Machine Learning-Driven Achieving Efficient Phosphorescent Carbon Nanodots in Aqueous Solution by Suppressing Triplet...

Area of Science:

• Materials Science
• Nanotechnology
• Photochemistry

Background:

• Matrix-assisted synthesis is common for solid-state phosphorescent carbon nanodots (CNDs).
• Efficient liquid-phase CND systems are difficult to achieve due to complex mechanisms and knowledge gaps.
• Translating phosphorescent CND systems between solid-state and liquid-phase configurations is hindered by these challenges.

Purpose of the Study:

• To elucidate the role of triplet electron leakage in reducing CND phosphorescence.
• To demonstrate the application of machine learning (ML) in suppressing triplet electron leakage.
• To achieve efficient phosphorescent CNDs in aqueous solution.

Main Methods:

• Developed an interpretable ML model using experimental data from hundreds of syntheses.
• Integrated SHapley Additive exPlanations (SHAP) analysis to understand feature-lifetime relationships.
• Systematically designed syntheses to gather data for ML model training.

Main Results:

• Identified triplet electron leakage as a key factor limiting phosphorescence.
• Established quantitative relationships between material features and phosphorescence lifetimes.
• Revealed an inverse correlation between matrix thickness and triplet electron leakage probability.
• Achieved efficient phosphorescent CNDs in aqueous solution with emission lifetimes >10 s and quantum yields >10%.

Conclusions:

• Demonstrated ML's capability to predict and tune phosphorescence lifetimes in CND systems.
• Established a framework for engineering high-performance liquid-phase phosphorescent nanomaterials.
• Successfully suppressed triplet electron leakage to create efficient aqueous CNDs.