(9f) Vapor-Liquid Equilibrium and Nucleation Studies of Water from First Principles-Based Machine Learning Models | AIChE

(9f) Vapor-Liquid Equilibrium and Nucleation Studies of Water from First Principles-Based Machine Learning Models


Car, R., Princeton University
Panagiotopoulos, A., Princeton University
Throughout the years, much attention has been given to water in its bulk liquid and ice phases. The vapor-liquid equilibrium of water is involved in several relevant industrial applications, such as in power generation. When liquid water is superheated past the liquid-vapor coexistence line, it becomes metastable. The transition from liquid to vapor occurs subsequently with the emergence of small embryos (bubbles) of the vapor phase in a process termed bubble nucleation, and their collective collapse may lead to structural damage in industrial machinery1. There are significant challenges in studying these processes both experimentally and theoretically, as evidenced by the discrepancy of results obtained over the years in studies for properties such as nucleation rates and cavitation pressures2. Computer simulations can be employed to obtain a realistic description of these phenomena; however, results can be highly dependent on the chosen representation of the intermolecular interactions. In this work, we use a neural network energy function constructed with the Deep Potential Molecular Dynamics (DPMD)3 framework, that accurately describes the SCAN approximation of quantum mechanical density functional theory. This model was recently used to predict the phase diagram of water’s liquid and ice phases, obtaining a good overall agreement with experiment4. We apply this model to perform direct coexistence molecular dynamics simulations of water’s vapor-liquid equilibrium and calculate properties such as coexistence densities, surface tensions and enthalpy of vaporization. Lastly, we describe ongoing efforts to perform a first principles bubble nucleation investigation of water to relate our results with previous studies.


1. Buckland, H. C., Masters, I., Orme, J. A. C. & Baker, T. Cavitation inception and simulation in blade element momentum theory for modelling tidal stream turbines. Proc. Inst. Mech. Eng. Part A J. Power Energy 227, 479–485 (2013).

2. Caupin, F. & Herbert, E. Cavitation in water: a review. Comptes Rendus Physique 7, 1000–1017 (2006).

3. Zhang, L., Han, J., Wang, H., Car, R. & E, W. Deep Potential Molecular Dynamics: A Scalable Model with the Accuracy of Quantum Mechanics. Phys. Rev. Lett. 120, 143001 (2018).

4. Zhang, L., Wang, H., Car, R. & E, W. Phase Diagram of a Deep Potential Water Model. Phys. Rev. Lett. 126, 236001 (2021).