Grand canonical Monte Carlo simulations of phase equilibria of pure silicon tetrachloride and its binary mixture with carbon dioxide

Grand canonical histogram-reweighting Monte Carlo simulations were used to obtain the phase behaviour of pure silicon tetrachloride and its binary mixture with carbon dioxide. Two new potential models for pure silicon tetrachloride were developed and parametrized to the vapour-liquid coexistence properties. The first model, with one exponential-6 site and fixed electrostatic charges on atoms, does not adequately reproduce the experimental phase behaviour due to its inability to represent orientational anisotropy in the liquid phase. The second potential model, with five exponential-6 sites for the repulsive and dispersive interactions plus partial charges, accurately reproduces experimental saturated liquid and vapour densities as well as vapour pressures and the second virial coefficient for pure silicon tetrachloride. This model was used in simulations of the phase behaviour of the binary mixture carbon dioxide-silicon tetrachloride. Two sets of combining rules (Lorentz-Berthelot and Kong [1973, J. chem. Phys., 59, 2464]) were used to obtain unlike-pair potential parameters. For the binary system, the predicted phase diagram is in good agreement with experiment when the Kong combining rules are used. The Lorentz-Berthelot rules significantly overestimate the solubility of carbon dioxide in silicon tetrachloride.     

Molecular simulation of phase equilibria for mixtures of polar and non-polar components

Grand canonical histogram-reweighting Monte Carlo simulations were used to obtain the phase behaviour of several binary mixtures. The main goal of this work was to test the predictive capabilities of recently developed intermolecular potential models that accurately reproduce the phase behaviour of pure components. These united-atom potentials utilize the exponential-6 functional form for repulsive and dispersion interactions and fixed point charges for electrostatic interactions. The mixtures studied were n-pentane—methane, ethane—CO₂, propane—CO₂, n-pentane-CO₂, H₂O-ethane, CH₃OH-n-hexane and CH₃OH-CO₂. The conventional Lorentz-Berthelot combining rules, as well as a set of combining rules due to Kong (1973, J. chem. Phys., 59, 2464) were used to obtain unlike-pair potential parameters. The Lorentz—Berthelot rules generally result in more attractive unlike-pair interactions than the Kong rules. For the n-alkane—CO₂ systems, predicted phase diagrams are in excellent agreement with experiment when the Kong combining rules are used. For mixtures with CH₃OH and H₂O, the Lorentz—Berthelot rules yield better agreement with experiment than the Kong rules, but statistically significant differences remain. Our results suggest that relatively simple intermolecular potential models can be used to predict the phase behaviour of broad classes of binary systems. For mixtures with large differences in polar character of the components, however, present models do not predict the phase behaviour in quantitative agreement with experiment. New models that include higher order interactions such as polarizability may be suitable for this purpose, a hypothesis that will need to be tested in the future.     

Excess properties of Lennard-Jones binary mixtures from computer simulation and theory

