Excited-state potential-energy surfaces of metal-adsorbed organic molecules from linear expansion Δ-self-consistent field density-functional theory (ΔSCF-DFT)

Propagated (fragment) Pipek-Mezey Wannier functions in real-time time-dependent density functional theory

Localization procedures are an important tool for analysis of complex systems in quantum chemistry, since canonical molecular orbitals are delocalized and can, therefore, be difficult to align with chemical intuition and obscure information at the local level of the system. This especially applies to calculations obeying periodic boundary conditions. The most commonly used approach to localization is Foster-Boys Wannier functions, which use a unitary transformation to jointly minimize the second moment of the orbitals. This procedure has proven to be robust and fast but has a side effect of often mixing σ- and π-type orbitals. σ/π-separation is achieved by the Pipek-Mezey Wannier function (PMWF) approach [Lehtola and Jónsson, J. Chem. Theory Comput. 10, 642 (2014) and Jónsson et al., J. Chem. Theory Comput. 13, 460 (2017)], which defines the spread functional in terms of partial charges instead. We have implemented a PMWF algorithm in the CP2K software package using the Cardoso-Souloumiac algorithm to enable their application to real-time time-dependent density functional theory. The method is demonstrated on stacked CO2 molecules, linear acetylenic carbon, boron and nitrogen co-doped graphene, and nitrogen-vacancy doped diamond. Finally, we discuss its computational scaling and recent efforts to improve it with fragment approaches.

Characterizing Molecule-Metal Surface Chemistry with Ab Initio Simulation of X-ray Absorption and Photoemission Spectra

X-ray photoemission and X-ray absorption spectroscopy are important techniques to characterize chemical bonding at surfaces and are often used to identify the strength and nature of adsorbate-substrate interactions. In this study, we judge the ability of X-ray spectroscopic techniques to identify different regimes of chemical bonding at metal-organic interfaces. To achieve this, we sample different interaction strength regimes in a comprehensive and systematic way by comparing two topological isomers, azulene and naphthalene, adsorbed on three metal substrates with varying reactivity, namely the (111) facets of Ag, Cu, and Pt. Using density functional theory, we simulate core-level binding energies and X-ray absorption spectra of the molecular carbon species. The simulated spectra reveal three distinct characteristics based on the molecule-specific spectral features which we attribute to types of surface chemical bonding with varying strength. We find that weak physisorption only leads to minor changes compared to the gas-phase spectra, weak chemisorption leads to charge transfer and significant spectral changes, and strong chemisorption leads to a loss of the molecule-specific features in the spectra. The classification we provide is aimed at assisting interpretation of experimental X-ray spectra for complex metal-organic interfaces.

Origin of Broadband Emission and Large Stokes Shift in Antimony Trisulfide

The antimony trisulfide (Sb₂S₃) has been theoretically predicted to have various merits in exploiting high-performance thin-film solar cells and attracted intense attention. However, the power conversion efficiency of Sb₂S₃-based solar cells is yet to be satisfactory in experiments and the origin of large open circuit voltage (V_OC) loss is still a controversial question. Based on first-principles calculations, we have systematically analyzed the excited state behavior and dynamics images of carriers in Sb₂S₃ materials. Our calculations showed that intrinsic defects like vacancy (V_Sb and V_S) and antisites (Sb_S and S_Sb) are energetically accessible. More importantly, we found that the sulfide vacancy-bound excitons can produce a large Stokes shift of ∼0.66 eV, which could well rationalize the experimental observations like the reduction of V_OC. These new findings suggest that the performance of Sb₂S₃-based solar cells might be largely enhanced by avoiding sulfide vacancy defects.

Spin-Orbit Couplings for Nonadiabatic Molecular Dynamics at the ΔSCF Level

A procedure for the calculation of spin-orbit coupling (SOC) at the delta self-consistent field (ΔSCF) level of theory is presented. Singlet and triplet excited electronic states obtained with the ΔSCF method are expanded into a linear combination of singly excited Slater determinants composed of ground electronic state Kohn-Sham orbitals. This alleviates the nonorthogonality between excited and ground electronic states and introduces a framework, similar to the auxiliary wave function at the time-dependent density functional theory (TD-DFT) level, for the calculation of observables. The ΔSCF observables of the formaldehyde system were compared to reference TD-DFT values. Our procedure gives all components (energies, gradients, nonadiabatic couplings, and SOC terms) at the ΔSCF level of theory for conducting efficient, full-atomistic nonadiabatic molecular dynamics with intersystem crossing, particularly in condensed phase systems.

