ABSTRACT
Two woven covalent organic framework materials (COF-505 and COF-506) have been synthesized since 2016, and the latter demonstrated the ability to take up dyes and other small molecules. This opens the door to applications such as separations, sensing, and catalysis. However, accelerating the design of future woven materials by changing the chemistry of the "threads" will require a computational model for these materials. Since no such atomic-scale model exists, we have developed a protocol for optimizing a force field for woven materials which can be used as the input to molecular dynamics simulations. Their high density and elasticity made these COFs challenging to model at a semiempirical level. Our modeling approach required simultaneous optimization of lattice parameters and elasticity using density functional theory-derived energy barriers and available experimental results. We used this force field, parameterized to fit COF-505, without change, to predict the structure of COF-506. This model allowed us to predict an anisotropy in 505's elasticity and preferred directions for diffusion which cannot be seen experimentally. The pore size distribution for 506 is dominated by small pores (80% <10 Å dia.), though 5% of the pores are up to 20 Å in diameter. We confirmed the experimental result that gases (barring helium) do not diffuse appreciably in COF-505. We validated our (unaltered) force field model to accurately predict experimental uptake data for tetrahydrofuran and methyl orange dye in COF-506. We proposed an atomic-scale mechanism by which COF-505 becomes metallated and demetallated. In addition, in advance of experimental studies, we determined the ability of 505 to incorporate other metals, such as Zn and Fe, which might be considered artificial photosynthesis agents. These predictions validate that Cu was a particularly appropriate choice of metal center for the synthesis, showcasing the ability of this model to play a role in designing woven materials a priori.
