Defect-Engineered WO<sub>3–<i>x</i></sub> Architectures Coupled with Random Forest Algorithm Enables Real-Time Seafood Quality Assessment
Ziqi Zhang, Junxuan Liang, Kai Liu, Weiliang Tian, Liang Xu, Kun Zhao, Kewei Zhang
Abstract
Reliable and real-time monitoring of seafood decay is attracting growing interest for food safety and human health, while it is still a great challenge to accurately identify the released triethylamine (TEA) from the complex volatilome. Herein, defect-engineered WO 3– x architectures are presented to design advanced TEA sensors for seafood quality assessment. Benefiting from abundant oxygen vacancies, the obtained WO 2.91 sensor exhibits remarkable TEA-sensing performance in terms of higher response (1.9 times), faster response time (2.1 times), lower detection limit (3.2 times), and higher TEA/NH 3 selectivity (2.8 times) compared with the air-annealed WO 2.96 sensor. Furthermore, the definite WO 2.91 sensor demonstrates long-term stability and anti-interference in complex gases, enabling the accurate recognition of TEA during halibut decay (0–48 h). Coupled with the random forest algorithm with 70 estimators, the WO 2.91 sensor enables accurate prediction of halibut storage with an accuracy of 95%. This work not only provides deep insights into improving gas-sensing performance by defect engineering but also offers a rational solution for reliably assessing seafood quality.