Synthesis of Machine Learning-Predicted Cs₂PbSnI₆ Double Perovskite Nanocrystals.

Area of Science:

• Materials Science
• Optoelectronics
• Computational Chemistry

Background:

• Halide perovskites are crucial for photonics and optoelectronics due to tunable emission and manufacturability.
• The full potential of halide perovskites is limited by the unmapped composition-emission energy landscape.
• Targeted material synthesis requires guided high-throughput screening methods.

Purpose of the Study:

• To develop a machine learning model for predicting halide perovskite band gaps.
• To identify promising halide perovskite compositions for optoelectronic applications.
• To validate the predictive model using experimental synthesis and characterization.

Main Methods:

• Literature data compilation for 10,920 halide perovskite compositions.
• Machine learning model development for band gap prediction.
• Synthesis and characterization of Cs2PbSnI6 nanocrystals.
• Ab initio GW band structure calculations.

Main Results:

• A machine learning model accurately predicts band gaps across a vast halide perovskite composition space.
• Experimental validation confirmed the model's predictions for Cs2PbSnI6.
• Synthesized Cs2PbSnI6 exhibits photoluminescence spectra consistent with predicted band gaps.

Conclusions:

• The developed machine learning model facilitates efficient exploration of halide perovskite materials.
• Elpasolite Cs2PbSnI6 is identified as a highly promising material for optoelectronic devices.
• This work bridges computational prediction with experimental validation for accelerated materials discovery.