American Mineralogist, Volume 87, pages 679–689, 2002

Surface structures, stabilities, and growth of magnesian calcites: A computational

investigation from the perspective of dolomite formation

NORA H. DE LEEUW*

Department of Chemistry, University of Reading, Whiteknights, Reading RG6 6AD, U.K.

ABSTRACT
Atomistic computer modeling methods are used to investigate the surface structures and stabili-
ties of magnesite MgCO3 and dolomite MgCa(CO3)2 in comparison with calcite, CaCO3. The sur-
faces are generally more stable than magnesite, but less so than calcite. Upon hydration, the magnesite
–
{1014} surface becomes more stable than either calcite or dolomite, due to the less favorable inter-
actions between dolomite surface ions and adsorbing water molecules. The high affinity of the sur-
face magnesium ions for adsorbing water molecules, shown by the large average hydration energy
over all surfaces for magnesite (134 kJ/mol), is negated in the case of dolomite by the inaccessibility
of the surface magnesium ions, due to surface relaxations and rotation of the carbonate groups. In
addition, the larger lattice spacing of dolomite compared to magnesite disrupts the extensive net-
work of hydrogen bonding between the adsorbed water molecules that are present on the latter,
leading to similar average hydration energies for calcite, dolomite, and magnesian calcite surfaces
(80–101 kJ/mol). Not only are the dolomite planes under aqueous conditions less stable than their
calcite counterparts, dolomite-like surfaces of disordered magnesian calcites are calculated to be
more stable than either pure calcite or dolomite surfaces. Extensive molecular dynamics simulations
of the growth of CaCO3 and MgCO3 at two experimentally observed calcite steps show that, on
thermodynamic grounds, Mg is easily incorporated into a growing calcite crystal, but will then
inhibit incorporation of further Ca. Furthermore, although the calculations suggest a small energetic
advantage (~10 kJ/mol) to ordering of the Ca and Mg ions on the obtuse growth step, a disordered
arrangement is favored at the acute growth step and incorporation of Mg is always energetically
more favorable than Ca, in accord with the occurrence of highly magnesian calcites in nature.

INTRODUCTION such as hot springs, in sediments of salt lakes, and in muds

Dolomite, an ordered Ca-Mg carbonate mineral, is found
extensively in ancient rocks in the Earth’s crust (e.g., Deer et
al. 1992) and calcite and dolomite together account for more
than 90% of the natural carbonates in rocks (Reeder 1990).
Despite several important occurrences in metamorphic rocks
(e.g., Letargo et al. 1995; Ferry 1996, 2000, 2001), dolomite is
typically a sedimentary mineral, yet the mechanism of its growth
under sedimentary conditions is unknown, apart from the fact
that there is no geologic evidence indicating that its formation
took place under unusual conditions of temperature or pres-
from salt lagoons undergoing strong solar evaporation, sedi-
mentary dolomite does not form in the normal present-day
marine environment and it is notoriously difficult to crystal-
lize under laboratory conditions (Krauskopf and Bird 1995).
Laboratory precipitation has been achieved at high tempera-
tures (e.g., Katz and Matthew 1977; Baker and Kastner 1981;
Sibley et al. 1994), at extreme supersaturations (e.g.,
Liebermann 1967; Nordeng and Sibley 1994) or from solu-
tions with pH values greater than 9.5 and containing high con-
centrations of SO2– –
4 and NO3 (Sibley et al. 1994; Krauskopf

