Aqueous Solvation of Hg(OH) - American Chemical Society

Article pubs.acs.org/JPCA

Aqueous Solvation of Hg(OH)2: Energetic and Dynamical Density Functional Theory Studies of the Hg(OH)2−(H2O)n (n = 1−24) Structures J. I. Amaro-Estrada,† L. Maron,‡ and A. Ramírez-Solís*,‡,§ †

Departamento de Física, Facultad de Ciencias, Universidad Autónoma del Estado de Morelos, Av. Universidad 1001, Cuernavaca, Morelos, 62209, México ‡ INSA Laboratoire de Physicochimie de Nano-Objets, Université de Toulouse, 135 Avenue de Rangueil, F31077 Toulouse, France S Supporting Information *

ABSTRACT: A systematic study of the hydration of Hg(OH)2 by the stepwise solvation approach is reported. The optimized structures, solvation energies, and incremental free energies of 1−24 water molecules interacting with the solute have been computed at the B3PW91 level using 6-31G(d,p) basis sets for the O and H atoms. The mercury atom was treated with the Stuttgart−Köln relativistic core potential in combination with an extended optimized valence basis set. One to three direct Hg−water interactions appear along the solvation process. The first solvation shell is fully formed with 24 water molecules. A stable pentacoordinated Hg trigonal bipyramid structure appears for n > 15. Density functional theory (DFT) Born− Oppenheimer molecular dynamics simulations showed the thermal stability of the Hg(OH)2−(H2O)24 structure at room temperature and the persistence of the trigonal bipyramid coordination around Hg. The Gibbs free energy for the first solvation shell is significantly larger for the fully solvated Hg(OH)2 than the one previously obtained for the HgCl2 case, due to σ-acceptor and π-donor properties involving the hydroxyl groups of the solute. This suggests that the transmembrane passage of Hg(OH)2 into the cell via simple diffusion is less favorable compared to the case when the metal is coordinated with two Cl groups.

I. INTRODUCTION The growing anthropogenic activity, mainly related to industrial manufacturing processes, has been recognized at the root of increasing concentrations of mercury levels in soils, plants, and aquatic environments.1 Due to the adverse effects of Hg on biota, it is crucial to find relevant information on the fate, transport, and transformation of mercury species in aquatic and terrestrial environments. An example of these efforts can be found in the reports concerning the studies of the Arctic environment.2 Approximately 90% of Hg in the atmosphere has been found as elemental Hg(0), although the ratio of free monatomic vapor or these adherent to airborne particulates varies continually.3,4 Once in the atmosphere, Hg(0) is oxidized to Hg2+ forms by different mechanisms, and the ion can reach the Earth’s surface by dry and wet deposition. Furthermore, the presence of radicals like Cl− and OH− in aqueous environments allows the formation of inorganic complexes, the most abundant being HgCl2, HgOHCl, and Hg(OH)2;5,6 note that all of them are stable neutral molecules. However, several uncontrolled environmental factors such as concentration of water and reactive species (e.g., Cl2, HF, OH−) in ambient air temperature, UV radiation, atmospheric pressure, and others, make it extremely difficult to determine the interconversion and degradation routes of these species. © 2013 American Chemical Society

In marine ecosystems, an important incursion via of mercury complexes into the food chain is through algae and bacteria, which are then ingested by higher organisms (fish, marine mammals, seabirds, etc.) leading to a process of mercury accumulation and biomagnification. The bioavailability of deposited Hg is followed by an important biochemical step in which one or several Hg-containing species cross a cell membrane by passive or active diffusion. The intrinsic selective properties of the cell membrane suggest that passage for Hgcontaining species is dependent on the specific type of ligands bonded to the Hg atom. In this direction we note that Gutknecht7 studied the permeability of several mercury complexes with a variety of inorganic and organic ligands (HgCl2, HgCl3−, HgCl42−, Hg(OH)2, and HgClOH) through laboratory-synthesized lipid bilayer membranes. The important result found by Gutknecht is that lipid membranes are highly permeable to HgCl2, showing a permeability 20 times larger than the permeability to water and more than a million times more permeable than to Na+, K+, and Cl−. Similar results were obtained by Barkay et al.,8 a significant fact for the present Received: June 4, 2013 Revised: August 16, 2013 Published: August 22, 2013 9069

