An electrodeposited nano-porous and neural network-like Ln@HOF film for SO2 gas quantitative detection via fluorescent sensing and machine learning

Abstract

The intelligent fluorescence detection of SO₂ gas has been a research focal point and machine learning (ML) for mining and analyzing data will be extensively applied during detecting chemicals in the big data era. Herein, a nano-porous and neural network-like Tb³⁺-functionalized HOF film (1) is successfully manufactured via electrophoretic deposition. Ligand-to-metal charge transfer (LMCT)-induced energy transfer (ET) from ME-IPA to Tb³⁺ ions makes 1 emit palpable green light. 1 as a fluorescent sensor can quantitatively distinguish SO₂ gas concentration with chromatic and ratiometric modes. The formation of noncovalent N⋯S interaction between amino and SO₂ molecules inhibiting TADF-assisted ET and LMCT-induced ET procedures can be responsible for the sensing mechanism of 1. The detection of derivatives SO₃²⁻ is also carried out in aqueous solution and serum systems. Moreover, a back propagation neural network (BPNN) model based on 1 has been firstly constructed, and real test data demonstrates that the BPNN can accurately discriminate SO₂ concentration by deep ML. This work not only proposes a facile pathway to fabricate a porous fluorescent HOF film as an excellent gas sensor, but also elaborates how to combine fluorescent sensing with deep ML to realize intelligent fluorescence detection of 1 toward SO₂.