Data-Driven Path Collective Variables.

Identifying optimal collective variables to model transformations using atomic-scale simulations is a long-standing challenge. We propose a new method for the generation, optimization, and comparison of collective variables that can be thought of as a data-driven generalization of the path collective variable concept. It consists of a kernel ridge regression of the committor probability, which encodes a transformation's progress. The resulting collective variable is one-dimensional, interpretable, and differentiable, making it appropriate for enhanced sampling simulations requiring biasing. We demonstrate the validity of the method on two different applications: a precipitation model and the association of Li+ and F- in water. For the former, we show that global descriptors such as the permutation invariant vector allow reaching an accuracy far from the one achieved via simpler, more intuitive variables. For the latter, we show that information correlated with the transformation mechanism is contained in the first solvation shell only and that inertial effects prevent the derivation of optimal collective variables from the atomic positions only.

Accelerating amorphous polymer electrolyte screening by learning to reduce errors in molecular dynamics simulated properties.

Polymer electrolytes are promising candidates for the next generation lithium-ion battery technology. Large scale screening of polymer electrolytes is hindered by the significant cost of molecular dynamics (MD) simulation in amorphous systems: the amorphous structure of polymers requires multiple, repeated sampling to reduce noise and the slow relaxation requires long simulation time for convergence. Here, we accelerate the screening with a multi-task graph neural network that learns from a large amount of noisy, unconverged, short MD data and a small number of converged, long MD data. We achieve accurate predictions of 4 different converged properties and screen a space of 6247 polymers that is orders of magnitude larger than previous computational studies. Further, we extract several design principles for polymer electrolytes and provide an open dataset for the community. Our approach could be applicable to a broad class of material discovery problems that involve the simulation of complex, amorphous materials.

Anion Specific Effects Drive the Formation of Li-Salt Based Aqueous Biphasic Systems.

Aqueous biphasic systems (ABSs) can form when mixing water with two compounds such as polymers, ionic liquids, or simple salts. While this phenomenon has been known for decades and found applications in various fields such as biology, recycling, or even more recently electrochemistry, the physics behind the formation of ABSs remains ill-understood. It was recently demonstrated that ABSs can be composed of two salts sharing the same cation (Li+) but different anions (sulfonamide and halide). Interestingly, their formation could not be explained by the position of the anions within the chaotropic/kosmotropic series and was rather proposed to originate from an anion size mismatch, albeit the size for these anions was never measured yet owing to the lack of a proper experimental methodology. Here, we combine experimental techniques and molecular simulations to assess the specific effects (size, shape, hydrophobic/hydrophilic character) of a series of anions and correlate them with the formation of ABSs. We demonstrate that while the anion size mismatch is a prerequisite for the formation of Li-salts based ABSs, their shape can also play an important role, providing general guidelines for forming new ABSs with potential future applications.

Importance of Equilibration Method and Sampling for Ab Initio Molecular Dynamics Simulations of Solvent-Lithium-Salt Systems in Lithium-Oxygen Batteries.

We examine the effect of equilibration methodology and sampling on ab initio molecular dynamics (AIMD) simulations of systems of common solvents and salts found in lithium-oxygen batteries. We compare two equilibration methods: (1) using an AIMD temperature ramp and (2) using a classical MD simulation followed by a short AIMD simulation both at the target simulation temperature of 300 K. We also compare two different classical all-atom force fields: PCFF+ and OPLS. By comparing the simulated association/dissociation behavior of lithium salts in different solvents with the experimental behavior, we find that equilibration with the classical force field that produces more physically accurate behavior in the classical MD simulations, namely, OPLS, also results in more physically accurate behavior in the AIMD runs compared to equilibration with PCFF+ or with the AIMD temperature ramp. Equilibration with OPLS outperforms even the pure AIMD equilibration because the classical MD equilibration is much longer than the AIMD equilibration (nanosecond vs picosecond timescales). These longer classical simulations allow the systems to find a more physically accurate initial configuration, and in the short simulation times available for the AIMD production runs, the initial configuration has a large impact on the system behavior. We also demonstrate the importance of averaging coordination number over multiple starting configurations and Li+ ions, as the majority of Li+ ions do not undergo a single association or dissociation event even in an ∼40 ps long simulation and thus do not sample a statistically significant portion of the phase space. These results show the importance of both equilibration method and sufficient independent sampling for extracting experimentally relevant quantities from AIMD simulations.

Low-frequency Raman spectrum of 2D layered perovskites: Local atomistic motion or superlattice modes?

