Leveraging Experimental Literature Data to Discover Novel Metal-Organic Frameworks and Mechanophores

The design of crystalline metal-organic frameworks and more disordered mechanophore-containing polymer networks are twin challenges that are both plagued by both synthetic accessibility and the large possible combinatorial space of novel materials. Metal-organic frameworks (MOFs) in particular have been widely studied for their ability to capture and store greenhouse gases. However, most chemical discovery efforts use computational study of hypothetical MOFs without consideration of their stability, limiting the practical application of novel materials. We have used natural language processing to extract experimental reports of numerous measures of stability of MOFs, including their thermal stability, stability upon activation (i.e., solvent removal), and in the presence of water or acid/base. I will show how we leverage these models to construct novel MOF candidates with vastly enriched stability over conventional hypothetical datasets. I will also describe how we pair these models with genetic algorithms to identify design principles for high mechanical stability and exceptional gas storage or for combined water and thermal stability. In the second half of my talk, I will describe our efforts to leverage experimental data to discover novel transition metal-containing mechanophores. Ferrocenes are particularly attractive targets as mechanophores due to their combination of high thermal stability and mechanochemical lability. However, the handful of demonstrated ferrocene mechanophores is sparse in comparison to several thousands of unique ferrocene complexes that have been synthesized. I will discuss our recent computational, machine learning guided discovery of synthesizable ferrocene mechanophores. I will describe how we identify over one hundred potential target ferrocene mechanophores with wide-ranging mechanochemical activity and use data-driven computational screening to identify a select number of promising complexes. I'll highlight design principles that we identified to alter mechanochemical activation of ferrocenes, including regio-controlled transition state stabilization through sterically bulky groups and a change in mechanism through non-covalent ligand–ligand interactions. I will discuss how we validated our computational screening experimentally both at the polymer strand level through sonication experiments and at the network level by mechanical testing. Experiments indicate that a computationally discovered ferrocene mechanophore cross-linker leads to greater than 4-fold enhancement in material tearing energy. Together, these efforts highlight the opportunities and outstanding challenges in computationally guided materials design.