Beyond the Status Quo: Density Functional Tight Binding and Neural Network Potentials as a Versatile Simulation Strategy to Characterize Host–Guest Interactions in Metal- and Covalent Organic Frameworks

Abstract

In recent years, research focused on synthesis, characterization, and application of metal–organic frameworks (MOFs) has attracted increased interest, from both an experimental as well as a theoretical perspective. Self-consistent charge density functional tight binding (SCC DFTB) in conjunction with a suitable constrained molecular dynamics (MD) simulation protocol provides a versatile and flexible platform for the study of pristine MOFs as well as guest@MOF systems. Although being a semi-empirical quantum mechanical method, SCC DFTB inherently accounts for polarization and many-body contributions, which may become a limiting factor in purely force field-based simulation studies. A number of examples such as CO2, indigo, and drug molecules embedded in various MOF hosts are discussed to highlight the capabilities of the presented simulation approach. Furthermore, a promising extension of the outlined simulation strategy toward the treatment of covalent organic frameworks utilizing state-of-the-art neural network potentials providing a description at DFT accuracy and force field cost is outlined.

Combining the D3 dispersion correction with the neuroevolution machine-learned potential

“…However, the dispersion interactions are important in describing e.g. the interlayer attractions in the so-called van-der-Waals (vdW) layered materials [3,4], structural transformation between a narrow-pore and large-pore phase in flexible metal-organic frameworks (MOFs) [5,6], and host-guest interactions in MOFs [7,8].…”

Towards hybrid quantum mechanical/molecular mechanical simulations of Li and Na intercalation in graphite – force field development and DFTB parametrisation

“…27 Although being developed in the context of (bio)organic chemistry, this DFTB parametrization proved surprisingly robust for a broad variety of system relevant in the regime of computational material sciences. Examples include the study of complex host–guest interactions in metal–organic frameworks 28–31 as well as the binding of anthraquinone to graphene/graphite. 32 These works provide a direct proof of concept that the applied SCC DFTB/3ob method is capable of treating solid-state systems including carbon-based layered materials.…”

Storage and Diffusion of Carbon Dioxide in the Metal Organic Framework MOF-5─A Semi-empirical Molecular Dynamics Study

“…In this work, a total of four neighboring temperature points were employed to determine the respective slope via finite differentiation ∂ a ∂ T ≈ ⟨ a 248 normalK ⟩ − 8 ⟨ a 273 normalK ⟩ + 8 ⟨ a 323 normalK ⟩ − ⟨ a 348 normalK ⟩ 12 normalΔ T …”

“…⟨ U MOF‑5 ⟩ and ⟨ U normalC O 2 ⟩ were determined over the last 0.25 ns of the simulation trajectory. While in this approach, all variations in the geometry of the host–guest systems, including variations in the lattice parameters, are still visible in the time evolution of U int , an analysis of the respective running average employing 500 data points enables a visual inspection of the equilibration period as well as changes in U int due to variations in the host–guest interaction …”

“…To overcome this limitation, an alternative semi-empirical approach using a set of approximations and extensive parametrization of interactions known as density functional tight binding (DFTB) , in combination with a suitable MD protocol has been employed. DFTB has shown an exceptional performance in the description of different MOF as well as guest@MOF system − in the past.…”