Machine learning and atomistic origin of high dielectric permittivity in oxides.

Discovering new stable materials with large dielectric permittivity is important for future energy storage and electronics applications. Theoretical and computational approaches help design new materials by elucidating microscopic mechanisms and establishing structure-property relations. Ab initio methods can be used to reliably predict the dielectric response, but for fast materials screening, machine learning (ML) approaches, which can directly infer properties from the structural information, are needed. Here, random forest and graph convolutional neural network models are trained and tested to predict the dielectric constant from the structural information. We create a database of the dielectric properties of oxides and design, train, and test the two ML models. Both approaches show similar performance and can successfully predict response based on the structure. The analysis of the feature importance allows identification of local geometric features leading to the high dielectric permittivity of the crystal. Dimensionality reduction and clustering further confirms the relevance of descriptors and compositional features for obtaining high dielectric permittivity.

Polymorphs of Titanium Dioxide: An Assessment of the Variants of Projector Augmented Wave Potential of Titanium on Their Geometric and Dielectric Properties.

Titanium dioxide (TiO2) is one of the important functional materials owing to its diverse applications in many fields of chemistry, physics, nanoscience, and technology. Hundreds of studies on its physicochemical properties, including its various phases, have been reported experimentally and theoretically, but the controversial nature of relative dielectric permittivity of TiO2 is yet to be understood. Toward this end, this study was undertaken to rationalize the effects of three commonly used projector augmented wave (PAW) potentials on the lattice geometries, phonon vibrations, and dielectric constants of rutile (R-)TiO2 and four of its other phases (anatase, brookite, pyrite, and fluorite). Density functional theory calculations within the PBE and PBEsol levels, as well as their reinforced versions PBE+U and PBEsol+U (U = 3.0 eV), were performed. It was found that PBEsol in combination with the standard PAW potential centered on Ti is adequate to reproduce the experimental lattice parameters, optical phonon modes, and the ionic and electronic contributions of the relative dielectric permittivity of R-TiO2 and four other phases. The origin of failure of the two soft potentials, namely, Ti_pv and Ti_sv, in predicting the correct nature of low-frequency optical phonon modes and ion-clamped dielectric constant of R-TiO2 is discussed. It is shown that the hybrid functionals (HSEsol and HSE06) slightly improve the accuracy of the above characteristics at the cost of a significant increase in computation time. Finally, we have highlighted the influence of external hydrostatic pressure on the R-TiO2 lattice, leading to the manifestation of ferroelectric modes that play a role in the determination of large and strongly pressure-dependent dielectric constant.

Challenges for fluoride superionic conductors: fundamentals, design, and applications.

Electronics, which harnesses the properties of electrons, has made remarkable progress since its inception and is a cornerstone of modern society. Ionics, which exploits the properties of ions, has also had a profound impact, as demonstrated by the award of the Nobel Prize in Chemistry in 2019 for achievements related to lithium-ion batteries (LIBs). Ionic conduction in solids is the flow of carrier ions through a solid owing to an electrical or chemical bias. Some ionic materials have been studied intensively because their ionic conductivities are higher than those of liquids, even though they are solids. Among various conductive species, fluoride ions are the most promising charge carriers for fluoride-ion batteries (FIBs) as post LIBs. Increasing fluoride-ion conductivity toward the superionic conductive region at room temperature would be a breakthrough for the room-temperature operation of all-solid-state FIBs. This review focuses on fluoride-ion conductors, from the general concept of ions to the characteristics of fluoride ions. Fluoride-ion conductors are classified according to material type and form, and our current understanding, identification of problems, and future directions are discussed from experimental and theoretical physics perspectives.

Permittivity boosting by induced strain from local doping in titanates from first principles.

We examine the effect of isovalent substitutions and co-doping on the ionic dielectric constant of paraelectric titanates (perovskite, Ruddlesden-Popper phases, and rutile) using density functional perturbation theory. Substitutions increase the ionic dielectric constant of the prototype structures, and new dynamically stable structures with εion ~ 102-104 are reported and analyzed. The boosting of ionic permittivity is attributed to local defect-induced strain, and maximum Ti-O bond length is proposed as a descriptor. The Ti-O phonon mode that is responsible for the large dielectric constant can be tuned by a local strain and symmetry lowering from substitutions. Our findings help explain the recently observed colossal permittivity in co-doped rutile, attributing its intrinsic permittivity boosting solely to the lattice polarization mechanism, without the need to invoke other mechanisms. Finally, we identify new perovskite- and rutile-based systems that can potentially display colossal permittivity.

