Utilizing renewable energy for green hydrogen production via electrolyzed seawater is a promising technology for the future. However, undesired chlorine evolution and the corrosive nature of seawater are crucial challenges for direct seawater splitting technology. In this work, heterojunctions of CoS₂/FeS₂ encapsulated in N-doped carbon nanocubes (denoted as CoS₂/FeS₂@NC) were designed by proposing the synchronous pyrolysis and vulcanization of polydopamine-coated coordination polymers. Such a synthetic strategy was demonstrated to be effective in increasing the favorable exposure of active sites, moderately regulating electronic structure, and remarkably facilitating charge transfer due to the controllable generation of unique core–shell structures with suitable carbon shells, leading to the excellent bifunctional electrocatalytic performance and enhanced stability of electrocatalysts. As a result, CoS₂/FeS₂@NC can be revealed as a superior water splitting catalyst, possessing a small voltage of 1.75 V and requiring 100.0 mA cm⁻² in 1 M KOH alkaline solution and 1.80 V for alkaline seawater media, with satisfactory long-term stability. This work presents fresh strategies for designing core–shell heterostructures and developing green technology for hydrogen production.

Heavy metals pollution originating from industrial activities poses a significant threat to aquatic ecosystems and human health. Electrochemical analysis have shown great potential in heavy metals detection due to their rapid response, high sensitivity, and ease of operation. However, the electrochemical signal interpretation of heavy metals in multi-ion solution still remains challenging due to interference among different types of heavy metals. In this study, an adaptive peak feature algorithm was developed for the automatic and precise acquisition of peak height, peak width, and peak area of Cd2+ and Pb2+ from electrochemical curves of differential pulse voltammetry (DPVs) in simulated seawater. More importantly, these extracted features of DPVs are used for establishing a prediction model based on a Backpropagation neural network optimized by the Ant Colony Optimization algorithm (ACO-BP) to address the nonlinear interference under the coexistence of Cd2+ and Pb2+ ions. As a result, the predictive coefficient of determination for Cd2+ and Pb2+ are 0.9850 and 0.9916 with low root mean square errors of 2.4083 µg/L and 1.7440 µg/L, respectively, which surpassed that of traditional linear models and other machine learning approaches without multidimensional feature fusion, indicating that the mutual interference between multiple heavy metals was suppressed. This study offers a robust solution for high-precision monitoring of heavy metals in multi-ion solution and facilitates the development of advanced seawater monitoring instrumentation.