Electrostatic and induction effects in the solubility of water in alkanes

Abstract

Experiments show that at 298 K and 1 atm pressure the transfer free energy, µ ex , of water from its vapor to liquid normal alkanes CnH2n+2 (n = 5 . . . 12) is negative. Earlier it was found that with the united-atom TraPPE model for alkanes and the SPC/E model for water, one had to artificially enhance the attractive alkane-water cross interaction to capture this behavior. Here we revisit the calculation of µ ex using the polarizable AMOEBA and the non-polarizable Charmm General (CGenFF) forcefields. We test both the AMOEBA03 and AMOEBA14 water models; the former has been validated with the AMOEBA alkane model while the latter is a revision of AMOEBA03 to better describe liquid water. We calculate µ ex using the test particle method. With CGenFF, µ ex is positive and the error relative to experiments is about 1.5 kBT . With AMOEBA, µ ex is negative and deviations relative to experiments are between 0.25 kBT (AMOEBA14) and 0.5 kBT (AMOEBA03). Quantum chemical calculations in a continuum solvent suggest that zero point effects may account for some of the deviation. Forcefield limitations notwithstanding, electrostatic and induction effects, commonly ignored in considerations of water-alkane interactions, appear to be decisive in the solubility of water in alkanes.

Monte Carlo Simulations of Fluid Phase Equilibria and Interfacial Properties for Water/Alkane Mixtures: An Assessment of Nonpolarizable Water Models and of Departures from the Lorentz–Berthelot Combining Rules

“…In contrast, the potential energy surfaces of the water–methane interactions described by the TIP4P/TraPPE–UA and SPC/E-1.7/TraPPE–UA combinations depend only on the oxygen–carbon distance (i.e., is invariant to changes in orientation) and yield binding energies of only 0.9 and 1.5 kJ/mol. This example illustrates the shortcomings of effective, nonpolarizable potentials (see also refs , , , , and ) but also indicates that these effective potentials should not strive to reproduce dimer energies.…”

“…Clearly, the Lorentz− Berthelot combining rules are far from optimal when considering unlike interactions for atoms that differ greatly in the number of electrons (e.g., helium/xenon mixtures 35 ) but may also not be the most suitable choice for water−alkane mixtures due to the large difference in the well depths. 72 There is no doubt that more complex force fields (explicithydrogen representation of the alkanes, Gaussian partial charges to reflect the charge distribution of water 73 and alkanes, explicit accounting for many-body induction, 30,74,75 coupling between fluctuating partial charges and dispersive terms, 76 allowance for charge transfer, 76 and inclusion of nuclear quantum effects 77,78 ) can yield a more accurate description of water/alkane mixtures, but each of these complexities comes with a large increase in the computational expense that are multiplicative. Akin-Ojo and Szalewicz 79 have used high-level electronic structure theory to determine the potential energy surface for the water−methane dimer and obtained a rugged surface with two distinct minima with binding energies of 4.3 and 3.0 kJ/mol for configurations with methane as the hydrogen-bond acceptor and as the donor, respectively.…”

“…From comparison to experimental solubility data, Ballal et al 28,29 concluded that there exists an anomalously strong interaction between alkanes and water in the alkane-rich phase that could be reproduced by increasing the unlike LJ well depth by about 70% and that arises primarily from induction interactions. 30 However, such a calculation of the solubility does not reflect the intermolecular interactions of the chosen water model in the neat water phase, and it also ignores any departure from infinite-dilution behavior in the alkane-rich phase and change in water's fugacity due to alkanes partitioning into the water-rich phase. Particularly, the last issue is disconcerting because a significantly increased strength of the unlike interactions may lead to a dramatic increase in the alkane solubility in the water-rich phase.…”

“…Using the calculated Henry’s law constant, the water solubility in the alkane-rich phase was then estimated using the fugacity of neat water from an accurate EOS. From comparison to experimental solubility data, Ballal et al , concluded that there exists an anomalously strong interaction between alkanes and water in the alkane-rich phase that could be reproduced by increasing the unlike LJ well depth by about 70% and that arises primarily from induction interactions . However, such a calculation of the solubility does not reflect the intermolecular interactions of the chosen water model in the neat water phase, and it also ignores any departure from infinite-dilution behavior in the alkane-rich phase and change in water’s fugacity due to alkanes partitioning into the water-rich phase.…”

Modeling Binary Mixtures of Water + Light Hydrocarbon Using the Perturbed-Chain Statistical Associating Fluid Theory with Induced Association: Improvement in Describing All Equilibrium Phases

“…However, from a PC-SAFT model and molecular dynamics simulations, the effective water Lennard–Jones energy was found much larger than that commonly used in simulation water models for describing the solubility of water in alkanes. This prompted the same research group to perform ab initio calculations, which revealed that nonpolar molecules, such as n -alkanes can be electrostatically induced by the presence of water molecules, consistent with other independent works. , They even conclude that the electrostatic and induction effects must be included to predict correctly the water content in alkanes …”