Advancement of capacitive deionization propelled by machine learning approach

The gradual deterioration of ecosystems and the exponential growth of the population have led to severe freshwater scarcity. Fetching available freshwater from seawater is expected to address freshwater scarcity. Capacitive deionization (CDI), a burgeoning adsorption technology, has demonstrated excellent desalination ability, with high-performance electrode materials playing an important role. However, traditional fabrication methods for electrode materials rely on “trial and error” principles, which are labor-intensive and time-consuming. Machine learning (ML), a derivative of the big data era, can effectively predict the desalination performance of electrode materials and guide the synthesis of novel electrode materials by training massive amounts of data, compensating for the shortcomings of traditional experiment methods. Moreover, ML can also analyze the effects of electrode properties, operational conditions and water quality on CDI performance, thereby accelerating the development and revolution of CDI. Despite its significance, there are currently no comprehensive reviews focusing on ML approaches for CDI applications. In this review, we detailed the different applications of ML in the CDI field, including the prediction of desalination performance, the analysis of feature contribution, etc. The future prospects of both ML and CDI were also discussed. This work provides a significant guidance for the development of CDI technology via ML-assisted method.