dx.doi.org/10.1021/jp405500f | J. Phys. Chem. A 2013, 117, 9069−9075

The Journal of Physical Chemistry A

Article

were performed at the DFT level with the hybrid B3PW91 functional without any symmetry restrictions, and the nature of the stationary points (all minima) was verified with analytical frequency calculations for all n. We followed the same stepwise solvation scheme as the one used to study the HgCl2−(H2O)n case.9 A very careful global search was performed to explore all the possible water arrangements in the Hg(OH) 2 −(H 2 O) n structures. In particular, we performed many optimizations for the three- to six-water (n = 3−6) molecule systems starting with radically different water positions around the previously optimized (n − 1) water complex; therefore we report the global minima for n ≤ 6. Since the number of degrees of freedom dramatically increases with the number of solvating water molecules, we stress that we are reporting optimized structures for n > 6 without the knowledge of their local or global minima character. However, for larger n, we have made many optimizations starting from structures where the position of the hydrogen bonds are modified with respect to the presently reported geometries, and the resulting optimized energies were only slightly higher (1−3 kcal/mol) than those for which we provide results. For this reason we believe the relative microsolvation free energies reported here are upper bounds to the true values for large n. The electronic density at the DFT level of the optimized structures has been analyzed using the natural population analysis (NPA) and the natural bond orbital (NBO) schemes15 using the same relativistic effective core potential and the basis set. The DFT and NPA calculations were performed with the Gaussian03 program.16 As usually done, the Gibbs free energies were calculated using the harmonic approximation, and the Wiberg bond indices have been calculated to further analyze the nature of the mercury−water interactions. To probe the thermal stability of the Hg(OH)2−(H2O)24 optimized structure, a Born−Oppenheimer DFT (B3PW91) molecular dynamics simulation (BO−DFT−MD) was carried out with the Geraldyn2.1 code,17 which uses the velocity-Verlet integration scheme18 and has been coupled to the electronic structure modules of the Gaussian03 code. Note that the trajectory was simulated starting from the optimized equilibrium structure at 0 K without any preferred velocity vector other than the thermal energy. The simulation time was 8 ps with a time step of 0.5 fs. A Nosé−Hoover chain of thermostats19,20 was used to control the temperature at 300 K. The electronic structure Kohn−Sham calculations involve 804 molecular orbitals and 278 electrons, so that the total MD simulation took 75 CPU days on eight [email protected] GHz running the Linux versions of Geraldyn2.1-G03.

study since these authors also reported high permeability rates for metal−OH complexes such as Hg(OH)2 and HgClOH. Recently, using electronic structure methods we have studied the aqueous solvation process of HgCl29,10 and HgClOH11 species. In particular, we are interested in some specific issues such as how many water molecules belong to their first solvation shells and what are the interaction energies with their aqueous environments. Otherwise, a crucial aspect related to the bioavailability of these complexes is whether these Hg neutral species can be considered as water-dressed molecules or not during the transmembrane transport process making them available to the cell interior. In this regard we addressed the solvation of HgCl2 through a density functional theory (DFT) study using stepwise cluster solvation including up to 24 water molecules,9 and through Monte Carlo simulations (MC) of the aqueous solution10 using very refined, ab initio derived, flexible, and polarizable interaction potentials. As an example of how detailed information at the molecular level can be of help to understand the main processes at work in these complex metal−membrane interactions, we recall that a refined study on the solvation mechanism of As(OH)3 has allowed us to better understand how this neutral toxic species can enter the cell via transmembranal aquaglyceroporines.12 We have found that the interaction of arsenious acid with the extracellular portion of aquaglyceroporines might be facilitated due to its singular amphipathic solvation pattern. Therefore, accurate information at the molecular level concerning the solvation pattern of Hg-containing molecules in aqueous environments can be used to provide new insights into the type of transmembrane cellular uptake of these toxic species. In this respect, finite temperature molecular dynamics studies of HgClOH in the gas phase have been recently published.13 In this work we shall address the structural and energetic features related to the hydration of Hg(OH)2 through a cluster aqueous microsolvation approach. In particular we are interested in comparing the specific metal−ligand coordination and the evolution of the solvation energies with hydration degree. Finally, to more realistically address the system in the gas phase, we also take into account the finite temperature effects through Born−Oppenheimer molecular dynamics simulations. Although our final goal is to study the liquid phase solvation of HgOH2, we emphasize that an adequate description of the gas phase microsolvation is important as a first step in the development of accurate interatomic interaction potentials needed for Monte Carlo simulations of the condensed phase, as we have previously done for HgCl2.10