sure (Krauskopf and Bird 1995). Some primary dolomites are
formed at the outset as evaporite deposits, but most dolomitic
sediments are produced from the reaction of calcium carbon-
ate deposits (e.g., isomorphous calcite) with magnesium solu-
tions. These secondary dolomites are either formed early on
from unconsolidated calcite sediments on the sea floor, or much
later when magnesian solutions have entered through faults and
joints in the calcium carbonate rock (e.g., Wenk 1993). How-
ever, although formation of dolomite as a primary precipitate
in nature has been reported in several unusual environments,
* E-mail: n.h.deleeuw@reading.ac.uk
0003-004X/02/0506–679$05.00
and Bird 1995), but problems persist when attempting to crys-
tallize dolomite under the usual mild conditions applicable to
the most common carbonate mineral calcite (Brady et al. 1996).
It is these extreme conditions represented by both the labora-
tory experiments and field occurrences, compared with usual
sedimentary environments, which makes the abundance of do-
lomite in sedimentary rocks so mysterious.
Not surprisingly, dolomite and its origins has been the sub-
ject of numerous investigations, both experimentally and theo-
retically. For example, dissolution experiments by Bertram et
al. (1991) showed that the solubility of synthetic high magne-
sian calcite decreases with increasing temperature, but that re-
679
680 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES
crystallization at low temperatures into dolomite and low-mag-
nesian calcite should be thermodynamically favored, while
experimental work on nucleation and growth kinetics of cal-
cium and magnesium carbonates by Nordeng and Sibley (1994)
showed that the surface free energy of high-magnesium calcite
is less than that of dolomite. Investigations of structural as-
pects of dolomite include Ross and Reeder’s (1992) X-ray in-
tensity data refining the structural parameters of stoichiometric
and ferroan dolomites and high pressure X-ray diffraction stud-
ies by Fiquet et al. (1994) to obtain bulk moduli for dolomite
and other carbonates, while Williams et al. (1992) measured
vibrational spectra of magnesite and dolomite. More recent stud-
ies of dolomites have investigated surface growth processes,
for example the transmission electron microscopy (TEM) stud-
ies by Paquette et al. (1999) of growth hillocks and surface
microtopography, and scanning microscopy investigations of
etch pits, layers, and islands on synthetic and natural dolomite
surfaces by Kessels et al. (2000).
Theoretical models based on experimental observations in-
clude the surface complexation model by Brady et al. (1996),
who proposed the formation of Mg-sulfate complexes as a pos-
sible path to dolomite growth and surface speciation models
for both magnesite and dolomite based on measurements of
surface charge as a function of pH (Pokrovsky et al. 1999a,
1999b). Previous computer modeling studies have addressed
the surface structures and stabilities of dehydrated dolomite
relative to calcite (Titiloye et al. 1998) while Wright et al. (2001)
–
compared the structure of the planar (1014) surface to those of
magnesite and calcite. Fisler et al. (2000) concentrated on the
ordering of Mg and Ca in the bulk mineral, following up ear-
lier experimental work on the behavior of Mg diffusion in bulk
calcite (Fisler and Cygan 1999).
In this article I use a combination of classical simulation
techniques, employing established potential models (Pavese et
al. 1996; de Leeuw and Parker 1998a, 2000). I investigate the
surface structures and stabilities of a range of dry and hydrated
dolomite surfaces, which are observed in the experimental mor-
phology (Deer et al. 1992). I compare the same with their coun-
terparts in the pure end-members calcite and magnesite and a
disordered magnesian calcite. In addition, as dolomite forma-
tion often occurs from the interaction of calcium carbonate with
Mg solutions (Krauskopf and Bird 1995), I also calculate the
energetics of incorporating Mg at two experimentally observed
–
growth steps on the calcite {1014} surface under aqueous con-
ditions, taking into account the ordering required to form the
dolomite crystal. My calculations provide insight into the ef-
fect of the different surface ions on the relative surface stabili-
ties of the carbonate phases, which has implications for the
mineral morphologies, and the relative enthalpies of growth of
ordered and disordered magnesian calcites.
THEORETICAL METHODS
Atomistic simulation techniques are based on the Born
model of solids (Born and Huang 1954), which assumes that
the ions in the crystal interact via long-range electrostatic forces
and short-range forces, including both repulsions and the van
der Waals attractions between neighboring electron charge
clouds. These short-range forces are described by simple ana-
lytical functions. The electronic polarizability of the ions is
included via the shell model of Dick and Overhauser (1958) in
which each polarizable ion, in this case the O atom ion, is rep-
resented by a core and a massless shell, connected by a spring.
The polarizability of the model ion is then determined by the
spring constant and the charges of the core and shell. When
necessary, angle-dependent forces are included to allow direc-
tionality of bonding as, for example, in the model of the cova-
lent carbonate anion developed by Pavese et al. (1996). For the
calcium carbonate materials, the present model uses the short-
range interactions derived empirically by Pavese et al. (1996)
in their study of the thermal dependence of structural and elas-
tic properties of calcite. Although this potential was fitted to
bulk properties, it is generally possible for ionic materials to
transfer potential parameters to surface calculations. In semi-
conductors, where the surface involves broken bonds, and met-
als, where the surface means a sudden change in the electron
density, there is often a problem with transferability from bulk
potential parameters to surfaces. However, in ionic materials
after relaxation, the Madelung potentials are 90% or more of
the bulk values and hence the change in ionic radii is negli-
gible. Oliver et al. (1996), for example, used bulk derived po-
tentials to model WO3 surfaces and found excellent agreement
between calculated and experimental surface structures, ob-
tained by scanning tunneling microscopy. Because the surfaces
considered in this work leave the carbonate groups intact, the
bulk derived potential model will be adequate. Indeed, the po-
tential model derived for calcite by Pavese et al. (1996) has
been used successfully in numerous surface simulations, not
only of calcite, but also of the less ubiquitous calcium carbon-
ate phases aragonite and vaterite (e.g., de Leeuw and Parker
1997, 1998a; Titiloye et al. 1998). The experimental surface
energies and structural features were reproduced accurately (de
Leeuw and Parker 1997), while the calculated morphologies
of the three calcium carbonate phases, which are dependent on
the relative stabilities of the various surfaces, agreed well with
experimentally determined crystal morphologies (de Leeuw and
Parker 1998a). The parameters used for the Mg-CO3 interac-
tions and those between Mg and water were derived in an ear-
lier work (de Leeuw and Parker 2000), using a combination of
static lattice energy minimization calculations of the magnes-
ite (MgCO3) crystal and molecular dynamics (MD) simulations
of the Mg2+ ion in liquid water. The intra- and intermolecular
interactions of the water molecules were derived using MD
simulations of liquid water, the details of which are described
in a previous paper on modeling MgO surfaces in liquid water
(de Leeuw and Parker 1998b). The interactions between water
and calcite surfaces were first derived in a previous work on
water adsorption at calcite surfaces (de Leeuw and Parker 1997).
The full potential model is given in Table 1.
The simulation methods used in this work are a combina-
tion of static lattice energy minimization techniques and MD
simulations. The energy minimization codes employed for the
calculation of the bulk crystal properties were PARAPOCS
(Parker and Price 1989) and GULP (Gale 1997), while the sur-
faces were modeled using the code METADISE (Watson et al.
1996), which is designed to model dislocations, interfaces, and
surfaces. Following the approach of Tasker (1979), the crystal
 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES 681