We report the low-frequency Raman spectrum (ω = 10 cm-1-150 cm-1) of a wide variety of alkylammonium iodide based 2D lead halide perovskites (2D LHPs) as a function of A-site cation (MA = methylammonium and FA = formamidinium), octahedral layer thickness (n = 2-4), organic spacer chain length (butyl-, pentyl-, hexyl-), and sample temperature (T = 77 K-293 K). Using density functional theory calculations under the harmonic approximation for n = 2 BA:MAPbI, we assign several longitudinal/transverse optical phonon modes between 30 cm-1 and 100 cm-1, the eigendisplacements of which are analogous to that observed previously for octahedral twists/distortions in bulk MAPbI. Additionally, we propose an alternative assignment for low-frequency modes below this band (<30 cm-1) as zone-folded longitudinal acoustic phonons corresponding to the periodicity of the entire layered structure. We compare measured spectra to predictions of the Rytov elastic continuum model for zone-folded dispersion in layered structures. Our results are consistent across the various 2D LHPs studied herein, with energetic shifts of optical phonons corresponding to microscopic structural differences between materials and energetic shifts of acoustic phonons according to changes in the periodicity and elastic properties of the perovskite/organic subphases. This study highlights the importance of both the local atomic order and the superlattice structure on the vibrational properties of layered 2D materials.

Quantitative Mapping of Molecular Substituents to Macroscopic Properties Enables Predictive Design of Oligoethylene Glycol-Based Lithium Electrolytes.

Molecular details often dictate the macroscopic properties of materials, yet due to their vastly different length scales, relationships between molecular structure and bulk properties can be difficult to predict a priori, requiring Edisonian optimizations and preventing rational design. Here, we introduce an easy-to-execute strategy based on linear free energy relationships (LFERs) that enables quantitative correlation and prediction of how molecular modifications, i.e., substituents, impact the ensemble properties of materials. First, we developed substituent parameters based on inexpensive, DFT-computed energetics of elementary pairwise interactions between a given substituent and other constant components of the material. These substituent parameters were then used as inputs to regression analyses of experimentally measured bulk properties, generating a predictive statistical model. We applied this approach to a widely studied class of electrolyte materials: oligo-ethylene glycol (OEG)-LiTFSI mixtures; the resulting model enables elucidation of fundamental physical principles that govern the properties of these electrolytes and also enables prediction of the properties of novel, improved OEG-LiTFSI-based electrolytes. The framework presented here for using context-specific substituent parameters will potentially enhance the throughput of screening new molecular designs for next-generation energy storage devices and other materials-oriented contexts where classical substituent parameters (e.g., Hammett parameters) may not be available or effective.

Atomic structure and defect dynamics of monolayer lead iodide nanodisks with epitaxial alignment on graphene.

Lead Iodide (PbI2) is a large bandgap 2D layered material that has potential for semiconductor applications. However, atomic level study of PbI2 monolayer has been limited due to challenges in obtaining thin crystals. Here, we use liquid exfoliation to produce monolayer PbI2 nanodisks (30-40 nm in diameter and > 99% monolayer purity) and deposit them onto suspended graphene supports to enable atomic structure study of PbI2. Strong epitaxial alignment of PbI2 monolayers with the underlying graphene lattice occurs, leading to a phase shift from the 1 T to 1 H structure to increase the level of commensuration in the two lattice spacings. The fundamental point vacancy and nanopore structures in PbI2 monolayers are directly imaged, showing rapid vacancy migration and self-healing. These results provide a detailed insight into the atomic structure of monolayer PbI2, and the impact of the strong van der Waals interaction with graphene, which has importance for future applications in optoelectronics.

Ionic Highways from Covalent Assembly in Highly Conducting and Stable Anion Exchange Membrane Fuel Cells.

A major challenge in the development of anion exchange membranes for fuel cells is the design and synthesis of highly stable (chemically and mechanically) conducting membranes. Membranes that can endure highly alkaline environments while rapidly transporting hydroxides are desired. Herein, we present a design using cross-linked polymer membranes containing ionic highways along charge-delocalized pyrazolium cations and homoconjugated triptycenes. These ionic highway membranes show improved performance. Specifically, a conductivity of 111.6 mS cm-1 at 80 °C was obtained with a low 7.9% water uptake and 0.91 mmol g-1 ion exchange capacity. In contrast to existing materials, ionic highways produce higher conductivities at reduced hydration and ionic exchange capacities. The membranes retain more than 75% of their initial conductivity after 30 days of an alkaline stability test. The formation of ionic highways for ion transport is confirmed by density functional theory and Monte Carlo studies. A single cell with platinum metal catalysts at 80 °C showed a high peak density of 0.73 W cm-2 (0.45 W cm-2 from a silver-based cathode) and stable performance throughout 400 h tests.

