Resource-Efficient Autonomous Lab for Multi-Objective Optimization of Mixed Halide Perovskite Nanocrystals

Metal halide perovskite (MHP) nanocrystals (NCs), conventionally pursued through time- and labor-intensive manual batch experimentation have shown significant potential to innovate in areas such as displays, LEDs, and solar cells. Despite their promising capabilities, the fast formation kinetics of MHP NCs have shown to cause variability in synthesis of MHP NCs in manual batch experiments. Continuous flow chemistry has shown to adapt well to their fast formation kinetics, and can even reduce the time and labor required for numerous experiments. Additionally, continuous experimentation, in-situ characterization, and reaction miniaturization are easily available to flow chemistry producing a superior synthesis platform for analyzing the input parameters of MHP NCs [1]. Although continuous flow chemistry addresses the aforementioned challenges, it suffers from wasting material while the system is waiting for steady-state to be achieved.<br/>Due to the immense experimental space of MHP NCs through their continuous (e.g., reaction temperature, reaction time) and discrete (e.g., different surface ligands and halide sources) parameters, an efficient way to conduct experiments to find the highest performing NCs is needed. Performing closed-loop experimentation integrating machine learning (ML)-guided experimentation with automated platforms known as self-driving labs (SDLs) has gained popularity in chemical and materials science [2]. MHP NCs have competing objectives: maximizing photoluminescence quantum yield (PLQY) and minimizing full width at half maximum (FWHM) at target emission wavelengths. These competing goals necessitate multi-objective optimization to determine the optimal input conditions for high-performing MHP NCs.<br/>In this work, we present a single-droplet fluidic platform that is able to synthesize mixed halide perovskite NCs using a 5 uL droplet per reaction condition. The single-droplet platform decreases material waste and time due to the decoupled precursor formulation and synthesis sections. After thoroughly characterizing, validating, and benchmarking the physical and digital components of the SDL, we utilize this advanced autonomous approach to perform multi-objective optimization of mixed halide MHP NCs. We then showcase the SDL’s capability for autonomously selecting synthetic routes to high-performing MHP NCs, achieving the highest PLQY and lowest FWHM through closed-loop Pareto-front mapping across a broad UV-Vis spectrum range.<br/><br/>References:<br/>[1] Antami, K.; Bateni, F.; Ramezani, M.; Hauke, C. E.; Castellano, F. N.; Abolhasani, M. CsPbI 3 Nanocrystals Go with the Flow: From Formation Mechanism to Continuous Nanomanufacturing. Adv Funct Materials 2022, 32 (6), 2108687. https://doi.org/10.1002/adfm.202108687.<br/><br/>[2] Abolhasani, M.; Kumacheva, E. The Rise of Self-Driving Labs in Chemical and Materials Sciences. Nat. Synth 2023, 2 (6), 483–492. https://doi.org/10.1038/s44160-022-00231-0.