consists of a series of charged planes parallel to the surface test runs with various masses for the O atom shell (from 0.05
and periodic in two dimensions. The crystal is divided into two to 0.5 au), 0.2 au was chosen which is small compared to the
blocks, each comprising two regions, region I and region II. mass of the hydrogen atom of 1.0 au, ensuring that there would
Region I contains those atoms near the surface, which are al- be no exchange of energy between vibrations of O atom core
lowed to relax to their mechanical equilibrium, while region II and shell with O atom and hydrogen vibrations (de Leeuw and
contains those atoms further away, which represent the rest of Parker 1998b). However, due to this small shell mass the MD
the crystal and are kept fixed at their bulk equilibrium posi- simulations needed to be run with a small timestep of 0.2 fs to
tion. Inclusion of region II is necessary to ensure that the po- keep the system stable. The stepped calcite surfaces were mod-
tential of an ion at the bottom of region I is modeled correctly. eled as a repeating slab, containing 120 CaCO3 units and a gap,
The DL_POLY code (Forester and Smith 1995) was used containing 48 water molecules, giving a total simulation cell
–
for the MD simulations of growth steps on the calcite {1014} of 1152 species. MD calculations of calcite dissolution (de
surface, where the integration algorithms are based around the Leeuw et al. 1999) showed that the size of these simulation
Verlet leap-frog scheme (Verlet 1967). The Nosé-Hoover algo- cells (where the steps are more than 10 Å apart) ensured that
rithm (Nosé 1984; Hoover 1985) was used for the thermostat there are no interactions between the ions on adjoining step
as this algorithm generates trajectories in both NVT ensembles edges.
(used for the surface simulations) and NPT ensembles (used
for the potential derivation), thus keeping the simulations con- RESULTS
sistent. The Nosé-Hoover parameters were set at 0.5 ps for both All three minerals have a rhombohedral crystal structure,
the thermostat and barostat relaxation times. In DL_POLY the which can be described by either a rhombohedral or hexagonal
shells are assigned a small mass and the choice of shell mass is axial system (Reeder 1990). As the two axial systems can make
a balance between keeping the mass as close to zero as pos- the description of the surfaces by their indices ambiguous, in
sible to ensure that no vibrational energy is exchanged with this work the hexagonal axial system and the resulting four
other species, notably hydrogen, and making the shells heavier, Miller-Bravais indices are used to describe each of the mineral
which allows a longer timestep during the simulations and hence surfaces. The pure end-members calcite CaCO3 and magnesite
–
computationally less expensive calculations. After a series of MgCO3 have the R3c space group, but the dolomite structure
comprises alternating planes of Mg and Ca ions (separated by
carbonate planes) in the c direction (Fig. 1), and as a result
–
TABLE 1. Potential model dolomite has the R3 space group, similar to calcite and magne-
Ion Core charge (e) Shell charge (e) Core-shell site, but with the loss of the c-glide. The experimental and cal-
interaction culated properties of the three minerals are listed in Table 2,
(eVÅ-2)
Ca, Mg +2.000 from which one can see that not only the calculated structural
C +1.135 but also the elastic properties agree well with the experimen-
H +0.400
carbonate oxygen (O) +0.587 –1.632 507.400000
tally determined values, even though the structure of dolomite
water oxygen (Ow) +1.250 –2.050 209.449602 and the elastic moduli for magnesite and dolomite were not
used in the derivation of the potential parameters. The poten-
Ion pair Buckingham potential
A (eV) r (Å) C (eVÅ6) tial model for MgCO3 was not only derived from the structure
Ca-O 1550.0 0.29700 0.0 of the magnesite crystal, but the parameters of the MgCO3 crys-
Ca-Ow 1186.6 0.29700 0.0 tal and its interactions with water were also fitted to the ex-
Mg-O 1092.2 0.27926 0.0
Mg-Ow 2290.0 0.22000 0.0
H-O 396.3 0.23000 0.0
H-Ow 396.3 0.25000 10.0
O-O 16372.0 0.21300 3.47
O-Ow 12533.6 0.21300 12.09

Lennard-Jones potential
A (eVÅ12) B (eVÅ6)
Ow-Ow 39344.98 42.15

Morse potential
D (eV) a (Å–1) r0 (Å)
C-O 4.710000 3.80000 1.18000
H-Ow 6.203713 2.22003 0.92376

Three-body potential
k (eV/rad2) Q0
Ocore-C-Ocore 1.69000 120.000000
H-Owshell-H 4.19978 108.693195

Four-body potential
k (eV/rad2) Q0
C-Ocore-Ocore-Ocore 0.11290 180.0

Coulombic subtraction (%) FIGURE 1. Bulk structure of dolomite, showing alternating planes
H0.4+-O0.8- 50 of calcium, carbonate, and magnesium (Ca = gray, Mg = black, CO3 =
H0.4+-H0.4+ 50 white three-spoked wheel).
682 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES

–
perimental enthalpies for the processes described in Equations {1120} surfaces of calcite, dolomite, and magnesite to investi-
1 and 2 below, which were calculated at +304.9 kJ/mol and gate how the relative stabilities of the surfaces depend on in-
+23.3 kJ/mol respectively [cf. exp. +311.5 and +35.2 kJ/mol creasing numbers of Mg ions in the structures and also how
(Lide 2000)]. The energies of the hydrated ions (Mg2+, Ca2+, they alter as water is adsorbed at the surface. To this end the
and CO32–) were obtained from Molecular Dynamics simula- surface energies of both dry and wet surfaces were calculated,
tions at constant pressure and temperature, where each ion was where the latter had a monolayer of water interacting with the
surrounded by 255 water molecules in a box, which was peri- surface [in this work a monolayer is defined to mean a surface
odically repeated in three dimensions. coverage of one water molecule per surface calcium or car-
bonate group, which was found to be the preferred configura-
–
CaCO3(s) + Mg(g)
2+
Æ MgCO3( s ) + Ca (2g+) (1) tion on the dominant {1014} surface (de Leeuw et al. 1999)].
The water molecules were placed above the surface, after which
CaCO3( s ) + Mg(2aq+ ) Æ MgCO3( s ) + Ca (2aq+ ) (2) both the surface and the adsorbing water molecules were al-
lowed to fully relax using a Newton Raphson energy minisation
The accuracy of the CaCO3-H2O interactions was verified technique (Watson et al. 1996). A host of initial configurations
by modeling the hydrous calcium carbonate phase ikaite, which for the water molecules was calculated, e.g., adsorbed to sur-
is a crystalline calcium carbonate hexahydrate, usually found face cations or O atoms or rotated in various positions, to en-
only at very low temperatures, and often in the presence of sure as far as possible that the lowest energy configuration had
calcite inhibititors (Bischoff et al. 1993; Clarkson et al. 1992). been obtained. In all cases the water molecules show a clear
Although thermodynamically stable with respect to the aque- preference for coordination by O ions, but where possible also
ous calcium and carbonate ions (DG = –37 kJ/mol), ikaite is hydrogen bonding to surface O atom ions. Where the same sur-
less stable than the anhydrous form calcite [by about 10 kJ/ faces were calculated for the three different minerals, the simu-
mol (Lide 2000)] and it is generally accepted as one of the pre- lations were performed on the same number of formula units
cursor phases in the precipitation of calcite from aqueous solu- in the simulation cells, so that the surface areas and surface
tions (Bischoff et al. 1993; Ito 1998). In an earlier work, good energies for the three minerals could be compared directly. The
agreement was found between the experimental and calculated simulation cells typically consisted of 40–80 unit cells, each
structural parameters of ikaite (de Leeuw and Parker 1997), containing six MCO3 formula units, giving a maximum of 3840
and a change in enthalpy for the dissocation of ikaite into cal- species per simulation cell. The surface energy g (Eq. 3) of a
cite and water of 47 kJ/mol per water molecule was calculated, particular surface is a measure of the stability of that surface
which compared very well with experimental values of 47–50 with a low positive value indicating high stability.
kJ/mol (Bischoff et al. 1993). In addition, when hydration of
Us - U b
the surfaces was taken into account, the experimental morpholo- g= (3)
gies of calcite and aragonite were reproduced accurately, which A
shows that the relative stabilities of the various surfaces were where Us is the energy of the surface block of the crystal, Ub is
calculated accurately (de Leeuw and Parker 1998a). Further- the energy of an equal number of atoms of the bulk crystal and
more, calcite surface features which had been observed by A is the surface area. When the surfaces are hydrated, the sur-
atomic force microscopy (AFM) were reproduced only when face energy is calculated as follows:
adsorbed water molecules were included in the surface calcu-
U h - ( U b + nU H 2 O )
lations (de Leeuw and Parker 1997), all of which indicate that gw = (4)
the water-solid interactions are modeled reliably. A
where Uh is the energy of the relaxed surface with n adsorbed
Structures and relative stabilities of planar surfaces water molecules and UH2O is the energy of a liquid water mol-
Static lattice energy minimization techniques were first ecule, which is the sum of the calculated self-energy of an iso-
– – –
employed to model the {1014}, {0001}, {1010}, {1011}, and lated water molecule and the condensation energy of water,
which were calculated using MD simulations at –43.0 kJ/mol
(de Leeuw and Parker 1998b). This is in excellent agreement
with the experimental value [–43.4 kJ/mol (Duan et al. 1995)],
TABLE 2. Calculated and experimental properties of calcite, mag- especially as the potential parameters were not fitted to the
nesite, and dolomite condensation energy.
Property Calcite Magnesite Dolomite The calculated surface and hydration energies for the vari-
exp. calc. exp. calc. exp. calc. ous surfaces are given in Tables 3 and 4 respectively. The
V (Å3) 367.8 348.0 279.1 260.1 323.6 300.9 –
(–5.4%) (–6.8%) (–7.0%)
{0001} and {1011} surfaces are dipolar and have more than
a,b (Å) 4.990 4.796 4.633 4.444 4.810 4.640 one surface termination, being either terminated by cations or
c (Å) 17.061 17.470 15.016 15.210 16.010 16.139 by carbonate groups. For the calcite and magnesite crystals, I
c/a 3.419 3.643 3.241 3.422 3.328 3.478
Bulk 73.5* 80 117† 123 91‡ 97 need only consider two surfaces, but in the case of dolomite
modulus there are two cation-terminated planes, either Ca or Mg, and in
(GPa) addition two carbonate terminated planes, one with Ca and the
* Redfern and Angel (1999).
† Ross (1997).
other with Mg ions in the second surface layer. All of these
‡ Martinez et al. (1996). different surface terminations need to be considered to calcu-
 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES 683

