Deep machine learning potentials for multicomponent metallic melts: Development, predictability and compositional transferability

ArcaNN: automated enhanced sampling generation of training sets for chemically reactive machine learning interatomic potentials

The emergence of artificial intelligence is profoundly impacting computational chemistry, particularly through machine-learning interatomic potentials (MLIPs). Unlike traditional potential energy surface representations, MLIPs overcome the conventional computational scaling limitations by offering an effective combination of accuracy and efficiency for calculating atomic energies and forces to be used in molecular simulations. These MLIPs have significantly enhanced molecular simulations across various applications, including large-scale simulations of materials, interfaces, chemical reactions, and beyond. Despite these advances, the construction of training datasets-a critical component for the accuracy of MLIPs-has not received proportional attention, especially in the context of chemical reactivity, which depends on rare barrier-crossing events that are not easily included in the datasets. Here we address this gap by introducing ArcaNN, a comprehensive framework designed for generating training datasets for reactive MLIPs. ArcaNN employs a concurrent learning approach combined with advanced sampling techniques to ensure an accurate representation of high-energy geometries. The framework integrates automated processes for iterative training, exploration, new configuration selection, and energy and force labeling, all while ensuring reproducibility and documentation. We demonstrate ArcaNN's capabilities through two paradigm reactions: a nucleophilic substitution and a Diels-Alder reaction. These examples showcase its effectiveness, the uniformly low error of the resulting MLIP everywhere along the chemical reaction coordinate, and its potential for broad applications in reactive molecular dynamics. Finally, we provide guidelines for assessing the quality of MLIPs in reactive systems.

Dynamical properties of hydrogen fluid at high pressures

The properties of the hydrogen fluid at high pressures are still of interest to the scientific community. The experimentally unreachable dynamical properties could provide new insights into this field. In 2020 [Cheng et al., Nature 585, 217-220 (2020)], the machine-learned approach allows the calculation of the self-diffusion coefficient in the warm dense hydrogen with higher precision. After that, the work [van de Bund et al., Phys. Rev. Lett. 126(22), 225701 (2021)] reports the ab initio treatment of isotopic effects on diffusion in H2/D2 and a significant increase in its value in the region of the phase transition. Both works indicate the anomalous growth of diffusion, but the reasons for this phenomenon are unclear. In the present work, we reveal the plasma-like behavior of the diffusion growth. We apply the classical molecular dynamics method using a machine learning potential developed on the ab initio modeling for the prediction of diffusion and shear viscosity coefficients. We consider dependencies of the vibrational spectrum, molecule lifetime, diffusion, and shear viscosity coefficients on density along the isotherms in the temperature range from 600 to 1100 K.

Transfer learning for accurate description of atomic transport in Al-Cu melts

Machine learning interatomic potentials (MLIPs) provide an optimal balance between accuracy and computational efficiency and allow studying problems that are hardly solvable by traditional methods. For metallic alloys, MLIPs are typically developed based on density functional theory with generalized gradient approximation (GGA) for the exchange-correlation functional. However, recent studies have shown that this standard protocol can be inaccurate for calculating the transport properties or phase diagrams of some metallic alloys. Thus, optimization of the choice of exchange-correlation functional and specific calculation parameters is needed. In this study, we address this issue for Al-Cu alloys, in which standard Perdew-Burke-Ernzerhof (PBE)-based MLIPs cannot accurately calculate the viscosity and melting temperatures at Cu-rich compositions. We have built MLIPs based on different exchange-correlation functionals, including meta-GGA, using a transfer learning strategy, which allows us to reduce the amount of training data by an order of magnitude compared to a standard approach. We show that r2SCAN- and PBEsol-based MLIPs provide much better accuracy in describing thermodynamic and transport properties of Al-Cu alloys. In particular, r2SCAN-based deep machine learning potential allows us to quantitatively reproduce the concentration dependence of dynamic viscosity. Our findings contribute to the development of MLIPs that provide quantum chemical accuracy, which is one of the most challenging problems in modern computational materials science.

Compositional transferability of deep learning potentials: a case study for LiCl-KCl melt

CONTEXT

One of the crucial issues related to machine learning potentials is the formation of representative dataset. For multicomponent systems, it is a general methodology to scan the composition range with a certain step. However, there is a lack of information on the compositional transferability of machine learning potentials. In this paper, we extend the knowledge in this area by studying the transferability of deep learning potential over the range of compositions of LiCl-KCl molten mixtures. The training dataset was formed using only the near-eutectic composition of 60% LiCl-40% KCl. Then, we tested the ability of the model to predict physicochemical properties of the melts far from the reference composition. It was found that for the composition range from 0 to 100% of LiCl, the calculated properties concur closely with those of other studies and ab initio calculations. Therefore, the model shows prominent non-intuitive compositional transferability. Moreover, the solid states and solid-liquid coexistence were reproduced. The calculated melting temperatures of LiCl and KCl show the errors of 6.6% and 0.4%, respectively. We argue that such good transferability stems from the local structure configurations that are typical both for pure LiCl and for pure KCl which are implicitly presented in the training dataset because of local fluctuations in composition.

METHODS

To collect the data for the initial dataset, density functional theory was employed. Then, the DeePMD package was used to train a neural network potential. To calculate the properties of the melts, standard equilibrium and non-equilibrium molecular dynamic approaches were utilized.

