High-Throughput Virtual Screening of Polymer Electrolytes with Molecular Dynamics Simulations

Polymer electrolytes have been proposed as solid-state electrolyte materials for Li-ion batteries, with preferrable mechanical and safety properties over current liquid electrolytes. However, polymer electrolytes have relatively low ionic conductivity at room temperature. Exploring the chemical design space of polymer electrolytes in the lab is cumbersome, typically requiring synthesis of new monomers, their polymerization and their evaluation as electrolytes. The number of potential monomers and potential polymers to explore is too large for pure experimental techniques.<br/><br/>Molecular dynamics simulations (MD) can be used to estimate the properties of hypothetical polymers before they are made, but they are rarely used in high-throughput. Here, we will discuss a discovery campaign to explore the scalability of MD simulations for self-driving computational discovery of polymer electrolytes. The workflow includes selection of force field parameters for novel monomers or salts, the assembly of homo- or hetero-polymer chains and the automated construction of amorphous simulation boxes. Then, MD simulations at the 100-ns are performed and ionic conductivity is estimated from analysis of the MD trajectories. Our platform successfully recapitulates the conductivities of over a dozen experimentally known polymers. Lastly, we explore the role of glass transition temperature (Tg), and particularly the discrepancy between Tg values measured experimentally and those estimate from MD simulations, as a source of error in the calculation of conductivity.<br/><br/>Polymer electrolytes have been proposed as solid-state electrolyte materials for Li-ion batteries, with preferrable mechanical and safety properties over current liquid electrolytes. However, polymer electrolytes have relatively low ionic conductivity at room temperature. Exploring the chemical design space of polymer electrolytes in the lab is cumbersome, typically requiring synthesis of new monomers, their polymerization and their evaluation as electrolytes. The number of potential monomers and potential polymers to explore is too large for pure experimental techniques.<br/><br/>Molecular dynamics simulations (MD) can be used to estimate the properties of hypothetical polymers before they are made, but they are rarely used in high-throughput. Here, we will discuss a discovery campaign to explore the scalability of MD simulations for self-driving computational discovery of polymer electrolytes. The workflow includes selection of force field parameters for novel monomers or salts, the assembly of homo- or hetero-polymer chains and the automated construction of amorphous simulation boxes. Then, MD simulations at the 100-ns are performed and ionic conductivity is estimated from analysis of the MD trajectories. Our platform successfully recapitulates the conductivities of over a dozen experimentally known polymers. Lastly, we explore the role of glass transition temperature (Tg), and particularly the discrepancy between Tg values measured experimentally and those estimate from MD simulations, as a source of error in the calculation of conductivity.