late the overall stability of these dipolar surfaces and to deter- interactions to surface O ions at approximately 1.8 Å. Molecu-
mine which surface termination will be the dominant plane. lar dynamics simulations of a monolayer of water on the cal-
–
The dipoles are removed by the creation of surface vacancies, cite {1014} surface at 300 K showed that there is no surface
where the resulting planes are 50% vacant with respect to ei- diffusion of the surface water molecules, which remain adsorbed
ther surface cations or carbonate groups (Tasker 1979), which in the same pattern as found here (de Leeuw et al. 1999). On
is why the nature of the underlying second surface layer can- the magnesite surface, the water molecules are adsorbed in an
not be ignored, as its ions are exposed underneath the carbon- equally regular pattern (Mg-Ow distance of 1.85 Å) but rotated
ate vacancy sites. with respect to the calcite surface. Due to the smaller area of
As can be seen from Table 3, the surface energies of the the surface unit cell, the rotated adsorption pattern ensures that
–
{1014} surfaces of the three minerals before hydration increase
with the increasing number of surface Mg ions, from g = 0.59
J/m2 for calcite to g = 0.64 J/m2 for dolomite and g = 0.76 J/m2
for magnesite, showing that the non-hydrated purely Ca sur-
face of calcite is the most stable and the purely Mg surface of
magnesite the least stable. However, once hydrated, the trend
is reversed with magnesite becoming the most stable surface (g
= 0.02 J/m2), followed by dolomite (g = 0.24 J/m2), and finally
calcite as the least stable surface (g = 0.33 J/m2). This indicates
that under aqueous conditions, formation of a structure con-
taining a complete Mg surface is energetically more favorable
than formation of the mixed Mg-Ca surface in dolomite. The
difference in stabilization of the three surfaces is due to the
dissimilar patterns of adsorbed water molecules (see Fig. 2).
On the calcite surface, the water molecules form a regular her-
ringbone pattern, closely coordinated to the surface. The water
molecules are coordinated by their O ions to surface Ca ions at
a distance of 2.4 Å, in good agreement with earlier calcula-
tions (de Leeuw and Parker 1997) and subsequent X-ray
reflectivity measurements by Fenter et al. (2000), who mea-
sured a Ca-OHx distance of 2.50 ± 0.12 Å (where x = 1, 2, or
3). In addition, the water molecules form hydrogen-bonded

TABLE 3. Surface energies of calcite, magnesite and dolomite (Jm–2)

Surface Calcite Magnesite Dolomite
dry wet dry wet dry wet
–
{101 4} 0.59 0.33 0.76 0.02 0.64 0.24
{0001}Ca 0.97 0.76 – – 1.04 0.83
{0001}Mg – – 1.41 0.82 1.29 0.77
{0001}CO3 (Ca) 0.99 0.70 – – 1.19 0.88
{0001}CO3 (Mg) – – 1.41 0.59 1.21 0.47
{1010}
– 0.97 0.82 1.26 0.80 1.13 0.84
{101–1}Ca 1.23 0.88 – – 1.28 0.83
{101–1}Mg – – 1.61 0.74 1.17 0.69
{101–1}CO3 (Ca) 1.24 0.95 – – 1.33 1.10
{101 1}CO3 (Mg) – – 1.42 0.99 1.30 0.79
{112–0} 1.39 0.80 1.90 0.70 1.28 0.82

