High-temperature viscosity of many-component glass melts

Abstract

In this article, we argue that 1) the activation energy for viscous flow becomes independent of temperature when the viscosity of molten glass is sufficiently low at high enough temperatures, such as those that exist in a glass-melting furnace, and 2) the intercept of the linear function lnη versus T−1 (η is the viscosity and T is the temperature) is independent of glass composition. This hypothesis, which is hardly new and is well supported by experimental data, allows minimization of the number of fitting parameters. A new dataset of meticulously measured viscosities of a large composition region of simulated nuclear waste glasses that recently became available provided an excellent opportunity to test this hypothesis to verify it again. Also, we used this dataset to demonstrate that some popular functions designed for representing the high-viscosity segment (where the activation energy changes with temperature) are not recommendable for approximating the low-viscosity segment (where the activation energy is constant). Fitting such functions produces overparameterization and leads to physically meaningless (or at least esthetically unsatisfactory) outcomes, or, if the functions are constrained by the glass-transition viscosity and the high-temperature asymptote, the result is a significant lack of fit.