Monte Carlo simulation and theory are used to calculate the excess thermodynamic properties of binary mixtures of spherical Lennard-Jones molecules. We study the excess functions of three binary mixtures characterized by the following size and dispersive energy ratios: (1) (σ₂₂/σ₁₁)³ = 2 and ?₂₂/?₁₁ = 2; (2) (σ₂₂/σ₁₁)³ = 1 and ?₂₂/?₁₁ = 1/2 and (3) (σ₂₂/σ₁₁)³ = 1/2 and ?₂₂/?₁₁ = 2. In all cases, the unlike size parameter, σ₁₂, is kept constant and equal to the value given by the Lorentz combining rule (σ₁₂ = (σ₁₁ + σ₂₂)/2). However, different unlike dispersive energy parameter values are considered through the following combining rules: (a) ?₁₂ = (?₁₁?₂₂)^1/2 (Berthelot rule); (b) ?₁₂ = ?₁₁ (association); and (c) ?₁₂ = ?₂₂ (solvation). The pressure and temperature dependence of the excess volume and excess enthalpy is studied using the NpT Monte Carlo simulation technique for all the systems considered. Additionally, the simplest conformal solution theory is used to check the adequacy of this approach in predicting the excess properties in a wide range of thermodynamic conditions and variety of binary mixtures. In particular, we have applied the van der Waals one-fluid theory to describe Lennard-Jones binary mixtures through the use of the Johnson et al. [1993, Molec. Phys., 78, 591] Helmholtz free energy. Agreement between simulation results and theoretical predictions is excellent in all cases and thermodynamic conditions considered. This work confirms the applicability of the van der Waals one-fluid theory in predicting excess thermodynamic properties of mixtures of spherical molecules. Furthermore, since binary mixtures of spherical Lennard-Jones molecules constitute the reference fluid to be used in perturbation theories for complex fluids, such as the statistical association fluid theory (SAFT), this work shows clearly the applicability of the conformal solution theory within the framework of SAFT for predicting excess functions.     

Molecular simulation of chemical reaction equilibria by Kinetic Monte Carlo

ABSTRACT

We describe a new algorithm for the molecular simulation of chemical reaction equilibria, which we call the Reactive Kinetic Monte Carlo (ReKMC) algorithm. It is based on the use of the equilibrium Kinetic Monte Carlo (eKMC) method (Ustinov et al., J. Colloid Interface Sci., 2012, 366, 216–223) to generate configurations in the underlying nonreacting system and to calculate the species chemical potentials at essentially zero marginal computational cost. We consider in detail the typical case of specified temperature, T and pressure, P, but extensions to other thermodynamic constraints are straightforward in principle. In the course of this work, we also demonstrate an alternative method for calculating simulation box volume changes in NPT ensemble simulations to achieve the specified P. We consider two sets of example reacting systems previously considered in the literature, and compare the ReKMC results and computational efficiencies with those of different implementations of the REMC algorithm (Turner et al., Molec. Simulation, 2008, 34, 119–146).     

Equation of state for systems of soft diatomic molecules

A new potential that is a modification of the BBL (Bratko, D.,Blum, L., and Luzar, A.,1985, J. chem. Phys., 83, 6367; Blum, L., Vericat, F., and Bratko, D., 1995, J. chem. Phys., 102, 1461) potential and of the one recently solved analytically by Blum and Vericat (BV) (1995, Molec. Phys., 86, 809; 1996, J. phys. Chem., 100, 1197) is studied by Monte Carlo simulation. The main feature of this potential is that it can be solved using only a small number of parameters (3 in the case treated by BV), and therefore produces a substantial simplification of earlier work. The new potential has an orientational octupole–octupole interaction term which is found necessary to reproduce the broad peak of the oxygen–oxygen structure function due to the tetrahedral position of the second nearest neighbour water molecule. This important feature was absent in the original BBL potential. This model agrees also with the experimental pair correlation functions for oxygen–hydrogen and hydrogen–hydrogen, and yields 42·6 kJ mol^-1 for the internal energy of water, also in agreement with experiment. The hard core central repulsion causes the sharpness of the first peaks in all three correlation functions. This is not necessary but convenient for an analytical solution.     

Is it possible to infer the equation of state of a mixture of hard discs from that of the one-component system?

Based on exact asymptotic properties of the composition-independent virial coefficients of a binary mixture of hard discs in the limits α = σ₂/σ₁ → 0, α → 1 and α → ∞, R. J. Wheatley (1998, Molec. Phys., 93, 965) has recently proposed an approximate interpolation equation for these coefficients. In this note, the equation of state equivalent to this interpolation is obtained, expressing the compressibility factor of the mixture in terms of that of the pure system. An extension to an arbitrary number of components is also given. The equation of state derived here is compared with another one recently proposed by following a different route (Santos, A., Yuste, S. B., and López de Haro, M., 1999, Molec. Phys., 96, 1) and with Monte Carlo simulation results. It is shown that the latter equation is more accurate than the former one, at least for not too disparate mixtures (0.7 < α < 1).     

