American Mineralogist, Volume 84, pages 877–883, 1999
Theoretical studies on the formation of mercury complexes in solution and the dissolution
and reactions of cinnabar
J.A. TOSSELL*
Department of Chemistry and Biochemistry, University of Maryland, College Park, Maryland 20742, U.S.A.
ABSTRACT
Expanding upon our previous studies of the properties of Au complexes, we present calculations
for several Hg2+ species in aqueous solution and for molecular models for cinnabar. Hydration effects
are treated with a combination of “supermolecule” calculations containing several explicit water
molecules and polarizable continuum calculations. We focus upon the following problems: (1) calculation of the stabilities of HgL2, L = F–, Cl– , OH–, SH–, and CN– and HgCl2–n
n n = 1–4; (2) development
of a molecular model for cinnabar of the form Hg3S2(SH)2; and (3) dissolution or adsorption reactions using this cinnabar model. The absolute and relative formation enthalpies of the HgL2 species
can be satisfactorily reproduced at the Hartree-Fock plus Moller-Plesset second order correlation
correction level using relativistic effective core potential basis sets if the hydration of neutral HgL 2 is
explicitly taken into account. Evaluating the energetics for the series of complexes HgCl2–n
n is more
difficult, because great accuracy is needed in the large hydration energies and some of the species are
highly nonspherical. The Hg3S2(SH)2 species shows an equilibrium structure very much like that in
cinnabar. The relative energetics for dissolution of cinnabar by H2O, H2S, SH–, and SH – + elemental
S are correctly reproduced using this model molecule. Calculations on Hg3S2ClI provide a model for
understanding the adsorption of I– ions on cinnabar surfaces in the presence of Cl–.
INTRODUCTION
The speciation of metal cations in aqueous solution strongly
influences mineral dissolution and precipitation processes
(Drever 1997) and the bioavailability of the metal (Mason et
al. 1996). Partly because of the significance of Hg as an environmental pollutant, its speciation in natural waters is of great
interest. The solubilities of Hg containing minerals in various
solutions and the concentrations of Hg2+ ions in equilibrium
with various ligands have been studied experimentally to deduce Hg speciation. These studies generally used methods that
determine the total concentrations of Hg containing species in
solution or the concentration of the Hg2+ hydrated cation. Such
studies have serious limitations: (1) the speciation models are
often ambiguous, with different researchers obtaining apparently different sets of species and formation constants, and (2)
the participation of chemical components whose activities or
concentrations cannot be varied experimentally are indeterminable, e.g., the participation of water in such species cannot be
determined in aqueous solution. Spectral studies could, in principle, determine which species were present, as well as their
concentrations if relative spectral intensities for the different
species could be assessed. However, in many cases the species
concentrations are so low that no usable spectra can be obtained. Although Hg speciation has been studied for years
(Schwarzenbach and Widmer 1963; Barnes et al. 1967; Shikina
et al. 1981) new results continue to emerge. For example,
Paquette and Helz (1995, 1998) established that cinnabar has
a higher solubility in solutions containing both elemental sul-
*E-mail: tossell@chem.umd.edu
0003-004X/99/0506–0877$05.00
fur and bisulfide than in those containing bisulfide alone and
identified a new Hg polysulfide species as the cause of this
enhanced solubility.
Recently, we have matched spectral properties calculated
quantum mechanically for various As, Sb, and Au species
(Tossell 1995, 1996a, 1997; Helz et al. 1995) with experimental spectral data to determine the predominant species present.
We have also calculated reaction energetics for these species,
but the low accuracy of our energetics prevented their use as
the main criterion to determine speciation. Rather the energetic calculations were offered as further confirmation of the
assignments made on spectral grounds. For example, for the
As hydroxides Tossell (1997) identified a trimeric species
As3O3(OH)3 existing in concentrated solutions primarily on the
basis of its Raman spectrum, but noted that its high calculated
thermodynamic stability provided additional, weaker evidence
for its existence. Quantum mechanical methods were also used
to model the structures, energetics, and spectral properties of
certain “molecular” minerals, e.g., realgar, As4S4 (Tossell
1996b). For strongly covalent sulfides with small coordination
numbers, such molecular approachs yielded accurate results
for structures, vibrational spectra, and energetics.
