Surface tension of liquid metals varies by over an order of magnitude across the periodic table. Density functional theory (DFT) can compute these values numerically, and Miedema's semi-empirical model achieves +/-10-20% accuracy using the Wigner-Seitz boundary electron density nws, but neither provides an electronic-structure descriptor that explains why metals with similar bulk densities (e. g. , Al and Zn, Delta rₛ = 2. 4%, Delta gamma = 46%) have vastly different surface tensions. Here we propose that the inverse participation ratio (IPR) of Kohn-Sham orbitals -- an existing quantity widely used in Anderson localization analysis -- provides this missing link when reinterpreted as a unified descriptor for interfacial properties. DFT calculations on 15 metal dimers show that a valence-only IPR-based delocalization index correlates with interatomic midpoint electron density at r = 0. 89 (p Fe (d6: 0. 37) > Cu (d10: 0. 33) is consistent with Friedel's model of d-band narrowing with filling. Maximally localized Wannier function (MLWF) calculations independently confirm this hierarchy. These results establish that Miedema's nws is a proxy for valence electron delocalization, answering the question raised by Williams, Gelatt, and Moruzzi (PRL 44, 429, 1980) regarding the physical meaning of nws.
Kazuyoshi Miyauchi (Sun,) studied this question.