Statistical mechanics of hydrogen bond networks

The static and dynamic properties of several hydrogen bond network models, based on thesquare ice model of Lieb [Phys. Rev.162, 162 (1967)] are studied. The two dimensionalsquare water (SW) model and the three-dimensioaalbrick water (BW) model were analyzed by means of Monte Carlo simulations. A simplified vesion of SW (simplified square water, SSW) can be solved exactly. All models yield similar thermodynamic results which can be derived-alternatively-from an independent bond approach due to Angell [J. Chem. Phys.75, 3698 (1971)]. We suggest the existence of a universality class of hydrogen bond networks that can be described by this theory, and which may include the liquid state of water. The mean lifetime of a hydrogen bond exhibits an Arrhenius temperature dependence. Comparison with experimental data on water provides an absolute time scale for the Monte Carlo simulations. The possible use of these models in simulations of protein-solvent systems is discussed.     

Mixtures of hard spherocylinders and hard spheres

Monte Carlo simulations have been performed for equimolar mixtures of hard prolate spherocylinders of length: breadth ratio 2:1 and hard spheres, in the fluid region. Two systems have been studied. In the first the breadth of the spherocylinder was equal to the hard sphere diameter, and in the second system both components were of equal molecular volume.

The compressibility factor, PV/NkT, has been obtained for both mixtures at four reduced densities (packing fractions) from 0·20 to 0·45. The results have been compared with the predictions of several analytical equations appropriate to mixtures of hard convex molecules, and an equation due to Pavlicek et al. was found to be very accurate. The results have been used to calculate the excess volumes of mixing at constant pressure, in an attempt to establish the relative importance of the effects of differences in molecular volume and shape on the thermodynamic properties.

The structural properties of the mixtures have also been investigated by calculating pair distribution functions for the three types of pair interactions present in these mixtures.     

Monte Carlo Simulation of Hard Hyperspheres in Six,Seven and Eight Dimensions for Low to Moderate Densities

The pair correlation function of hard hyperspheres in six, seven and eight dimensions is obtained from Monte Carlo simulations. The value of the pair correlation function at contact is compared with the results from molecular dynamics calculations and a variety of theoretical approaches. Remarkably good agreement is found with the simple, closed-form equations of Y. Song, E. A. Mason and R. M. Stratt, J. Phys. Chem., 93:6916–6919 (1989). The Monte Carlo results for the equation of state are compared with the theoretical expressions of M. Baus and J. L. Colot, Phys. Rev. A, 36:3912 (1987), M. Luban and J. P. J. Michels, Phys. Rev A, 41:6796 (1990), and high order virial expansions. In addition, in seven dimensions, comparisons are made with the exact PY solution provided by M. Robles, M. L. de Haro and A. Santos, J. Chem. Phys., 120:9113 (2004). Very good agreement was observed between theory and computer simulation in all dimensions.     

Perturbation theory for the square-well dimer fluid

An analytical equation of state is presented for the square-well dimer fluid of variable well width (1 ≤ λ ≥ 2) based on Barker-Henderson perturbation theory using the recently developed analytical expression for radial distribution function of hard dimers. The integral in the first- and the second-order perturbation terms utilizes the Tang, Y and Lu, B. C.-Y., 1994, J. chem. Phys., 100, 6665 formula for the Hilbert transform. To test the equation of state, NVT and Gibbs ensemble Monte Carlo simulations for square-well dimer fluids are performed for three different well widths (λ = 1.3, 1.5 and 1.8). The prediction of the perturbation theory is also compared with that of thermodynamic perturbation theory in which the equation of state for the square-well dimer is written in terms of that of square-well monomers and the contact value of the radial distribution function.     

