Critical-point of the Lennard-Jones fluid: A finite-size scaling study

Is water one liquid or two?

The idea that water is a mixture of two distinct states is analyzed in some detail. It is shown that the known compressibility of water is in fact sufficiently small that for a volume of water of size 1 nm³, the density fluctuations are of order 4% of the average density. This is much smaller than the ≈25% density fluctuations that would be required for significant regions of high and low density water to occur on this volume scale. It is also pointed out that the density fluctuations in water are, if anything, smaller than those that occur in other common liquids which do not have the anomalous properties of water. It is shown that if the distribution of density fluctuations is unimodal, the system is in the one-phase region, and if bimodal, it is in the two-phase region. None of the liquid or amorphous phases of water explored in this work give any sign of being in the two-phase region. Existing neutron and X-ray scattering data on water in the amorphous phases, and in the stable liquid phases as a function pressure and temperature, are subject to a new set of empirical potential structure refinement simulations. These simulations are interrogated for their configurational entropy, using a spherical harmonic reconstruction of the full orientational pair correlation function. It is shown that the excess pair entropy derived from this function, plus the known perfect gas contributions, give a reasonable account of the total entropy of water, within the likely errors. This estimated entropy follows the expected declining trend with decreasing temperature. Evidence that higher density water will have higher entropy than lower density water emerges, in accordance with what is expected from the negative thermal expansion coefficient of water at low temperatures. However, this entropy increase is not large and goes through a maximum before declining at yet higher densities and pressures, in a manner reminiscent of what has been previously observed in the diffusion coefficient as a function of pressure. There is no evidence that ambient water can be regarded as patches of high density, high entropy and low density, low entropy liquid, as some have claimed, since high density water has a similar entropy to low density water. There is some evidence that the distinction between these two states will become more pronounced as the temperature is lowered. Extensive discussion of the use of order parameters to describe water structure is given, and it is pointed out that these indices generally cannot be used to infer two-state behavior.

Critical fluctuations and slowing down of chaos

Fluids cooled to the liquid–vapor critical point develop system-spanning fluctuations in density that transform their visual appearance. Despite a rich phenomenology, however, there is not currently an explanation of the mechanical instability in the molecular motion at this critical point. Here, we couple techniques from nonlinear dynamics and statistical physics to analyze the emergence of this singular state. Numerical simulations and analytical models show how the ordering mechanisms of critical dynamics are measurable through the hierarchy of spatiotemporal Lyapunov vectors. A subset of unstable vectors soften near the critical point, with a marked suppression in their characteristic exponents that reflects a weakened sensitivity to initial conditions. Finite-time fluctuations in these exponents exhibit sharply peaked dynamical timescales and power law signatures of the critical dynamics. Collectively, these results are symptomatic of a critical slowing down of chaos that sits at the root of our statistical understanding of the liquid–vapor critical point.

It is well known that fluids become opaque at the liquid–vapor critical point, but a description of the underlying mechanical instability is still missing. Das and Green leverage nonlinear dynamics to quantify the role of chaos in the emergence of this critical phenomenon.

Chemical potential perturbation: a method to predict chemical potentials in periodic molecular simulations

A new method, called chemical potential perturbation (CPP), has been developed to predict the chemical potential as a function of density in periodic molecular simulations. The CPP method applies a spatially varying external force field to the simulation, causing the density to depend upon position in the simulation cell. Following equilibration the homogeneous (uniform or bulk) chemical potential as a function of density can be determined relative to some reference state after correcting for the effects of the inhomogeneity of the system. We compare three different methods of approximating this correction. The first method uses the van der Waals density gradient theory to approximate the inhomogeneous Helmholtz free energy density. The second method uses the local pressure tensor to approximate the homogeneous pressure. The third method uses the Triezenberg-Zwanzig definition of surface tension to approximate the inhomogeneous free energy density. If desired, the homogeneous pressure and Helmholtz free energy can also be predicted by the new method, as well as binodal and spinodal densities of a two-phase fluid region. The CPP method is tested using a Lennard-Jones (LJ) fluid at vapor, liquid, two-phase, and supercritical conditions. Satisfactory agreement is found between the CPP method and an LJ equation of state. The efficiency of the CPP method is compared to that for Widom's method under the tested conditions. In particular, the new method works well for dense fluids where Widom's method starts to fail.

