Real-space observation of non-covalent interactions in planar and non-planar molecules using scanning probe microscopy

With the rapid advancement of supramolecular materials, the study of non-covalent interactions (NCIs) has become increasingly critical. These interactions are fundamental to the forming, stability, and functionality of biological and synthetic materials. Scanning probe microscopy (SPM) offers an unparalleled platform for investigating NCIs within low-dimensional supramolecular self-assemblies on surfaces, allowing direct visualization and overcoming ensemble-averaging errors. The emergence of electrospray deposition (ESD) in SPM has significantly broadened its applicability to non-volatile, fragile biomolecules. However, accurately imaging and confirming the precise 3D structures of such molecules remains a considerable challenge. Addressing this difficulty requires the development of efficient structure search and reconstruction methods, which are still relatively underexplored. This thesis investigates the structural and functional implications of NCIs in low-dimensional supramolecular nanostructures on surfaces, progressing from planar molecules to non-planar small molecules and flexible biomolecules. This systematic approach demonstrates how to overcome the increasing complexity of the systems studied. The discussion begins with planar self-assemblies of DNA base molecules, which create dynamic confined environments that stabilize water dimers, revealing their influence in causing mismatched hydrogen bonding. It then advances to integrating machine learning (ML) to predict the structures of disordered water nanoclusters on metal surfaces, enabling the determination of complex hydrogen-bond patterns. Finally, the investigation focuses on glycosidic bond stereochemistry in carbohydrate self-assemblies, highlighting its critical role in regulating the charge distribution of groups attached to the non-anomeric carbon and governing chirality transfer in self-assemblies, using data-efficient, multi-fidelity structure search methods integrated with ML. Through these studies, this thesis provides insights into addressing the challenges posed by increasingly complex systems, paving the way for more efficient methods to study NCIs and their impact on supramolecular materials at atomic resolution.