A high-dimensional neural network potential for finite-temperature phenomena in NiTi martensite

Machine-learning approach for the phase stability and mechanical properties of disordered alloys at finite temperature

The prediction of stable alloys forming solid-state solutions across large portions of the composition space is a serious theoretical challenge, since one has to evaluate the Gibbs free energy, including both configurational and vibrational contributions. This requires an energy theory capable of extremely high throughput. By taking the Ni-Pd system as prototype, we construct an efficient Jacobi-Legendre machine-learning potential based on density-functional-theory data, which provides accurate energies and forces across the entire composition space. Based on a cluster expansion up to three-body terms and only 873 trainable parameters, this allows us to compute the partition function by directly integrating all accessible microstates, differing for composition, atomic configuration and thermal agitation. We confirm that Ni and Pd are fully miscible, forming an $fcc$ solid-state solution. This is only metastable at room temperature, while becomes thermodynamically stable at around 600~K, with the stability achieved first at the Pd-rich end of the composition range. Interestingly, entropy and heat capacity analysis reveal a competition between the solid-state solution and two intermetallic phases with long-period L1$_0$ structure for NiPd and NiPd$_3$. All in all, our approach offers a powerful and high-throughput workflow for the study of disordered alloys, an approach that can be extended to multi-component systems such as high-entropy alloys.

Training and validation of the interatomic potential for Al-Sc alloy based on deep-learning framework

Thermodynamics and phase stability of the AlTiCrMoW high-entropy alloy simulated using machine-learned interatomic potentials

High-entropy alloys (HEAs) possess exceptional mechanical properties and thermal stability, yet their vast compositional space and complex atomic arrangements present significant challenges for property prediction and design. To overcome this, we have trained a machine-learned interatomic potential (MLIP) for the AlTiCrMoW HEA using the message-passing atomic cluster expansion (MACE) framework. This new MLIP is trained on a series of high-accuracy density functional theory (DFT) calculations that covers the available configuration space of the HEA. The training set includes key configurations with linear-response-informed short-range ordered configurations, which are generated using a concentration wave analysis within the coherent potential approximation (CPA) and subsequently optimizedd using conventional DFT calculations. Phonon bispersion ands (PDBs) for the HEA are generated using the MACE model, and several physical properties are extracted for the HEA, such as the elastic constants and the bulk modulus. The fine-tuned model reproduces DFT structural and elastic benchmarks with high accuracy, at a significantly reduced computational cost, allowing for larger simulations to be performed, which can more accurately probe the configuration space and composition of the alloy. Here, we perform Monte Carlo based simulations to identify the ordering transition of equiatomic AlTiCrMoW, where we match the experimental value of 1273 K. Vacancy relaxation volumes are investigated for AlTiCrMoW, as an example of a property analysis, which would be computationally intractable via DFT.

Multiscale modelling of ferroelectrics using a physics-informed neural network driven by molecular dynamics data: parameter identification and field reconstruction

A physics-informed neural network (PINN) framework is developed, in which the loss function consists of MD-data penalty and partial difference equation (PDE) constraints, which provides an effective methodology for multiscale bridging between atomistic and continuum descriptions for perovskite ferroelectric materials.

Atomistic simulations of high-entropy alloys: from density functional theory to machine-learning interatomic potentials

This review aims to provide a comprehensive perspective on the ongoing transition from conventional DFT-based simulations toward scalable, statistically rigorous, and predictive atomistic modeling frameworks for HEAs and related compositionally complex materials.

Influence of Interface Energy Anisotropy on the Solid-state Instability in Ni-based Superalloy: A Multiscale Study

The microstructural stability of nickel-based superalloys critically depends on the morphology and evolution of $\gamma'$-precipitates, which is governed by elastic and interfacial anisotropies at the atomic scale. Here, we present a novel quantitative multiscale framework that, for the first time, directly incorporates atomistically computed interface energy anisotropy into mesoscale phase-field simulations to elucidate morphological selection and instability in the Ni--Al system. We employ density functional theory (DFT) to accurately predict the orientation-dependent $\gamma/\gamma'$ interface energies for key crystallographic planes. A rigorous analytic mapping is then developed to systematically reduce the three-dimensional (3D) interface anisotropy landscape to the two-dimensional (2D) simulation plane. This enables quantitative transfer of DFT-informed anisotropy parameters into a continuum phase-field model that also accounts for elastic inhomogeneity and eigenstrain. Our simulations demonstrate that the explicit inclusion of DFT-based interface energy anisotropy fundamentally alters precipitate morphological evolution, robustly suppressing instability and faceting phenomena otherwise promoted by supersaturation and elastic effects. The framework bridges atomic- to mesoscale modeling, enabling predictive control of precipitate shapes and providing new insights into the interplay of elastic and interfacial contributions in Ni-based superalloys. This approach paves the way for quantitative microstructural design in advanced high-temperature alloys via first-principles-guided multiscale simulation.