Machine Learning and Molecular Simulations Reveal Mechanisms of ZIFs Polymorph Selection

TL;DR

This study uses machine learning and molecular simulations to uncover how different polymorphs of Zn(imidazolate)$_2$ MOFs are selected during synthesis, revealing early-stage determinants.

Contribution

It introduces a combined simulation and neural network approach to identify the stage at which polymorph selection occurs in MOF formation.

Findings

Pre-nucleation clusters and amorphous intermediates are polymorph-dependent.

Polymorph selection occurs early, at the pre-nucleation cluster stage.

Simulations and neural networks effectively classify transient structures.

Abstract

Zn(imidazolate)$_2$ metal-organic frameworks (MOFs) exhibit a remarkable degree of polymorphism. Because of their promising industrial applications, many research groups have investigated phase transitions, phase diagram and relative stability of these polymorphs. There is now wide consensus in the research community that these MOFs are solvothermally formed via non-classical nucleation mechanisms, in which pre-nucleation clusters are first formed, followed by an intermediate amorphous structure that subsequently reorganizes to yield the final crystalline MOF. However, no study up to date has uncovered which part of the synthesis process determines the final polymorph obtained. In this work, path collective variable metadynamics simulations performed with a partially reactive force field give insights into mechanistic and thermodynamic aspects of the self-assembly of these MOFs. Databases of transient and intermediate synthesis structures are built from the simulations. By developing and applying neural network classifiers over these databases, it is found that both pre-nucleation clusters and the amorphous intermediate structures are polymorph-dependent. These results suggest that polymorph selection happens as early as the pre-nucleation cluster stage.