Session K59: First Principles Modeling of Excited-State Phenomena in Materials: Materials and Applications

Nonhomogeneous ultrafast dynamics of anti- and ferromagnetic transient metallic states formation in pulse-excited monoclinic vanadium sesquioxide

Several experiments point to the presence of striped monoclinic (insulating) domains in the antiferromagnetic (AFM) phase of V₂O₃. These domains accommodate the birth of metallic nanodroplets that grow after photoexcitation and transform the state of the system into a metallic phase without lattice reconstruction (insulator-to-metal transition). Using the combination of time-dependent density-functional theory and dynamical mean-field theory we analyze the microscopic details of the initial (0-1ps) growth of the domains in V₂O₃ excited by a laser pulse by tracking the spatially resolved population of the excited V d-orbitals. We find that the preferred expansion of the metallic domains is along the “AFM” vanadium hexagonal planes, in which the photoinduced inter-plane hybridization of the d-states plays an important role. We also analyze details of the growth of the FM metallic domains created by a circularly polarized pulse. In both AFM and FM cases we establish the critical initial domain size and critical pulse fluence that results in a ps domain expansion moving the system into a metastable metallic phase. We compare the results with available recent experimental data and discuss possible application of the findings in terahertz electronic and magnetic switching technologies.   

Photogalvanic effect in multiferroic breathing Kagome lattice

Photogalvanic effect, as a second-order nonlinear optical response, can generate an electric current in non-centrosymmetric materials when exposed to light irradiation. In this talk, we investigate monolayer Nb3I8, a recently fabricated two-dimensional multiferroic breathing Kagome material. Our first-principles calculations reveal a remarkable sensitivity of nonlinear photocurrent to variations in the magnetic order in monolayer Nb3I8. Furthermore, nonlinear photocurrent can also be manipulated by switching the electric polarization. In contrast, the linear optical response almost remains unaffected by changes in the magnetic and ferroelectric orders. Our findings offer several avenues to control second-order photocurrent and the potential to utilize multiferroic breathing Kagome materials for magnetic-field sensors.   

Unraveling Optical Pico-polarization of Zinc-Blend Materials

The dynamics of electron waves and their corresponding dispersion in crystalline materials have shaped our understanding of materials science and condensed matter physics. On the contrary, the lattice-level description of optical waves has long been unresolved. Here, we put forth a pico-optical band theory of solids which completely characterizes the dynamics of optical polarization waves within a crystal lattice. We unravel the optical pico-polarization of 14 distinct Group IV, III-V, and II-VI materials. We show that our optical pico-polarization indices generalize the concept of the refractive index of the medium to determine the polarization texture and crowding throughout the crystal lattice. The optical pico-polarization indices serve as symmetry indicators and provide a signature for identifying optical topological phases in natural materials. Our work establishes a foundational crystallographic feature to discover novel pico-optical phases of matter.   

Predicting the color polymorphism of ROY using optimally-tuned screened range-separated hybrid functionals

5-methyl-2-((2-nitrophenyl)amino)thiophene-3-carbonitrile is known to crystallize in various polymorphs that can exhibit different shades of red, orange, or yellow, earning it the nickname ROY. While spectacular, this color polymorphism corresponds to relatively modest spectral changes and is a long-standing challenge for time-dependent density functional theory. Here, we show that the experimentally observed ROY colors can be predicted using non-empirical optimally-tuned screened range-separated hybrid functionals.   

Ab Initio Design Principles for Excited-State Desorption in Heterogeneous Catalysis

Heterogeneous photocatalysis has drawn attention in recent decades due to its ability to dramatically speed up reaction rates, enhance selectivity amongst multiple products, and open new reaction pathways. Despite significant experimental progress, the accurate theoretical predictions about these phenomena are difficult to make due to their chemical complexity. Here, apply many-body perturbation theory (MBPT) to accurately compute excited state potential energy surfaces (PESs) for the proton desorption reaction from a rutile TiO2 (110) surface. We find sensitive dependence of the excited states on the position of the adsorbate, including sharp variations in the excited state PESs and intersections of the ground and excited states. The excited state PESs are qualitatively different from the ground state and greatly favor desorption with activation barriers as low as half the ground state barrier. Through analysis of mean-field and excited-state wavefunctions, we deduce general design principles for desorption reactions in excited-state catalysis. This work demonstrates the applicability of MBPT to the study of heterogeneous photocatalysts, and provides the framework to rigorously analyze reaction dynamics in the excited state from first principles.   

