npj Computational Materials最新文献

Tungsten-bronze-type material Ba_6-3xRE_8+2xTi₁₈O₅₄, (RE = rare earth elements) is an important microwave dielectric that has shown great promises for future miniaturization of microwave devices because of its high dielectric constant, low loss, and tunabilities, and there is still much room for improvement. With their proven predictive power, first-principles calculations may greatly help accelerate materials optimization by reducing or eliminating the expensive and time-consuming experimental trial-and-error process. However, microwave dielectrics such as the tungsten-bronze-type materials are rather complex systems with unit cells containing hundreds or thousands of atoms, making ab initio calculations prohibitively expensive. In this work, we propose an elemental-unit decomposition (EUD) technique that can drastically reduce the computational effort of predicting the properties of complex microwave dielectrics and demonstrate its accuracy and efficiency. Our approach facilitates first-principles prediction and design of complex microwave dielectric materials that would otherwise be extremely difficult.

The ground state electron density — obtainable using Kohn-Sham Density Functional Theory (KS-DFT) simulations — contains a wealth of material information, making its prediction via machine learning (ML) models attractive. However, the computational expense of KS-DFT scales cubically with system size which tends to stymie training data generation, making it difficult to develop quantifiably accurate ML models that are applicable across many scales and system configurations. Here, we address this fundamental challenge by employing transfer learning to leverage the multi-scale nature of the training data, while comprehensively sampling system configurations using thermalization. Our ML models are less reliant on heuristics, and being based on Bayesian neural networks, enable uncertainty quantification. We show that our models incur significantly lower data generation costs while allowing confident — and when verifiable, accurate — predictions for a wide variety of bulk systems well beyond training, including systems with defects, different alloy compositions, and at multi-million-atom scales. Moreover, such predictions can be carried out using only modest computational resources.

The second law of thermodynamics asserts that energy diffuses from hot to cold. The resulting temperature gradients drive the efficiencies and failures in a plethora of technologies. However, as the dimensionalities of materials shrink to the nanoscale regime, proper heat dissipation strategies become more challenging since the mean free paths of phonons become larger than the characteristic length scales. This leads to temperature gradients that are dependent on interfaces and boundaries, which ultimately can lead to severe thermal bottlenecks. Herein, we uncover a phenomenon which we refer to as ‘phonon funneling’, that allows the control of phonon transport to preferentially direct phonon energy away from geometrically confined interfacial thermal bottlenecks and into localized colder regions. This phenomenon supersedes heat diffusion based on the macroscale temperature gradients, thus introducing a nanoscale regime in which boundary scattering increases the phonon thermal conductivity of thin films, an opposite effect than what is traditionally realized. This work advances the fundamental understanding of phonon transport at the nanoscale and the role of efficient scattering methods for enhancing thermal transport.

We demonstrate a machine learning-based approach which predicts the properties of crystal structures following relaxation based on the unrelaxed structure. Use of crystal graph singular values reduces the number of features required to describe a crystal by more than an order of magnitude compared to the full crystal graph representation. We construct machine learning models using the crystal graph singular value representations in order to predict the volume, enthalpy per atom, and metal versus semiconductor/insulator phase of DFT-relaxed organic salt crystals based on randomly generated unrelaxed crystal structures. Initial base models are trained to relate 89,949 randomly generated structures of salts formed by varying ratios of 1,3,5-triazine and HCl with the corresponding volumes, enthalpies per atom, and phase of the DFT-relaxed structures. We further demonstrate that the base model is able to be extended to related chemical systems (isomers, pyridine, thiophene and piperidine) with the inclusion of 2000 to 10,000 crystal structures from the additional system. After training a single model with a large number of data points, extension can be done at significantly lower cost. The constructed machine learning models can be used to rapidly screen large sets of randomly generated organic salt crystal structures and efficiently downselect the structures most likely to be experimentally realizable. The models can be used as a stand-alone crystal structure predictor, but may serve CSP efforts best as a filtering step in more sophisticated workflows.

Density fluctuation potential induced by a screening of the photohole scatters the photoelectron and generally causes its emission delay from the scattering matrix in the photoemission spectroscopy, where the photoemission delay usually quantifies the extrinsic loss of the photoelectron depending on the atomic orbital. Without the potential scattering, however, the photoemission from the coherent two-state mixture created by the laser driving is found to undergo the unexpected photoemission delay, which originates from the mixed photoemission matrix. Using the Haldane model, we analytically calculate such coherent mixing induced photoemission delay in an angle-resolved mode, which is found to reveal the local Berry curvature structure as long as the coherent mixing is sustained. This finding is confirmed through the streaking computation for the photoemission delay by solving the time-dependent Schrödinger equation and suggests that the photoemission delay be a new spectroscopic diagnosis of the material topology of two-dimensional semiconductors.

