A machine learning potential construction based on radial distribution function sampling.

Sampling reference data is crucial in machine learning potential (MLP) construction. Inadequate coverage of local configurations in reference data may lead to unphysical behaviors in MLP-based molecular dynamics (MLP-MD) simulations. To address this problem, this study proposes a new on-the-fly reference data sampling method called radial distribution function (RDF)-based data sampling for MLP construction. This method detects and extracts anomalous structures from the trajectories of MLP-MD simulations by focusing on the shapes of RDFs. The detected structures are added to the reference data to improve the accuracy of the MLP. This method allows us to realize a reasonable MLP construction for liquid water with minimal additional data. We prepare data from an H2O molecular cluster system and verify whether the constructed MLPs are practical for bulk water systems. MLP-MD simulations without RDF-based data sampling show unphysical behaviors, such as atomic collisions. In contrast, after applying this method, we obtain MLP-MD trajectories with features, such as RDF shapes and angle distributions, that are comparable to those of ab initio MD simulations. Our simulation results demonstrate that the RDF-based data sampling approach is useful for constructing MLPs that are robust to extrapolations from molecular cluster systems to bulk systems without any specialized know-how.

An approach for full space inverse materials design by combining universal machine learning potential, universal property model, and optimization algorithm.

We present a full space inverse materials design (FSIMD) approach that fully automates the materials design for target physical properties without the need to provide the atomic composition, chemical stoichiometry, and crystal structure in advance. Here, we used density functional theory reference data to train a universal machine learning potential (UPot) and transfer learning to train a universal bulk modulus model (UBmod). Both UPot and UBmod were able to cover materials systems composed of any element among 42 elements. Interfaced with optimization algorithm and enhanced sampling, the FSIMD approach is applied to find the materials with the largest cohesive energy and the largest bulk modulus, respectively. NaCl-type ZrC was found to be the material with the largest cohesive energy. For bulk modulus, diamond was identified to have the largest value. The FSIMD approach is also applied to design materials with other multi-objective properties with accuracy limited principally by the amount, reliability, and diversity of the training data. The FSIMD approach provides a new way for inverse materials design with other functional properties for practical applications.

Atomistic Simulation of HF Etching Process of Amorphous Si3N4 Using Machine Learning Potential.

An atomistic understanding of dry-etching processes with reactive molecules is crucial for achieving geometric integrity in highly scaled semiconductor devices. Molecular dynamics (MD) simulations are instrumental, but the lack of reliable force fields hinders the widespread use of MD in etching simulations. In this work, we develop an accurate neural network potential (NNP) for simulating the etching process of amorphous Si3N4 with HF molecules. The surface reactions in diverse local environments are considered by incorporating several types of training sets: baseline structures, reaction-specific data, and general-purpose training sets. Furthermore, the NNP is refined through iterative comparisons with the density functional theory results. Using the trained NNP, we carry out etching simulations, which allow for detailed observation and analysis of key processes such as preferential sputtering, surface modification, etching yield, threshold energy, and the distribution of etching products. Additionally, we develop a simple continuum model, built from the MD simulation results, which effectively reproduces the surface composition obtained with MD simulations. By establishing a computational framework for atomistic etching simulation and scale bridging, this work will pave the way for more accurate and efficient design of etching processes in the semiconductor industry, enhancing device performance and manufacturing precision.

Disorder-Dependent Li Diffusion in Li6PS5Cl Investigated by Machine-Learning Potential.

Solid-state electrolytes with argyrodite structures, such as Li6PS5Cl, have attracted considerable attention due to their superior safety compared to liquid electrolytes and higher ionic conductivity than other solid electrolytes. Although experimental efforts have been made to enhance conductivity by controlling the degree of disorder, the underlying diffusion mechanism is not yet fully understood. Moreover, existing theoretical analyses based on ab initio molecular dynamics (MD) simulations have limitations in addressing various types of disorder at room temperature. In this study, we directly investigate Li-ion diffusion in Li6PS5Cl at 300 K using large-scale, long-term MD simulations empowered by machine-learning potentials (MLPs). To ensure the convergence of conductivity values within an error range of 10%, we employ a 25 ns simulation using a 5 × 5 × 5 supercell containing 6500 atoms. The computed Li-ion conductivity, activation energies, and equilibrium site occupancies align well with experimental observations. Notably, Li-ion conductivity peaks when Cl ions occupy 25% of the 4c sites rather than at 50% where the disorder is maximized. In addition, Li-ion diffusion shows non-Arrhenius behavior, leading to different activation energies at high temperatures (>400 K). These phenomena are explained by the interplay between inter- and intracage jumps. By elucidation of the key factors affecting Li-ion diffusion in Li6PS5Cl, this work paves the way for optimizing ionic conductivity in the argyrodite family.