Theoretical and Experimental Studies of KLi6TaO6 as a Li-Ion Solid Electrolyte.

We study a hexagonal oxide KLi6TaO6 (KLTO), proposed as a Li-ion solid electrolyte, by using a recently developed screening method. First-principles calculations predict that KLTO presents a good Li-ion conductivity (σLi) and a low activation energy (Ea). Li migration is enhanced by the presence of excess Li ions in the interstitial region via a kick-out mechanism. Our experimental results demonstrate that Sn-doped KLTO presents a conductivity of 1 × 10-5 S cm-1, a σLi of 6 × 10-6 S cm-1, and a relatively low Ea of 36 kJ mol-1, which confirm the validity of the proposed screening method. Conversely, detailed analyses of the microstructure and X-ray diffraction patterns of KLTO samples indicate that a stable Li-excess condition is not achieved, therefore leaving potential improvement of the performance of KLTO as a Li-ion solid electrolyte by optimizing extrinsic doping and fabrication processes.

On-the-Fly Active Learning of Interatomic Potentials for Large-Scale Atomistic Simulations.

The on-the-fly generation of machine-learning force fields by active-learning schemes attracts a great deal of attention in the community of atomistic simulations. The algorithms allow the machine to self-learn an interatomic potential and construct machine-learned models on the fly during simulations. State-of-the-art query strategies allow the machine to judge whether new structures are out of the training data set or not. Only when the machine judges the necessity of updating the data set with the new structures are first-principles calculations carried out. Otherwise, the yet available machine-learned model is used to update the atomic positions. In this manner, most of the first-principles calculations are bypassed during training, and overall, simulations are accelerated by several orders of magnitude while retaining almost first-principles accuracy. In this Perspective, after describing essential components of the active-learning algorithms, we demonstrate the power of the schemes by presenting recent applications.

Charge Trapping Process in Photoexcited Nitrogen-Doped Titanium Oxides.

We present a first-principles study on the structural changes induced by charge trapping that occurs after photoexcitation in nitrogen-doped titanium oxide (N-TiO2). The charge trapping site and the corresponding K edge EXAFS spectra of Ti atoms were predicted and compared with those obtained by an experiment under ultraviolet (UV) light excitation. The results indicate that charge trapping occurs in the neighborhood of the oxygen vacancy (O-vac) sites. Furthermore, our calculations show that the O-vac site significantly affects the EXAFS spectra, while substitutional nitrogen doping for an oxygen site in the vicinity of the O-vac site is insensitive in the EXAFS spectra. Based on this observation combined with the knowledge from previous experiments, we propose a charge trapping process where the UV light-excited electron migrates at the O-vac site in bulk (∼300 ps) while the visible light-excited electron (N 2p → Ti 3d) is immediately trapped at the O-vac site neighboring the N site (∼1 ps).

Descriptors representing two- and three-body atomic distributions and their effects on the accuracy of machine-learned inter-atomic potentials.

When determining machine-learning models for inter-atomic potentials, the potential energy surface is often described as a non-linear function of descriptors representing two- and three-body atomic distribution functions. It is not obvious how the choice of the descriptors affects the efficiency of the training and the accuracy of the final machine-learned model. In this work, we formulate an efficient method to calculate descriptors that can separately represent two- and three-body atomic distribution functions, and we examine the effects of including only two- or three-body descriptors, as well as including both, in the regression model. Our study indicates that non-linear mixing of two- and three-body descriptors is essential for an efficient training and a high accuracy of the final machine-learned model. The efficiency can be further improved by weighting the two-body descriptors more strongly. We furthermore examine a sparsification of the three-body descriptors. The three-body descriptors usually provide redundant representations of the atomistic structure, and the number of descriptors can be significantly reduced without loss of accuracy by applying an automatic sparsification using a principal component analysis. Visualization of the reduced descriptors using three-body distribution functions in real-space indicates that the sparsification automatically removes the components that are less significant for describing the distribution function.

Highly efficient evaluation of diffusion networks in Li ionic conductors using a 3D-corrugation descriptor.

