(696g) Autonomous Synthesis of Metal Halide Perovskite Nanocrystals

Metal halide perovskite (MHP) nanocrystals (NCs), owing to their facile solution-processibility and tunable optoelectronic properties, are considered an emerging class of active materials for next-generation clean energy technologies. [1] Metal cation doping of MHP NCs with transition-metal ions (e.g., manganese, Mn²⁺) is an attractive route to not only substitute heavy metal ions (Pb²⁺) with less toxic impurity dopants, but also introduce new optoelectronic and magnetic properties into the host NCs. [2]

Discovery and development of metal cation-doped MHP NCs is conventionally achieved using batch reactors (flasks) with offline characterization. [1] However, flask-based NC synthesis is time-, material- and labor-intensive, and suffers from batch-to-batch variations, non-uniform heat/mass transfer rates, and difficulty of integration with in situ NC characterization probes. [1] In addition, the complex and vast design space of metal cation-doped MHP NCs further complicate their fundamental mechanistic studies and hampers the search for finding the optimal synthetic route using the trial-and-error one at a time experimentation approach. [3]

The emergence of artificial intelligence (AI)-assisted experimental space exploration strategies have provided an exciting opportunity to accelerate navigation through the high-dimensional experimental space of emerging advanced functional materials through convergence of AI with colloidal nanoscience and automated experimentation. The result of such convergence is ‘self-driving laboratories’ for closed-loop autonomous exploration and/or exploitation of the experimental space. [3] In particular, deep neural networks (DNNs) have proven to be an effective machine learning (ML) technique for accelerated fundamental and applied studies of materials with high-dimensional design space, including semiconductor NCs.[3] Microreactors with their reduced chemical consumption and waste generation, reproducible and enhanced heat/mass transfer rates, and facile integration with in situ NC characterization techniques are an ideal reactor type for controlled synthesis of MHP NCs. [1]

In this work, we present a self-driving laboratory using modular flow reactors for accelerated design space exploration, synthetic route discovery, and fundamental mechanistic studies of metal cation-doped MHP NCs. Specifically, we developed a two-stage sequential halide exchange and cation doping of MHP NCs without the need for an intermediate washing stage. Next, we built an AI model of the two-stage flow synthesis platform (i.e., digital twin of the experimental platform) using 60 autonomously conducted NC synthesis experiments. We then utilized the developed digital twin of the self-driving lab to: (i) identify the key experimental input parameters controlling the metal cation doping of MHP NCs, (ii) unveil the mechanism of the metal cation doping of MHP NCs , and (iii) synthesize Mn-doped MHP NCs on-demand with a targeted doping level. The developed self-driving lab can rapidly identify the optimal synthetic route for in-flow synthesis of MHP NCs with a desired Mn doping level in less than 90 min. The self-driving lab detailed in this work, presents a generic modular framework for autonomous formulation discovery, synthesis, optimization, and continuous manufacturing of novel advanced functional materials for applications in clean energy technologies.

References:

[1] Abdel-Latif, K.; Bateni, F.; Crouse, S.; Abolhasani, M. Flow Synthesis of Metal Halide Perovskite Quantum Dots: From Rapid Parameter Space Mapping to AI-Guided Modular Manufacturing. Matter 2020, 3 (4), 1053–1086. https://doi.org/10.1016/j.matt.2020.07.024.

[2] Bateni, F.; Epps, R. W.; Abdel-latif, K.; Dargis, R.; Han, S.; Volk, A. A.; Ramezani, M.; Cai, T.; Chen, O.; Abolhasani, M. Ultrafast Cation Doping of Perovskite Quantum Dots in Flow. Matter 2021, S2590238521002174. https://doi.org/10.1016/j.matt.2021.04.025.

[3] Bateni, F.; Epps, R. W.; Antami, K.; Dargis, R.; Bennett, J. A.; Reyes, K. G.; Abolhasani, M. Autonomous Nanocrystal Doping by Self‐Driving Fluidic Micro‐Processors. Advanced Intelligent Systems 2022, 2200017. https://doi.org/10.1002/aisy.202200017.