Abstract

Flow Reactors for Sustainable Colloidal Synthesis of Nanocrystals

Noah Malmstadt, Professor of Chemical Engineering and Materials Science, University of Southern California

Nanocrystal materials including metals, metal carbides and phosphides, and perovskites have broad applications in the transition to sustainable energy. In particular, they can serve as next-generation catalysts for carbon dioxide conversion, fuel cell membranes, and biofuel upgrading. While there are well-established routes to the colloidal synthesis of these materials, they are highly sensitive to local reaction environment, and it has been challenging to scale their production using traditional chemical manufacturing technologies. On the other hand, millifluidic flow reactors, which can deliver excellent reaction environment uniformity, are a promising route to the production of colloidal nanocrystals. Recent work has demonstrated that scaling millifluidic reactors via parallelization can approach industrially relevant product throughput. Flow reactors are also powerful tools for reaction discovery. Here, we present two examples of how flow reactor systems can be used to understand the parameter space of nanocrystal synthesis reactions and identify targeted reaction conditions. The first of these examples is the production of Pt nanoparticles (NPs) in ionic liquids (ILs). Ionic liquid (IL) solvents represent a special class of low-volatility, generally safe solvents that are particularly easy to recycle. While the capacity to produce metallic NPs in ILs has been known for decades, we know little about the mechanism of these reactions and in particular how solvent choice can guide this mechanism. To discover the mechanism of Pt NP fabrication in ILs, we have constructed a flow reactor with in-line spectrophotometric monitoring of the products. To determine reaction component concentration from the complex spectral data, we have implemented a machine learning (ML) algorithm that can determine concentration. By measuring product concentration as a function of residence time, we are able to determine the IL solvent-dependent reaction kinetics. The second example involves synthesizing photoactive perovskite nanocrystals in a parallel flow reactor system. By controlling hydrodynamic resistance across the channel network, we are able to rapidly screen composition space for the reactants. Analyzing these high throughput data with a neural network facilitates the construction of a map between reactant composition space and product crystal phase space, allowing for manufacturing to target a desired product phase.