Heterogeneous ice nucleation is ubiquitous in nature, yet the microscopic factors governing its kinetics are complex and remain incompletely understood. A widely adopted view is that an effective ice nucleator should exhibit minimal lattice mismatch with ice when other factors are identical. Here, using forward flux sampling simulations, we show that this criterion breaks down for graphene substrates: The highest ice nucleation rate occurs at a lattice constant that matches neither bulk ice nor the critical ice nucleus. We demonstrate that this counterintuitive behavior is attributed to the density difference between interfacial water and ice in the first contact layer. Enhanced surface hydrophilicity induces a negative lattice mismatch that minimizes this density difference, thus maximizing the nucleation rate. Strikingly, the interfacial water-ice density difference exhibits a global correlation with heterogeneous ice nucleation rates across a wide range of surface hydrophilicities and lattice mismatches. These results establish density difference, rather than lattice match, as a fundamental and transferable descriptor of ice nucleation efficiency.
Zhao et al. (Wed,) studied this question.