Insight into Interfacial Heat Transfer of ß-Ga2O3/Diamond Heterostructures via the Machine Learning Potential.

ß-Ga2O3 is an ultrawide-band gap semiconductor with excellent potential for high-power and ultraviolet optoelectronic device applications. Low thermal conductivity is one of the major obstacles to enable the full performance of ß-Ga2O3-based devices. A promising solution for this problem is to integrate ß-Ga2O3 with a diamond heat sink. However, the thermal properties of the ß-Ga2O3/diamond heterostructures after the interfacial bonding have not been studied extensively, which are influenced by the crystal orientations and interfacial atoms for the ß-Ga2O3 and diamond interfaces. In this work, molecular dynamics simulations based on machine learning potential have been adopted to investigate the crystal-orientation-dependent and interfacial-atom-dependent thermal boundary resistance (TBR) of the ß-Ga2O3/diamond heterostructure after interfacial bonding. The differences in TBR at different interfaces are explained in detail through the explorations of thermal conductivity value, thermal conductivity spectra, vibration density of states, and interfacial structures. Based on the above explorations, a further understanding of the influence of different crystal orientations and interfacial atoms on the ß-Ga2O3/diamond heterostructure was achieved. Finally, insightful optimization strategies have been proposed in the study, which could pave the way for better thermal design and management of ß-Ga2O3/diamond heterostructures according to guidance in the selection of the crystal orientations and interfacial atoms of the ß-Ga2O3 and diamond interfaces.

Interfacial Optimization for AlN/Diamond Heterostructures via Machine Learning Potential Molecular Dynamics Investigation of the Mechanical Properties.

AlN/diamond heterostructures hold tremendous promise for the development of next-generation high-power electronic devices due to their ultrawide band gaps and other exceptional properties. However, the poor adhesion at the AlN/diamond interface is a significant challenge that will lead to film delamination and device performance degradation. In this study, the uniaxial tensile failure of the AlN/diamond heterogeneous interfaces was investigated by molecular dynamics simulations based on a neuroevolutionary machine learning potential (NEP) model. The interatomic interactions can be successfully described by trained NEP, the reliability of which has been demonstrated by the prediction of the cleavage planes of AlN and diamond. It can be revealed that the annealing treatment can reduce the total potential energy by enhancing the binding of the C and N atoms at interfaces. The strain engineering of AlN also has an important impact on the mechanical properties of the interface. Furthermore, the influence of the surface roughness and interfacial nanostructures on the AlN/diamond heterostructures has been considered. It can be indicated that the combination of surface roughness reduction, AlN strain engineering, and annealing treatment can effectively result in superior and more stable interfacial mechanical properties, which can provide a promising solution to the optimization of mechanical properties, of ultrawide band gap semiconductor heterostructures.

Phase Transition in Silicon from Machine Learning Informed Metadynamics.

Investigating reconstructive phase transitions in large-sized systems requires a highly efficient computational framework with computational cost proportional to the system size. Traditionally, widely used frameworks such as density functional theory (DFT) have been prohibitively expensive for extensive simulations on large systems that require long-time scales. To address this challenge, this study employed well-trained machine learning potential to simulate phase transitions in a large-size system. This work integrates the metadynamics simulation approach with machine learning potential, specifically deep potential, to enhance computational efficiency and accelerate the study of phase transition and consequent development of grains and dislocation defects in a system. The new method is demonstrated using the phase transitions of bulk silicon under high pressure. This approach has revealed the transition path and formation of polycrystalline silicon systems under specific stress conditions, demonstrating the effectiveness of deep potential-driven metadynamics simulations in gaining insights into complex material behaviors in large-sized systems.

The validity of learning potential for predicting educational adjustment in preschoolers with Down syndrome: A longitudinal study.

Learning Potential tests aim to assess cognitive functioning using mediation strategies by observing subsequent changes in learning patterns. In this study, this methodology was applied with the Preschool Learning Potential and Abilities Scale and two additional tests, the Kaufman Brief Intelligence Test and the Battery of Aptitudes for School Learning I, to a total of 58 children with Down Syndrome, at four and six years of age. The results demonstrate improvements in general intelligence, learning potential, and school aptitudes between the two timepoints. There was a significant, positive relationship between the variables measured in the different tests, as well as on the predictive variables of school aptitudes, in these children at six years of age. There is evidence that supports the use of this dynamic evaluation methodology, opening new fields of action in child evaluation processes.

