Nanocrystals (NCs) are key materials in catalysis, optoelectronics, and medicine, where their performance critically depends on composition, size, and shape. Colloidal chemistry offers exceptional control over these features. Yet, the link between reaction chemistry and final NC features (i.e., composition, size, shape) remains poorly understood, limiting rational synthesis design.
This thesis links reaction chemistry to shape and composition control by focusing on colloidal Cu and Ni NCs. These NCs were chosen for the impact of shape and composition in sustainable catalytic processes, including electrochemical CO2 conversion.
Chapter 3 demonstrates that increasing precursor reactivity enables unprecedented control over Cu NC shapes within one same synthetic approach, from single-crystalline to twinned and stacking fault-lined NCs. Replacing trioctylphosphine with diphenylphosphine is crucial for this result as the forming complex lowers the Cu NC formation temperature and enables shape tunability. In situ X-ray absorption spectroscopy reveals that the mechanism involves a molecular complex between the Cu precursor and the ligand containing a P-H bond.
Chapter 4 expands on the importance of ligand chemistry in increasing precursor reactivity in Cu NC synthesis through a ligand-library screening of phosphines (PR3), arsines (AsR3), and stibines (SbR3), analyzed via 191 ligand descriptors. Density functional theory supports a mechanism in which ligands act as electron shuttles during Cu(I) disproportionation. The partial charge on the ligand central atom in the radical anion species correlates with precursor reactivity, which dictates the formation temperature of the NCs. This parameter can then be used as a predictive descriptor for ligand reactivity, enabling a priori design of low-temperature Cu NC syntheses.
Chapter 5 extends the study of ligand chemistry to Ni-based NCs, showing that ligand electronic properties dictate composition. Weak Ï -donors favor nickel pnictides (NixEy, E = P, As, Sb), while strong Ï -donors yield metallic Ni NCs. These outcomes originate from different reducibility of the Ni(II) centers leading to Ni-E cluster and to Ni(I) complex intermediates for NixEy and Ni NCs, respectively. This chemistry lesson enables the synthesis of diverse Ni-based NCs, including multiply twinned, flower-like, and cubic Ni NCs as well as Ni2P, Ni12P5, Ni5P2, Ni5As2 and NiSb NCs.
Chapter 6 explores how machine learning can accelerate NC synthesis understanding, leveraging on the experimental datasets from previous chapters. A low-data regime toolbox is developed to predict NC shape from reaction parameters and suggest reaction conditions for a target morphology. This approach successfully predicts and yields a previously unreported Cu rhombic dodecahedral shape.
Finally, Chapter 7 offers an outlook on future developments, including the translation from batch to flow chemistry, the expansion of data-driven methods and the synthesis of
