Related Experiment Video
Updated: May 12, 2025
10:13
A Technical Guide for Performing Spectroscopic Measurements on Metal-Organic Frameworks
Published on: April 28, 2023
2.2K
A Practical Application of Machine Learning for the Development of Metallole-Based Fluorescent Materials.
Yusuke Kanematsu1,2, Akiyoshi Ohta1, Shunya Nagai1
1Smart Innovation Program, Graduate School of Advanced Science and Engineering, Hiroshima University, Higashi-Hiroshima 739-8527, Japan.
Molecules (Basel, Switzerland)
|May 7, 2025
Summary
We developed a prediction model for fluorescence quantum yields in metalloles. The model accurately identifies weakly fluorescent molecules, aiding in the efficient discovery of novel fluorescent materials.
Area of Science:
- Materials Science
- Photochemistry
- Computational Chemistry
Background:
- Predicting fluorescence quantum yields (FQYs) is crucial for designing efficient organic light-emitting materials.
- Metalloles are a class of compounds with potential applications in optoelectronics due to their photophysical properties.
Purpose of the Study:
- To develop and validate a predictive model for the fluorescence quantum yields of metallole compounds.
- To assess the model's accuracy in classifying FQYs and its utility in screening candidate molecules.
Main Methods:
- A prediction model for FQYs of metalloles was constructed.
- Ten fluorescent metallole molecules were synthesized based on the model's predictions.
- Fluorescence quantum yields were experimentally measured to validate the model's performance.
Main Results:
- The prediction model achieved a classification accuracy of 0.7 for FQYs.
- The model demonstrated perfect prediction accuracy for low quantum yields, effectively screening out weakly fluorescent compounds.
- A precision of 0.5 was observed, attributed to a bias in the training dataset towards high-FQY fluorine-containing molecules.
Conclusions:
- The developed prediction model is a useful tool for screening weakly fluorescent metalloles.
- Dataset bias can impact model precision, necessitating careful data curation.
- Revision of the model with a candidate molecular structure generator and expanded dataset improved applicability for fluorescent material development.
