npj Computational Materials最新文献

Si and its oxides have been extensively explored in theoretical research due to their technological importance. Simultaneously describing interatomic interactions within both Si and SiO₂ without the use of ab initio methods is considered challenging, given the charge transfers involved. Herein, this challenge is overcome by developing a unified machine learning interatomic potentials describing the Si/SiO₂/O system, based on the moment tensor potential (MTP) framework. This MTP is trained using a comprehensive database generated using density functional theory simulations, encompassing diverse crystal structures, point defects, extended defects, and disordered structure. Extensive testing of the MTP is performed, indicating it can describe static and dynamic features of very diverse Si, O, and SiO₂ atomic structures with a degree of fidelity approaching that of DFT.

As analogies to magnetic skyrmions, polar skyrmions in ferroelectric superlattices and multilayers have garnered widespread attention for their non-trivial topology and novel properties like negative capacitance and nonlinear optical effect. So far, they have only been theoretically predicted to be able to assemble ordered hexagonal skyrmion lattices (SkLs) in ferroelectric thin films. Here, based on phase-field simulations, we report the critical roles of flexoelectricity playing in the stabilization and transformation of polar SkLs. Different polar SkL patterns can emerge in the ferroelectric thin films, including tetragonal-SkL, and hexagonal-SkLs with diverse orientations, as summarized by phase diagrams. These emergent SkL states are attributed to the material anisotropy modified by the flexoelectric effect. Interestingly, we further found that the hexagonal-SkLs can be rotated by applying strain gradient or in-plane electric field to the films. Moreover, a nonreciprocal bending response of tetragonal-SkL is also induced by the flexoelectric effect. Our results provide useful guidelines for the implementation of polar skyrmion lattices in experiments.

Transition metal d-electron oxides with an odd number of electrons per unit cell are expected to form metals with partially occupied energy bands, but exhibit in fact a range of behaviors, being either insulators, or metals, or having insulator-metal transitions. Traditional explanations involved predominantly electron-electron interactions in fixed structural symmetry. The present work focuses instead on the role of symmetry breaking local structural motifs. Viewing the previously observed V-V dimerization in VO₂ as a continuous knob, reveals in density functional calculations the splitting of an isolated flat band from the broad conduction band. This leads past a critical percent dimerization to the formation of the insulating phase while lowering the total energy. In VO₂ this transition is found to have a rather low energy barrier approaching the thermal energy at room temperature, suggesting energy-efficient switching in neuromorphic computing. Interestingly, sufficient V-V dimerization suppresses magnetism, leading to the nonmagnetic insulating state, whereas magnetism appears when dimerization is reduced, forming a metallic state. This study opens the way to design novel functional quantum materials with symmetry breaking-induced flat bands.

Knowing the rate at which particle radiation releases energy in a material, the “stopping power,” is key to designing nuclear reactors, medical treatments, semiconductor and quantum materials, and many other technologies. While the nuclear contribution to stopping power, i.e., elastic scattering between atoms, is well understood in the literature, the route for gathering data on the electronic contribution has for decades remained costly and reliant on many simplifying assumptions, including that materials are isotropic. We establish a method that combines time-dependent density functional theory (TDDFT) and machine learning to reduce the time to assess new materials to hours on a supercomputer and provide valuable data on how atomic details influence electronic stopping. Our approach uses TDDFT to compute the electronic stopping from first principles in several directions and then machine learning to interpolate to other directions at a cost of 10 million times fewer core-hours. We demonstrate the combined approach in a study of proton irradiation in aluminum and employ it to predict how the depth of maximum energy deposition, the “Bragg Peak,” varies depending on the incident angle—a quantity otherwise inaccessible to modelers and far outside the scales of quantum mechanical simulations. The lack of any experimental information requirement makes our method applicable to most materials, and its speed makes it a prime candidate for enabling quantum-to-continuum models of radiation damage. The prospect of reusing valuable TDDFT data for training the model makes our approach appealing for applications in the age of materials data science.

The frontier molecular orbitals of organic semiconductor materials play a crucial role in the performance of photoelectric devices, including organic photovoltaics (OPVs), organic light-emitting diodes (OLEDs), and organic photodetectors (OPDs). In this work, a model for predicting frontier molecular orbital of organic materials, including HOMO and LUMO levels, is established with the extreme gradient boosting algorithm and Klekota-Roth fingerprints. The correlation coefficients of HOMO or LUMO energy levels in the testing set are 0.75 and 0.84 in the transfer model from 11,626 DFT data in Harvard Energy database to 1198 experimental data in literature. The difference between the ML predicted value and the experimental value is smaller than the difference between ML prediction and DFT calculation, always less than 10%. Moreover, based on correlation and SHAP interpretability analysis, 13 key structural fragments influencing energy levels are selected to further verify the effective regulation of the frontier molecular orbital by the key structural fragments in practical applications. Considering the completely opposite regulatory functions of key structural fragments on HOMO and LUMO energy levels, four new Y6 derivatives, Y-PCP, Y-P6F, Y-PCF, and Y-P4FC, are designed to flexibly modify the HOMO and LUMO energy levels. The prediction trends of ML align closely with the computational trends from DFT. It is worth noting that the accuracy of LUMO energy level prediction by the prediction model makes up for the instability of DFT calculation on LUMO energy level. This work offers a cost-effective method to accelerate the acquisition of electronic properties of organic materials.