This paper focuses more upon the energetics of reactions
in solution, to determine how well we can distinguish the stabilities of different complexes and the energetics for various
dissolution reactions. Structural and spectral results are presented which may prove useful in characterizing the complexes.
Two different types of reactions are considered: (1) complex
formation, in which I start with isolated hydrated ions and form
(hydrated) complexes, and (2) dissolution reactions, in which I
start with a molecular model for the solid mineral and consider
877
878
TOSSELL: THEORY OF REACTIONS OF HGS
the dissolution caused by species in solution. Each type of reaction presents its own problems for quantum mechanical calculation. In the association reactions the gas-phase reactants and
products are fairly simple and can be treated at a high quantum
mechanical level. The much more serious problem is the description of the hydration processes. For dissolution reactions, particularly where solids produce neutral solution species, the
hydration problem is less serious, but studying a sufficiently large
molecular model for the solid may not be computationally feasible. In general, small molecular models of solids represent the
real solid more accurately if the solid shows small coordination
numbers and substantial covalency. Such a approach is therefore
feasible for HgS, cinnabar, a helical chain structure with twofold-coordinated Hg (1950) and for HgCl2, which is basically a
molecular crystal with twofold-coordinated Hg (Wells 1975). As
discussed by Kaupp and von Schnering (1994a) and more qualitatively by Tossell and Vaughan (1981), the small coordination
numbers observed in Hg compounds are a consequence of relativistic effects. The impact of relativistic effects on the structures of heavy element compounds has also been discussed by
Pyykko (1988). Other researchers have recently studied Hg complexes (Stromberg et al. 1991; Swerdtfeger et al. 1993; Kaupp
and Schnering 1994), establishing the importance of relativistic
effects and calculating structures and energetics at a fairly high
quantum mechanical level.
COMPUTATIONAL METHODS
We use Hartree-Fock based techniques (Hehre et al. 1986).
The basis sets used are of the relativistic effective core potential type, as implemented by Stevens, Basch, and Krauss (SBK)
(Stevens et al. 1992). Polarization functions of d type only are
added to each of the non-H atoms. We used the programs
GAMESS (Schmidt et al. 1993) and GAUSSIAN94 (Frisch et
al. 1994). In addition to the Hartree-Fock calculations we have
utilized the Moller-Plesset 2nd order perturbation theory method
(MP2) (Pople et al. 1976) for evaluation of the bond distances
and energetics of many of the molecules studied, particularly
those containing a single Hg atom. For most of the molecules
considered we have also calculated the Hessian matrix to establish that we have found an equilibrium geometry and to determine the vibrational spectrum, allowing determination of the
zero-point vibrational energy.
The calculation of hydration effects upon chemical reactions is currently a very active area within quantum chemistry,
with a number of different approachs being used by various
researchers (Wiberg et al. 1996; Cramer and Truhlar 1994). To
evaluate hydration enthalpies we use a multipart approach. For
monatomic ions, such as Hg2+ and Cl–, and for the common
small polyatomics, such as OH– or SH–, we use “experimental”
hydration energies from the tables of Rashin and Honig (1985).
Of course the hydration energy of a neutral compound cannot
be uniquely divided into contributions from cation and anion,
but by chosing the hydration energy of one particular species
as a reference, most all other hydration energies can be defined
with respect to it, obtaining quite consistent values. For neutral
molecules, we evaluate hydration enthalpies by performing
quantum mechanical supermolecule calculations with as many
as 6 water molecules surrounding the solute molecule. The
choice of six water molecules is somewhat arbitrary and is dictated in part by computational considerations and in part by
the coordination numbers commonly observed for cations using solution X-ray diffraction or EXAFS. For a few of the species considered, e.g., Hg+2 and HgCl2, we have considered
complexation with four, six, and eight water molecules, and
have established that the greatest stability occurs for complexation with six water molecules. For polyatomic ions we evaluate the first hydration sphere contribution to the hydration
energy by performing such a supermolecule calculation and
determine the longer range contribution by performing a selfconsistent-reaction-field (SCRF) calculation (Wiberg et al.