Using ground-state and excited-state DFT to decipher 3d dopant defects in GaN

Predictive calculations of electronic excitations do not (not always) require theoretical frameworks that go beyond ground-state density functional theory (DFT). Using ground-state together with excited-state DFT in size-converged supercells, we decipher 3d defects in GaN from limited experimental data for defect levels and more common optical data probing excited states. Applying a local moment counter charge (LMCC) approach to avoid jellium-neutralization errors, defect levels in GaN:3d do not suffer a band gap problem and accurately predict observed levels. For self-consistent calculations of excited states, we implement an occupation-constrained DFT (occDFT) in a ground-state DFT code. The occDFT approach, used with a standard functional PBE, yields optical transitions quantitatively comparable (0.1-0.2 eV) to measurements. As a specific example, PBE-LMCC/occDFT predicts 1.28 eV (adiabatic) for the observed GaN:Mn(0) 1.42 eV photoabsorption. A partnered ground-state/excited-state analysis enables more confident defect identification—chemical fingerprinting. The results mandate extensive reinterpretation of experiments, e.g., the long-standing assignment of the 1.19 eV photoluminescence to Cr(1+) is incorrect.  We predict design of alternate d² centers that are possible candidates for optically controlled quantum applications. A simple, fast occDFT approach is often effective and accurate for excited states in wide band gap systems.   

Many-body effects on the ground-state properties of monolayer 1T' WTe2

In the monolayer limit, 1T' WTe₂ is a two-dimensional (2D) topological insulator exhibiting the quantum spin Hall effect and is believed to host an exciton insulator ground state. However, theoretical analysis is complicated by the difficulty of obtaining consistent descriptions of the single-quasiparticle band structure within conventional first-principles techniques. Previous band structure calculations using the Perdew-Burke-Ernzerhof (PBE) functional and a one-shot GW approximation result in a semimetallic band structure, while calculations with hybrid functionals appear to open a bandgap. Here, we demonstrate that self-consistently updating wavefunctions within a static GW approximation (static COHSEX) can reproduce the band structure experimentally observed by angle-resolved photoemission spectroscopy (ARPES) without resorting to mechanisms beyond the quasiparticle picture. Finally, Bethe–Salpeter equation (BSE) calculation on top of self-consistent GW hint at previously observed negative exciton frequencies, leaving open the possibility of exciton condensation in 1T' monolayer WTe₂.   

Band structure and excitonic properties of monolayer WSe2 in an all-electron QSGW^ approach

In this work we study the electronic band structure and optical absorption spectrum of monolayer WSe₂ using an all-electron quasiparticle self-consistent GW approach, QSGW^, in which the screened Coulomb interaction W^ is calculated by including ladder diagrams representing electron-hole interactions. We employ the Bethe-Salpeter equation to calculate the screened Coulomb interaction W^ both in the quasiparticle band structure and in the macroscopic dielectric function. The convergence of the quasiparticle band gap and lowest optical peak position is studied as a function of the spatial separation of the monolayers when using periodic boundary conditions. The quasiparticle gap is found to scale with 1/L where L is the size of the vacuum separation, while the lowest peak reaches convergence much faster. In addition, the Bethe-Salpeter equation provides us with information on the excitonic spectrum. We focus on the first few excitonic states, determining their brightness and further analyzing their wave vectors projected onto momentum and real space, and compare them to the hydrogenic model.   

