in complex realistic morphologies.   

Cumulative Surface Hopping: Faster and More Reproducible

Photochemistry simulations powered by trajectory surface hopping (TSH) dynamics have become an essential tool for understanding light-driven reactivity. In TSH, electronic transitions are mimicked using stochastic hops between electronic surfaces and the statistics of hopping are sampled using independent trajectories. In the conventional approach to TSH, hops are determined by comparing the instantaneous hopping probability to a random number at each time step. Here, we describe an alternative scheme using the cumulative hopping probability. We will show how this seemingly simple alteration opens the door to semi-stochastic sampling algorithms that drastically reduce computational time, and new strategies to numerically evaluate propagation algorithms, all while improving computational reproducibility. We argue that using cumulative hopping probabilities has myriad advantages over the conventional approach and no apparent disadvantages. Thus, we recommend cumulative surface hopping as the default approach in all TSH implementations.   

Fast all-electron real-time TDDFT calculations for electronic stopping power

Swift ions traveling through matter are slowed down due to electronic excitations. The electronic stopping power quantifies this phenomenon and is central to the aging of functional materials in nuclear or space environments [1]. Confronted with the heavy computational cost of real-time time-dependent density-functional theory (RT-TDDFT) approaches, we have developed a fast alternative using localized Gaussian-type orbitals [2, 3]. Our program has been successfully applied to both metal [3] and insulator [4] targets, however only for projectile ions totally stripped of their electrons.

We will present here our recent progresses in the inclusion of projectile electrons. Having a time-dependent basis is a theoretical and a computational challenge. We will also show our latest results for proton and α-particle impinging metallic targets.   

Withdrawn

The recent proposal of a Floquet theory of electronic stopping along periodic directions in crystalline solids indicates the existence of stationary solutions for the electronic stopping problem. Some properties are predicted to be time periodic (or stroboscopically  constant, as the electronic stopping power itself), while other have time-periodic time derivatives (spectral decompositions). Taking advantage of large-cell real-time time-dependent density-functional theory simulations, we explore such properties in the electronic stopping of protons in large supercells of diamond and compare them to metal hosts.  We explore the velocity and direction dependence of electronic stopping power in this crystal.    

First-Principles Methods for Simulating Inelastic Scattering of High Energy Electron Beams by Materials

 

Recent work relying on the convergent electron beams found in aberration-corrected scanning transmission electron microscopes (STEMs) has demonstrated that highly localized chemical reactions can be controllably induced in materials via electron irradiation.  In certain cases, electron beam-induced reactions can be promoted with true atomic precision.  Such reactions are understood to be driven largely by elastic electron-nucleus scattering processes, through which beam electrons can transfer significant linear momentum to individual nuclei.  However, inelastic scattering —which results in coupled electronic and vibrational excitations in the material — can also occur simultaneously with the elastic scattering by the nuclei.  Whether changes in the potential energy landscape of materials due to such excitations nontrivially influence the reaction kinetics remains an open question.  Our group has recently developed a suite of first-principles tools capable of simulating the electronic and vibrational response of materials to electron beams, both in real- and momentum-space.  The methods are first overviewed, and their application to some simple systems of experimental interest is then presented.  Finally, we provide some perspective on the use of these simulations to guide experimental efforts in the area of beam-induced materials chemistries, and how machine learning approaches can provide a bridge between the disparate timescales of these STEM experiments and ab initio simulations.   

Non-Linear Dynamics of Plasmon Excitation at Metal-Semiconductor Interfaces from First Principles Theory

We apply first-principles electronic structure calculation to study the process of plasmon-induced charge transfer that takes place between a plasmonic metal nanoparticle and semiconductor surface. Through analysis of the time-dependent maximally localized Wannier functions in real-time time-dependent density functional theory, we describe the excited behavior of the plasmon and the resulting effect on the interface. In this work we focus specifically on the plasmonic Ag20 nanocluster interfaced with a hydrogenated silicon surface.   

Analytical nonadiabatic exchange-correlation potentials for nonequilibrium studies of strongly correlated materials

The quest for a microscopic understanding of the ultrafast processes in strongly correlated materials hinges on the fact that it captures the role of electron correlations. To accomplish this challenging task, a way forward is through derivation of appropriate exchange-correlation (XC) potentials for implementation in time-dependent density-functional theory (TDDFT). We present here analytical expressions for such (nonadiabatic) potentials in the limits of strong and weak electron correlations, obtained by using the Sham-Schlueter equation approach. The XC potentials are obtained from the local-in-space electron self-energy in the one-band Hubbard model. To test the potentials, we employ them in TDDFT calculations for the one-band Hubbard model and compare the results with the nonequilibrium dynamical mean-field theory solution which shows good agreement. We discuss strategies for formulation of a universal XC potential valid also for intermediate strengths of electron correlations. The net outcome is a reliable and computationally efficient XC potential that takes into account both memory and correlation effects and can be applied to examine ultrafast properties of materials at any strength of the perturbing field.    

