Elucidating the Formation of Perovskite Nanocrystals through Machine Learning a Physical Model based on Automated and In-Situ Synthesis Experiments
Advances in automation and data analytics can aid exploration of the complex chemistry of nanoparticles. Lead halide perovskite colloidal nanocrystals provide an interesting proving ground: there are reports of many different phases and transformations, which has made it hard to form a coherent conceptual framework for their controlled formation through traditional methods. In this work, we systematically explore the portion of Cs–Pb–Br synthesis space in which many optically distinguishable species are formed using high-throughput robotic synthesis to understand their formation reactions. We deploy an automated method that allows us to determine the relative amount of absorbance that can be attributed to each species in order to create maps of the synthetic space. Based on these maps, we test potential transformation routes between perovskite nanocrystals of different shapes and phases. We demonstrate a dynamic equilibrium between complexes, monolayers, and nanocrystals of lead bromide, with substantial impact on the reaction outcomes. This allows us to construct a chemical reaction network that qualitatively explains our results as well as previous reports and can serve as a guide for those seeking to prepare a particular composition and shape. Using these experiments as well as in-situ measurements in a stopped flow setup, we are developing a numerical kinetic model that can be fit to experimental data using developments and tools from machine learning.
Jakob’s work focuses on understanding the physical chemistry of nanocrystal synthesis in lead halide and gold nanocrystals. While it is possible to make different sizes and shapes nanocrystals, scientists have only a rough, qualitative understanding of the chemical processes. Jakob uses in-situ observations in the Alivisatos lab and high-throughput experiments in the Chan lab to generate 1000s of datapoints on nanocrystal synthesis. Machine learning techniques help analyze this data. Currently, he is working with the Limmer and Persson groups to develop chemistry-based machine learning models that will quantify and clarify our understanding of the complex reaction networks involved in forming nanocrystals. The goal is to enable robust, chemical-model based predictions for nanocrystal yield, size, shape and composition for synthesis procedures that have not previously been attempted. Ideally, this will both speed up nanocrystal synthesis exploration and lead to enhanced understanding of the underlying chemical processes.