Machine learning-assisted Te-CdS@Mn3O4 nano-enzyme induced self-enhanced molecularly imprinted ratiometric electrochemiluminescence sensor with smartphone for portable and visual monitoring of 2,4-D

Abstract

Here, a novel and portable machine learning-assisted smartphone-based visual molecularly imprinted ratiometric electrochemiluminescence (MIRECL) sensing platform was constructed for highly selective sensitive detection of 2,4-Dichlorophenoxyacetic acid (2,4-D) for the first time. Te doped CdS-coated Mn₃O₄ (Te-CdS@Mn₃O₄) with catalase-like activity served as cathode-emitter, while luminol as anode luminophore accompanied H₂O₂ as co-reactant, and Te-CdS@Mn₃O₄ decorated molecularly imprinted polymers (MIPs) as recognition unit, respectively. Molecular models were constructed and MIP band and binding energies were calculated to elucidate the luminescence mechanism and select the best functional monomers. The peroxidase activity and the large specific surface area of Mn₃O₄ and the electrochemical effect can significantly improve the ECL intensity and analytical sensitivity of Te-CdS@Mn₃O₄. 2,4-D-MIPs were fabricated by in-situ electrochemical polymerization, and the rebinding of 2,4-D inhibits the binding of H₂O₂ to the anode emitter, and with the increase of the cathode impedance, the ECL response of Te-CdS@Mn₃O₄ decreases significantly. However, the blocked reaction of luminol on the anode surface also reduces the ECL response. Thus, a double-reduced MIRECL sensing system was designed and exhibited remarkable performance in sensitivity and selectivity due to the specific recognition of MIPs and the inherent ratio correction effect. Wider linear range in the range of 1 nM-100 μM with a detection limit of 0.63 nM for 2,4-D detection. Interestingly, a portable and visual smartphone-based MIRECL analysis system was established based on the capture of luminescence images by smartphones, classification and recognition by convolutional neural networks, and color analysis by self-developed software. Therefore, the developed MIRECL sensor is suitable for integration with portable devices for intelligent, convenient, and fast detection of 2,4-D in real samples.

Keywords: Machine learning; Molecularly imprinted polymers; Ratiometric electrochemiluminescence; Smartphone; Visual.