On the role of the definition of potential models in Gibbs ensemble phase equilibria simulations of the H2S-pentane mixture

The effect of treating explicitly the coulombic and polarization interactions is investigated through the calculation of the coexistence curve of the pentane-H₂S binary mixture. In this work, potential models have been developed for hydrogen sulphide and pentane, which include electrostatic sites—estimated from ab initio calculations—and polarizable sites—estimated from experimental data—in addition to Lennard-Jones sites. Compared to existing models, these new models have the same number of fitting parameters to experimental thermodynamic data. They are shown to correctly describe the coexistence curve of the pure compounds. When applied to the case of mixtures, together with Lorentz-Berthelot combining rules, they allow one to obtain a more accurate prediction of the coexisting compositions of the mixture. Furthermore, it is shown that the interaction energy in this kind of mixture cannot be properly described by using effective potential models usually considered for pure compounds.     

Effect of short- and long-range forces on the properties of fluids. III. Dipolar and quadrupolar fluids

Using realistic pair potential models for acetone and carbon dioxide, both the spatial and orientational structure of these two typical multipolar (i.e. dipolar and quadrupolar, respectively) fluids is investigated in detail by computing the complete set of the site-site correlation functions, multipole-multipole correlation functions, and selected 2D correlation functions. The effect of the range of interactions on both the structural and thermodynamic properties of these fluids is studied by decomposing the potential into short- and long-range parts in the same manner as for water [Kolafa, J. and Nezbeda, I., 2000, Molec. Phys., 98, 1505; Nezbeda, I. and Lísal, M., 2001, Molec. Phys., 99, 291]. It is found that the spatial arrangement of the molecules is only marginally affected by the long-range forces. The effect of the electrostatic interactions is significant at short separations and cannot be neglected but nevertheless the overall structure of the short-range and full systems is similar as well as their dielectric constants. These findings are also reflected in the dependence of the thermodynamic properties on the potential range with the short-range models providing a very good approximation to those of the full system.     

Recoil growth algorithm for chain molecules with continuous interactions

The recoil growth (RG) scheme is a dynamic Monte Carlo algorithm that has been suggested as an improvement over the configurational bias Monte Carlo (CBMC) method (Consta, S., Wilding, N. B., Frenkel, D. and Alexandrowicz, Z., 1999, J. chem. Phys., 110, 3220). The RG method had originally been tested for hard core polymers on a lattice, and it was found that RG outperforms CBMC for dense systems and long chain molecules. In the present paper, the RG scheme is extended to the practically more relevant case of off-lattice chain molecules with continuous interactions. It is found that for longer chain molecules RG becomes over an order of magnitude more efficient than CBMC. However, other schemes are better suited to the computation of the excess chemical potential. Moreover, it is more difficult to parallelize RG than CBMC.     

Monte Carlo simulations of primitive models for ionic systems using the Wolf method

Thermodynamic and structural properties of primitive models for electrolyte solutions and molten salts were studied using NVT and NPT Monte Carlo simulations. The Coulombic interactions were simulated using the Wolf method [D. Wolf, Phys. Rev. Lett. 68, 3315 (1992); D. Wolf, P. Keblinnski, S. R. Phillpot, and J. Eggebrecht, J. Chem. Phys. 110, 8254 (1999)]. Results for 1?:?1 and 2?:?1 charge ratio electroneutral systems are presented, using the restricted and non-restricted primitive models, as well as a soft PM pair potential for a monovalent salt [J.-P. Hansen and I. R. McDonald, Phys. Rev. A 11, 2111 (1975)] that has also been used to model 2?:?12 and 1?:?20 asymmetric colloidal systems, with size ratios 1?:?10 and 2?:?15, respectively [B. Hribar, Y. V. Kalyuzhnyi, and V. Vlachy, Molec. Phys. 87, 1317 (1996)]. We present the predictions obtained for these systems using the Wolf method. Our results are in very good agreement with simulation data obtained with the Ewald sum method as well as with integral-equation theories results. We discuss the relevance of the Wolf method in the context of variable-ranged potentials in molecular thermodynamic theories for complex fluids.     