Robust ΔSCF calculations with direct energy functional minimization methods and STEP for molecules and materials

The direct energy functional minimization method using the orbital transformation (OT) scheme in the program package CP2K has been employed for Δ self-consistent field (ΔSCF) calculations. The OT method for non-uniform molecular orbitals occupations allows us to apply the ΔSCF method for various kinds of molecules and periodic systems. Vertical excitation energies of heteroaromatic molecules and condensed phase systems, such as solvated ethylene and solvated uracil obeying periodic boundary conditions, are reported using the ΔSCF method. In addition, a Re-phosphate molecule attached to the surface of anatase (TiO₂) has been investigated. Additionally, we have implemented a recently proposed state-targeted energy projection ΔSCF algorithm [K. Carter-Fenk and J. M. Herbert, J. Chem. Theory Comput. 16(8), 5067-5082 (2020)] for diagonalization based SCF in CP2K. It is found that the OT scheme provides a smooth and robust SCF convergence for all investigated excitation energies and (non-)periodic systems.

The ΔSCF method for non-adiabatic dynamics of systems in the liquid phase

Computational studies of ultrafast photoinduced processes give valuable insights into the photochemical mechanisms of a broad range of compounds. In order to accurately reproduce, interpret, and predict experimental results, which are typically obtained in a condensed phase, it is indispensable to include the condensed phase environment in the computational model. However, most studies are still performed in vacuum due to the high computational cost of state-of-the-art non-adiabatic molecular dynamics (NAMD) simulations. The quantum mechanical/molecular mechanical (QM/MM) solvation method has been a popular model to perform photodynamics in the liquid phase. Nevertheless, the currently used QM/MM embedding techniques cannot sufficiently capture all solute-solvent interactions. In this Perspective, we will discuss the efficient ΔSCF electronic structure method and its applications with respect to the NAMD of solvated compounds, with a particular focus on explicit quantum mechanical solvation. As more research is required for this method to reach its full potential, some challenges and possible directions for future research are presented as well.

The photodissociation of solvated cyclopropanone and its hydrate explored via non-adiabatic molecular dynamics using ΔSCF

The decay of cyclopropanone is a typical example of a photodecomposition process. Ethylene and carbon monoxide are formed following the excitation to the first singlet excited state through a symmetrical or asymmetrical pathway. The results obtained with non-adiabatic molecular dynamics (NAMD) using the delta self-consistent field (ΔSCF) method correspond well to previous experimental and multireference theoretical studies carried out in the gas phase. Moreover, this efficient methodology allows NAMD simulations of cyclopropanone in aqueous solution to be performed, which reveal analogue deactivation mechanisms, but a shorter lifetime and reduced photodissociation as compared to the gas-phase. The excited state dynamics of cyclopropanone hydrate, an enzyme inhibitor, in an aqueous environment are reported as well. Cyclopropanone hydrate strongly interacts with the surrounding solvent via the formation of hydrogen bonds. Excitation to the first singlet excited state shows an asymmetric pathway with cyclopropanone hydrate and propionic acid as the main photoproducts.

The lifetime and photodissociation of cyclopropanone are reduced in aqueous solution, while the excitation of solvated cyclopropanone hydrate yields a range of photoproducts.

First-principles study of exciton self-trapping and electric polarization in one-dimensional organic lead halide perovskites

Revealing the origin of self-trapped excitons is a prerequisite for further improving the photoluminescence efficiency of low-dimensional organic perovskite. Here, the microscopic formation mechanism of intrinsic self-trapped excitons in one-dimensional...

Machine Learning for Electronically Excited States of Molecules

Electronically excited states of molecules are at the heart of photochemistry, photophysics, as well as photobiology and also play a role in material science. Their theoretical description requires highly accurate quantum chemical calculations, which are computationally expensive. In this review, we focus on not only how machine learning is employed to speed up such excited-state simulations but also how this branch of artificial intelligence can be used to advance this exciting research field in all its aspects. Discussed applications of machine learning for excited states include excited-state dynamics simulations, static calculations of absorption spectra, as well as many others. In order to put these studies into context, we discuss the promises and pitfalls of the involved machine learning techniques. Since the latter are mostly based on quantum chemistry calculations, we also provide a short introduction into excited-state electronic structure methods and approaches for nonadiabatic dynamics simulations and describe tricks and problems when using them in machine learning for excited states of molecules.

Machine Learning for Electronically Excited States of Molecules

Electronically excited states of molecules are at the heart of photochemistry, photophysics, as well as photobiology and also play a role in material science. Their theoretical description requires highly accurate quantum chemical calculations, which are computationally expensive. In this review, we focus on not only how machine learning is employed to speed up such excited-state simulations but also how this branch of artificial intelligence can be used to advance this exciting research field in all its aspects. Discussed applications of machine learning for excited states include excited-state dynamics simulations, static calculations of absorption spectra, as well as many others. In order to put these studies into context, we discuss the promises and pitfalls of the involved machine learning techniques. Since the latter are mostly based on quantum chemistry calculations, we also provide a short introduction into excited-state electronic structure methods and approaches for nonadiabatic dynamics simulations and describe tricks and problems when using them in machine learning for excited states of molecules.

