RESUMO
Ion transport through electrolytes critically impacts the performance of batteries and other devices. Many frameworks used to model ion transport assume hydrodynamic mechanisms and focus on maximizing conductivity by minimizing viscosity. However, solid-state electrolytes illustrate that non-hydrodynamic ion transport can define device performance. Increasingly, selective transport mechanisms, such as hopping, are proposed for concentrated electrolytes. However, viscosity-conductivity scaling relationships in ionic liquids are often analyzed with hydrodynamic models. We report data-centric analyses of hydrodynamic transport models of viscosity-conductivity scaling in ionic liquids by merging three databases to bridge physical properties and computational descriptors. With this expansive database, we constrained scaling analyses using ion sizes defined from simulated volumes, as opposed to estimating sizes from activity coefficients. Remarkably, we find that many ionic liquids exhibit positive deviations from the Nernst-Einstein model, implying ions move faster than hydrodynamics should allow. We verify these findings using microrheology and conductivity experiments. We further show that machine learning tools can improve predictions of conductivity from molecular properties, including predictions from solely computational features. Our findings reveal that many ionic liquids exhibit super-hydrodynamic viscosity-conductivity scaling, suggesting mechanisms of correlated ion motion, which could be harnessed to enhance electrochemical device performance.