Machine learning-assisted ratiometric Zr-MOF-based fluorescent probe for intelligent detection of uranyl ions

Abstract

Uranium contamination in aquatic environments poses potential threats to ecosystems and human health due to the high radioactive and chemical toxicity of uranyl ions (UO22+), necessitating the development of accurate and rapid detection strategies for the monitoring of UO22+. Herein, a zirconium-based metal-organic framework (NH2-PCN-134) was constructed through a one-pot hydrothermal method and employed as a ratiometric fluorescent probe for the detection of UO22+. The ratiometric fluorescence response of NH2-PCN-134 toward UO22+ featured emission quenching at 655nm and enhancement at 515nm, enabling quantitative analysis based on the I515/I655 intensity ratio. Furthermore, the probe exhibited a broad linear response range of 0–200μM with a low detection limit of 1.23μM, and excellent selectivity toward UO22+ in the presence of various interfering ions. The ratiometric fluorescence response was attributed to the synergistic effects of photoinduced electron transfer (PET) and molecular interactions between NH2-PCN-134 and UO22+. Moreover, the color features extracted from fluorescent images were integrated with multiple machine learning models for quantitative prediction, which exhibited strong agreement between the predicted and experimental values even in the presence of interfering ions, demonstrating their effectiveness for quantitative analysis. This study provides a machine learning-assisted ratiometric fluorescence sensing strategy for intelligent detection of UO22+, highlighting its potential for environmental monitoring.