Implementation and Validation of an OpenMM Plugin for the Deep Potential Representation of Potential Energy.

Machine learning potentials, particularly the deep potential (DP) model, have revolutionized molecular dynamics (MD) simulations, striking a balance between accuracy and computational efficiency. To facilitate the DP model's integration with the popular MD engine OpenMM, we have developed a versatile OpenMM plugin. This plugin supports a range of applications, from conventional MD simulations to alchemical free energy calculations and hybrid DP/MM simulations. Our extensive validation tests encompassed energy conservation in microcanonical ensemble simulations, fidelity in canonical ensemble generation, and the evaluation of the structural, transport, and thermodynamic properties of bulk water. The introduction of this plugin is expected to significantly expand the application scope of DP models within the MD simulation community, representing a major advancement in the field.

The Relationship between Learning Potential in Preschool Children and their Cognitive Abilities / La relación entre el potencial de aprendizaje en niños de preescolar y sus habilidades cognitivas

Learning potential can be used to assess and provide direction for action in diverse populations in modern educational settings. This study examined the relationship between the EHPAP dynamic evaluation scale and cognitive abilities (linguistic development, specific cognitive abilities, and executive functioning) in typically developing preschool children. Linear regression models indicated that children’s cognitive abilities formed part of the prediction of their independent performance in tasks before and after mediation. The abilities that formed part of the predictions were related to general linguistic development, specific cognitive functions, and executive functioning. Age as a predictive factor only contributed to verbal planning. This study examined the contribution of these variables in different EHPAP scales. (AU)

El potencial de aprendizaje se puede utilizar para evaluar y orientar la acción en diversas poblaciones en entornos educativos modernos. Este estudio analiza la relación entre la escala de evaluación dinámica EHPAP y las habilidades cognitivas (desarrollo lingüístico, habilidades cognitivas específicas y funcionamiento ejecutivo) en niños en edad preescolar con un desarrollo típico. Los modelos de regresión lineal indican que las habilidades cognitivas de los niños forman parte de la predicción de su desempeño independiente en tareas antes y después de la mediación. Las habilidades que formaban parte de las predicciones estaban relacionadas con el desarrollo lingüístico general, las funciones cognitivas específicas y el funcionamiento ejecutivo. La edad como factor predictivo solo contribuye a la planificación verbal. Este estudio analiza la contribución de estas variables en diferentes escalas EHPAP. (AU)

Training machine learning potentials for reactive systems: A Colab tutorial on basic models.

In the last several years, there has been a surge in the development of machine learning potential (MLP) models for describing molecular systems. We are interested in a particular area of this field - the training of system-specific MLPs for reactive systems - with the goal of using these MLPs to accelerate free energy simulations of chemical and enzyme reactions. To help new members in our labs become familiar with the basic techniques, we have put together a self-guided Colab tutorial (https://cc-ats.github.io/mlp_tutorial/), which we expect to be also useful to other young researchers in the community. Our tutorial begins with the introduction of simple feedforward neural network (FNN) and kernel-based (using Gaussian process regression, GPR) models by fitting the two-dimensional Müller-Brown potential. Subsequently, two simple descriptors are presented for extracting features of molecular systems: symmetry functions (including the ANI variant) and embedding neural networks (such as DeepPot-SE). Lastly, these features will be fed into FNN and GPR models to reproduce the energies and forces for the molecular configurations in a Claisen rearrangement reaction.

Long Time CO2 Storage Under Ambient Conditions in Isolated Voids of a Porous Coordination Network Facilitated by the "Magic Door" Mechanism.

A coordination network containing isolated pores without interconnecting channels is prepared from a tetrahedral ligand and copper(I) iodide. Despite the lack of accessibility, CO2 is selectively adsorbed into these pores at 298 K and then retained for more than one week while exposed to the atmosphere. The CO2 adsorption energy and diffusion mechanism throughout the network are simulated using Matlantis, which helps to rationalize the experimental results. CO2 enters the isolated voids through transient channels, termed "magic doors", which can momentarily appear within the structure. Once inside the voids, CO2 remains locked in limiting its escape. This mechanism is facilitated by the flexibility of organic ligands and the pivot motion of cluster units. In situ powder X-ray diffraction revealed that the crystal structure change is negligible before and after CO2 capture, unlike gate-opening coordination networks. The uncovered CO2 sorption and retention ability paves the way for the design of sorbents based on isolated voids.