Topology Effects in Molecular Organic Electronic Materials: Pyrene and Azupyrene*

Pyrene derivatives play a prominent role in organic electronic devices, including field effect transistors, light emitting diodes, and solar cells. The flexibility in the desired properties has previously been achieved by variation of substituents at the periphery of the pyrene backbone. In contrast, the influence of the topology of the central π‐electron system on the relevant properties such as the band gap or the fluorescence behavior has not yet been addressed. In this work, pyrene is compared with its structural isomer azupyrene, which has a π‐electron system with non‐alternant topology. Using photoelectron spectroscopy, near edge X‐ray absorption fine structure spectroscopy, and other methods, it is shown that the electronic band gap of azupyrene is by 0.72 eV smaller than that of pyrene. The difference of the optical band gaps is even larger with 1.09 eV, as determined by ultraviolet–visible absorption spectroscopy. The non‐alternant nature of azupyrene is also associated with a more localized charge distribution. Further insight is provided by density functional theory (DFT) calculations of the molecular properties and ab initio coupled cluster calculations of the optical transitions. The concept of aromaticity is used to interpret the major topology‐related differences.

The nuts and bolts of core-hole constrainedab initiosimulation forK-shell x-ray photoemission and absorption spectra

X-ray photoemission (XPS) and near edge x-ray absorption fine structure (NEXAFS) spectroscopy play an important role in investigating the structure and electronic structure of materials and surfaces.Ab initiosimulations provide crucial support for the interpretation of complex spectra containing overlapping signatures. Approximate core-hole simulation methods based on density functional theory (DFT) such as the delta-self-consistent-field (ΔSCF) method or the transition potential (TP) method are widely used to predictK-shell XPS and NEXAFS signatures of organic molecules, inorganic materials and metal-organic interfaces at reliable accuracy and affordable computational cost. We present the numerical and technical details of our variants of the ΔSCF and TP method (coined ΔIP-TP) to simulate XPS and NEXAFS transitions. Using exemplary molecules in gas-phase, in bulk crystals, and at metal-organic interfaces, we systematically assess how practical simulation choices affect the stability and accuracy of simulations. These include the choice of exchange-correlation functional, basis set, the method of core-hole localization, and the use of periodic boundary conditions (PBC). We particularly focus on the choice of aperiodic or periodic description of systems and how spurious charge effects in periodic calculations affect the simulation outcomes. For the benefit of practitioners in the field, we discuss sensible default choices, limitations of the methods, and future prospects.

Ultrafast Dynamics of Electronic Resonances in Molecules Adsorbed on Metal Surfaces: A Wave Packet Propagation Approach

We present a wave packet propagation-based method to study the electron dynamics in molecular species in the gas phase and adsorbed on metal surfaces. It is a very general method that can be employed to any system where the electron dynamics is dominated by an active electron and the coupling between the discrete and continuum electronic states is of importance. As an example, one can consider resonant molecule-surface electron transfer or molecular photoionization. Our approach is based on a computational strategy allowing incorporating ab initio inputs from quantum chemistry methods, such as density functional theory, Hartree-Fock, and coupled cluster. Thus, the electronic structure of the molecule is fully taken into account. The electron wave function is represented on a three-dimensional grid in spatial coordinates, and its temporal evolution is obtained from the solution of the time-dependent Schrödinger equation. We illustrate our method with an example of the electron dynamics of anionic states localized on organic molecules adsorbed on metal surfaces. In particular, we study resonant charge transfer from the π* orbitals of three vinyl derivatives (acrylamide, acrylonitrile, and acrolein) adsorbed on a Cu(100) surface. Electron transfer between these lowest unoccupied molecular orbitals and the metal surface is extremely fast, leading to a decay of the population of the molecular anion on the femtosecond timescale. We detail how to analyze the time-dependent electronic wave function in order to obtain the relevant information on the system: the energies and lifetimes of the molecule-localized quasistationary states, their resonant wavefunctions, and the population decay channels. In particular, we demonstrate the effect of the electronic structure of the substrate on the energy and momentum distribution of the hot electrons injected into the metal by the decaying molecular resonance.