A highly efficient computational approach for the screening of Li ion conducting materials is presented and its performance is demonstrated for olivine-type oxides and thiophosphates. The approach is based on a topological analysis of the electrostatic (Coulomb) potential obtained from a single density functional theory calculation augmented by a Born-Mayer-type repulsive term between Li ions and the anions of the material. This 3D-corrugation descriptor enables the automatic determination of diffusion pathways in one, two, and three dimensions and reproduces migration barriers obtained from density functional theory calculations using nudged elastic band method within approximately 0.1 eV. Importantly, it correlates with Li ion conductivity. This approach thus offers an efficient tool for evaluating, ranking, and optimizing materials with high Li-ion conductivity.

Phosphorescent Material Search Using a Combination of High-Throughput Evaluation and Machine Learning.

High-throughput experiments including combinatorial chemistry are useful for generating large amounts of data within a short period of time. Machine learning can be used to predict the regularity of a response variable using a statistical model of a data set. Because a combination of these methods can accelerate the material development, we applied such a combination to a search of semiconducting thin films prepared on an Eu and Dy codoped SrAl2O4-based phosphorescent material to improve the lifetime of its afterglow. Oxide targets MgO, GeO2, Ga2O3, ZnO, Bi2O3, Ta2O5, TiO2, and Y2O3 were deposited to form a thin film on a SrAl2O4 substrate as a combinatorial library with a systematical change in these ratios. The sample was calcined under several conditions, and a data set of 800 examples was obtained using a high-throughput evaluation. The 800 examples were then randomly divided into training and test data sets. The lifetime of the afterglow was interpolated through machine learning using the film thickness of each element and the calcined condition of the training data set as explanatory variables. The accuracy of the interpolation was evaluated using a correlation coefficient and the root mean squared error of the predicted values with respect to the experimental values of the test data set. As a result, it was found that a MgO thin film is effective at improving the lifetime of the afterglow and that its optimum condition is a film thickness of approximately 100 nm with calcination at 400-600 °C in air.

First-principles prediction of high oxygen-ion conductivity in trilanthanide gallates Ln3GaO6.

We systematically investigated trilanthanide gallates (Ln3GaO6) with the space group Cmc21 as oxygen-ion conductors using first-principles calculations. Six Ln3GaO6 (Ln = Nd, Gd, Tb, Ho, Dy, or Er) are both energetically and dynamically stable among 15 Ln3GaO6 compounds, which is consistent with previous experimental studies reporting successful syntheses of single phases. La3GaO6 and Lu3GaO6 may be metastable despite a slightly higher energy than those of competing reference states, as phonon calculations predict them to be dynamically stable. The formation and the migration barrier energies of an oxygen vacancy (V O) suggest that eight Ln3GaO6 (Ln = La, Nd, Gd, Tb, Ho, Dy, Er, or Lu) can act as oxygen-ion conductors based on V O. Ga plays a role of decreasing the distances between the oxygen sites of Ln3GaO6 compared with those of Ln2O3 so that a V O migrates easier with a reduced migration barrier energy. Larger oxygen-ion diffusivities and lower migration barrier energies of V O for the eight Ln3GaO6 are obtained for smaller atomic numbers of Ln having larger radii of Ln3+. Their oxygen-ion conductivities at 1000 K are predicted to have a similar order of magnitude to that of yttria-stabilized zirconia.

Development of a High-Throughput Screening Method for Oxide-Ion Conductors and Its Application to Bismuth-Based Oxide Library Thin Films.

To accelerate material discovery, we develop a screening method for oxide-ion conductors that comprises combinatorial synthesis using chemical-solution deposition and high-throughput measurements using X-ray diffraction and conductivity. The present method allows us to form an arbitrary and uniform composition within an evaluation area at an arbitrary position in the library on a substrate. This screening method is applied to ABi2Zr x(Nb1- yTa y)1- xO9 bismuth-layered compounds, which are known to have relatively high oxide-ion conductivities but are yet to be examined thoroughly. By making systematic thin-film libraries for A = Sr or Ca, we aim to find the optimized composition. The total time required for synthesis, phase identification, and conductivity measurements is found to be significantly shorter than that with the conventional method, and the maximum oxide-ion conductivity of this compound in the libraries reaches 10-3 S/cm at 800 °C.

Oxygen conduction mechanism in Ca3Fe2Ge3O12 garnet-type oxide.

