Homologous-Constrained Machine Learning Enables Amino Acid Additive Screening for Water-Structure-Regulated Aqueous Zinc-Ion Batteries.

Aqueous zinc-ion batteries (AZIBs) are widely regarded as a compelling technology for grid-scale energy storage owing to their intrinsic safety, cost-effectiveness, and the natural abundance of Zn. Nevertheless, their practical deployment is largely impeded by water-induced hydrogen evolution and dendritic growth on the Zn anode. To address these challenges, a machine-learning-assisted design paradigm based on a homologous-constraint strategy was developed, in which a linear-kernel support vector regression model combined with leave-one-out cross-validation was employed to screen and identify two amino acid-derived additives with outstanding water-structure-regulating capability, namely, dl-glutamic acid and dl-histidine. Experimental investigations reveal that these additives reconfigure the local hydrogen-bonding network of the electrolyte, effectively suppressing the hydrogen evolution reaction originating from highly reactive water and markedly lowering the interfacial desolvation energy barrier of Zn2+. As a consequence, Zn deposition evolves from disordered dendritic growth to a uniform two-dimensional nucleation process. Benefiting from the coupled optimization of thermodynamic and kinetic factors, the Zn||Zn symmetric cell achieves stable operation for over 4900 h in the modified electrolyte. This study not only establishes a novel data-driven strategy for designing electrolyte additives in AZIBs but also provides a broadly applicable framework for the rational molecular engineering of complex battery systems.