• The change of SiO x content leads to a shift in capacitive deionization performance. • CDI device positive and negative electrodes simultaneously remove anions and cations. • The biochar electrode itself achieves up to 90.71 % heavy metal removal efficiency. • The electrode maintained high removal efficiency in real aqueous matrices. • Machine learning models predicted a fluoride adsorption capacity of 79.28 mg g -1 . The composition of biochar was a critical factor governing its performance. Our findings revealed that the presence of SiO x in biochar significantly affected capacitive deionization efficiency. Nevertheless, the underlying mechanisms through which SiO x content in biochar influenced capacitive deionization performance remained to be comprehensively elucidated. This study investigated the capacitive deionization effect of SiO x on anions and cations using biochar treated with varying concentrations of NaOH. A series of characterization and electrochemical analyses were performed. Structural analysis revealed that while the macroscopic surface morphology remained largely unchanged, increasing NaOH concentrations induced significant alterations in nanoscale aggregation states and silicon distribution within the microstructure. Electrochemical and electro-adsorption experiments demonstrated that biochar with moderate Si content exhibited optimal electrochemical performance and superior capacitive deionization (CDI) activity. Beyond the electric double-layer mechanism of CDI, the removal of F - primarily occurred via coordination exchange between Si-O/Si-OH groups and F - , whereas Cu 2+ removal was governed by complexation reactions. At moderate Si content, simultaneous removal efficiencies of 46.50 % for F - and 86.70 % for Cu 2+ were achieved. Furthermore, machine learning models predicted a fluoride adsorption capacity of 79.28 mg g -1 in practical wastewater systems. This study proposed a universal strategy for optimizing the capacitive deionization performance of silicon-enriched biochar materials.
Song et al. (Thu,) studied this question.