Machine Learning Accelerates Discovery of Optimal Colloidal Quantum Dot Synthesis.

Area of Science:

• Materials Science
• Nanotechnology
• Quantum Chemistry

Background:

• Colloidal quantum dots (CQDs) offer tunable bandgaps for optoelectronic applications.
• Improving CQD monodispersity is crucial for enhancing solar cell efficiency, reducing lasing thresholds, and improving photodetector performance.

Purpose of the Study:

• To leverage machine learning to optimize the synthesis of monodisperse PbS colloidal quantum dots.
• To identify synthetic parameters leading to record monodispersity in PbS quantum dots.

Main Methods:

• Utilized a machine-learning-in-the-loop approach to analyze experimental data and propose synthesis parameters.
• Introduced a growth-slowing precursor (oleylamine) to control nucleation and growth dynamics.

Main Results:

• Achieved record-large-bandgap PbS nanoparticles (611 nm exciton) with a narrow excitonic absorption peak (145 meV hwhm).
• Obtained improved monodispersity at longer wavelengths: 55 meV hwhm at 950 nm and 24 meV hwhm at 1500 nm.
• Surpassed previous state-of-the-art monodispersity values at 950 nm (75 meV) and 1500 nm (26 meV).

Conclusions:

• Machine learning is effective for navigating complex synthetic parameter spaces in nanomaterial synthesis.
• Controlling nucleation over growth via precursor addition is a viable strategy for high-quality CQD synthesis.
• The developed method enables the production of PbS CQDs with unprecedented monodispersity for advanced optoelectronic devices.