Deciphering the controlling factors for phase transitions in zeolitic imidazolate frameworks

Zeolitic imidazolate frameworks (ZIFs) feature complex phase transitions, including polymorphism, melting, vitrification, and polyamorphism. Experimentally probing their structural evolution during transitions involving amorphous phases is a significant challenge, especially at the medium-range length scale. To overcome this challenge, here we first train a deep learning-based force field to identify the structural characteristics of both crystalline and non-crystalline ZIF phases. This allows us to reproduce the structural evolution trend during the melting of crystals and formation of ZIF glasses at various length scales with an accuracy comparable to that of ab initio molecular dynamics, yet at a much lower computational cost. Based on this approach, we propose a new structural descriptor, namely, the ring orientation index, to capture the propensity for crystallization of ZIF-4 (Zn(Im)2, Im = C3H3N2-) glasses, as well as for the formation of ZIF-zni (Zn(Im)2) out of the high-density amorphous phase. This crystal formation process is a result of the reorientation of imidazole rings by sacrificing the order of the structure around the zinc-centered tetrahedra. The outcomes of this work are useful for studying phase transitions in other metal-organic frameworks (MOFs) and may thus guide the development of MOF glasses.

Active learning prediction and experimental confirmation of atomic structure and thermophysical properties for liquid Hf_{76}W_{24} refractory alloy

The determination of liquid atomic structure and thermophysical properties is essential for investigating the physical characteristics and phase transitions of refractory alloys. However, due to the stringent experimental requirements and underdeveloped interatomic potentials, acquiring such information through experimentation or simulation remains challenging. Here, an active learning method incorporating a deep neural network was established to generate the interatomic potential of the Hf_{76}W_{24} refractory alloy. Then the achieved potential was applied to investigate the liquid atomic structure and thermophysical properties of this alloy over a wide temperature range. The simulation results revealed the distinctive bonding preferences among atoms, that is, Hf atoms exhibited a strong tendency for conspecific bonding, while W atoms preferred to form an interspecific bonding. The analysis of short-range order (SRO) in the liquid alloy revealed a significant proportion of icosahedral (ICO) and distorted ICO structures, which even exceeded 30% in the undercooled state. As temperature decreased, SRO structures demonstrated an increase in larger coordination number (CN) clusters and a decrease in smaller CNs. The alterations of the atomic structure indicated that the liquid alloy becomes more ordered, densely packed, and energetically favorable with decreasing temperature, consistent with the obtained fact: Both density and surface tension increase linearly. The simulated thermophysical properties were close to experimental values with minor deviations of 2.8% for density and 3.4% for surface tension. The consistency of the thermophysical properties further attested to the accuracy and reliability of active learning simulation.

Development and Validation of Versatile Deep Atomistic Potentials for Metal Oxides

Machine learning interatomic potentials powered by neural networks have been shown to readily model a gradient of compositions in metallic systems. However, their application to date on ionic systems tends to focus on specific compositions and oxidation states owing to their more heterogeneous chemical nature. Herein we show that a deep neural network potential (DNP) can model various properties of metal oxides with different oxidation states without additional charge information. We created and validated DNPs for Ag_xO_y, Cu_xO_y Mg_xO_y, Pt_xO_y, and Zn_xO_y, whereby each system was trained without any limitations on oxidation states. We illustrate how the database can be augmented to enhance the DNP transferability for a new polymorph, surface energies, and thermal expansion. In addition, we show that these potentials can correctly interpolate significant pressure and temperature ranges, exhibit stability over long molecular dynamics simulation time scales, and replicate nonharmonic thermal expansion, consistent with experimental results.

Two-temperature molecular dynamics simulations of crystal growth in a tungsten supercooled melt

In this work we use the two-temperature model (TTM) coupled to molecular dynamics (MD) with sinks at the boundaries of the electronic subsystem to study crystal-growth rate in a quasi-one-dimensional tungsten crystal into a supercooled melt. The possibility of varying the extension of the electronic grid along with the sinks allows a more realistic description of the electronic thermal transport away from the system, providing a considerable heat dissipation from the crystallization front. Based on this approach, our results regarding crystal-growth rates are not affected even if the size of the system is changed. Moreover, comparisons are established with respect to MD and standard TTM simulations. For these comparisons between models, something remarkable is found, and it is that the temperature and the value of the maximum growth rate are the same. In contrast, the inclusion of sinks has a great impact with respect to the standard approaches specially reflected at low temperatures, where a frustration of the liquid-crystal interface dynamics is seen until a state of zero crystal growth is reached, which is not possible to characterize quantitatively since a kind of stochastic behavior is present.

Kinetics of rapid growth and melting of Al50Ni50alloying crystals: phase field theoryversusatomistic simulations revisited. JOURNAL OF PHYSICS. CONDENSED MATTER : AN INSTITUTE OF PHYSICS JOURNAL 2022;34:494002. [PMID: 36228604 DOI: 10.1088/1361-648x/ac9a1c] [Citation(s) in RCA: 0] [Impact Index Per Article: 0]

A revised study of the growth and melting of crystals in congruently melting Al₅₀Ni₅₀alloy is carried out by molecular dynamics (MDs) and phase field (PF) methods. An embedded atom method (EAM) potential of Purja Pun and Mishin (2009Phil. Mag.89 3245) is used to estimate the material's properties (density, enthalpy, and self-diffusion) of the B2 crystalline and liquid phases of the alloy. Using the same EAM potential, the melting temperature, density, and diffusion coefficient become well comparable with experimental data in contrast with previous works where other potentials were used. In the new revision of MD data, the kinetics of melting and solidification are quantitatively evaluated by the 'crystal-liquid interface velocity-undercooling' relationship exhibiting the well-known bell-shaped kinetic curve. The traveling wave solution of the kinetic PF model as well as the hodograph equation of the solid-liquid interface quantitatively describe the 'velocity-undercooling' relationship obtained in the MD simulation in the whole range of investigated temperatures for melting and growth of Al₅₀Ni₅₀crystals.