npj Computational Materials最新文献

The quasiharmonic approximation (QHA) in combination with density-functional theory is the main computational method used to calculate thermodynamic properties under arbitrary temperature and pressure conditions. QHA can predict thermodynamic phase diagrams, elastic properties and temperature- and pressure-dependent equilibrium geometries, all of which are important in various fields of knowledge. The main drawbacks of QHA are that it makes spurious predictions for the volume and other properties in the high temperature limit due to its approximate treatment of anharmonicity, and that it is unable to model dynamically stabilized structures. In this work, we propose an extension to QHA that fixes these problems. Our approach is based on four ingredients: (i) the calculation of the n-th order force constants using randomly displaced configurations and regularized regression, (ii) the calculation of temperature-dependent effective harmonic frequencies ω(V, T) within the self-consistent harmonic approximation (SCHA), (iii) Allen’s quasiparticle (QP) theory, which allows the calculation of the anharmonic entropy from the effective frequencies, and (iv) a simple Debye-like numerical model that enables the calculation of all other thermodynamic properties from the QP entropies. The proposed method is conceptually simple, with a computational complexity similar to QHA but requiring more supercell calculations. It allows incorporating anharmonic effects to any order. The predictions of the new method coincide with QHA in the low-temperature limit and eliminate the QHA blowout at high temperature, recovering the experimentally observed behavior of all thermodynamic properties tested. The performance of the new method is demonstrated by calculating the thermodynamic properties of geologically relevant minerals MgO and CaO. In addition, using cubic SrTiO₃ as an example, we show that, unlike QHA, our method can also predict thermodynamic properties of dynamically stabilized phases. We expect this new method to be an important tool in geochemistry and materials discovery.

We present an algorithm for the high-throughput computational discovery of intermetallic compounds in systems with a large number of components. It is particularly important for high entropy alloys (HEAs), where multiple principal elements can form numerous potential intermetallic compounds during the condensation process, making it challenging to predict the dominant phase. Our algorithm is based on a brute-force evaluation of candidate structures with a fixed underlying lattice (FCC or BCC) accelerated by machine-learning interatomic potentials. The algorithm takes a set of chemical elements and a crystal lattice type as inputs and produces structures on and near the convex hull of thermodynamically stable structures. The candidate structures are evaluated using the low-rank potential (LRP), trained to reproduce energies of structures equilibrated with density functional theory (DFT). Thanks to extreme computational effectiveness of the LRP, it is feasible to evaluate hundreds of thousands of structures per second, per CPU core. Thus, our algorithm screens a complete set of candidate structures for a given system without missing any configurations. We validated our method on systems with BCC (Nb-W, Nb-Mo-W, V-Nb-Mo-Ta-W) and FCC (Cu-Pt, Cu-Pd-Pt, Cu-Pd-Ag-Pt-Au) lattices and discovered 268 new alloys not reported in the AFLOW database¹, which we used as a benchmark.

Dislocations in ceramics are increasingly recognized for their promising potential in applications such as toughening intrinsically brittle ceramics and tailoring functional properties. However, the atomistic simulation of dislocation plasticity in ceramics remains challenging due to the complex interatomic interactions characteristic of ceramics, which include a mix of ionic and covalent bonds, and highly distorted and extensive dislocation core structures within complex crystal structures. These complexities exceed the capabilities of empirical interatomic potentials. Therefore, constructing neural network potentials (NNPs) emerges as the optimal solution. Yet, creating a training dataset that includes dislocation structures proves difficult due to the complexity of their core configurations in ceramics and the computational demands of density functional theory for large atomic models containing dislocation cores. In this work, we propose a training dataset from properties that are easier to compute via high-throughput calculation. Using this dataset, we have successfully developed NNPs for dislocation plasticity in ceramics, specifically for three typical functional ceramics: ZnO, GaN, and SrTiO₃. These NNPs effectively capture the nonstoichiometric and charged core structures and slip barriers of dislocations, as well as the long-range electrostatic interactions between charged dislocations. The effectiveness of this dataset was further validated by measuring the similarity and uncertainty across snapshots derived from large-scale simulations, alongside extensive validation across various properties. Utilizing the constructed NNPs, we examined dislocation plasticity in ceramics through nanopillar compression and nanoindentation, which demonstrated excellent agreement with experimental observations. This study provides an effective framework for constructing NNPs that enable the detailed atomistic modeling of dislocation plasticity, opening new avenues for exploring the plastic behavior of ceramics.

Predicting the properties of non-fullerene acceptors (NFAs), complex organic molecules used in organic solar cells (OSCs), poses a significant challenge. Some existing approaches primarily focus on atom-level information and may overlook high-level molecular features, including the subunits of NFAs. While other methods that effectively represent subunit information show improved prediction performance, they require labor-intensive data labeling. In this paper, we introduce an efficient molecular description method that automatically extracts molecular information at both the atom and subunit levels without any labor-intensive data labeling. Inspired by the Word2Vec method, our Ring2Vec method treats the “rings” in organic molecules as analogous to “words” in sentences. We achieve fast and accurate predictions of the energy levels of NFA molecules, with a minimal prediction error of merely 0.06 eV. Furthermore, our method can potentially have broad applicability across various domains of molecular description and property prediction, owing to the efficiency of the Ring2Vec model.

