Two-state thermodynamics and the possibility of a liquid-liquid phase transition in supercooled TIP4P/2005 water

Abstract

Water shows intriguing thermodynamic and dynamic anomalies in the supercooled liquid state. One possible explanation of the origin of these anomalies lies in the existence of a metastable liquid-liquid phase transition (LLPT) between two (high and low density) forms of water. While the anomalies are observed in experiments on bulk and confined water and by computer simulation studies of different water-like models, the existence of a LLPT in water is still debated. Unambiguous experimental proof of the existence of a LLPT in bulk supercooled water is hampered by fast ice nucleation which is a precursor of the hypothesized LLPT. Moreover, the hypothesized LLPT, being metastable, in principle cannot exist in the thermodynamic limit (infinite size, infinite time). Therefore, computer simulations of water models are crucial for exploring the possibility of the metastable LLPT and the nature of the anomalies. In this work, we present new simulation results in the NVT ensemble for one of the most accurate classical molecular models of water, TIP4P/2005. To describe the computed properties and explore the possibility of a LLPT, we have applied two-structure thermodynamics, viewing water as a non-ideal mixture of two interconvertible local structures ("states"). The results suggest the presence of a liquid-liquid critical point and are consistent with the existence of a LLPT in this model for the simulated length and time scales. We have compared the behavior of TIP4P/2005 with other popular water-like models, namely, mW and ST2, and with real water, all of which are well described by two-state thermodynamics. In view of the current debate involving different studies of TIP4P/2005, we discuss consequences of metastability and finite size in observing the liquid-liquid separation. We also address the relationship between the phenomenological order parameter of two-structure thermodynamics and the microscopic nature of the low-density structure.

Two-structure thermodynamics for the TIP4P/2005 model of water covering supercooled and deeply stretched regions

“…Experiments [22][23][24] and simulations 13,17,21 support the existence of two distinct, interconvertible local structures in cold and supercooled water, as well as in water-like models. In particular, the TIP4P/2005, 13,17 TIP5P, 13 and mW 21 models of water show an increase in the number of molecules with four nearest neighbors and in local tetrahedral arrangements upon cooling.…”

“…Several versions of the TSEOS were successfully used for the description of the thermodynamic anomalies in supercooled water, 12,14 as well as in different models of water: mW, 15 ST2, 16 and TIP4P/2005. 17 Sufficient non-ideality in the mixing of these two alternative structures could lead to a liquid-liquid phase transition (as in ST2 16,[18][19][20] and, possibly, TIP4P/2005 17 ). The existence of a low-density, low entropy structure accounts for the density anomaly upon cooling, as well as for the increase in compressibility and isobaric heat capacity.…”

“…1,37,38 However, the debates on the behavior of the stability limits in doubly metastable water are far from over, especially in view of a "critical-point-free" scenario, the possibility of continuation of the first-order LLPT down to the absolute stability of the liquid state. 9,35,39,41 In the present work, we have applied two-structure thermodynamics to the description of recently published 17,42,43 and new, previously unpublished extensive simulation data from the Princeton group on the thermodynamic properties of the TIP4P/2005 classical water model over a wide range of temperatures and pressures. TIP4P/2005 is one of the best available models of water, and, in particular, it reproduces well the thermodynamic anomalies of real water at low temperatures, 44 and the sound velocity in stretched water.…”

Markov state model of the two-state behaviour of water

“…Also for TIP4P/2005 water, evidence for a liquid-to-liquid phase transition has been provided, [37][38][39][40] albeit much less thoroughly than for ST2 water. Furthermore, the criticism that the observation of a liquid-to-liquid phase transition is an artifact of non-equilibrated MD simulation [11][12][13][14] is probably even more relevant in the case of TIP4P/2005 water, since the hypothesized phase transition point is much lower in temperature, i.e., at ≈180-190 K. [37][38][39][40] It will be shown that the MSM can resolve two-state behaviour also for TIP4P/2005 water in a temperature range that covers the density maximum, with a shift in equilibrium between HDL and LDL domains that is exactly as expected for the two-state model of water.…”

Communication: Diffusion constant in supercooled water as the Widom line is crossed in no man’s land

“…[33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49][50][51] However, there is a growing consensus that many reasonable simulation models do show an LLCP. 11,30,[51][52][53][54] For example, simulations with the TIP4P/2005 model are consistent with an LLCP at about 1.7 kbar and 182 K. 54 We have been working with a related model that includes explicit three-body interactions called the E3B3 model 55 10 emanates from the LLCP and extends into the one-phase region. Also shown are the homogeneous nucleation line (green), 18 the temperature of maximum compressibility at 1 bar 28 (orange square), and the states sampled by Wyslouzil and co-workers (solid red), 26 Nilsson and co-workers (solid blue), 27 this model at various state points in the metastable region of the p-T plane: at pressures (1 bar to 1 kbar) much lower than the (theoretical) critical pressure, we find that these properties change smoothly as one crosses the theoretical Widom line, but at higher pressures (1.5 to 2 kbar), these properties show dramatic changes, consistent with the expectation described above.…”