Predicting phase stability in high entropy alloys (HEAs), such as phase fractions as functions of composition and temperature, is essential for understanding alloy properties and screening desirable materials. Traditional methods like CALPHAD are computationally intensive for exploring high-dimensional compositional spaces. To address such a challenge, this study explored and compared the effectiveness of random forests (RF) and deep neural networks (DNN) for accelerating materials discovery by building surrogate models of phase stability prediction. For interpolation scenarios (testing on the same order of system as trained), RF models generally produce smaller errors than DNN models. However, for extrapolation scenarios (training on lower-order systems and testing on higher order systems), DNNs generalize more effectively than traditional ML models. DNN demonstrate the potential to predict topologically relevant phase composition when data were missing, making it a powerful predictive tool in materials discovery frameworks. The study uses a CALPHAD dataset of 480 million data points generated from a custom database, available for further model development and benchmarking. Experiments show that DNN models are data-efficient, achieving similar performance with a fraction of the dataset. This work highlights the potential of DNNs in materials discovery, providing a powerful tool for predicting phase stability in HEAs, particularly within the Cr-Hf-Mo-Nb-Ta-Ti-V-W-Zr composition space.

Hydrogen in metals is a significant research area with far-reaching implications, encompassing diverse fields such as hydrogen storage, metal-insulator transitions, and the recently emerging phenomenon of room-temperature superconductivity under high pressure. Hydrogen atoms pose challenges in experiments as they are nearly invisible, and they are considered within ideal crystalline structures in theoretical predictions, which hampers research on the formation of metastable hydrides. Here, we propose pressure-induced hydrogen migration from tetrahedral (T-) site to octahedral (O-) site, forming ({{rm{LaH}}}_{x}^{{rm{O}}}{{rm{H}}}_{2-x}^{{rm{T}}}) in cubic LaH_2. Under decompression, it retains ({{rm{H}}}_{x}^{{rm{O}}}) occupancy, and is dynamically stable even at ambient pressure, enabling a synthesis route of metastable dihydrides via compression-decompression process. We predict that the electron-phonon coupling strength of ({{rm{LaH}}}_{x}^{{rm{O}}}{{rm{H}}}_{2-x}^{{rm{T}}}) is enhanced with increasing x, and the associated T_c reaches up to 10.8 K at ambient pressure. Furthermore, we calculated stoichiometric hydrogen migration threshold pressure (P_c) for various lanthanides dihydrides (RH₂, where R = Y, Sc, Nd, and Lu), and found an inversely linear relation between P_c and ionic radii of R. We propose that the highest T_c in the face-centered-cubic dihydride system can be realized by optimizing the O/T-site occupancies.

We report high throughput computational screening for magnetic ground state order in 2D materials. The workflow is based on spin spiral calculations and yields the magnetic order in terms of a two-dimensional ordering vector Q. We then include spin-orbit coupling to extract the easy and hard axes for collinear structures and the orientation of spiral planes in non-collinear structures. Finally, for all predicted ferromagnets we compute the Dzyaloshinskii-Moriya interactions and determine whether or not these are strong enough to overcome the magnetic anisotropy and stabilise a chiral spin spiral ground state. We find 58 ferromagnets, 21 collinear anti-ferromagnets, and 85 non-collinear ground states of which 15 are chiral spin spirals driven by Dzyaloshinskii-Moriya interactions. The results show that non-collinear order is in fact as common as collinear order in these materials and emphasise the need for detailed investigation of the magnetic ground state when reporting magnetic properties of new materials.

Machine learning has been widely used to guide the search for new materials by learning the patterns underlying available data. However, the pure prediction-oriented search is often biased to interpolation due to the limited data in a large unexplored space. Here we present a sampling framework towards extrapolation, that integrates unsupervised clustering, interpretable analysis, and similarity evaluation to sample target candidates with improved properties from a vast search space. Using the design of superalloys with improved ({gamma }^{{prime} })-phase solvus temperature (({T}_{{gamma }^{{prime} }})) as a model case, we start with sparse data, and by a few experiments, we find nine new superalloys with chemistries distinct to those in the training data. Three of them show improved ({T}_{{gamma }^{{prime} }}) by about 50 °C, a large enhancement for superalloys. Moreover, we find two features characterizing mismatch in atomic size and mixing enthalpy linearly effect. This work demonstrates the capability of unsupervised learning to search for new materials when limited data is available.

To accurately capture the dynamic behavior of small nanoparticles in scanning transmission electron microscopy, high-quality data and advanced data processing is needed. The fast scan rate required to observe structural dynamics inherently leads to very noisy data where machine learning tools are essential for unbiased analysis. In this study, we develop a workflow based on two U-Net architectures to automatically localize and classify atomic columns at particle-support interfaces. The model is trained on non-physical image simulations, achieves sub-pixel localization precision, high classification accuracy, and generalizes well to experimental data. We test our model on both in situ and ex situ experimental time series recorded at 5 frames per second of small Pt nanoparticles supported on CeO₂(111). The processed movies show sub-second dynamics of the nanoparticles and reveal site-specific movement patterns of individual atomic columns.