Structure elucidation of amorphous materials and microcrystalline solids in powder form is one of the key challenges in chemistry today. While techniques such as single crystal diffraction and cryo-electron microscopy are generally not able to characterize such materials, we will show how an approach based on measured NMR chemical shifts in combination with methods for large scale computation of shifts can rapidly determine full three-dimensional structures from powders.

For example, using a machine learning model of chemical shifts, we will demonstrate the complete atomic-level structure determination of amorphous pharmaceutical forms by combining dynamic nuclear polarization (DNP) enhanced solid-state NMR experiments with chemical shifts predicted using machine learning for MD simulations of large systems.
From these amorphous structures we then extract and analyze preferred conformations, H-bonding motifs, and intermolecular interactions in the amorphous sample in terms of the stabilization of the amorphous form of the drugs.