Tuning the Potential Energy Landscape to Suppress Ostwald Ripening in Surface-Supported Catalyst Systems.

Rational control of nanoparticle (NP) size distribution during operation is crucial to improve catalytic performance and noble metal sustainability. Herein, we explore the Ostwald ripening (OR) of metal atoms on zeolite surfaces by a coupled theoretical-experimental approach. Zeolites with the same structure (ZSM-5) but different concentrations of aluminum doped into the matrix were observed to yield systematic differences in supported nanoparticle size distributions. Our first-principles simulations suggest that NP stability at high temperature is governed by both geometric constraints and the roughness of the energetic landscape. Calculated adatom migration paths across the zeolite surface and desorption paths from the supported NPs lend insight into the modified OR sintering processes with the emergence of different binding configurations as the aluminum concentration increases from pristine to heavily doped ZSM-5. These findings reveal the potential for the rational design of support structures to suppress OR sintering.

Highly efficient evaluation of diffusion networks in Li ionic conductors using a 3D-corrugation descriptor.

A highly efficient computational approach for the screening of Li ion conducting materials is presented and its performance is demonstrated for olivine-type oxides and thiophosphates. The approach is based on a topological analysis of the electrostatic (Coulomb) potential obtained from a single density functional theory calculation augmented by a Born-Mayer-type repulsive term between Li ions and the anions of the material. This 3D-corrugation descriptor enables the automatic determination of diffusion pathways in one, two, and three dimensions and reproduces migration barriers obtained from density functional theory calculations using nudged elastic band method within approximately 0.1 eV. Importantly, it correlates with Li ion conductivity. This approach thus offers an efficient tool for evaluating, ranking, and optimizing materials with high Li-ion conductivity.

Graph dynamical networks for unsupervised learning of atomic scale dynamics in materials.

Understanding the dynamical processes that govern the performance of functional materials is essential for the design of next generation materials to tackle global energy and environmental challenges. Many of these processes involve the dynamics of individual atoms or small molecules in condensed phases, e.g. lithium ions in electrolytes, water molecules in membranes, molten atoms at interfaces, etc., which are difficult to understand due to the complexity of local environments. In this work, we develop graph dynamical networks, an unsupervised learning approach for understanding atomic scale dynamics in arbitrary phases and environments from molecular dynamics simulations. We show that important dynamical information, which would be difficult to obtain otherwise, can be learned for various multi-component amorphous material systems. With the large amounts of molecular dynamics data generated every day in nearly every aspect of materials design, this approach provides a broadly applicable, automated tool to understand atomic scale dynamics in material systems.

Correlations from Ion Pairing and the Nernst-Einstein Equation.

We present a new approximation to ionic conductivity well suited to dynamical atomic-scale simulations, based on the Nernst-Einstein equation. In our approximation, ionic aggregates constitute the elementary charge carriers, and are considered as noninteracting species. This approach conveniently captures the dominant effect of ion-ion correlations on conductivity, short range interactions in the form of clustering. In addition to providing better estimates to the conductivity at a lower computational cost than exact approaches, this new method allows us to understand the physical mechanisms driving ion conduction in concentrated electrolytes. As an example, we consider Li^{+} conduction in poly(ethylene oxide), a standard solid-state polymer electrolyte. Using our newly developed approach, we are able to reproduce recent experimental results reporting negative cation transference numbers at high salt concentrations, and to confirm that this effect can be caused by a large population of negatively charged clusters involving cations.

Revealing the Cluster-Cloud and Its Role in Nanocrystallization.

Elucidating the early stages of crystallization from supersaturated solutions is of critical importance, but remains a great challenge. An in situ liquid cell transmission electron microscopy study reveals an intermediate state of condensed atomic clusters during Pd and Au crystallizations, which is named a "cluster-cloud." It is found that nucleation is initiated by the collapse of a cluster-cloud, first forming a nanoparticle. The subsequent particle maturation proceeds via multiple out-and-in relaxations of the cluster-cloud to improve crystallinity: from a poorly crystallized phase, the particle evolves into a well-defined single-crystal phase. Both experimental investigations and atomistic simulations suggest that the cluster-cloud-mediated nanocrystallization involves an order-disorder phase separation and reconstruction, which is energetically favored compared to local rearrangements within the particle. This finding grants new insights into nanocrystallization mechanisms, and provides useful information for the improvement of synthesis pathways of nanocrystals.