Ab initio intermolecular potentials of methane,nitrogen and methane + nitrogen and their use in Monte Carlo simulations of fluids and fluid mixtures

Intermolecular pair potentials of methane and of methane + nitrogen have been calculated by quantum chemical ab initio methods. The repulsive and electrostatic parts were determined pointwise for various distances and orientations of the dimers (supermolecule approach) by self-consistent field (SCF) calculations including the counterpoise correction. Gaussian basis functions of triple-zeta quality plus one set of polarization functions on all atoms were used. The dispersion energy, which cannot be calculated at the SCF level, has been added by a semi-empirical estimate. These potentials and the pair potential for nitrogen (of similar quality, taken from the literature) were fitted to analytical functions and used for NVT Monte Carlo simulations of thermodynamic properties of the fluids and their mixture over a wide temperature/density area. Comparison with measurements and with Monte Carlo results from the literature (pressure, internal energy, radial distribution function) obtained with other pair potentials indicates the quality of the present calculations.     

The virial coefficients of hard hypersphere binary mixtures

The third, fourth and fifth virial coefficients of hard hypersphere binary mixtures with dimensionality d = 4, 5 have been calculated for size ratios R ≥ 0.1, R = ≡ σ₂₂/σ₁₁, where σ _ii is the diameter of component i. The composition independent partial virial coefficients have been evaluated by Monte Carlo integration of the corresponding Mayer modified star diagrams. The results are compared with the predictions of Santos, S., Yuste, S. B., and Lopez de Haro, M., 1999, Molec. Phys., 96, 1 of the equation of state of a multicomponent mixture of hard hyperspheres, and the good agreement gives strong support to the validity of that recipe.     

Monte Carlo simulations of neon and argon using ab initio potentials

Gibbs ensemble Monte Carlo simulations of neon and argon have been performed with pair potentials taken from literature as well as with new ab initio potentials from just above the triple point to close to the critical point. The densities of the coexisting phases, their pair correlation functions, the vapour pressure and the enthalpy and entropy of vaporization have been calculated. The influence of the potential choice and of the addition of the Axilrod-Teller (AT) three-body potential on the above mentioned properties have been investigated. It turns out that an accurate ab initio two-body potential in connection with the AT potential yields very good results for thermodynamic properties of phase equilibria.     

Phase coexistence and interface structure of a two-component Lennard-Jones fluid in porous media: application of Born—Green—Yvon equation

The Born-Green-Yvon equation with smoothed density approximation is used to calculate the liquid-liquid density profiles of a symmetric Lennard-Jones fluid in a hard sphere disordered matrix. The phase diagrams are evaluated for model systems characterized by different matrix densities and compared with the results of theoretical predictions and the Monte Carlo simulations of Gordon, P. A., and Glandt, E. D., 1996, J. chem. Phys., 105, 4257. It was found that increasing the matrix packing fraction reduces the magnitude of the miscibility gap and smooths the density profiles between two coexisting phases.     

A Monte Carlo study of dipolar hard spheres The pair distribution function and the dielectric constant

This paper is a systematic investigation of the effects of boundary conditions upon Monte Carlo calculations for dipolar fluids. Results are reported for the minimum image (MI), spherical cut-off (SC) and uniform reaction field methods. All three approximations are shown to give different pair distribution functions, g(12), and none yields the infinite system result. It is concluded that theories giving g(12) for an infinite system should not be evaluated by direct comparison with Monte Carlo results. Two alternative methods by which meaningful comparisons can be made are described in the text. The dependence of the thermodynamic properties upon boundary conditions is important only at large values of the dipole moment. For small to moderate dipoles both MI and SC are found to give reasonable estimates of the dielectric constant.