Self-trapping at the liquid-vapor critical point: a path-integral study

Experiments suggest that localization via self-trapping plays a dominant role in the behavior of a low-mass particle, e.g., an electron, positron, or positronium atom, in both liquids and supercritical fluids. In the latter case, the behavior is dominated by the liquid-vapor critical point. However, because of its large isothermal compressibility, the critical point is difficult to probe both experimentally and theoretically. Here we present the results of path-integral computations of the characteristics of a generic self-trapped particle at the critical point of a Lennard-Jones fluid for a positive particle-atom scattering length. We carefully investigate the dependence of the properties of both the self-trapped quantum particle and the proximal fluid on the range of the direct particle-atom interaction. To the extent that the generic particle mimics the behavior of ortho-positronium, qualitative information is provided on the pick-off decay rate. In general, compared with self-trapping at higher temperature, we find that the localized quantum defect has a much larger range of influence on the host fluid. In particular, it appears that long-range density oscillations are induced in the fluid surrounding the defect. The results also suggest that, even at the critical point, there is a minimum interaction range below which self-trapping does not occur.

Soft core thermodynamics from self-consistent hard core fluids

In an effort to generalize the self-consistent Ornstein-Zernike approximation (SCOZA)-an accurate liquid state theory that has been restricted so far to hard core systems-to arbitrary soft core systems we study a combination of SCOZA with a recently developed perturbation theory. The latter was constructed by Ben-Amotz and Stell [J. Phys. Chem. B 108, 6877 (2004)] as a reformulation of the Weeks-Chandler-Andersen [J. Chem. Phys. 54, 5237 (1971)] perturbation theory directly in terms of an arbitrary hard sphere reference system. We investigate the accuracy of the combined approach for the Lennard-Jones fluid in comparison with simulation data and pure perturbation theory predictions and determine the dependence of the thermodynamic properties and the phase behavior on the choice of the effective hard core diameter of the reference system.

Self-consistent Ornstein-Zernike approximation for molecules with soft cores

The self-consistent Ornstein-Zernike approximation (SCOZA) is an accurate liquid state theory. So far it has been tied to interactions composed of hard core repulsion and long-range attraction, whereas real molecules have soft core repulsion at short distances. In the present work, this is taken into account through the introduction of an effective hard core with a diameter that depends upon temperature only. It is found that the contribution to the configurational internal energy due to the repulsive reference fluid is of prime importance and must be included in the thermodynamic self-consistency requirement on which SCOZA is based. An approximate but accurate evaluation of this contribution relies on the virial theorem to gauge the amplitude of the pair distribution function close to the molecular surface. Finally, the SCOZA equation is transformed by which the problem is reformulated in terms of the usual SCOZA with fixed hard core reference system and temperature-dependent interaction.

Critical point estimation of the Lennard-Jones pure fluid and binary mixtures

The apparent critical point of the pure fluid and binary mixtures interacting with the Lennard-Jones potential has been calculated using Monte Carlo histogram reweighting techniques combined with either a fourth order cumulant calculation (Binder parameter) or a mixed-field study. By extrapolating these finite system size results through a finite size scaling analysis we estimate the infinite system size critical point. Excellent agreement is found between all methodologies as well as previous works, both for the pure fluid and the binary mixture studied. The combination of the proposed cumulant method with the use of finite size scaling is found to present advantages with respect to the mixed-field analysis since no matching to the Ising universal distribution is required while maintaining the same statistical efficiency. In addition, the accurate estimation of the finite critical point becomes straightforward while the scaling of density and composition is also possible and allows for the estimation of the line of critical points for a Lennard-Jones mixture.

Energy of the Quasi-Free Electron in Supercritical Krypton near the Critical Point

Field ionization measurements of high-n CH(3)I and C(2)H(5)I Rydberg states doped into krypton are presented as a function of krypton number density along the critical isotherm. These data exhibit a decrease in the krypton-induced shift of the dopant ionization energy near the critical point. This change in shift is modeled to within +/-0.2% of experiment using a theory that accounts for the polarization of krypton by the dopant ion, the polarization of krypton by the quasi-free electron that arises from field ionization of the dopant, and the zero point kinetic energy of the free electron. The overall decrease in the shift of the dopant ionization energy near the critical point of krypton, which is a factor of 2 larger than that observed in argon, is dominated by the increase in the zero point kinetic energy of the quasi-free electron.

