Deeply supercooled water exhibits complex dynamics with large density fluctuations, ice coarsening and characteristic time scales extending from picoseconds to milliseconds. Here, we discuss implications of these time scales as they pertain to two-ph
ase coexistence and to molecular simulations of supercooled water. Specifically, we argue that it is possible to discount liquid-liquid criticality because the time scales imply that correlation lengths for such behavior would be bounded by no more than a few nanometers. Similarly, it is possible to discount two-liquid coexistence because the time scales imply a bounded interfacial free energy that cannot grow in proportion to a macroscopic surface area. From time scales alone, therefore, we see that coexisting domains of differing density in supercooled water can be no more than nano-scale transient fluctuations.
The recent paper cited above claims that a molecular simulation of one specific model of supercooled water establishes a stable interface separating two metastable liquid phases, which would imply the existence of metastable two-liquid criticality fo
r that model. Here, we note that this claim conflicts with fundamental principles and with earlier work published in the textit{Journal of Chemical Physics}, and we show that the claim is unjustified by the data put forward to support the conclusion. Other technical problems are also noted.
This article is a brief Retrospective on the life and work of Robert W. Zwanzig, who formulated nonequilibrium statistical mechanics and who passed away in May of this year.
