Autonomous De Novo Lead Halide CsPbBr₃ Perovskite Quantum Dots Synthesis Platform With Transfer Learning Accelerated...

Area of Science:

• Materials Science
• Chemical Engineering
• Artificial Intelligence

Background:

• Traditional material synthesis relies on expert-driven methods, which can be time-consuming and inefficient.
• Emerging self-driving laboratories offer an autonomous paradigm for enhanced research and development (R&D).

Purpose of the Study:

• To develop an autonomous system for the on-demand synthesis of cesium lead bromide (CsPbBr3) quantum dots (QDs).
• To improve the R&D efficiency of functional material synthesis using artificial intelligence and flow chemistry.

Main Methods:

• Development of a micro Transfer learning accelerated Bayesian Optimization driven reaction System (µTRBOS).
• Utilizing Ligand-Assisted RePrecipitation (LARP) method for QD synthesis.
• Implementing autonomous operation without human supervision.

Main Results:

• Successfully synthesized high-quality CsPbBr3 QDs with user-specified emission wavelengths (455-505 nm) and <2 nm error.
• Achieved autonomous synthesis of QDs with particle sizes ranging from 2.5 to 7.4 nm.
• Optimized synthetic conditions in fewer than six experiments on average, leveraging transfer learning.

Conclusions:

• The µTRBOS system demonstrates efficient and autonomous synthesis of functional materials.
• Transfer learning significantly reduces the number of experiments needed for optimization.
• Optimal conditions highlight the complex role of temperature in the LARP method for QD synthesis.