The local point symmetry of the short-range order in simple monatomic liquids remains a fundamental open question in condensed-matter science. For more than 40 years it has been conjectured that liquids with centrosymmetric interactions may be composed of icosahedral building blocks. But these proposed mobile, randomly orientated structures have remained experimentally inaccessible owing to the unavoidable averaging involved in scattering experiments, which can therefore determine only the isotropic radial distribution function. Here we overcome this limitation by capturing liquid fragments at a solid-liquid interface, and observing the scattering of totally internally reflected (evanescent) X-rays, which are sensitive only to the liquid structure at the interface. Using this method, we observe five-fold local symmetry in liquid lead adjacent to a silicon wall, and obtain an experimental portrait of the icosahedral fragments that are predicted to occur in all close-packed monatomic liquids. By shedding new light on local bond order in disordered structures such as liquids and glasses, these results should lead to a better microscopic understanding of melting, freezing and supercooling.

X-ray reflectivity measurements, modeled on atomic scale by a dynamic approach, reveal a smooth Au top layer and subsequent Cu/ Au layering at the (001) surface of CuAu crystals at high temperatures, enforced by a large surface field h 1. Approaching the first order bulk transition the ordered near surface film grows but does not completely wet its disordered bulk phase as predicted. In contrast to earlier wetting experiments on heterogeneous systems which were dominated by surface roughness, we found strong lattice relaxations and incomplete wetting due to strain.

We report a new type of short-range order correlations at the (001) surface of Cu3Au which no longer produces the 2k(F)-splitting characteristic for the bulk short-range order scattering. We present the temperature dependence of this phenomenon and a theoretical interpretation of its origin. We argue that this new surface effect is caused by a drastic change of the strain-induced interactions at the surface.