ColabFit exchange: Open-access datasets for data-driven interatomic potentials.

Data-driven interatomic potentials (IPs) trained on large collections of first principles calculations are rapidly becoming essential tools in the fields of computational materials science and chemistry for performing atomic-scale simulations. Despite this, apart from a few notable exceptions, there is a distinct lack of well-organized, public datasets in common formats available for use with IP development. This deficiency precludes the research community from implementing widespread benchmarking, which is essential for gaining insight into model performance and transferability, and also limits the development of more general, or even universal, IPs. To address this issue, we introduce the ColabFit Exchange, the first database providing open access to a large collection of systematically organized datasets from multiple domains that is especially designed for IP development. The ColabFit Exchange is publicly available at https://colabfit.org, providing a web-based interface for exploring, downloading, and contributing datasets. Composed of data collected from the literature or provided by community researchers, the ColabFit Exchange currently (September 2023) consists of 139 datasets spanning nearly 70 000 unique chemistries, and is intended to continuously grow. In addition to outlining the software framework used for constructing and accessing the ColabFit Exchange, we also provide analyses of the data, quantifying the diversity of the database and proposing metrics for assessing the relative diversity of multiple datasets. Finally, we demonstrate an end-to-end IP development pipeline, utilizing datasets from the ColabFit Exchange, fitting tools from the KLIFF software package, and validation tests provided by the OpenKIM framework.

Bayesian, frequentist, and information geometric approaches to parametric uncertainty quantification of classical empirical interatomic potentials.

In this paper, we consider the problem of quantifying parametric uncertainty in classical empirical interatomic potentials (IPs) using both Bayesian (Markov Chain Monte Carlo) and frequentist (profile likelihood) methods. We interface these tools with the Open Knowledgebase of Interatomic Models and study three models based on the Lennard-Jones, Morse, and Stillinger-Weber potentials. We confirm that IPs are typically sloppy, i.e., insensitive to coordinated changes in some parameter combinations. Because the inverse problem in such models is ill-conditioned, parameters are unidentifiable. This presents challenges for traditional statistical methods, as we demonstrate and interpret within both Bayesian and frequentist frameworks. We use information geometry to illuminate the underlying cause of this phenomenon and show that IPs have global properties similar to those of sloppy models from fields, such as systems biology, power systems, and critical phenomena. IPs correspond to bounded manifolds with a hierarchy of widths, leading to low effective dimensionality in the model. We show how information geometry can motivate new, natural parameterizations that improve the stability and interpretation of uncertainty quantification analysis and further suggest simplified, less-sloppy models.

Atomic and electronic reconstruction at the van der Waals interface in twisted bilayer graphene.

Control of the interlayer twist angle in two-dimensional van der Waals (vdW) heterostructures enables one to engineer a quasiperiodic moiré superlattice of tunable length scale1-8. In twisted bilayer graphene, the simple moiré superlattice band description suggests that the electronic bandwidth can be tuned to be comparable to the vdW interlayer interaction at a 'magic angle'9, exhibiting strongly correlated behaviour. However, the vdW interlayer interaction can also cause significant structural reconstruction at the interface by favouring interlayer commensurability, which competes with the intralayer lattice distortion10-16. Here we report atomic-scale reconstruction in twisted bilayer graphene and its effect on the electronic structure. We find a gradual transition from an incommensurate moiré structure to an array of commensurate domains with soliton boundaries as we decrease the twist angle across the characteristic crossover angle, θc ≈ 1°. In the solitonic regime (θ < θc) where the atomic and electronic reconstruction become significant, a simple moiré band description breaks down and the secondary Dirac bands appear. On applying a transverse electric field, we observe electronic transport along the network of one-dimensional topological channels that surround the alternating triangular gapped domains. Atomic and electronic reconstruction at the vdW interface provide a new pathway to engineer the system with continuous tunability.

Mapping the stochastic response of nanostructures.

Nanostructures are technological devices constructed on a nanometer length scale more than a thousand times thinner than a human hair. Due to the unique properties of matter at this scale, such devices offer great potential for creating novel materials and behaviors that can be leveraged to benefit mankind. This paper addresses a particular challenge involved in the design of nanostructures--their stochastic or apparently random response to external loading. This is because fundamentally the function that relates the energy of a nanostructure to the arrangement of its atoms is extremely nonconvex, with each minimum corresponding to a possible equilibrium state that may be visited as the system responds to loading. Traditional atomistic simulation techniques are not capable of systematically addressing this complexity. Instead, we construct an equilibrium map (EM) for the nanostructure, analogous to a phase diagram for bulk materials, which fully characterizes its response. Using the EM, definitive predictions can be made in limiting cases and the spectrum of responses at any desired loading rate can be obtained. The latter is important because standard atomistic methods are fundamentally limited, by computational feasibility, to simulations of loading rates that are many orders of magnitude faster than reality. In contrast, the EM-based approach makes possible the direct simulation of nanostructure experiments. We demonstrate the method's capabilities and its surprisingly complex results for the case of a nanoslab of nickel under compression.

