Machine-Learning-Guided Molecular Design for Dynamic Self-Activating Interface toward Ah-Level Anode-Free Zn–I2 Pouch Cells

Abstract Anode-free Zn batteries are attractive for zero-excess-Zn operation and maximized anode utilization, but reversible high-areal-capacity Zn plating/stripping over large hostless interfaces remains difficult, especially in Ah-level Zn–I2 pouch cells where polyiodide shuttling accelerates capacity decay and corrodes the nascent anode interface. Here, we integrate machine-learning-guided polyiodide regulation with dynamic interfacial reconstruction to establish a self-activating Zn deposition pathway for reversible anode-free Zn–I2 cells. DFT-supervised screening reveals multidentate weak O–C–H···I interactions and chain-assisted confinement as key molecular design principles for polyiodide confinement, guiding the selection of polyethylene glycol 4000 (PE4000). More importantly, PE4000 gates trace reversible Sn/Sn2+ interfacial conversion, enabling dynamic Zn–Sn reconstruction, persistent zincophilic sites, and orientation-regulated Zn nucleation/growth during repeated plating/stripping. Consequently, the optimized anode-free Zn–I2 cells sustain 16,000 cycles at 10 mA cm–2. Single-layer pouch cells operate for over 3000 cycles at 4.65 mAh cm–2, with recovered Sn remaining reusable for another 1000 cycles. Ah-level pouch cells further deliver 2390 Ah cumulative capacity over 2800 cycles. This work establishes a system-level interface-matching strategy for practical high-utilization anode-free batteries.