We investigate the oxygen conduction mechanism in a garnet-type oxide, Ca3Fe2Ge3O12, for the first time in detail by first-principle calculations. The nudged elastic band results confirm that this oxide has a lower migration barrier energy (0.45 eV) for an oxygen interstitial (Oi) with the kick-out mechanism than that (0.76 eV) for an oxygen vacancy. The migration paths for Oi are delocalized and connected to the neighboring cells in three-dimensional space. This oxide does not have a very low formation energy of Oi when the Fermi level is near the lowest unoccupied molecular orbital at a high temperature, which implies the possibility of electron doping by high-valence cations. These theoretical results suggest that the doping of Ca3Fe2Ge3O12 for generation of excess Oi provides a good oxygen-ion conductivity, along with the electronic conductivity.

Charge transfer dynamics at the boron subphthalocyanine chloride/C60 interface: non-adiabatic dynamics study with Libra-X.

We report a study on the non-adiabatic molecular dynamics (NA-MD) of the charge transfer (CT) process in the boron subphtalocyanine chloride (SubPc)/fullerene (C60) interface using our newly implemented Libra-X software package, which is based on an interface of the Libra NA-MD library and the GAMESS electronic structure software. In particular, we address the following aspects of the simulation protocol: (a) the choice of the potential used to treat interatomic interactions and its effect on the structures of the complex and CT rates; (b) the choice of the electronic structure methodology used; and (c) the choice of the trajectory surface hopping (TSH) methodology used. From our analysis of the electronic structure, we suggest that the distortion of the SubPc conical structure affects orbital localization and that the "breathing" motion of SubPc drives the CT process in SubPc/C60. This study illustrates that the choice of the TSH methodology and electronic decoherence are crucial for the CT simulation. We extend our analysis of CT in SubPc/(C60)n models by increasing the number of C60 molecules up to n = 4. We find that the details of the interfacial SubPc/(C60)n geometry determine the CT rate. Finally, we find the computed CT timescale to be in the range of 2.2-5.0 ps, which is in agreement with the experimentally determined timescale in the order of magnitude of ∼10 ps. The developed open-source Libra-X package is freely available on the Internet at https://github.com/Quantum-Dynamics-Hub/Libra-X.

Band bending and dipole effect at interface of metal-nanoparticles and TiO2 directly observed by angular-resolved hard X-ray photoemission spectroscopy.

This paper describes the observation of band bending and band edge shifts at the interfaces between nanoscale metals and TiO2 film over a wide depth range by angular-resolved hard X-ray photoemission spectroscopy (HAXPES). The HAXPES results indicate strong electrostatic interactions between the TiO2 semiconductor and metal nanoparticles, while density functional theory (DFT) calculations suggest that these interactions are primarily associated with charge transfer leading to electric dipole moments at the interface in the ground state. The effects of these dipole moments are not limited to the surface but also occur deep in the bulk of the semiconductor, and are highly dependent on the coverage of the metal nanoparticles on the semiconductor species.

Discovery of zirconium dioxides for the design of better oxygen-ion conductors using efficient algorithms beyond data mining.

It is important to find crystal structures with low formation (E v) and migration-barrier (E m) energies for oxygen vacancies for the development of fast oxygen-ion conductors. To identify crystal structures with lower E v and E m than those of ground-state ZrO2, we first reoptimize the crystal structures of various oxides reported in the database, and then directly construct them using an evolutionary algorithm. For efficient searching, we employ the linearized ridge regression model for E v using descriptors obtained from density functional theory calculations of the unit cells and apply the predicted E v as a fitness value in the evolutionary algorithm. We also find a correlation between the E v and E m for the crystal structures of ZrO2. On the basis of this correlation, we confirm that the newly constructed crystal structures, as well as certain reoptimized structures from the database, that possess low E v also have E m lower than that of ground-state ZrO2. Our successful strategy consisting of a combination of the evolutionary algorithm, first-principles calculations, and machine-learning techniques may be applicable to other oxide systems in finding crystal structures with low E v and E m.

A Universal 3D Voxel Descriptor for Solid-State Material Informatics with Deep Convolutional Neural Networks.