Molecular dynamics at constant Cauchy stress.

The Parrinello-Rahman algorithm for imposing a general state of stress in periodic molecular dynamics simulations is widely used in the literature and has been implemented in many readily available molecular dynamics codes. However, what is often overlooked is that this algorithm controls the second Piola-Kirchhoff stress as opposed to the true (Cauchy) stress. This can lead to misinterpretation of simulation results because (1) the true stress that is imposed during the simulation depends on the deformation of the periodic cell, (2) the true stress is potentially very different from the imposed second Piola-Kirchhoff stress, and (3) the true stress can vary significantly during the simulation even if the imposed second Piola-Kirchhoff is constant. We propose a simple modification to the algorithm that allows the true Cauchy stress to be controlled directly. We then demonstrate the efficacy of the new algorithm with the example of martensitic phase transformations under applied stress.

Entropically stabilized dislocations.

Dislocations are line defects that play a key role in the plasticity of crystalline materials and affect their thermal, chemical, and electrical properties. Typically dislocations are treated as stable defects; e.g., the equilibrium core structure of a dislocation is obtained by minimizing the crystal potential energy with respect to atom positions. Here we show for the first time the possibility of "entropically stabilized dislocations" that exist due to entropic effects without a corresponding potential energy well. An entropically stabilized dislocation was discovered in an accelerated multiscale quasicontinuum simulation. Its entropic nature was verified with fully atomistic free energy calculations and explained by a simple continuum-based model. This result has important consequences for the study of dislocations as well as for temporal multiscale methods that use information from the potential energy surface to accelerate time in molecular simulations.

Type Label Framework for Bonded Force Fields in LAMMPS.

New functionality is added to the LAMMPS molecular simulation package, which increases the versatility with which LAMMPS can interface with supporting software and manipulate information associated with bonded force fields. We introduce the "type label" framework that allows atom types and their higher-order interactions (bonds, angles, dihedrals, and impropers) to be represented in terms of the standard atom type strings of a bonded force field. Type labels increase the human readability of input files, enable bonded force fields to be supported by the OpenKIM repository, simplify the creation of reaction templates for the REACTER protocol, and increase compatibility with external visualization tools, such as VMD and OVITO. An introductory primer on the forms and use of bonded force fields is provided to motivate this new functionality and serve as an entry point for LAMMPS and OpenKIM users unfamiliar with bonded force fields. The type label framework has the potential to streamline modeling workflows that use LAMMPS by increasing the portability of software, files, and scripts for preprocessing, running, and postprocessing a molecular simulation.

Accurate Force Fields for Atomistic Simulations of Oxides, Hydroxides, and Organic Hybrid Materials up to the Micrometer Scale.

The simulation of metals, oxides, and hydroxides can accelerate the design of therapeutics, alloys, catalysts, cement-based materials, ceramics, bioinspired composites, and glasses. Here we introduce the INTERFACE force field (IFF) and surface models for α-Al2O3, α-Cr2O3, α-Fe2O3, NiO, CaO, MgO, ß-Ca(OH)2, ß-Mg(OH)2, and ß-Ni(OH)2. The force field parameters are nonbonded, including atomic charges for Coulomb interactions, Lennard-Jones (LJ) potentials for van der Waals interactions with 12-6 and 9-6 options, and harmonic bond stretching for hydroxide ions. The models outperform DFT calculations and earlier atomistic models (Pedone, ReaxFF, UFF, CLAYFF) up to 2 orders of magnitude in reliability, compatibility, and interpretability due to a quantitative representation of chemical bonding consistent with other compounds across the periodic table and curated experimental data for validation. The IFF models exhibit average deviations of 0.2% in lattice parameters, <10% in surface energies (to the extent known), and 6% in bulk moduli relative to experiments. The parameters and models can be used with existing parameters for solvents, inorganic compounds, organic compounds, biomolecules, and polymers in IFF, CHARMM, CVFF, AMBER, OPLS-AA, PCFF, and COMPASS, to simulate bulk oxides, hydroxides, electrolyte interfaces, and multiphase, biological, and organic hybrid materials at length scales from atoms to micrometers. The nonbonded character of the models also enables the analysis of mixed oxides, glasses, and certain chemical reactions, and well-performing nonbonded models for silica phases, SiO2, are introduced. Automated model building is available in the CHARMM-GUI Nanomaterial Modeler. We illustrate applications of the models to predict the structure of mixed oxides, and energy barriers of ion migration, as well as binding energies of water and organic molecules in outstanding agreement with experimental data and calculations at the CCSD(T) level. Examples of model building for hydrated, pH-sensitive oxide surfaces to simulate solid-electrolyte interfaces are discussed.