TABLE 4. Hydration energies of calcite, magnesite, and dolomite

surfaces (kJmol–1)
Surface Calcite Magnesite Dolomite
–
{101 4} –73.7 –114.5 –85.8
{0001}Ca –68.6 – –67.6
{0001}Mg – –104.7 –101.9
{0001}CO3 (Ca) –67.2 – –78.5
{0001}CO3 (Mg) – –127.9 –127.5
{1010}
– –81.9 –137.4 –107.9
{101–1}Ca –54.2 – –113.7 –
FIGURE 2. Plan view of the hydrated {1014} surfaces of (a) calcite,
{101–1}Mg – –164.8 –118.2
{101–1}CO3 (Ca) –74.9 – –80.1
(b) magnesite, and (c) dolomite, showing the different adsorption
{101 patterns water molecules (MCO3 as framework: Ca = gray, Mg = black,
– 1}CO3 (Mg) – –102.8 –122.5
{112 0} –130.5 –184.5 –103.1 CO3 = white, water space filling: Owater = black, H = white).
684 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES
the water molecules not only form hydrogen bonds to surface
O ions, but they are now also capable of forming a network of
hydrogen bonds inter-molecularly between the water molecules,
which is not possible on calcite because of the larger lattice
spacing. However, on the mixed Mg-Ca dolomite surface, the
smaller Mg ions have receded into the surface with the Ca ions
protruding above the surface. As a result the Mg ions are less
accessible to adsorbing water molecules and the regular pat-
tern of water molecules found in calcite and magnesite is dis-
turbed. The water molecules adsorb in pairs, where there is
hydrogen-bonded coordination between the two paired water
molecules, but no extensive intermolecular network as was
found in magnesite. Furthermore, because of the lesser acces-
sibility of the Mg ions, the water is less strongly coordinated to
surface cations and as a result the hydrated dolomite surface is
less stable than its pure calcium or magnesium counterpart.
Despite the strong affinity of Mg for water, the surface energy
of the hydrated dolomite plane is similar to calcite rather than
magnesite.
In general, the other dehydrated surfaces follow the same
–
trend as the {1014} surfaces, where the surface energies for
magnesite are higher than those for calcite, but now they re-
main similar to calcite even when the surfaces are hydrated,
which shows that the extremely low surface energy of the hy-
–
drated magnesite {1014} surface is a special case, due to the
fact that the lattice spacing enables the formation of an unusu-
ally stable network of hydrogen-bonded coordinations between
the water molecules and to the surface (described above). Of
– – –
the dolomite surfaces, the {1014}, {1010}, and {1120} planes
–
have mixed Mg/Ca layers, while the {0001} and {1011} sur-
faces contain alternating planes of Mg or Ca ions, separated by
carbonate layers. When I compare the surface energies of the
dolomite surfaces, I see that for the dehydrated {0001} surface
the planes where the first cation layer consists of Mg are less
stable than the planes terminated by Ca ions in the first cation
–
layer, while the reverse is the case for the {1011} surface, where
the Mg-terminated surfaces are more stable. Once hydrated,
the Mg-terminated planes are the more stable planes for both
– –
{0001} and {1011} surfaces. Similar to the {1014} surfaces,
–
the surface energy of the dehydrated dolomite {1010} surface
(1.13 J/m2) falls between that for calcite (0.97 J/m2), and that
–
for magnesite (1.26 J/m2), but the dehydrated dolomite {1120}
surface is more stable than either its calcite or magnesite coun-
terpart. Once hydrated, some of the dolomite surfaces are less
stable than the equivalent magnesite surface, but the Mg-ter-
–
minated {0001} and {1011} planes have lower surface ener-
gies than their magnesite counterparts. Again, I see that the
mixed Ca/Mg surfaces are less stabilized by hydration than the
purely Mg surfaces, due to the strong interactions between sur-
face Mg ions and the water molecules and the reduced accessi-
bility of the Mg ions to the water molecules, once Ca is
incorporated in the surfaces as well.
Comparing the hydration energies for the different miner-
als (Table 4), the hydration energies for the magnesite surfaces
are generally larger than those for calcite. The weaker binding
–
of water to the dolomite {1014} surface, compared to the mag-
–
nesite {1014} surfaces, is shown quantitatively in the hydra-
tion energy. For dolomite this resembles the hydration energy
–
of the calcite {1014} surface, despite the presence of Mg ions
in the dolomite surface layer. On the other two mixed sur-
– –
faces, {1010} and {1120}, the hydration energies are dissimi-
–
lar to both calcite and magnesite, and hydration of the {1120}
surface releases only 103 kJ/mol, compared to 131 and 185 kJ/
mol for calcite and magnesite, respectively. The hydrated
–
{1120} surfaces of magnesite, calcite, and dolomite are shown
in Figure 3. The magnesite surface is virtually bulk-terminated,
without detectable relaxation of the carbonate groups, and the
water molecules are closely coordinated to the surface Mg ions
in addition to interacting by their hydrogen atoms to O ions of
neighboring water molecules and in the surface. In the calcite
surface, however, although the cation sublattice remains unaf-
fected, extensive rotation of the carbonate groups occurs in the
first three surface layers, and half the topmost carbonate groups
now lie almost flat in the plane. Calcite has a greater lattice
–
spacing (surface area {1120} = 48.4 Å2) than magnesite (sur-
2
face area = 39.1 Å ), and even after rotation of the surface car-
bonate groups, the sub-surface Ca ions are still accessible to
the water molecules, which bond to the Ca ions by their O at-
oms. These water molecules also form some hydrogen-bonded
interactions to surface O ions, although to a lesser extent than
on magnesite, and hence the hydration energy is lower for cal-
–
cite. The carbonate groups in the dolomite {1120} surface (sur-
face area = 43.3 Å2) rotate to a limited extent in the surface
layer. However, this partial rotation makes the Mg ions far less
accessible to the water molecules than in the magnesite sur-
face. In addition, the smaller lattice spacing results in less ad-
sorption of water to the Ca ions, and both effects combine to
lessen the hydration energy of this dolomite surface with re-
spect to the pure end-members.
The hydration energies of the dolomite {0001} planes are
very similar to their counterparts in magnesite and calcite, but
–
the hydration pattern on the {1011} planes is again disturbed.
The Ca-terminated dolomite plane has a much higher hydra-
tion energy than calcite while hydration of the Mg-terminated
magnesite plane releases far more energy than the dolomite
plane, due to the loss of an extensive intermolecular network
of hydrogen-bonded interactions between the water molecules
on the smaller magnesite surface (surface area = 46.5 Å2 com-
pared to 51.5 Å2 on dolomite). Rotation of the carbonate groups
on the carbonate-terminated surfaces leads to a similar hydra-
tion energy for dolomite compared to calcite, but not compared
–
to its magnesite counterpart. Contrary to the {1120} surface,
where rotation of the surface carbonate groups made the sur-
face Mg ions less accessible to the adsorbing water molecules,
–
the surface carbonate groups on the carbonate-terminated {101
1} planes rotate away from the underlying cations. These cat-
ions are now more accessible to the water molecules and hence,
this time, the hydration energy for dolomite is larger than for
–
magnesite, where again bulk termination of the {1011} sur-
face occurs without rotation of surface carbonate groups.
In addition to the fully ordered dolomite structure, a collec-
tion of less-ordered magnesian calcites was studied as well,
where Mg ions are introduced at various depths in the calcite
surfaces, leading to substitutions of Ca by Mg ions of between
6 to 25%. The surface and hydration energies of some surfaces
of these magnesian calcites are collected in Table 5, where only
 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES 685