1996) for the supermolecule immersed in a polarizable continuum with a dielectric constant equal to that in bulk H2O.
The Born radius needed for the SCRF calculation is determined
using the Rashin and Honig (1985) semiempirical prescription
(not an essentially arbitrary electron density surface criterion
as in Wiberg et al. 1996). The use of the RH effective Born
radius value, rather than the radius calculated from the molecular volume by GAUSSIAN, is potentially significant because of the large magnitude of the Born term. In the
GAUSSIAN94 implementation of the SCRF scheme only the
dipole term is included, so that for a supermolecule with zero
–2
dipole moment, e.g., Hg(OH2)+2
6 or HgCl4(OH2)6 , the Born term
is the only non-zero term. Other potentially more accurate
methods, such as the isodensity polarized continuum method
(IPCM), are not yet implemented for the case of effective core
potentials.
The general procedure of combining a supermolecule calculation with a Born model or SCRF calculation is simple and
intuitively appealing and has been used by a number of different researchers (e.g., Claverle et al. 1978; Freitas et al. 1992;
Parchment et al. 1996) at various levels of sophistication. It
basically relies upon the idea that there should be a smooth
convergence of the local hydration energy for a cluster of solvent molecules to the bulk solvation energy as the number of
solvent molecules is increased. Coe (1994) has demonstrated
that this is approximately true, although the change in solvation energy with number of solvent molecules may not be completely continuous.
It may be worthwhile to describe in detail the procedure
used for the cases of Hg+2, Cl–, HgCl +1, and HgCl2. First, for
Hg+2 and Cl– we use the hydration energies of -0.700 and -0.136
Hartrees (H), respectively, tabulated by Rashin and Honig
(1985). An alternative would be to calculate these directly using our combined supermolecule, Born model approach. For
+2
the formation of Hg(OH2)+2
and six H2O mol6 for free Hg
ecules we calculate an energy change of -0.420 H, using the
HF method and the polarized SBK basis and correcting for basis
set superposition error (Boys and Bernardi 1970). The calculated Hg-O distance in Hg(OH 2)+2
6 is 2.428 Å (Th symmetry assumed), consistent with the experimental X-ray value of 2.40
Å obtained by Johansson (1971). Adding the Rashin and Honig
value of 1.495 Å for the “O-H” bond radius to the calculated
Hg-O distance we get a radius for the Born model calculation
of 3.92 Å, giving a Born energy of –0.271 H, for a total hydration energy of –0.691 H, certainly in good agreement with the
RH value of –0.700 H. For the case of Cl– the formation energy
TOSSELL: THEORY OF REACTIONS OF HGS
for Cl(OH 2)6–1, is –0.091 H (after basis set superposition correction), and the Cl-O distance has a range from 3.17 to 3.60 Å
and an average value of 3.54 Å. This gives a Born energy of
-0.053 H, for a total hydration energy of –0.144 H, fairly close
to the the RH “exp.” value of –0.136 H. Note that in our calculations we have required the Cl– to lie at the center of the cluster, which is apparently not its lowest energy position with a
small cluster (Caldwell and Kollman 1992; Coe 1994) but is
appropriate for bulk solution. However, we emphasize that the
reaction energies (below) are not based upon such calculated
hydration energies for the monotomic and small ions, but utilize the RH experimental values.