Finding the true pathway for reversible isomerization of a single azobenzene molecule tumbling on Au(111) surface

Switchable trans-cis isomerization of azobenzene (AB) and its derivatives on metallic surfaces have offered rich possibilities to functionalize molecular devices. However, the lack of a good understanding of the isomerization pathway has severely limited our ability for rational design. One of the long-debated issues is the cis configuration of the parental AB on the Au(111) surface, for which the experimentally inferred structure differs from the theoretically predicted global minimum. Here, we theoretically identify a new in situ metastable configuration for cis-AB on Au(111) that can reproduce all the observations reported in the scanning tunneling microscopy experiments. It reveals that the bistability of AB on the Au(111) surface is attributed to the significantly increased kinetic stability of the newly discovered cis-AB isomer. A fascinating tumbling pathway that overcomes two energy barriers stimulated by tunneling electrons for the trans-cis AB isomerization on Au(111) has been verified, suggesting a new type of molecular motion based on the AB systems.

Modeling nonadiabatic dynamics in condensed matter materials: some recent advances and applications

This review focuses on recent developments in the field of nonadiabatic molecular dynamics (NA-MD), with particular attention given to condensed-matter systems. NA-MD simulations for small molecular systems can be performed using high-level electronic structure (ES) calculations, methods accounting for the quantization of nuclear motion, and using fewer approximations in the dynamical methodology itself. Modeling condensed-matter systems imposes many limitations on various aspects of NA-MD computations, requiring approximations at various levels of theory-from the ES, to the ways in which the coupling of electrons and nuclei are accounted for. Nonetheless, the approximate treatment of NA-MD in condensed-phase materials has gained a spin lately in many applied studies. A number of advancements of the methodology and computational tools have been undertaken, including general-purpose methods, as well as those tailored to nanoscale and condensed matter systems. This review summarizes such methodological and software developments, puts them into the broader context of existing approaches, and highlights some of the challenges that remain to be solved.

Computational design of metal-supported molecular switches: transient ion formation during light- and electron-induced isomerisation of azobenzene

In molecular nanotechnology, a single molecule is envisioned to act as the basic building block of electronic devices. Such devices may be of special interest for organic photovoltaics, data storage, and smart materials. However, more often than not the molecular function is quenched upon contact with a conducting support. Trial-and-error-based decoupling strategies via molecular functionalisation and change of substrate have in many instances proven to yield unpredictable results. The adsorbate-substrate interactions that govern the function can be understood with the help of first-principles simulation. Employing dispersion-corrected density-functional theory (DFT) and linear expansion delta-self-consistent-field DFT, the electronic structure of a prototypical surface-adsorbed functional molecule, namely azobenzene adsorbed to (1 1 1) single crystal facets of copper, silver and gold, is investigated and the main reasons for the loss or survival of the switching function upon adsorption are identified. The light-induced switching ability of a functionalised derivative of azobenzene on Au(1 1 1) and azobenzene on Ag(1 1 1) and Au(1 1 1) is assessed based on the excited-state potential energy landscapes of their transient molecular ions, which are believed to be the main intermediates of the experimentally observed isomerisation reaction. We provide a rationalisation of the experimentally observed function or lack thereof that connects to the underlying chemistry of the metal-surface interaction and provides insights into general design strategies for complex light-driven reactions at metal surfaces.

Non-adiabatic molecular dynamics with ΔSCF excited states

Accurate modelling of nonadiabatic transitions and electron-phonon interactions in extended systems is essential for understanding the charge and energy transfer in photovoltaic and photocatalytic materials. The extensive computational costs of the advanced excited state methods have stimulated the development of many approximations to study the nonadiabatic molecular dynamics (NA-MD) in solid-state and molecular materials. In this work, we present a novel ▵SCF-NA-MD methodology that aims to account for electron-hole interactions and electron-phonon back-reaction critical in modelling photoinduced nuclear dynamics. The excited states dynamics is described using the delta self-consistent field (▵SCF) technique within the density functional formalism and the trajectory surface hopping. The technique is implemented in the open-source Libra-X package freely available on the Internet (https://github.com/Quantum-Dynamics-Hub/Libra-X). This work illustrates the general utility of the developed ▵SCF-NA-MD methodology by characterizing the excited state energies and lifetimes, reorganization energies, photoisomerization quantum yields, and by providing the mechanistic details of reactive processes in a number of organic molecules.

Spin Manipulation by Creation of Single-Molecule Radical Cations

All-trans-retinoic acid (ReA), a closed-shell organic molecule comprising only C, H, and O atoms, is investigated on a Au(111) substrate using scanning tunneling microscopy and spectroscopy. In dense arrays single ReA molecules are switched to a number of states, three of which carry a localized spin as evidenced by conductance spectroscopy in high magnetic fields. The spin of a single molecule may be reversibly switched on and off without affecting its neighbors. We suggest that ReA on Au is readily converted to a radical by the abstraction of an electron.