CHARMM-GUI Nanomaterial Modeler for Modeling and Simulation of Nanomaterial Systems.

Molecular modeling and simulation are invaluable tools for nanoscience that predict mechanical, physicochemical, and thermodynamic properties of nanomaterials and provide molecular-level insight into underlying mechanisms. However, building nanomaterial-containing systems remains challenging due to the lack of reliable and integrated cyberinfrastructures. Here we present Nanomaterial Modeler in CHARMM-GUI, a web-based cyberinfrastructure that provides an automated process to generate various nanomaterial models, associated topologies, and configuration files to perform state-of-the-art molecular dynamics simulations using most simulation packages. The nanomaterial models are based on the interface force field, one of the most reliable force fields (FFs). The transferability of nanomaterial models among the simulation programs was assessed by single-point energy calculations, which yielded 0.01% relative absolute energy differences for various surface models and equilibrium nanoparticle shapes. Three widely used Lennard-Jones (LJ) cutoff methods are employed to evaluate the compatibility of nanomaterial models with respect to conventional biomolecular FFs: simple truncation at r = 12 Å (12 cutoff), force-based switching over 10 to 12 Å (10-12 fsw), and LJ particle mesh Ewald with no cutoff (LJPME). The FF parameters with these LJ cutoff methods are extensively validated by reproducing structural, interfacial, and mechanical properties. We find that the computed density and surface energies are in good agreement with reported experimental results, although the simulation results increase in the following order: 10-12 fsw <12 cutoff < LJPME. Nanomaterials in which LJ interactions are a major component show relatively higher deviations (up to 4% in density and 8% in surface energy differences) compared with the experiment. Nanomaterial Modeler's capability is also demonstrated by generating complex systems of nanomaterial-biomolecule and nanomaterial-polymer interfaces with a combination of existing CHARMM-GUI modules. We hope that Nanomaterial Modeler can be used to carry out innovative nanomaterial modeling and simulations to acquire insight into the structure, dynamics, and underlying mechanisms of complex nanomaterial-containing systems.

Torsional periodic lattice distortions and diffraction of twisted 2D materials.

Twisted 2D materials form complex moiré structures that spontaneously reduce symmetry through picoscale deformation within a mesoscale lattice. We show twisted 2D materials contain a torsional displacement field comprised of three transverse periodic lattice distortions (PLD). The torsional PLD amplitude provides a single order parameter that concisely describes the structural complexity of twisted bilayer moirés. Moreover, the structure and amplitude of a torsional periodic lattice distortion is quantifiable using rudimentary electron diffraction methods sensitive to reciprocal space. In twisted bilayer graphene, the torsional PLD begins to form at angles below 3.89° and the amplitude reaches 8 pm around the magic angle of 1. 1°. At extremely low twist angles (e.g. below 0.25°) the amplitude increases and additional PLD harmonics arise to expand Bernal stacked domains separated by well defined solitonic boundaries. The torsional distortion field in twisted bilayer graphene is analytically described and has an upper bound of 22.6 pm. Similar torsional distortions are observed in twisted WS2, CrI3, and WSe2/MoSe2.

Holey Substrate-Directed Strain Patterning in Bilayer MoS2.

Key properties of two-dimensional (2D) layered materials are highly strain tunable, arising from bond modulation and associated reconfiguration of the energy bands around the Fermi level. Approaches to locally controlling and patterning strain have included both active and passive elastic deformation via sustained loading and templating with nanostructures. Here, by float-capturing ultrathin flakes of single-crystal 2H-MoS2 on amorphous holey silicon nitride substrates, we find that highly symmetric, high-fidelity strain patterns are formed. The hexagonally arranged holes and surface topography combine to generate highly conformal flake-substrate coverage creating patterns that match optimal centroidal Voronoi tessellation in 2D Euclidean space. Using TEM imaging and diffraction, as well as AFM topographic mapping, we determine that the substrate-driven 3D geometry of the flakes over the holes consists of symmetric, out-of-plane bowl-like deformation of up to 35 nm, with in-plane, isotropic tensile strains of up to 1.8% (measured with both selected-area diffraction and AFM). Atomistic and image simulations accurately predict spontaneous formation of the strain patterns, with van der Waals forces and substrate topography as the input parameters. These results show that predictable patterns and 3D topography can be spontaneously induced in 2D materials captured on bare, holey substrates. The method also enables electron scattering studies of precisely aligned, substrate-free strained regions in transmission mode.