Ferroelectric Domain and Switching Dynamics in Curved In2Se3: First-Principles and Deep Learning Molecular Dynamics Simulations.

Despite its prevalence in experiments, the influence of complex strain on material properties remains understudied due to the lack of effective simulation methods. Here, the effects of bending, rippling, and bubbling on the ferroelectric domains are investigated in an In2Se3 monolayer by density functional theory and deep learning molecular dynamics simulations. Since the ferroelectric switching barrier can be increased (decreased) by tensile (compressive) strain, automatic polarization reversal occurs in α-In2Se3 with a strain gradient when it is subjected to bending, rippling, or bubbling deformations to create localized ferroelectric domains with varying sizes. The switching dynamics depends on the magnitude of curvature and temperature, following an Arrhenius-style relationship. This study not only provides a promising solution for cross-scale studies using deep learning but also reveals the potential to manipulate local polarization in ferroelectric materials through strain engineering.

Extremely Large Response of Phonon Coherence in Twisted Penta-NiN2 Bilayer.

Twisting has recently been demonstrated as an effective strategy for tuning the interactions between particles or quasi-particles in layered materials. Motivated by the recent experimental synthesis of pentagonal NiN2 sheet [ACS Nano 2021, 15, 13539], for the first time, the response of phonon coherence to twisting in bilayer penta-NiN2 , going beyond the particle-like phonon transport is studied. By using the unified theory of phonon transport and high order lattice anharmonicity, together with the self-consistent phonon theory, it is found that the lattice thermal conductivity is reduced by 80.6% from 33.35 to 6.47 W m-1 K-1 at 300 K when the layers are twisted. In particular, the contribution of phonon coherence is increased sharply by an order of magnitude, from 0.21 to 2.40 W m-1 K-1 , due to the reduced differences between the phonon frequencies and enhanced anharmonicity after the introduction of twist. The work provides a fundamental understanding of the phonon interaction in twisted pentagonal sheets.

Enhancing Immediate Memory, Potential Learning, and Working Memory with Transcranial Direct Current Stimulation in Healthy Older Adults.

BACKGROUND: Transcranial direct current stimulation (tDCS) has emerged as a prevention method or minimizer of the normal cognitive deterioration that occurs during the aging process. tDCS can be used to enhance cognitive functions such as immediate memory, learning, or working memory in healthy subjects. The objective of this study was to analyze the effect of two 20-min sessions of anodal transcranial direct stimulation on immediate memory, learning potential, and working memory in healthy older adults. METHODS: A randomized, single-blind, repeated-measures, sham-controlled design was used. The sample is made up of 31 healthy older adults, of whom 16 were in the stimulation group and 15 were in the sham group. The anode was placed on position F7, coinciding with the left dorsolateral prefrontal cortex region, and the cathode was placed on Fp2, the right supraorbital area (rSO). RESULTS: When comparing the results of the treatment group and the sham group, differences were observed in working memory and learning potential; however, no differences in immediate memory were found. CONCLUSION: The results showed that tDCS is a non-invasive and safe tool to enhance cognitive processes in healthy older adults interested in maintaining some cognitive function.

Exploring the Effects of Ionic Defects on the Stability of CsPbI3 with a Deep Learning Potential.

Inorganic metal halide perovskites, such as CsPbI3 , have recently drawn extensive attention due to their excellent optical properties and high photoelectric efficiencies. However, the structural instability originating from inherent ionic defects leads to a sharp drop in the photoelectric efficiency, which significantly limits their applications in solar cells. The instability induced by ionic defects remains unresolved due to its complicated reaction process. Herein, to explore the effects of ionic defects on stability, we develop a deep learning potential for a CsPbI3 ternary system based upon density functional theory (DFT) calculated data for large-scale molecular dynamics (MD) simulations. By exploring 2.4 million configurations, of which 7,730 structures are used for the training set, the deep learning potential shows an accuracy approaching DFT-level. Furthermore, MD simulations with a 5,000-atom system and a one nanosecond timeframe are performed to explore the effects of bulk and surface defects on the stability of CsPbI3 . This deep learning potential based MD simulation provides solid evidence together with the derived radial distribution functions, simulated diffraction of X-rays, instability temperature, molecular trajectory, and coordination number for revealing the instability mechanism of CsPbI3 . Among bulk defects, Cs defects have the most significant influence on the stability of CsPbI3 with a defect tolerance concentration of 0.32 %, followed by Pb and I defects. With regards to surface defects, Cs defects have the largest impact on the stability of CsPbI3 when the defect concentration is less than 15 %, whereas Pb defects act play a dominant role for defect concentrations exceeding 20 %. Most importantly, this machine-learning-based MD simulation strategy provides a new avenue to explore the ionic defect effects on the stability of perovskite-like materials, laying a theoretical foundation for the design of stable perovskite materials.

