MISPR: A High-Throughput Multi-Scale Infrastructure for Automating Materials Science Computations

Developing the capability to accurately predict the macroscopic properties of complex multicomponent solutions from microscopic features of molecular species is a grand challenge spanning chemistry, materials science, and engineering. Despite significant efforts undertaken in the past, the existing studies fall short of predicting the properties of multicomponent solutions due to multiple length and time scales of functional properties and a vast chemical and parameter space, which cannot be efficiently explored using trial and error based experimental approaches, brute-force computational approaches, and/or high-throughput computational approach that typically focus on just a single scale.<br/><br/>Driven by these needs, we developed a high-throughput computational framework coined <b>MISPR (Materials Informatics for Structure-Property Relationships - </b>https://github.com/molmd/mispr<b>) </b>that seamlessly integrates density functional theory (DFT) calculations with classical molecular dynamics (MD) simulations to robustly predict molecular and ensemble properties in complex multi-component liquid solutions. Functionalities of MISPR include <i>(i) </i>full automation of DFT and MD simulations, <i>(ii) </i>creation of computational databases for establishing structure-property relationships and maintaining data provenance and reproducibility, <i>(iii) </i>automatic error detection and handling, <i>(iv) </i>support for flexible and well-tested DFT workflows for computing properties such as bond dissociation energy, binding energy, and redox potentials, and <i>(v) </i>derivation of ensemble properties such radial distribution functions, ionic conductivity, and residence time. The infrastructure allows running 100-1000s of calculations in parallel by minimizing manual interference and generates high-fidelity databases of computational properties and force field parameters.<br/><br/>In this talk, I will demonstrate the unique features of MISPR by highlighting its different automated workflows and their application in designing optimal electrolytes for next-generation energy storage devices.¹ I will then show (1) a novel <i>“</i><b><i>DFT-CMD-DFT”</i></b> approach to accurately predict various stable species present in a multi-component solution by analyzing experimental nuclear magnetic resonance (NMR) spectra³ and (2) how MISPR can help screen thousands of potential solvents and salts to identify optimal electrolyte systems.<br/><br/><b>References </b><br/>1. Atwi, R.; Bliss, M.; Makeev, M.; Rajput, N. N., MISPR: an open-source package for high-throughput multiscale molecular simulations. <i>Scientific Reports </i><b>2022,</b> <i>12</i> (1), 15760.<br/>2. Atwi, R.; Chen, Y.; Han, K. S.; Mueller, K. T.; Murugesan, V.; Rajput, N. N., An automated framework for high-throughput predictions of NMR chemical shifts within liquid solutions. <i>Nature Computational Science </i><b>2022</b>.<br/>&lt;!--![endif]----&gt;