Molecular dynamics simulations of metal ions in solution

Metal ions solvated in aqueous or non-aqueous solvents and in mixtures of solvents play an important role in electro-chemical applications and solvent separation technologies. Solvated ions also appear in high concentrations in living organisms, where their presence or absence can fundamentally alter the functions of life. In fact, the structure and dynamics of the solvation shells have a large impact on any chemical reaction of metal ions in solution.

Some molecular-level questions that we want to answer in this project are for example: How far out in the solution do the solvent molecules feel the strong influence from the positive cation? And, conversely, how much are the ions affected by the surrounding solvent molecules? In a mixed solvent, which type of solvent molecules will win the competition for the places around the ion? We are trying to find the answers to questions like these using theoretical simulations of different kinds.

Below we will discuss two types of simulations: 1) The preferential solvation of Li+, Na+, and Ag+ in water and acetonitrile mixtures, and 2) Some CPMD simulations of ions in water.

1) The preferential solvation of Lithium, Sodium, and Silver cations in water and acetonitrile mixtures

Experimentally it is known that the monovalent coinage metal cations are unusual in that they are strongly preferentially solvated by acteonitrile in water/acetonitrile mixtures, in sharp contrast to most other metal ions.

Here, these systems were studied by developing many-body forcefields by fitting to data from high level ab initio cluster calculations (MP2, triple-zeta basis). These forcefields were subsequentually used in classical MD-simulations of a single cation and 512 solvent molecules. The solvent molecules are described by a polarizable forcefield also fitted to ab initio cluster calculations. In the description of the interaction of the silver cation with the solvent molecules it was necessary to include four-body terms in the forcefield.

The pictures below illustrates the large differences in the bonding of sodium and silver ions to water and acetonitrile. This large difference is due to covalent binding between the silver ion and the solvent molecules, and necessitates the inclusion of explicit higher-than-pair potential terms.

Successive binding energies of Na(solvent)+
Successive binding energies of Ag(solvent)+

Analyses of the configurations from the Molecular Dynamics simulations (which were on the order of 1 ns in length) in 50/50 mixtures of water and acetonitrile reveals that most of the neighbours to the Sodium and Lithium ions are water molecules, whereas most of the neighbours of the Silver ions are acetonitrile molecules.

The animated images below show more or less representative snapshots taken from the Molecular Dynamics simulations of the 50/50 simulations for Lithium and Silver. In both cases there is only a single kind of solvent molecule coordinating the ion. Around the Lithium ion there are four water molecules, and around the Silver ion there are four acetonitrile molecules. During the simulations both the number of solvent molecules and the kind of solvent molecules surrounding the ion fluctuate.

Nearest neighbours around a Lithium ion in a 50/50 mixture of water and acetonitrile.
Nearest neighbours around a Silver ion in a 50/50 mixture of water and acetonitrile.

The situation around the Sodium ion is similar to the one around the Lithium ion.Simulations of mixtures take a fairly long time to reach equilibrium. In these simulations, the starting configurations were always selected randomly. To ensure that proper statistics was obtained at the end of each simulation, the distribution of the number of solvent molecules was computed separately for the first and second half of the simulation, and the results were compared to ensure that they were essentially the same.

Finally, the preferential solvation of the ions in three different mixture compositions were studied: 10, 50, and 90% water. The images below show how common it is for an ion to be surrounded by a specific number of solvent molecules.

Preferential solvation of lithium ions in different mixture compositions of water and acetonitrile. The large red bar shows the most common configuration. From left to right: In 10% water, it is most commonly found that the lithium ion is surrounded by three water molecules and a single acetonitrile molecule. In both the 50% and 90% compositions, the lithium ion is most commonly surrounded by four water molecules.
The preferential solvation of sodium ions follows the same trend as the lithium ion, although, as the sodium ion is larger than the lithium ion, there is room for more solvent molecules. The distribution is also broader. From left to right: In 10% water, it is most commonly found that the sodium ion is surrounded by four water molecules and three acetonitrile molecules. In the 50% mixture, this changes to five water molecules and a single acetonitrile molecule. In the mixture containing 90% water, the sodium ion is most commonly surrounded by six water molecules.
Finally, for the silver ion, in all mixtures the most common configuration is the one with four acetonitrile molecules directly coordinating the silver ion, with no water molecules in the first solvation shell.

Conclusions: It is the difference in the ion-solvent bonding strengths which is responsible for the difference in solvation properties between the silver ion and the lithium and sodium ions. It is very interesting to note that in the gas phase, all three ions bind more strongly to acetonitrile, pointing both to the importance of more than the closest neighbours, as well as finite temperature effects (essentially entropy). Although, the "special" bonding characteristics for the silver ion is present for both water and acetonitrile, the difference in the ion-acetonitrile and ion-water bonding strengths of actetonitrile compared to water is substantially larger for the silver ion than for the lithium and sodium ions.

2.Some CPMD simulations of ions in water.

We have performed Car-Parrinello molecular dynamics (CPMD) simulations of some multivalent transition metal ions in D2O water at 300 K, namely Al3+(aq), Fe3+(aq), and Cu2+(aq).

OD vibrations and hydration structure in an Al3+(aq) solution from CPMD simulation

S. Amira, D. Spångberg, and K. Hermansson, J. Chem. Phys. 124, 104501-1 - 104501-13 (2006).

Car-Parrinello Molecular Dynamics simulaton of Fe3+(aq)
S. Amira et al., J. Phys. Chem. B 109, 14235-14242 (2005).

Distorted fivefold coordination of Cu2+(aq) from a CPMD simulation
S. Amira, D. Spångberg, and K. Hermansson, PCCP 7, 2874-2880 (2005).

For Cu2+(aq), there is an animated debate in the literature concerning the correct coordination number and coordination figure. Our CPMD (BLYP, US-PP) simulations five five-fold coordination with four equidistant equatorial water molecules at 2.00 Å and one axial water molecule at 2.45 Å from the cation. A "hole" without water molecules is found on the opposite side of the axial water molecule.

For Fe3+(aq), CPMD simulations for a small MD box (1 ion + 32 waters) and classical force-field MD simulations with the same small box as well as a bigger box (1 ion + 512 waters) were performed. The small box size is found to affect, for example, the first-shell OD vibrations and the second-shell water orientations. However, the water angle opening in the first shell in the CPMD simulations appears to be a robust result and in agreement with experiment (the classical FF simulations give an angle closing of some 10 deg.).

For Al3+(aq), we find that both our calculated OD vibrational frequencies and the gas-to-liquid frequency shifts have rather large systematic errors, which arise from the BLYP functional (and the US-PP), our choice of the "fictitious electron mass", box-size effects and the neglect of anharmonicity in the vibrational analysis.