the calculated values are shown for the planes that are directly
comparable to the dolomite surfaces. Magnesium is either in-
troduced in the topmost cation layer or in the second layer in
the same ratio as in dolomite. In a previous work (de Leeuw
and Parker 2000) we find that Mg introduced to the surface in
the presence of water, does not segregate into the bulk mineral,
but only dissolves into the bulk when the surface water is re-
moved. This study is primarily concerned with the possible
dolomitization of existing calcite and in the absence of diffu-
sion of Mg into the bulk, I found that introducing Mg deeper
into the structure (third and fourth sub-surface layers) did not
significantly affect the structures and energies compared to pure
calcite. Hence, the structures of the disordered magnesian cal-
cites, given in Table 5, all consist of an underlying bulk calcite
structure, but with a dolomite-type ordering of Mg and Ca ions
in the topmost surface layer(s). These magnesian calcites may
resemble, on a much smaller scale, some dolomite-like miner-
als, which are found to be richer in Ca than pure dolomite and
have a superstructure of larger lattice dimensions (Schubel et
al. 2000).
When comparing the surface energy of the dehydrated
–
{1014} surface of the magnesian calcite with those for calcite,
magnesite, and dolomite, this disordered structure is equally
stable as the fully ordered dolomite structure despite the un-
derlying calcite lattice. Hydration of the magnesian calcite leads
to a very similar pattern of adsorption of water molecules as on
the magnesite surface (Fig. 2b), even though the distribution
of Mg ions in the surface layer of the disordered material is the
same as in dolomite. However, because the bulk lattice is that
of calcite rather than dolomite, the surface Mg ions are not
obscured by the Ca ions to a similar extent as in dolomite, and
they thus remain accessible to the water molecules. However,
due to the larger lattice spacing of the magnesian calcite com-
pared to magnesite, the extensive network of hydrogen-bonded
interactions between the adsorbing water molecules is not
present in the magnesian calcite, which is reflected in the
smaller hydration energy for the disordered material compared
to that for magnesite (Table 5).
All of the other surface energies for the magnesian calcites
are lower than those for dolomite, both when dehydrated and
when water is present at the surface. As the only difference
between the surfaces of dolomite and magnesian calcite is the
increased lattice spacing of the latter (due to the underlying

TABLE 5. Surface and hydration energies of dolomite-like surfaces

of magnesian calcites
Surface Surface energies (Jm–2) Hydration energies (kJmol–1)
Dry Wet
–
{101 4} 0.65 0.25 –90.5
{0001}Ca 0.73 0.54 –65.6
{0001}Mg 1.03 0.59 –96.8
– {0001}CO3 (Ca) 0.87 0.42 –79.4
FIGURE 3. Elevation view of the {1120} surfaces of (a) magnesite, {0001}CO3 (Mg) 0.96 0.42 –83.2
(b) calcite, and (c) dolomite, showing little relaxation of the surface {1010}
– 0.98 0.72 –108.8
cations, but rotation of the carbonate groups in calcite and dolomite, {101–1}Ca 1.11 0.52 –111.9
{101–1}Mg 1.01 0.73 –75.8
together with the different patterns of adsorption of the water molecules {101–1}CO3 (Ca) 1.02 0.75 –90.4
(MCO3 as framework: Ca = gray, Mg = black, CO3 = white, water {101
– 1}CO3 (Mg) 1.08 0.65 –117.9
space filling: Owater = black, H = white). {112 0} 1.25 0.78 –111.4
686 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES

bulk calcite structure), rather than the cation distribution in the

surface, the stability of the surfaces is clearly influenced to a
large extent by the more open structure of the magnesian cal-
cite. This open structure has a twofold effect; first it results in
a more effective relaxation of the Mg-containing surface lay-
ers, hence lowering the surface energies of the dehydrated sur-
faces and, secondly, it enhances the accessibility of the surface
Mg ions to the adsorbing water molecules, hence further low-
ering the surface energies of the hydrated surfaces. As the hy-
dration energies of the magnesian calcite surfaces are actually
either similar or less than those for the equivalent surfaces in
dolomite (Table 5), the first effect is the more important in de-
termining the stabilities of the surfaces.
It is clear from the above calculations of surface energies of
dolomite, calcite, and magnesite that many of the dolomite sur-
faces in an aqueous environment are thermodynamically less
stable than the surfaces of its constituent carbonates calcite and
magnesite. In addition, a disordered magnesian calcite is also
energetically preferred over the ordered dolomite structure. As the
magnesian calcites were modeled on a calcite bulk structure, the
next step was to investigate the growth of such materials, by mod-
–
eling the incorporation of MgCO3 into the calcite {1014} surface.
FIGURE 4. Schematic representation of the modeled growth process
at the calcite step edges, where the black shaded box is the newly
incorporated CaCO3 or MgCO3 unit.
–
{1014} surface of dolomite is such that if this surface is to
grow from the calcite steps, the edge of each growth step should
consist of a regular array of alternating CaCO3 and MgCO3
groups.
I first calculated the energetics of adding isolated CaCO3
and MgCO3 units at either step (step 1 in Fig. 4), introducing
the first kink sites at the perfect step edges, then I calculated
the addition of CaCO3 and MgCO3 units next to other CaCO3
units on a growing edge (steps 2 and 3 in Fig. 4), and finally I
calculated the addition of the last unit into a step, annihilating
all kink sites and recreating a perfect step edge (step 4 in Fig.
4). Figure 6 shows graphs of the energetics of these steps for