II. METHOD AND COMPUTATIONAL DETAILS Since we are interested in a systematic study of large Hg(OH)2−(H2O)n (with n > 20) clusters, for the oxygen and hydrogen atoms we have employed the 6-31G(d,p) doubleζ basis sets, as we did for the HgCl2−(H2O)n case.9 Again, the mercury atom is treated with the Stuttgart−Köln relativistic effective core potential14 in combination with its adapted valence basis set; we have recently improved this basis with a set of s, p, and d diffuse functions (for details on the exponents and coefficients see ref 11). To determine the exchangecorrelation functional to be used we have performed a careful calibration using as reference the MP2/aug-cc-pVTZ energetic and structural results for HgCl2 with one to three water molecules.9 The hybrid B3PW91 functional turned out to be the best among 10 other GGA, hybrid, and meta exchangecorrelation functionals. Therefore the geometry optimizations

III. RESULTS AND DISCUSSION A. Stepwise Solvation. The optimized geometry of the isolated Hg(OH)2 molecule (see Figure 1) leads to a quasilinear O−Hg−O (175.8°) angle and Hg−O distances of 1.98 Å. Based on natural population analysis, the atomic charges are +0.98 for mercury and −0.98 for both lateral oxygen atoms. As the number of solvating water molecules increases, the Hg−O(H) distances grow up to 2.09 Å and the O−Hg−O angle shows a large change from quasi-linearity at 166° for n = 12. Table 1 shows the main optimized geometrical parameters for the Hg(OH)2−(H2O)n clusters, with particular emphasis on the variation of the distances and the angle between Hg and lateral O(H) moieties of the solute, as well as the distances of the water oxygen atoms (Ow) directly coordinated to Hg. 9070

dx.doi.org/10.1021/jp405500f | J. Phys. Chem. A 2013, 117, 9069−9075

The Journal of Physical Chemistry A

Article

Table 2. Gibbs Microsolvation Free Energies, Incremental Microsolvation Free Energies (kcal/mol), and Number of Direct Hg−O(water) Interactions and Hydrogen Bonds vs the Number of Water Molecules in Each Clustera water molecules

direct Hg−Ow interactions

hydrogen bonds

ΔG°solv

incremental energy

1 2 3 4 5 6 7 8 12 15 16 24

0 1 1 2 2 2 2 2 2 2 3 3

1 2 4 5 7 8 10 12 19 24 27 41

−0.30 −4.42 −6.53 −11.04 −13.63 −12.43 −21.41 −24.71 −32.82 −35.45 −38.68 −60.93

0.30 −4.12 −2.11 −4.51 −2.59 +1.20 −8.98 −3.3 −8.11 −2.63 −3.23 −22.25

Figure 1. Optimized geometry of bare Hg(OH)2 and solvated with a water molecule.