Influence of Confinement on the Electrostatic Interaction between Charged Colloids: a (N,V,T) Monte Carlo Study within Hyperspherical Geometry

Monte Carlo simulations within closed hyperspherical geometry are used to analyze the ionic distribution around two confined charged colloids to determine the origin of the net attraction recently reported in the literature. A scaling procedure is used to compare our numerical results obtained with small ideal colloids with the conclusion of the measurements performed with large silica colloids. Although no electrostatic attraction is detected under confinement, our simulations exhibit a significant reduction of the electrostatic repulsion between charged colloids confined between two weakly charged walls. After rescaling to reproduce the electrostatic repulsion between large confined colloids, our numerical results are qualitatively consistent with the reported attraction because we reasonably expect a reduction of the electrostatic force between such confined colloids below the order of magnitude of their van der Waals attraction.

Monte Carlo simulations in generalized isobaric-isothermal ensembles

We present three generalized isobaric-isothermal ensemble Monte Carlo algorithms, which we refer to as the multibaric-multithermal, multibaric-isothermal, and isobaric-multithermal algorithms. These Monte Carlo simulations perform random walks widely in volume space and/or in potential energy space. From only one simulation run, one can calculate isobaric-isothermal-ensemble averages in wide ranges of pressure and temperature. We demonstrate the effectiveness of these algorithms by applying them to the Lennard-Jones 12-6 potential system with 500 particles.

A finite-size scaling study of a model of globular proteins

Grand canonical Monte Carlo simulations are used to explore the metastable fluid-fluid coexistence curve of the modified Lennard-Jones model of globular proteins of ten Wolde and Frenkel [Science, 277, 1975 (1997)]. Using both mixed-field finite-size scaling and histogram-reweighting methods, the joint distribution of density and energy fluctuations is analyzed at coexistence to accurately determine the critical-point parameters. The subcritical coexistence region is explored using the recently developed hyper parallel tempering Monte Carlo simulation method along with histogram reweighting to obtain the density distributions. The phase diagram for the metastable fluid-fluid coexistence curve is calculated in close proximity to the critical point, a region previously unattained by simulations.

Statistical substantiation of the van der Waals theory of inhomogeneous fluids

Computer experiments on simulation of thermodynamic properties and structural characteristics of a Lennard-Jones fluid in one- and two-phase models have been performed for the purpose of checking the base concepts of the van der Waals theory. Calculations have been performed by the method of molecular dynamics at cutoff radii of the intermolecular potential r(c,1)=2.6sigma and r(c,2)=6.78sigma. The phase equilibrium parameters, surface tension, and density distribution have been determined in a two-phase model with a flat liquid-vapor interface. The strong dependence of these properties on the value of r(c) is shown. The p,rho,T properties and correlation functions have been calculated in a homogeneous model for a stable and a metastable fluid. An equation of state for a Lennard-Jones fluid describing stable, metastable, and labile regions has been built. It is shown that at T > or =1.1 the properties of a flat interface within the computer experimental error can be described by the van der Waals square-gradient theory with an influence parameter kappa independent of the density. Taking into account the density dependence of kappa through the second moment of the direct correlation function will deteriorate the agreement of the theory with data of computer simulation. The contribution of terms of a higher order than (nablarho)(2) to the Helmholtz free energy of an inhomogeneous system has been considered. It is shown that taking into account terms proportional to (nablarho)(4) leaves no way of obtaining agreement between the theory and simulation data, while taking into consideration of terms proportional to (nablarho)(6) makes it possible to describe with adequate accuracy all the properties of a flat interface in the temperature range from the triple to the critical point.

Temperature and density extrapolations in canonical ensemble monte carlo simulations

We show how to use the multiple histogram method to combine canonical ensemble Monte Carlo simulations made at different temperatures and densities. The method can be applied to study systems of particles with arbitrary interaction potential and to compute the thermodynamic properties over a range of temperatures and densities. The calculation of the Helmholtz free energy relative to some thermodynamic reference state enables us to study phase coexistence properties. We test the method on the Lennard-Jones fluids for which many results are available.