Dielectrics are crucial for technologies like flash memory, CPUs, photovoltaics, and capacitors, but public data on these materials are scarce, restricting research and development. Existing machine learning models have focused on predicting scalar polycrystalline dielectric constants, neglecting the directional nature of dielectric tensors essential for material design. This study leverages multi-rank equivariant structural embeddings from a universal neural network potential to enhance predictions of dielectric tensors. We develop an equivariant readout decoder to predict total, electronic, and ionic dielectric tensors while preserving O(3) equivariance, and benchmark its performance against state-of-the-art algorithms. Virtual screening of thermodynamically stable materials from Materials Project for two discovery tasks, high-dielectric and highly anisotropic materials, identifies promising candidates including Cs₂Ti(WO₄)₃ (band gap E_g = 2.93eV, dielectric constant ε = 180.90) and CsZrCuSe₃ (anisotropic ratio α_r = 121.89). The results demonstrate our model’s accuracy in predicting dielectric tensors and its potential for discovering novel dielectric materials.

First principles approaches have revolutionized our ability in using computers to predict, explore, and design materials. A major advantage commonly associated with these approaches is that they are fully parameter-free. However, numerically solving the underlying equations requires to choose a set of convergence parameters. With the advent of high-throughput calculations, it becomes exceedingly important to achieve a truly parameter-free approach. Utilizing uncertainty quantification (UQ) and linear decomposition we derive a numerically highly efficient representation of the statistical and systematic error in the multidimensional space of the convergence parameters for plane wave density functional theory (DFT) calculations. Based on this formalism we implement a fully automated approach that requires as input the target precision rather than convergence parameters. The performance and robustness of the approach are shown by applying it to a large set of elements crystallizing in a cubic fcc lattice.

Charge density wave (CDW) is discovered within a wide interval in solids, however, its microscopic nature is still not transparent in most realistic materials, and the recently studied chiral ones with chiral structural distortion remain unclear. In this paper, we try to understand the driving forces of chiral CDW transition by chiral phonons from the electron-phonon coupling scenario. We use the prototypal monolayer 1T-TiSe₂ as a case study to unveil the absence of chirality in the CDW transition and propose a general approach, i.e., symmetry-breaking stimuli, to engineer the chirality of CDW in experiments. Inelastic scattering patterns are also studied as a benchmark of chiral CDW (CCDW, which breaks the mirror/inversion symmetry in 2D/3D systems). We notice that the anisotropy changing of Bragg peak profiles, which is contributed by the soft chiral phonons, can show a remarkable signature for CCDW. Our findings pave a path to understanding the CCDW from the chiral phonon perspective, especially in van der Waals materials, and provides a powerful way to manipulate the chirality of CDW.

We demonstrate the ability of a large language model to perform evolutionary optimization for materials discovery. Anthropic’s Claude 3.5 model outperforms an active learning scheme with handcrafted surrogate models and an evolutionary algorithm in selecting monomer sequences to produce targeted morphologies in macromolecular self-assembly. Utilizing pre-trained language models can potentially reduce the need for hyperparameter tuning while offering new capabilities such as self-reflection. The model performs this task effectively with or without context about the task itself, but domain-specific context sometimes results in faster convergence to good solutions. Furthermore, when this context is withheld, the model infers an approximate notion of the task (e.g., calling it a protein folding problem). This work provides evidence of Claude 3.5’s ability to act as an evolutionary optimizer, a recently discovered emergent behavior of large language models, and demonstrates a practical use case in the study and design of soft materials.

We present a comprehensive and user-friendly framework built upon the pyiron integrated development environment (IDE), enabling researchers to perform the entire Machine Learning Potential (MLP) development cycle consisting of (i) creating systematic DFT databases, (ii) fitting the Density Functional Theory (DFT) data to empirical potentials or MLPs, and (iii) validating the potentials in a largely automatic approach. The power and performance of this framework are demonstrated for three conceptually very different classes of interatomic potentials: an empirical potential (embedded atom method - EAM), neural networks (high-dimensional neural network potentials - HDNNP) and expansions in basis sets (atomic cluster expansion - ACE). As an advanced example for validation and application, we show the computation of a binary composition-temperature phase diagram for Al-Li, a technologically important lightweight alloy system with applications in the aerospace industry.

Directed atomic fabrication using an aberration-corrected scanning transmission electron microscope (STEM) opens new pathways for atomic engineering of functional materials. In this approach, the electron beam is used to actively alter the atomic structure through electron beam induced irradiation processes. One of the impediments that has limited widespread use thus far has been the ability to understand the fundamental mechanisms of atomic transformation pathways at high spatiotemporal resolution. Here, we develop a workflow for obtaining and analyzing high-speed spiral scan STEM data, up to 100 fps, to track the atomic fabrication process during nanopore milling in monolayer MoS₂. An automated feedback-controlled electron beam positioning system combined with deep convolution neural network (DCNN) was used to decipher fast but low signal-to-noise datasets and classify time-resolved atom positions and nature of their evolving atomic defect configurations. Through this automated decoding, the initial atomic disordering and reordering processes leading to nanopore formation was able to be studied across various timescales. Using these experimental workflows a greater degree of speed and information can be extracted from small datasets without compromising spatial resolution. This approach can be adapted to other 2D materials systems to gain further insights into the defect formation necessary to inform future automated fabrication techniques utilizing the STEM electron beam.