Overcoming Computational Barriers

Predicting interfacial reactions in rechargeable lithium batteries using small yet highly accurate atomistic simulations


by Lieven Bekaert, Ashish Raj, Jean-François Gohy, Annick Hubin, Frank De Proft, and Mesfin H. Mamme


​When studying the reactivity of chemical systems, atomistic simulations such as Density Functional Theory and Molecular Dynamics techniques provide unique insights into a size and timescale which is often out of reach for experiments. Take for instance rechargeable lithium-ion batteries, in which the migration of billions of lithium ions as we use our electronic devices causes significant detrimental structural and chemical changes. These processes reduce the lifespan and performance of batteries, and understanding and tackling these challenges is therefore critical to developing the rechargeable batteries of the future.


But with highly accurate simulations comes a price, which is their high computational cost. As a result, the accessible simulation size and time scales are very limited, often to only several hundreds of atoms and several picoseconds, making it a serious challenge to extrapolate the atomistic findings to the macroscopic scale and experiments. This is why in this study a new approach is introduced which makes it possible to predict reactions without actually seeing them take place. This can be achieved by statistically analyzing subtle changes in the distributions, rather than instantaneous values, of bond lengths inside molecules.

This research would not have been possible without the Tier-1 infrastructure provided by the VSC thanks to which we were able to perform the computationally expensive ab-initio molecular dynamics (AIMD) simulations. We were very pleased with the short submission queue times and their easy-to-use infrastructure and Tier-1 application process.

We applied this approach to study the reactivity of a solid polymer electrolyte at the negative electrode interface in rechargeable lithium batteries. In doing so we were able to identify the most reactive bond types and functional groups, and how their reactivity changed in function of the environment such as different electrode materials (graphite, silicon, lithium), temperatures, and electric field strengths. Furthermore, in the presence of an electric field and under special conditions, bonds were found to selectively change their orientation, which could be used as a new design principle in optimizing the interfacial stability.


The size and length of simulations, therefore, doesn’t need to be a limitation: by statistically analyzing subtle changes, reliable predictions can be made of the events which will take place far beyond those accessible with modern computational limitations.


This research would not have been possible without the Tier-1 infrastructure provided by the VSC thanks to which we were able to perform the computationally expensive ab-initio molecular dynamics (AIMD) simulations. We were very pleased with the short submission queue times and their easy-to-use infrastructure and Tier-1 application process.


The full publication can be consulted below:

Bekaert, L., Raj, A., Gohy, J. F., Hubin, A., De Proft, F., & Mamme, M. H. (2022). Assessing the Long-Term Reactivity to Achieve Compatible Electrolyte–Electrode Interfaces for Solid-State Rechargeable Lithium Batteries Using First-Principles Calculations. The Journal of Physical Chemistry C. https://doi.org/10.1021/acs.jpcc.2c01144