Material informatics (MI) is a promising approach to liberate us from the time-consuming Edisonian (trial and error) process for material discoveries, driven by machine-learning algorithms. Several descriptors, which are encoded material features to feed computers, were proposed in the last few decades. Especially to solid systems, however, their insufficient representations of three dimensionality of field quantities such as electron distributions and local potentials have critically hindered broad and practical successes of the solid-state MI. We develop a simple, generic 3D voxel descriptor that compacts any field quantities, in such a suitable way to implement convolutional neural networks (CNNs). We examine the 3D voxel descriptor encoded from the electron distribution by a regression test with 680 oxides data. The present scheme outperforms other existing descriptors in the prediction of Hartree energies that are significantly relevant to the long-wavelength distribution of the valence electrons. The results indicate that this scheme can forecast any functionals of field quantities just by learning sufficient amount of data, if there is an explicit correlation between the target properties and field quantities. This 3D descriptor opens a way to import prominent CNNs-based algorithms of supervised, semi-supervised and reinforcement learnings into the solid-state MI.

Study of the conformation of polyelectrolyte aggregates using coarse-grained molecular dynamics simulations.

The conformation of polyelectrolyte aggregates as a function of the backbone rigidity is investigated by coarse-grained molecular dynamics simulation. The polyelectrolyte is represented by a bead-spring chain with charged side chains. The simulations start from the uniform distributions of the polyelectrolytes, and the resultant polyelectrolyte conformation after a few microseconds exhibits spherical self-aggregates, clusters, or bending bundle-like aggregates, depending on the backbone rigidity. The interaggregate structures on a large scale are featured by the static structure factor (SSF). The simulated SSFs of the bending bundle-like aggregates are consistent with those of the small angle X-ray scattering (SAXS) measurement so we successfully assign the microscopic structures of polyelectrolytes to the SAXS measurement. The power-law of the SSFs for the bundle conditions is steeper than that of the conventional cylinder model. The present study finds that such discrepancy in the power-law results from the bending of the bundle-like aggregates. In addition, the relaxation behavior includes slow dynamics. The present study proposes that such slow dynamics results from diffusion-limited aggregation and from gliding processes to reduce local metastable folding within the aggregates.

Predicting Catalytic Activity of Nanoparticles by a DFT-Aided Machine-Learning Algorithm.

Catalytic activities are often dominated by a few specific surface sites, and designing active sites is the key to realize high-performance heterogeneous catalysts. The great triumphs of modern surface science lead to reproduce catalytic reaction rates by modeling the arrangement of surface atoms with well-defined single-crystal surfaces. However, this method has limitations in the case for highly inhomogeneous atomic configurations such as on alloy nanoparticles with atomic-scale defects, where the arrangement cannot be decomposed into single crystals. Here, we propose a universal machine-learning scheme using a local similarity kernel, which allows interrogation of catalytic activities based on local atomic configurations. We then apply it to direct NO decomposition on RhAu alloy nanoparticles. The proposed method can efficiently predict energetics of catalytic reactions on nanoparticles using DFT data on single crystals, and its combination with kinetic analysis can provide detailed information on structures of active sites and size- and composition-dependent catalytic activities.

What Makes the Photocatalytic CO2 Reduction on N-Doped Ta2O5 Efficient: Insights from Nonadiabatic Molecular Dynamics.

Recent experimental studies demonstrated that photocatalytic CO2 reduction by Ru catalysts assembled on N-doped Ta2O5 surface is strongly dependent on the nature of the anchor group with which the Ru complexes are attached to the substrate. We report a comprehensive atomistic analysis of electron transfer dynamics in electroneutral Ru(di-X-bpy) (CO)2Cl2 complexes with X = COOH and PO3H2 attached to the N-Ta2O5 substrate. Nonadiabatic molecular dynamics simulations indicate that the electron transfer is faster in complexes with COOH anchors than in complexes with PO3H2 groups, due to larger nonadiabatic coupling. Quantum coherence counteracts this effect, however, to a small extent. The COOH anchor promotes the transfer with significantly higher frequency modes than PO3H2, due to both lighter atoms (C vs P) and stronger bonds (double vs single). The acceptor state delocalizes onto COOH, but not PO3H2, further favoring electron transfer in the COOH system. At the same time, the COOH anchor is prone to decomposition, in contrast to PO3H2, making the former show smaller turnover numbers in some cases. These theoretical predictions are consistent with recent experimental results, legitimating the proposed mechanism of the electron transfer. We emphasize the role of anchor stability, nonadiabatic coupling, and quantum coherence in determining the overall efficiency of artificial photocatalytic systems.