MgCO3 growth at calcite steps

Although diffusion and Ca-Mg cation exchange may play a
role in dolomite formation, an alternative mechanism is by the
incorporation of Mg in a growing calcite crystal. Calcite crys-
tal growth occurs at steps. Hence, in contrast to the above com-
parison of the planar surfaces of calcite, magnesite, dolomite,
and magnesian calcite, the next task was to investigate the up-
–
take of Mg2+ ions at steps on the dominant {1014} surface.
Molecular dynamic simulations were used, enabling inclusion
of temperature in the calculations, to calculate the energetics
of incorporation of Mg ions in a growing calcite crystal (at 300
K). Each step in the growth process, namely the incorporation
of a new MgCO3 or CaCO3 unit at the growing step, was mod-
eled by a full molecular dynamic simulation including relax-
ation of the solid surface and the water molecules. Figure 4
shows a schematic representation of the modeled growth pro-
cess, where the dark shaded box is the newly incorporated
CaCO3 or MgCO3 unit.
Calcite growth is experimentally found to occur from two
–
monatomic steps on the {1014} surface (e.g., Gratz et al. 1993;
Liang et al. 1996a, 1996b). The absorption of Mg ions was
therefore studied at these two growth steps, to investigate pos-
sible dolomitization of unconsolidated calcium carbonates, in-
cluding solvent effects to mimic sedimentary conditions. One
of the growth steps is acute, i.e., the carbonate groups on the
edge of the step overhang the plane below the step with a cal-
culated angle between relaxed step wall and plane of 80∞ [cf.
exp. 78∞ (Park et al. 1996)] (Fig. 5a), while the other step is
obtuse, i.e., the carbonate groups on the step edge lean back
with respect to the plane below, forming an angle between re-
laxed step wall and surface plane of 105∞ [cf. exp. 102∞ (Park
et al. 1996)] (Fig. 5b). For a mixed Mg-Ca surface to be formed, FIGURE 5. Side view of the (a) acute and (b) obtuse calcite growth
it should be energetically feasible to incorporate Mg ions at Ca –
steps, showing the {1014} planes, offset from each other by monatomic
carbonate steps and calcium ions at steps decorated with mag- steps (MCO3 as framework: Ca = gray, Mg = black, CO3 = white, water
nesium carbonate. The arrangement of Ca and Mg ions in the space filling: Owater = black, H = white).
 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES 687

the addition of either a MgCO3 or a CaCO3. In all cases (steps by a second MgCO3 unit, and finally completing the edge with
1–4 in Fig. 4), addition of an MgCO3 unit is thermodynami- another CaCO3 unit. The enthalpies of these processes are also
cally favorable, whether incorporated onto a complete edge displayed in Figure 8, which shows that there is a large penalty
(step 1) or next to a CaCO3 corner site (steps 2–4). However, in energy in step 2 of the process. On the obtuse step, incorpo-
the addition of CaCO3 units is energetically unfavorable when ration of this CaCO3 group costs 34 kJ/mol, while adding an-
incorporated onto a complete edge or next to an isolated CaCO3 other MgCO3 unit releases 51 kJ/mol, a difference of 85 kJ/
unit. It is only when two (acute step) or three (obtuse step) mol. The same processes on the acute step are energetically
neighboring CaCO3 units are already present that the addition even more divergent, requiring and releasing 19 and 82 kJ/mol,
of a further CaCO3 unit is an exothermic process. In the case of
the obtuse edge this means that only the annihilation of the
kink sites and re-formation of a complete edge is energetically
favorable. As these calculations suggest that the growth of
MgCO3 onto the calcite steps is energetically preferred over
CaCO3 growth, the incorporation of further MgCO3 units next
to the first and subsequent MgCO3 units was also calculated, in
effect growing a complete MgCO3 edge onto both calcite steps.
–
Figure 7 shows a plan view of the {1014} surface with com-
TOP
pletely Mg-decorated acute step edges, which are offset from
each other in height by one monatomic layer (as found experi-
mentally by Hillner et al. 1993), where for clarity the water
molecules are not shown. There is some relaxation of the ions
MIDDLE
on the step edge, notably the carbonate groups, which have
rotated to best accommodate the Mg ions, but the Ca ions and
carbonate groups on the terraces remain virtually in their bulk-
BOTTOM
terminated positions. It is clear from the graph in Figure 8 that
further growth of MgCO3 units onto both edges is also ener-
getically very favorable, and far more exothermic than CaCO3
growth (energies shown in Fig. 6). However, the formation of
dolomite requires the presence of both Mg and Ca at the sur-
–
face and the arrangement of the Mg and Ca ions in the {1014}
surface of dolomite is such that I need to incorporate alternat-
ing CaCO3 and MgCO3 units at both growth steps to create a
surface that has the dolomitic cation arrangement. The calcu- –
lated energies for this alternating arrangement of Mg and Ca FIGURE 7. Plan view of three calcite {1014} planes (top, middle,
and bottom), offset from each other by acute steps, showing fully
ions at the growing edges are displayed in Figure 8. As the
magnesium-decorated step edges, where for clarity the water molecules
incorporation of an MgCO3 unit at the perfect edge is the ener- are not shown (Ca = gray, Mg = black, CO3 = white).
getically more favorable process, this MgCO3 unit was taken
as a starting point, subsequently adding a CaCO3 unit, followed

