Deep-Learning Interatomic Potential Connects Molecular Structural Ordering to the Macroscale Properties of Polyacrylonitrile.

Polyacrylonitrile (PAN) is an important commercial polymer, bearing atactic stereochemistry resulting from nonselective radical polymerization. As such, an accurate, fundamental understanding of governing interactions among PAN molecular units is indispensable for advancing the design principles of final products at reduced processability costs. While ab initio molecular dynamics (AIMD) simulations can provide the necessary accuracy for treating key interactions in polar polymers, such as dipole-dipole interactions and hydrogen bonding, and analyzing their influence on the molecular orientation, their implementation is limited to small molecules only. Herein, we show that the neural network interatomic potentials (NNIPs) that are trained on the small-scale AIMD data (acquired for oligomers) can be efficiently employed to examine the structures and properties at large scales (polymers). NNIP provides critical insight into intra- and interchain hydrogen-bonding and dipolar correlations and accurately predicts the amorphous bulk PAN structure validated by modeling the experimental X-ray structure factor. Furthermore, the NNIP-predicted PAN properties, such as density and elastic modulus, are in good agreement with their experimental values. Overall, the trend in the elastic modulus is found to correlate strongly with the PAN structural orientations encoded in the Hermans orientation factor. This study enables the ability to predict the structure-property relations for PAN and analogues with sustainable ab initio accuracy across scales.

Chemistry Informed Machine Learning-Based Heat Capacity Prediction of Solid Mixed Oxides.

Knowing heat capacity is crucial for modeling temperature changes with the absorption and release of heat and for calculating the thermal energy storage capacity of oxide mixtures with energy applications. The current prediction methods (ab initio simulations, computational thermodynamics, and the Neumann-Kopp rule) are computationally expensive, not fully generalizable, or inaccurate. Machine learning has the potential of being fast, accurate, and generalizable, but it has been scarcely used to predict mixture properties, particularly for mixed oxides. Here, we demonstrate a method for the generalizable prediction of heat capacity of solid oxide pseudobinary mixtures using heat capacity data obtained from computational thermodynamics and descriptors from ab initio databases. Models trained through this workflow achieved an error (mean absolute error of 0.43 J mol-1 K-1) lower than the uncertainty in differential scanning calorimetry measurements, and the workflow can be extended to predict other properties derived from the Gibbs free energy and for higher-order oxide mixtures.

Tracing mechanistic pathways and reaction kinetics toward equilibrium in reactive molten salts.

In the dynamic environment of multi-component reactive molten salts, speciation unfolds as a complex process, involving multiple competing reaction pathways that are likely to face free energy barriers before reaching the reaction equilibria. Herein, we unravel intricate speciation in the AlCl3-KCl melt compositions with rate theory and ab initio molecular dynamics simulations. We find that the compositions with 100 and 50 mol% AlCl3 exclusively comprise neutral Al2Cl6 dimers and charged AlCl4- monomers, respectively. In intermediate AlCl3-KCl compositions, the chemical speciation proves to be a very complex process, requiring over 0.5 nanosecond to reach an equilibrium distribution of multiple species. It is a consequence of the competitive formation and dissociation of additional species, including charged Al dimers, trimers, and tetramers. Here, the species formation occurs through ion exchange events, which we explain by computing free energy landscapes and employing a Marcus-like rate theory. We show that both interspecies and intraspecies ion exchanges are probable and are dictated by the local structural reorganization reflected in the change of local coulombic fields. The species distributions are validated by comparing computed Raman spectra and neutron structure factors with the available experimental data. We find an excellent simulation-experiment agreement in both cases. Nevertheless, Raman spectroscopy turns out to be particularly advantageous for distinguishing between unique species distributions because of the distinct vibrational signatures of different species. The mechanistic insight into reaction dynamics gained in this study will be essential for the advancement of molten salts as reactive media in high-temperature energy applications.

Transferable Deep Learning Potential Reveals Intermediate-Range Ordering Effects in LiF-NaF-ZrF4 Molten Salt.

LiF-NaF-ZrF4 multicomponent molten salts are promising candidate coolants for advanced clean energy systems owing to their desirable thermophysical and transport properties. However, the complex structures enabling these properties, and their dependence on composition, is scarcely quantified due to limitations in simulating and interpreting experimental spectra of highly disordered, intermediate-ranged structures. Specifically, size-limited ab initio simulations and accuracy-limited classical models used in the past are unable to capture a wide range of fluctuating motifs found in the extended heterogeneous structures of liquid salt. This greatly inhibits our ability to design tailored compositions and materials. Here, accurate, efficient, and transferable machine learning potentials are used to predict structures far beyond the first coordination shell in LiF-NaF-ZrF4. Neural networks trained at only eutectic compositions with 29% and 37% ZrF4 are shown to accurately simulate a wide range of compositions (11-40% ZrF4) with dramatically different coordination chemistries, while showing a remarkable agreement with theoretical and experimental Raman spectra. The theoretical Raman calculations further uncovered the previously unseen shift and flattening of bending band at ∼250 cm-1 which validated the simulated extended-range structures as observed in compositions with higher than 29% ZrF4 content. In such cases, machine learning-based simulations capable of accessing larger time and length scales (beyond 17 Å) were critical for accurately predicting both structure and ionic diffusivities.