Compared to traditional solid porous materials, such as zeolites, metal organic frameworks (MOFs) exhibit great structural tunability. Using different metal centers and ligands, myriads of MOF structures with different topologies can be synthesized. While such a huge structural library allows for many possible variations in properties, it is intrinsically impossible to exhaustively explore this diversity experimentally.
Hence, for a certain application, such as CO2 sequestration, theoretical tools can be extraordinarily useful in predicting optimal MOF structures and properties. To date, using an appropriate force field and Monte Carlo simulation technique, researchers are able to screen hundreds to thousands of these structures and identify the best one in a reasonable amount of time. However, this scheme is largely limited by the speed of current simulation methods, and employing force fields with accurate long range interactions can be slow and unaffordable. Therefore, the large-scale screening process is currently utilized mainly on spherically symmetric molecules, such as rare gases or CH4, where adsorption is dispersion dominated. Implementation of long range electrostatic interactions requires more advanced GPU based algorithms to reduce simulation times, and many-body induction potentials are simply intractable for these traditional methods.
In order to solve this problem, we have developed a new lattice model based simulation technique. This lattice model is constructed and parameterized based on atomistic scale simulations, and is rigorous in the low pressure limit. We show that with appropriate coarse-grained solute-solute interactions, we are able to quantitatively reproduce the atomistic simulation results. The errors on various adsorption isotherms are typically below 15% and kinetic properties such as diffusivities can also be obtained with a fairly high accuracy. For a typical force field employing Ewald electrostatics, our method is 2~3 orders of magnitude faster than the traditional GCMC approach. Moreover, this method also enables the use of more complicated polarizable force fields for large scale screening, leading to more robust results for differing electrostatic environments.