Atomic Structure and Dynamics of Self-Limiting Sub-Nanometer Pores in Monolayer WS2.

We reveal a self-limiting mechanism during the formation of a specific type of circular nanopore in monolayer WS2 that limits its diameter to sub-nm. A single W atom vacancy (triangular nanopore) is transformed into the self-limiting nanopore (SLNP) through the atomic restructuring of S atoms around the area, reducing the number of dangling bonds at the nanopore edge by shifting them further in-plane with W-W bonding instead. Bond rotations in WS2 help accommodate the electron beam induced atomic loss and ensure the stability of the SLNP. The SLNP shows significant improvement in diameter stability during electron beam irradiation compared to other triangular nanopores in WS2 that typically continue to expand in diameter during atom loss. The atomic structure of these SLNPs is studied using aberration-corrected scanning transmission electron microscopy with an in situ heating holder, revealing that the SLNPs are mostly formed at a temperature of ∼500 °C, which is a balance between thermally activated S vacancy diffusion and sufficient S vacancy density to initiate local atomic reconstruction. At higher temperatures ( i. e., 1000 °C), S vacancies quickly migrate away into long line vacancies, resulting in low S vacancy density and rapidly expanding holes generated at the edges of the line vacancies. At room temperature, S vacancy migration is low and vacancy density is very high, which limits atomic reconstruction, and instead many small holes open up. These results provide insights into the factors that lead to uniform sized nanopores in the sub-nm range in transition-metal dichalcogenides.

Ab initio parameterization of a charge optimized many-body forcefield for Si-SiO2: Validation and thermal transport in nanostructures.

In an effort to extend the reach of current ab initio calculations to simulations requiring millions of configurations for complex systems such as heterostructures, we have parameterized the third-generation Charge Optimized Many-Body (COMB3) potential using solely ab initio total energies, forces, and stress tensors as input. The quality and the predictive power of the new forcefield are assessed by computing properties including the cohesive energy and density of SiO2 polymorphs, surface energies of alpha-quartz, and phonon densities of states of crystalline and amorphous phases of SiO2. Comparison with data from experiments, ab initio calculations, and molecular dynamics simulations using published forcefields including BKS (van Beest, Kramer, and van Santen), ReaxFF, and COMB2 demonstrates an overall improvement of the new parameterization. The computed temperature dependence of the thermal conductivity of crystalline alpha-quartz and the Kapitza resistance of the interface between crystalline Si(001) and amorphous silica is in excellent agreement with experiment, setting the stage for simulations of complex nanoscale heterostructures.

Thermal properties of amorphous/crystalline silicon superlattices.

Thermal transport properties of crystalline/amorphous silicon superlattices using molecular dynamics are investigated. We show that the cross-plane conductivity of the superlattices is very low and close to the conductivity of bulk amorphous silicon even for amorphous layers as thin as ≃ 6 Å. The cross-plane thermal conductivity weakly increases with temperature which is associated with a decrease of the Kapitza resistance with temperature at the crystalline/amorphous interface. This property is further investigated considering the spatial analysis of the phonon density of states in domains close to the interface. Interestingly, the crystalline/amorphous superlattices are shown to display large thermal anisotropy, according to the characteristic sizes of elaborated structures. These last results suggest that the thermal conductivity of crystalline/amorphous superlattices can be phonon engineered, providing new directions for nanostructured thermoelectrics and anisotropic materials in thermal transport.

Atomistic amorphous/crystalline interface modelling for superlattices and core/shell nanowires.

In this paper we present a systematic and well controlled procedure for building atomistic amorphous/crystalline interfaces in silicon, dedicated to the molecular dynamics simulations of superlattices and core/shell nanowires. The obtained structures depend on the technique used to generate the amorphous phase and their overall quality is estimated through comparisons with structural information and interfacial energies available from experimental and theoretical results. While most of the related studies focus on a single planar interface, we consider here both the generation of multiple superlattice planar interfaces and core/shell nanowire structures. The proposed method provides periodic homogeneous and reproducible, atomically sharp and defect free interface configurations at low temperature and pressure. We also illustrate how the method may be used to predict the thermal transport properties of composite crystalline/amorphous superlattices.