The rapid advancement of machine learning (ML) technology across diverse domains has provided a framework for discovering and rationalising materials and photovoltaic devices. This study introduces a five-step methodology for implementing ML models in fabricating hole transport layer (HTL) free carbon-based PSCs (C-PSC). Our approach leverages various prevalent ML models, and we curated a comprehensive dataset of 700 data points using SCAPS-1D simulation, encompassing variations in the thickness of the electron transport layer (ETL) and perovskite layers, along with bandgap characteristics. Our results indicate that the ANN-based ML model exhibits superior predictive accuracy for C-PSC device parameters, achieving a low root mean square error (RMSE) of 0.028 and a high R-squared value of 0.954. The novelty of this work lies in its systematic use of ML to streamline the optimisation process, reducing the reliance on traditional trial-and-error methods and providing a deeper understanding of the interdependence of key device parameters.

We report a novel deep learning (DL) method for classifying inorganic compounds using 3D electron density data. We transform Density Functional Theory (DFT)-derived CHGCAR files from the Materials Project (MP) and experimental data from the Inorganic Crystal Structure Database (ICSD) into point clouds and sparse tensors, optimized for use in DL models such as PointNet and Sparse 3D CNN. This approach effectively overcomes the limitations of handling the dense 3D data, a common challenge in DL. Contrasting with traditional 1D or 2D X-ray diffraction (XRD) patterns that necessitate complex reciprocal space analysis, our method utilizes 3D density data for direct interpretation in real lattice space. This shift significantly enhances classification accuracy, outperforming traditional XRD-driven DL methods. We achieve accuracies of 97.28%, 90.77%, and 90.10% for crystal system, extinction group, and space group classifications, respectively. Our 3D electron density-based DL approach not only showcases improved accuracy but also contributes a more intuitive and effective framework for materials discovery.

Atomistic simulations are crucial for predicting material properties and understanding phase stability, essential for materials selection and development. However, the high computational cost of density functional theory calculations challenges the design of materials with complex structures and composition. This study introduces new data acquisition strategies using Bayesian-Gaussian optimization that efficiently integrate the geometry of the convex hull to optimize the yield of batch experiments. We developed uncertainty-based acquisition functions to prioritize the computation tasks of configurations of multi-component alloys, enhancing our ability to identify the ground-state line. Our methods were validated across diverse materials systems including Co-Ni alloys, Zr-O compounds, Ni-Al-Cr ternary alloys, and a planar defect system in intermetallic (Ni_1−x, Co_x)₃Al. Compared to traditional genetic algorithms, our strategies reduce training parameters and user interaction, cutting the number of experiments needed to accurately determine the ground-state line by over 30%. These approaches can be expanded to multi-component systems and integrated with cost functions to further optimize experimental designs.

Charge density waves (CDWs) in transition metal dichalcogenides are the subject of growing scientific interest due to their rich interplay with exotic phases of matter and their potential technological applications. Here, using density functional theory with advanced meta-generalized gradient approximations (meta-GGAs) and linear response time-dependent density functional theory (TDDFT) with state-of-the-art exchange-correlation kernels, we investigate the electronic, vibrational, and optical properties in 1T-TiSe₂ with and without CDW. In both bulk and monolayer TiSe₂, the electronic bands and phonon dispersions in either normal or CDW (semiconducting) phase are described well via meta-GGAs, which separate the valence and conduction bands just as HSE06 does but with significantly more computational feasibility. The experimentally observed humps of electron energy loss spectroscopy are successfully reproduced in TDDFT. Our work opens the door to simulating these complexities in CDW compounds from first principles by revealing meta-GGAs as an accurate low-cost alternative to HSE06.

Powder crystallography is the experimental science of determining the structure of molecules provided in crystalline-powder form, by analyzing their x-ray diffraction (XRD) patterns. Since many materials are readily available as crystalline powder, powder crystallography is of growing usefulness to many fields. However, powder crystallography does not have an analytically known solution, and therefore the structural inference typically involves a laborious process of iterative design, structural refinement, and domain knowledge of skilled experts. A key obstacle to fully automating the inference process computationally has been formulating the problem in an end-to-end quantitative form that is suitable for machine learning, while capturing the ambiguities around molecule orientation, symmetries, and reconstruction resolution. Here we present an ML approach for structure determination from powder diffraction data. It works by estimating the electron density in a unit cell using a variational coordinate-based deep neural network. We demonstrate the approach on computed powder x-ray diffraction (PXRD), along with partial chemical composition information, as input. When evaluated on theoretically simulated data for the cubic and trigonal crystal systems, the system achieves up to 93.4% average similarity (as measured by structural similarity index) with the ground truth on unseen materials, both with known and partially-known chemical composition information, showing great promise for successful structure solution even from degraded and incomplete input data. The approach does not presuppose a crystalline structure and the approach are readily extended to other situations such as nanomaterials and textured samples, paving the way to reconstruction of yet unresolved nanostructures.