For ions such as HgCl + no tabulated hydration energies are
available, so we utilize the combined supermolecule, SCRF
model approach. We first calculated energies for formation of
+
HgCl(OH 2)+1
n , n = 4–6, from HgCl and H 2O, establishing that
the most stable species had n = 5. This species is shown in
Figure 1. The calculated Hg-O distances are 2.34, 2.60, 2.60,
2.64, and 2.64, while the Hg-Cl distance is 2.38 Å. Adding the
RH radii of 1.94 and 1.495 Å for Cl– and OH–, respectively, we
get radii along the bond directions of 3.84, 4.10, 4.10, 4.14,
4.14, and 4.32 Å, for an average of 4.11 Å. This average radius
gives a Born energy of -0.0647H. The SCRF calculation as
implemented in GAUSSIAN94, gives an additional polarization energy of -0.0134 from interaction of the dipole moment
of HgCl(OH 2)+1
5 with the solvent polarizable continuum, for a
total hydration energy of –0.251 H. For the HgCl2 case, six
H2O molecules attach with an energy of –0.0475 H (four H2O
molecules give a supermolecule formation energy of –0.0375
H). There is no Born energy because the species is neutral and
there is no SCRF polarization energy because the dipole term
is zero for HgCl2(OH2)6. The neglect of higher multipole terms
probably systematically underestimates the stability of such
symmetric species. However, using an average Born radius and
treating lower symmetry species such as HgCl(OH 2)5+1 within a
spherical approximation may also systematically underestimate
their stability.
879
RESULTS
Structures and vibrational spectra
As is generally observed for Hg compounds (Kaupp and
Schnering 1994b) calculated bond distances (Table 1) are substantially longer than experiment at the Hartree-Fock level (even
when relativistic effects are included) but are considerably
improved by MP2, although they are still a bit too long. The
errors are similar in magnitude for the neutral species HgCl2
and Hg 3S2(SH)2 but seem somewhat larger for the HgCl4–2 anion. If we add the Born hydration energy in aqueous solution
for HgCl –2
4 to the HF energy of the free anion the equilibrium
distance is somewhat reduced, because shorter Hg-Cl distances
give a larger Born stabilization. The calculated equilibrium distance drops from 2.63 to 2.58 Å, giving better agreement with
experiment (although the experimental values is for a different
condensed-phase environment). For HgS our Hg3S2(SH)2 model
reproduces not only the Hg-S bond distance, but S-Hg-S and
Hg-S-Hg angles, and the helical structure of HgS, cinnabar, as
shown in Figure 2. Note that the calculations on the trimeric
model Hg3S2(SH)2 are already extremely demanding, so that
we have not considered any larger oligomers.
Vibrational frequencies calculated at the HF level (Table 2)
are similar to those of Kaupp and Schnering (1994b) and establish that unscaled HF values reasonably fit experiment for
neutral molecules or ions with small charges. That is, for these
heavy element compounds the Hartree-Fock method with conventional basis sets does not give the systematic underestimation of bond distances and overestimation of vibrational
frequencies observed for light element compounds (Hehre et
al. 1986). For the case of HgCl–2
4 , the calculated frequences are
seriously underestimated, presumably for the same reason that
the bond distance was overestimated for free HgCl–2
4 , compared
to the condensed-phase experimental data. The spectrum of the
molten HgCl2, KCl system is rather complicated (Janz and
James 1963) and some of the spectral assignments may be in
error. An extremely demanding calculation on the vibrational
spectrum of HgCl4–2...6H2O yielded no significant change in
the calculated frequences, indicating that explicit consideration
of the Born model term would probably be a better approach.
TABLE 1. Comparison of calculated and experimental Hg-X distances (in Å)
Calculated
HF
MP2
2.33
2.30
2.33, 2.36 2.25 × 2
3.31
–
HgCl 4–2
2.63 (2.58§) 2.58
+2
Hg(OH 2)6
2.43
–
HgS
2.36
2.27
Hg2 S(SH) 2
2.39
2.36
Hg3 S2(SH) 2
2.39–2.41
Purchase answer to see full
attachment