Relation between diffusion, viscosity and packing fraction in metallic melts

Abstract

In liquids, the self-diffusion D and the viscosity η can be often linked by the phenomenological Stokes-Einstein relation (SER), which is based on a model of a mesoscopic sphere immersed in a viscous fluid. For a multicomponent, undercooled glass forming melts, different species in the liquid may exhibit different dynamics, so that in some cases the SER does not provide a good description of the relation between D and η. These deviations can be thus attributed to the so-called ”dynamical decoupling”. However, in the equilibrium and undercooled melt of Zr64Ni36 it has been experimentally observed that D · η = const. in contrast to the SER [1]. This cannot be explained by dynamical decoupling, as the self-diffusion coefficients of Zr and Ni are almost equal [Basuki PRB 2017]. One explanation according to Schober [2] is that in densely packed systems a collective motion of the diffusing particles occurs, whereas the SER is based on a model of uncorrelated motions of individual species. Thus, the packing fraction of a melt appears to be a key parameter. In order to check the validity of the SER in relation to the packing fraction, reliable measurements on metallic melts with different packing fractions have been carried out. DNi, η and ρ of liquid Ni66.7B33.3 and Ge66.7Ni33.3 have been measured via quasielastic neutron scattering experiments, via electrostatic levitation combined with oscillating drop method and via oscillating cup technique, respectively. For the densely packed liquid Ni66.7B33.3, it was found D · η = const., which is contradicting the SER. For the less densely packed Ge66.7Ni33.3, it was found that D · η ∝ T, which is in reasonable agreement to the predictions of the SER. [1] J. Brillo, A. I. Pommrich, and A. Meyer, Phys. Rev. Lett. 107, 165902 (2011), [2] H. R. Schober, Physics 4, 80 (2011).