Improving the accuracy of the neuroevolution machine learning potential for multi-component systems.

In a previous paper Fanet al(2021Phys. Rev.B104, 104309), we developed the neuroevolution potential (NEP), a framework of training neural network based machine-learning potentials using a natural evolution strategy and performing molecular dynamics (MD) simulations using the trained potentials. The atom-environment descriptor in NEP was constructed based on a set of radial and angular functions. For multi-component systems, all the radial functions between two atoms are multiplied by some fixed factors that depend on the types of the two atoms only. In this paper, we introduce an improved descriptor for multi-component systems, in which different radial functions are multiplied by different factors that are also optimized during the training process, and show that it can significantly improve the regression accuracy without increasing the computational cost in MD simulations.

[School-related attitudes of children from educationally disadvantaged backgrounds in the corona pandemic]. / Schulbezogene Einstellungen von Kindern aus bildungsfernen Milieus in der Corona-Pandemie.

There is widespread consensus that distance learning in lockdown discriminates children and young people from educationally disadvantaged backgrounds in particular. The gap between the privileged and the left-behind, which is already wide open in Germany, is widening more and more as a result of the COVID 19 pandemic. Empirical evidence for this pandemic-related effect is, however, still scarce, and the corresponding studies rarely include students themselves, and even less so those who are taught at schools in challenging situations. This paper reports on a quantitative study of elementary schools in particularly disadvantaged settings during the lockdown in spring 2021. It asks about individual attitudes and learning experiences of students during homeschooling. The goal was not to define a group of adolescents as disadvantaged and thus as disconnected, but to empirically investigate which individual factors, framework conditions, and mechanisms of action lead to some of these students being more successful in learning during the lockdown than others with initially equal starting conditions. The data analyses revealed four stable clusters, which are based on different forms of student agency and show potentials and opportunities to support students from educationally disadvantaged backgrounds individually-in homeschooling and in face-to-face teaching.

High Thermal Conductivity of Wurtzite Boron Arsenide Predicted by Including Four-Phonon Scattering with Machine Learning Potential.

Materials with high thermal conductivity are of great importance to the thermal management of modern electronic devices. Recently, it was found that cubic boron arsenide (c-BAs) is a high thermal conductivity (κ) material with a value of ∼1300 W/(m·K) at room temperature (RT), where four-phonon scattering plays a crucial role in limiting the κ. In this work, with four-phonon scattering included, we find that the κ of wurtzite BAs (w-BAs) reaches as high as 1036 W/(m·K) along the a-b plane at RT, decreasing by 43% compared to the calculation without considering four-phonon scattering. The similar phonon transport properties between c-BAs and w-BAs can be understood in terms of similar projected density of states and scattering rates, which have the origin in crystal structural resemblance. To accelerate the calculation, the moment tensor potential derived from machine learning is adopted and proven to be a reliable and efficient method to obtain high-order interatomic force constants.

High-temperature phonon transport properties of SnSe from machine-learning interatomic potential.

As a promising thermoelectric material, tin selenide (SnSe) is of relatively low thermal conductivity. However, the phonon transport mechanisms in SnSe are not fully understood due to the complex phase transition, dynamical instability, and strong anharmonicity. In this work, we perform molecular dynamics simulations with a machine-learning interatomic potential to explore the thermal transport properties of SnSe at different temperatures. The developed interatomic potential is parameterized using the framework of moment tensor potential, exhibiting satisfactory predictions on temperature-dependent lattice constants and phonon dispersion, as well as phase transition temperature. From equilibrium molecular dynamics simulations, we obtained the thermal conductivity tensor from 200 K to 900 K. The origins of temperature-dependent thermal conductivity anisotropy and the roles of four-phonon scatterings are identified. The obtained interatomic potential can be utilized to study the mechanical and thermal properties of SnSe and related nanostructures in a wide range of temperatures.