When a single water molecule is added to Hg(OH)2, the optimized structure is qualitatively different to the one computed by Castro et al. for the HgCl2 molecule9 since no direct Hg−Ow interaction is found; however, it is similar to one obtained for HgClOH11 at the MP2 level. An attractive interaction of the water hydrogen atom with the electronegative part of the solute (one of the oxygen atoms) arises, producing a slight stretching in the Hg−O bond (2.01 Å). The first direct interaction between Hg and a water oxygen (2.59 Å) was found for n = 2, explaining the fact that the solvation free energy when n = 1 (−0.3 kcal/mol, see Table 2) is smaller than the value of −2.7 kcal/mol we reported for the HgCl 2−H2 O case.9 The nature of the water−mercury interaction was investigated by computing the Wiberg bond indexes. Interestingly, the Wiberg index value (0.1463) for the n = 2 case is larger compared with the 0.0927 value when n = 1 in line with a stronger orbital interaction. For n = 2 the O−Hg−O angle reaches its maximum value (179.1°), but a decrease of 5.3° appears for n = 3 due to the symmetric hydrogen bonds that appear on both sides of the solute (see Figure 2, right). Although as expected, the microsolvation free energies increase monotonically, the incremental microsolvation energies show an interesting sawtooth pattern as the number of water molecule increases. We have found that this pattern is related to two factors: the Hg−Ow interactions and the particular way in which the hydrogen bonding between water molecules is established for each n. Particularly, the absolute value of the incremental energy for the trimer (2.11 kcal/mol for n = 3) is lower compared with the value obtained for the dimer (4.12 kcal/mol for n = 2), despite of having two new hydrogen bonds in the network. This suggests a strong

a ΔG°solv = [G°(system with n water molecules) − nG°(H2O) − G°(Hg(OH)2)].

Figure 2. Optimized geometries of Hg(OH)2 solvated by two (left) and three (right) water molecules. Note the dramatic change of orientation of the right-side hydrogen atom of the solute.

contribution from the direct mercury−water interaction to the solvation energy for n = 2. This phenomenon also appears when n changes from 4 to 5 and when n changes from 7 to 8. A remarkable case is when n increases from 5 to 6, where the change even shows a sign reversal. A careful examination of the global minima for these structures reveals the reason of this apparent anomaly in the incremental microsolvation energy trend. We found that two structural features explain this fact: the water−water distances are larger in the hydrogen bond

Table 1. Main Optimized Geometrical Parameters of Gaseous and Solvated Hg(OH)2 Systemsa no. water molecules

Hg−O(H)

Hg−O(H)

O(H)−Hg−O(H) angle

Hg−Ow d1

Hg−Ow d2

Hg−Ow d3

0 1 2 3 4 5 6 8 12 15 16 24 24b

1.98 2.01 2.02 2.02 2.02 2.04 2.04 2.04 2.06 2.07 2.09 2.04 2.03

1.98 1.98 1.99 2.02 2.02 2.03 2.02 2.04 2.05 2.03 2.09 2.09 2.04

175.8 177.7 179.1 173.8 174.0 178.2 178.7 172.7 166.4 177.8 175.7 177.3 171.7

2.82 2.59 2.52 2.63 2.68 2.65 2.59 2.66 2.58 2.63 2.75 2.97

2.97 2.70 2.75 3.03 2.72 2.71 2.66 2.79 3.02

3.19 3.00 3.22 2.68 2.80 3.06

a

Water oxygen atoms (Ow) are those directly linked to Hg. Distances in Å and angle in degrees. bAverage structure obtained with the BO−DFT− MD simulation at 300 K. 9071

dx.doi.org/10.1021/jp405500f | J. Phys. Chem. A 2013, 117, 9069−9075

The Journal of Physical Chemistry A

Article

network for n = 6, and most importantly, a rather large reorientation of one of the solute hydrogen atoms (i.e., a significantly modified O−Hg−O-H dihedral angle with respect to the solute geometry when n = 5) contributes to a decrease in the incremental free microsolvation energy when n goes from 5 to 6. A second Hg−O(water) interaction is observed with the addition of a fourth water molecule (see Figure 3), although Figure 5. Optimized geometries of Hg(OH)2 solvated by 12 (left) and 16 (right) water molecules.