Probabilistic forecast of nonlinear dynamical systems with uncertainty quantification

Data-driven modeling is useful for reconstructing nonlinear dynamical systems when the underlying process is unknown or too expensive to compute. Having reliable uncertainty assessment of the forecast enables tools to be deployed to predict new scenarios unobserved before. In this work, we first extend parallel partial Gaussian processes for predicting the vector-valued transition function that links the observations between the current and next time points, and quantify the uncertainty of predictions by posterior sampling. Second, we show the equivalence between the dynamic mode decomposition and the maximum likelihood estimator of the linear mapping matrix in the linear state space model. The connection provides a probabilistic generative model of dynamic mode decomposition and thus, uncertainty of predictions can be obtained. Furthermore, we draw close connections between different data-driven models for approximating nonlinear dynamics, through a unified view of generative models. We study a few numerical examples, including the time-dependent adiabatic GW (TD-aGW) method for understanding quantum many-body systems far from equilibrium, and Lorenz 96 model for simulating chaotic behaviors in nonlinear dynamical systems. The examples indicate that uncertainty of forecast can be properly quantified, whereas model or input misspecification can degrade the accuracy of uncertainty quantification.   

Breaking barriers: A machine learning approach to efficiently explore the free energy surface of protein-surface systems

Understanding protein-surface interactions is crucial for the rational design of biotechnologies such as medical implants, biological sensors, and drug-delivery vehicles. Molecular dynamics (MD) simulations are excellent for this task due to their atomistic resolution, but the sizable free energy barriers between stable adsorbed protein conformations can limit sampling. To address this challenge, a framework developed by Wei et. al, titled Molecular Enhanced Sampling with Autoencoders (MESA), accelerates the sampling of protein conformations in isotropic environments (e.g., bulk solvent) using a machine learning architecture called an autoencoder. The model learns low-dimensional representations of the configurations sampled in MD simulations and directs future simulations to sample unexplored regions of the configuration space. A challenge in directly implementing this method to surface systems is that the model cannot distinguish conformations based on the orientation of the protein with the surface. In this work, we extend the MESA framework to efficiently sample adsorbed protein conformations through a modified training procedure and simulation protocol. We demonstrate the approach through studies of peptides on graphene and silica surfaces. We identify the various molecular driving forces governing peptide-surface interactions. Collectively, our method highlights the power of combining ML with molecular simulations in studying biomolecular systems near surfaces.   

A Novel Machine Learning Framework for More Accurate Coarse Grained Free Energy Models

Coarse graining is an essential tool for computational materials science. In systems with long time- and length-scale dynamics, standard all-atom resolution methodologies often become too expensive. State of the art approaches to coarse graining are bottom-up techniques that target the reproduction of a coordinate-dependent free energy surface, the potential of mean force (PMF). Despite their appeal for their ability to accurately reproduce thermodynamics of the all-atom system, accurately capturing all structural correlations and other thermodynamic behavior remains a challenge.

In this work, we address this issue by introducing a new conceptual framework for designing coarse grained models with machine learning. We review our recently developed on-the-fly active learning framework for building coarse grained potentials, which is an important starting point for the demonstration of our new methods. We demonstrate the benefit and validity of our novel techniques in the context of learning relative energies, the generalizability to multiple machine learning architectures, and discuss future opportunities for studying coarse grained models enabled by these methods.    

Bayesian and Equivariant Force Fields for the Description of Metals in their Bulk, Surface, and Nano-Scale Forms

Understanding atomic-level processes in surface science and heterogeneous catalysis is complicated by the wide range of time scales and length scales needed for simulations. To accelerate molecular dynamics calculations, we rely on machine learning methods to capture interatomic interactions with quantum accuracy. We then implement and deploy these models on parallel GPUs to reach billions of atoms in size or microseconds in time. One method is a family of equivariant interatomic potential models (NequIP and Allegro) based on symmetry-preserving layer architectures that we use to achieve state-of-the-art accuracy and training efficiency for simulating atomistic. Another method (FLARE) enables autonomous selection of training sets for reactive systems, based on adaptive closed-loop algorithm that constructs accurate and uncertainty-aware Bayesian force fields on-the-fly from a molecular dynamics simulation. We examine the current limitations of machine learning models and highlight the usefulness of ML-accelerated MD simulations to study dislocation dynamics, surface reconstructions, direct heterogeneous reactions, and nanoparticle shape-changes.