Multiobjective Platform for Autonomous Property Targeting and Optimization of Colloidal Lead Halide Perovskite Quantum Dots

The optimization of colloidal quantum dot (CQD) materials, synthesis routes, and processing methods are complex challenges that are ripe for automation and artificial intelligence (AI) to have a great impact. These optimization challenges are seldom oriented to a single target; therefore, it is vital that autonomous systems can handle multiple objectives. In this work, we present an autonomous CQD synthesis system that successfully performs multiobjective optimization (MOO) via Bayesian optimization-based algorithms. We demonstrate the efficacy of the system through three distinct synthesis challenges, based on one, two, and three objective optimization problems, in the synthesis of cesium lead halide perovskite CQDs. Objectives included maximizing fluorescence brightness, minimizing particle size dispersity, and targeting of a specific optical band gap and particle diameter. The triobjective challenge achieved simultaneous targeting of specific CQD sizes and band gaps independently via reaction tuning and halide doping, while minimizing the particle size dispersity. This work demonstrates AI-assisted multiobjective targeting and dynamic synthesis of targeted colloidal CQDs using exciton energy analysis of absorption spectra to infer both size and optical band gap. It presents an accessible, automated, and data-driven platform for CQD discovery and optimization (both for single and multiple objectives), highlighting the promise of widespread integration of AI-guided strategies into CQD R&D.