Nonequilibrium molecular dynamics simulations of the thermal conductivity of water: A systematic investigation of the SPC/E and TIP4P/2005 models

Abstract

We report an extensive non equilibrium molecular dynamics investigation of the thermal conductivity of water using two of the most accurate rigid non polarizable empirical models available, SPC/E and TIP4P/2005. Our study covers liquid and supercritical states. Both models predict the anomalous increase of the thermal conductivity with temperature and the thermal conductivity maximum, hence confirming their ability to reproduce the complex anomalous behaviour of water. The performance of the models strongly depends on the thermodynamic state investigated, and best agreement with experiment is obtained for states close to the liquid coexistence line and at high densities and temperatures. Considering the simplicity of these two models the overall agreement with experiments is remarkable. Our results show that explicit polarizability and molecular flexibility are not needed to reproduce the anomalous heat conduction of water.

Molecular alignment in molecular fluids induced by coupling between density and thermal gradients

“…The same thermal gradient induces a TMO effect 2.4 times larger in the size asymmetric system studied here compared with the mass asymmetric system. The strength of the TMO signal varies both with the thermodynamic conditions as well as the model parameters, as has been noted previously [8,11,12,14,23,28]. …”

Polarisation of polar dumbbell fluids in thermal gradients: the importance of the treatment of electrostatic interactions

“…It is also worth noting that in our setup the system is in mechanical equilibrium (∇P = 0), and on average the pressure is constant in the simulation box. Equations of state obtained from local temperatures and densities across the simulation box in this method match results from equilibrium simulations quite well [25], as do quantities such as thermal conductivities computed from a Green-Kubo approach in equilibrium simulations versus a direct computation based on the heat flux in NEMD [26].…”

Non-equilibrium molecular dynamics simulations of the thermal transport properties of Lennard-Jones fluids using configurational temperatures

“…In this work we employed a simple velocity rescaling approach, where the new velocity of particle i, v i,new = v i,old , is given by the velocity before rescaling (v i,old ) and the rescaling constant, = p K t /K old , which is defined by the ratio of the old and target kinetic energies. This implementation is similar to the thermostatting BD-NEMD scheme [19].…”

“…There are two main methods to perform non-equilibrium simulations under thermal gradients, the synthetic approach [16], which relies on the calculation of correlation functions under the ⇤ Electronic address: n.jackson12@imperial.ac.uk † Electronic address: f.bresme@imperial.ac.uk influence of a fictitious field, and the boundary driven non-equilibrium molecular dynamics (BD-NEMD) approach, in which a thermal gradient is explicitly simulated by inducing a heat flux along the simulation box. The flux is normally driven by adding/withdrawing energy [17,18] or by directly thermostatting the molecules in predefined boundary regions [19]. The BD-NEMD approach has become a popular choice, as it is easy to implement, and no a priori knowledge of the microscopic heat flux equation (MHFE) is required, since the heat flux can be extracted from an analysis of the rate of energy exchange.…”