First-principles study of shift current mechanism in low-dimensional inversion-broken systems

Bulk photovoltaic effect, characterized by the generation of a steady photocurrent without the aid of external p-n junction, has attracted a lot of attention because of its potential as a high-performance solar energy harvester. We perform ab initio simulations to obtain the real-time photocurrent, and investigate underlying electronic origin of the photocurrent generation mechanism in organic molecular solids (TTF-CA), hybrid halide perovskite (MAPbI3 and FAPbI3), and transition metal dichalcogenides (TMDs). We show the photovoltaic nature can be associated with the interchain charge shifting mechanism in TTF-CA. For hybrid halide perovskite, though both the ferroelectric polarization and the nonzero bulk photocurrent are prototypical manifestation of the broken inversion symmetry, we investigate that the photovoltaic nature is not necessarily associated with the ferroelectricity, but is rather determined by the intrinsic electronic band properties near the Fermi level. By considering one-dimensional TMD nanotubes we examine the effect of dimensionality, beyond the broken inversion symmetry, on the generation mechanism of photocurrent. We find that the nanotube structure can produce substantially larger shift current as observed in experiment. We suggest that the real-time analysis of the photocurrent can provide a better understanding of the bulk photovoltaics, beyond the widely used perturbation schemes.   

Real time electron emission dynamics of graphene under proton irradiation

Graphene has a wide range of applications in novel electronics devices, such as transistors, biosensors, and solar cells. However, since its properties can be sensitively influenced by its defects, a precise characterization technique is often required for materials fabrication. Focused ion beams have been widely employed in this context, but their parameters must be carefully selected, which requires a detailed understanding of the underlying excited electron dynamics. Here, we apply a combination of real-time time dependent density functional theory and Ehrenfest dynamics to qualitatively investigate the intensity and duration of secondary electron pulses for graphene under proton irradiation and its dependence on proton trajectory and velocity. We reveal that centroid trajectories yield 38-53% more secondary electrons than channeling trajectories on both sides of the graphene. Also, we find that the duration of secondary electron pulses increases as the proton velocity increases. Recapture of electrons can be observed after 1 fs for protons with kinetic energy near 6.1 keV but after 3 fs for protons with kinetic energy more than 25 keV. This comprehensive description of secondary electron pulses provides critical insight into optimizing the parameters of ion beams.    

First-principles studies of point defects in semiconductors using time-dependent density functional theory

First-principles calculations can be used to accelerate the discovery of point defects in semiconductors and insulators, with applications for solid-state quantum technologies. However, the simulation of the optical properties of point-defects, e.g., of photoluminescence (PL) spectra [1], is still a difficult task because it is challenging to obtain an accurate description of multi-configurational excited potential energy surfaces of defects in solids. Here we discuss the calculations of excited states and PL spectra of defects in diamond and hexagonal boron nitride, carried out using an efficient implementation of forces at the level of time-dependent hybrid density functional theory (TDDFT) for periodic systems. The comparison of our results with experiments suggests that TDDFT with hybrid functionals can be used for robust predictions of the excited state properties of point defects in semiconductors.

[1] Y. Jin, M. Govoni, G. Wolfowicz, S. E. Sullivan, F. J. Heremans, D. D. Awschalom, and G. Galli, Phys. Rev. Mater. 5, 084603 (2021)   

First-Principles Demonstration of Nonadiabatic Thouless Pumping of Electrons in a Molecular System

We demonstrate nonadiabatic Thouless pumping of electrons in trans-polyacetylene in the framework of Floquet engineering using first-principles theory. We identify the regimes in which the quantized pump is operative with respect to the driving electric field for a time-dependent Hamiltonian. By employing the time-dependent maximally localized Wannier functions in real-time time-dependent density functional theory simulation, we connect the winding number, a topological invariant, to a molecular-level understanding of the quantized pumping. Using a gauge-invariant formulation called dynamical transition orbitals, an alternative viewpoint on the nonequilibrium dynamics is obtained in terms of the particle-hole excitation. A single time-dependent transition orbital is found to be largely responsible for the observed quantized pumping. Further, robustness of the nonadiabatic Thouless pumping is examined by introducing different types of chemical modifications.   

Low-energy electron projectiles in water: Linear vs non-linear energy loss

Low-energy electrons as secondary projectiles are behind a significant part of the radiation damage affecting living matter. Research in the optimisation of ion radiotherapy and treatment of radiation poisoning in nuclear-energy environments or outer space make extensive use of Monte Carlo-based cascade simulations (see e.g. [1]) using scattering cross sections and stopping power as input. The ones for electron stopping in liquid water are key input for those simulations. They are customarily obtained from linear-response and phenomenology. Here we present first principles results for the quantum friction for low-energy electron projectiles in water using time-dependent density-functional calculations, both in linear response (frequency domain) and beyond linear, the latter based on explicit real-time simulations. The comparison reveals important deviations, non-linear vs linear, around the Bragg peak, which might be significant for modelling.

[1] K. P. Chatzipapas, P. Papadimitroulas, D. Emfietzoglou, S. A, Kalospyros, M. Hada, A. G. Georgakilas, and G. C. Kagadis, Cancers 12(4), 799 (2020).