Molecular Simulation of per or Polyfluoroalkyl Substances for the Development of an Ideal Separation Media

The purpose of this research is to utilize computer simulation to study the adsorption and separation characteristics of per (or poly) fluoroalkyl substances (PFAS) inside of metal-organic frameworks (MOFs) in the search of an ideal adsorbent for water purification purposes. PFAS are a diverse class of highly stable compounds with a myriad of commercial and industrial applications, though recently concerningly high concentrations have been detected in drinking water globally and the adverse effects on human health are yet to be fully understood. In recent years, researchers have looked toward MOFs as a potential adsorbent as they are highly tunable and can plausibly create an environment specific enough to selectively adsorb different PFAS. Computer simulations allow us to test the relative ability of different MOFs to perform this separation through molecular dynamics (MD) and Monte Carlo (MC) techniques. PFAS simulation files were found online via the Automated Topology Builder (ATB) database and imported into LAMMPS, where we calculated diffusion coefficients. To ensure our molecular models are accurate, it is essential to compare the diffusion coefficients to results from the literature. Current results indicate our simulation set up is accurate as our simulated diffusion coefficients are within at least an order of magnitude of those reported in the literature, which is accurate for comparing diffusivity between materials. Looking forward, we hope to first compare simulation results at intracrystalline and extracrystalline MOF sites to directly analyze differences in diffusive behavior and characteristics that result in increased separation capacity. The end goal of this work is the development of a machine learning algorithm that can generate, test, and improve MOF design over multiple iterations to identify ideal candidates for PFAS separation. Overall, PFAS contamination poses a serious threat to the global population and is already nearly ubiquitous. Utilizing the modern benefits of machine learning and computer simulation will allow us to rapidly generate and test potential adsorbents to address and ideally mitigate this growing public health concern.