Figure 3. Optimized geometries of Hg(OH)2 solvated by four (left) and five (right) water molecules.

this new direct interaction is more evident for n = 5 as shown by the Wiberg indexes of 0.1381 and 0.1313. The variation of the geometrical parameters in the solute seems to play an important role on the evolution of the water hydrogen bond network when the size of the cluster is increased. When n changes from 4 to 5, a slight increase of 4.2° in the O−Hg−O angle allows the entrance of an additional water molecule into the cluster (see Figure 3, right). Each water molecule interacts with other two neighboring water molecules, leading to an important increase of the ΔG°solv value (−13.63 kcal/mol) as a consequence of the cumulative interactions between them in the solvating network. Figure 4 presents the optimized geometries of the solute solvated by six and eight water molecules. In the former case,

Figure 6. Optimized geometry of Hg(OH)2 solvated by 24 water molecules at 0 K. We highlight the three water oxygens in the equatorial plane of the trigonal bipyramid (blue, green, and black atoms), and the solute oxygen atoms are shown in gold.

for n = 24 the direct coordination Hg−Ow optimized distances are 2.75, 2.79, and 2.80 Å (optimized atomic coordinates can be found in the Supporting Information). Unlike the HgCl2− (H2O)24 case, where both the equatorial and the apical coordination patterns were found as stable structures,9 the fully solvated Hg(OH)2 presents only the apical structure, where the solute oxygen atoms are in the apical positions of the triangular bipyramid pattern (O−Hg−O angle of 177.3°). We explored different hydration configurations for n = 24 to find a solute-equatorial structure, but in all cases, this search lead to axial-solute stable structures, all of these very close to our reported geometry (which has the lowest energy). The microsolvation free energy for n = 24 is found to be −60.93 kcal·mol−1. In Figure 7 we summarize the results of the population analysis, showing the evolution of natural atomic charges for the Hg and O atoms of the solute as function of the number of solvating water molecules. Note that mercury charge oscillates between 0.98 and 1.2, while in the case of the solute oxygen atoms the atomic charges show very similar behavior with increasing solvation (−1.06 and −1.08 for large n). More details can be found in the Supporting Information. Finally, as part of a complete analysis of the first solvation shell, we carried out a Born−Oppenheimer molecular dynamics simulation at the B3PW91 level and 300 K starting from the Hg(OH)2−(H2O)24 optimized structure (see Figure 6). Two important dynamical results must be highlighted. The first one is that the solvation shell remains stable throughout the 8 ps of the simulation despite the non-negligible thermal effects at room temperature. Second, while a clathrate-like structure is

Figure 4. Optimized geometries of Hg(OH)2 solvated by six (left) and eight (right) water molecules.

two Hg−O(water) direct interactions (2.65, 2.75 Å) and eight hydrogen bonds appear in the network while, in the latter case, an additional Hg−O(water) weak interaction is observed at 3.19 Å. As a consequence, the O−Hg−O angle decreases (172.7°), and the free microsolvation energy is now −24.71 kcal/mol. This clearly highlights the energetic importance of the direct Hg−water interactions over the formation of hydrogen bonds in the solvating network. From n = 16 onward the third direct interaction Hg−Ow appears, which is again supported by the Wiberg indices of 0.1179, 0.1204, and 0.1097 obtained with the NBO scheme. This is in line with orbitaldriven interactions. We have found that from n = 16 up to the complete the first solvation shell (n = 24), the coordination geometry of mercury is a trigonal bipyramid, as shown in Figures 5 and 6. Note that 9072

dx.doi.org/10.1021/jp405500f | J. Phys. Chem. A 2013, 117, 9069−9075

The Journal of Physical Chemistry A

Article

Figure 9. Hg−Ow distance evolution for the three equatorial water molecules in the trigonal bipyramid solvation pattern. The blue, green, and black curves correspond to the atoms highlighted in Figures 6 and 8. Figure 7. Calculated NPA charges of Hg (violet) and O (red and green) atoms of the solute in gaseous and solvated Hg(OH)2 systems. Average Hg charge (orange).

strong enough to keep the trigonal bipyramid motif throughout the simulation. Figure 10 shows the temporal evolution and the average value of the O−Hg−O angle of the solute throughout the

obtained at the end of the simulation (see Figure 8), the trigonal bypiramid pattern remains with the three equatorial waters oscillating around their equilibrium positions coordinated to Hg.

Figure 10. O−Hg−O angle evolution (red) and the average value (dashed line) at 300 K.

simulation. It can be seen that large amplitude oscillations (>17°) occur every 1.0−1.5 ps, while small amplitude oscillations (