FIGURE 6. Energies of incorporation of MgCO3 and CaCO3 units FIGURE 8. Energies of sequential growth of MgCO3 units at acute
onto the acute and obtuse calcite growth steps at the different stages and obtuse calcite steps, together with energies of growth of dolomite
–
of the growth process shown in Figure 4. edges at the acute and obtuse steps on the calcite {1014} surface.
688 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES
respectively, a difference in energy of over 100 kJ/mol. The
last two steps in the formation of a dolomite edge are exother-
mic, but in view of the large energies required to incorporate
the first CaCO3 unit, especially in competition with Mg ions
present in the solution, dolomite growth is not expected to oc-
cur at ambient temperatures. For completeness sake, energies
of mixed Mg and Ca edges with less regular arrangements of
the two cations were also calculated and it was found that once
MgCO3 and CaCO3 units had been incorporated at the obtuse
step, there is a small energetic advantage (10 kJ/mol) for the
next MgCO3 unit to grow next to the CaCO3 group rather than
next to the first MgCO3 unit. Hence, if a CaCO3 group did grow
next to the first MgCO3 unit, the next step would indeed lead to
a dolomite-type arrangement on the obtuse edge. However, on
the acute edge, the reverse happens and the second MgCO3
unit would preferentially grow next to the first MgCO3, rather
than next to the CaCO3 unit (by about 14 kJ/mol). In addition,
the last unit to be incorporated on both edges should be a CaCO3
group, but it is energetically far preferable to incorporate an-
other MgCO3 unit (–150 and –205 kJ/mol for the acute and
obtuse edges, respectively) than another CaCO3 unit (–51 and
–22 kJ/mol for the acute and obtuse edges, respectively). Thus,
these molecular dynamics calculations of dolomite growth at
two experimentally observed calcite growth steps show that
growth of MgCO3 at the steps is preferred over incorporation
of CaCO3 at all stages of the growth process and it would there-
fore be very unlikely for an ordered structure like dolomite to
be formed in this way. It is, however, clear that on thermody-
namic grounds MgCO3 growth on calcite will proceed very
easily, which explains why highly magnesian calcites occur
widely in natural sediments.
DISCUSSION
The calculated structural and elastic properties of the three
minerals are in good agreement with experimental values, even
though bulk moduli were not used in the derivation of the po-
tential parameters. The relative enthalpies of formation of the
calcite and magnesite crystals were fitted to experimental en-
thalpies and are therefore reproduced well, which was a prerequi-
site for the calculations of the varying stabilities of the pure and
disordered minerals and the energies of the growth processes.
The calculated surface energies for the three carbonate min-
erals show that, in general the dolomite surfaces are thermody-
namically more stable than their counterparts in magnesite, in
an aqueous environment the dolomite surfaces are less stable
than the surfaces of its constituent carbonates calcite and mag-
nesite. In addition, a disordered magnesian calcite is also cal-
culated to be energetically preferred over the ordered dolomite
structure compared to magnesite. The lesser stability of the hy-
–
drated {1014} dolomite surface is in part due to the strong Mg-
H2O interactions, which upon hydration lower the surface
energies of the magnesite surface far more than the calcite or
dolomite planes, together with the fact that the Mg ions present
in the dolomite surface are less accessible as a result of surface
relaxations and/or rotations of the carbonate groups. They are,
however, more accessible in the magnesian calcites, due to the
larger lattice spacing of the underlying calcite structure. The
–
ratio of surface energies of the dominant hydrated {1014} sur-
faces is gcal:gmag:gdol:gdisordered = 17:1:12:12, clearly showing the
unusual stabilities of magnesite (which is also due to the ex-
tensive network of hydrogen-bonded interactions between the
adsorbed water molecules) especially compared to calcite.
The hydration energies of the pure minerals are largest for
magnesite (at an average of 134 kJ/mol over all surfaces), fol-
lowed by dolomite and calcite (average 101 and 79 kJ/mol re-
spectively). However, whereas the average hydration energy
of the pure dolomite falls between magnesite and calcite, the
average hydration energy for the disordered magnesian calcite
is less at 94 kJ/mol. This lower average hydration energy for
the disordered structure is partly due to the greater lattice spac-
ing of magnesian calcite compared to both magnesite and do-
lomite, which makes the formation of a network of
hydrogen-bonded interactions between the adsorbed water
molecules impossible, and partly due to the more effective re-
laxation of the anhydrous disordered surfaces, which therefore
gain less on hydration than the magnesite and dolomite sur-
faces.
Molecular dynamics simulations of the competitive growth
of CaCO3 and MgCO3 at the two experimentally observed cal-
cite steps show that, at all stages, incorporation of MgCO3 is
more favorable than CaCO3, indicating that if enough Mg is
present a complete MgCO3 plane will form. When investigat-
ing the formation of dolomite-type layers from the steps, I find
a small energetic advantage for an ordered arrangement of Ca
and Mg ions on the obtuse step (10 kJ/mol), but not on the
acute step. As the obtuse step is experimentally found to grow
and dissolve more rapidly than the acute step (MacInnis and
Brantley 1992; Liang et al. 1996a, 1996b), this may show a route
to at least partial dolomite formation. Even so, however, the incor-
poration of MgCO3 rather than CaCO3 is far more exothermic and
hence, a mixed Mg-Ca carbonate would only form if Mg is present
at very small concentrations.
ACKNOWLEDGMENTS
N.H.dL. thanks NERC, grant no. NER/M/S/2001/00068, and the Royal
Society, grant no. 22292, for financial support.
REFERENCES CITED
Baker, P.A. and Kastner, M. (1981) Constraints on the formation of sedimentary
dolomite. Science, 213, 114–116.
Bertram, M.A., Mackenzie, F.T., Bishop, F.C., and Bischoff, W.D. (1991) Influence
of temperature on the stability of magnesian calcite. American Mineralogist,
76, 1889–1896.
Bischoff, J.L., Fitzpatrick, J.A., and Rosenbauer, R.J. (1993) The solubility and
stabilization of ikaite (CaCO3·6H2O) from 0 ∞C to 25 ∞C—environmental and
paleoclimatic implications for thinolite tufa. Journal of Geology, 101, 21–33.
Born, M. and Huang, K. (1954) Dynamical Theory of Crystal Lattices. Oxford Uni-
versity Press, Oxford.
Brady, P.V., Krumhansi, J.L., and Papenguth, H.W. (1996) Surface complexation
clues to dolomite growth. Geochimica et Cosmochimica Acta, 60, 727–731.
Clarkson, J.R., Price, T.J., and Adams, C.J. (1992) Role of metastable phases in the
spontaneous precipitation of calcium-carbonate. Journal of the Chemical Soci-
ety Faraday Transactions, 88, 243–249.
de Leeuw, N.H. and Parker, S.C. (1997) Atomistic simulation of the effect of mo-
lecular adsorption of water on the surface structure and energies of calcite sur-
faces. Journal of the Chemical Society Faraday Transactions, 93, 467–475.
———(1998a) Surface structure and morphology of calcium carbonate polymor-
phs calcite, aragonite and vaterite: an atomistic approach. Journal of Physical
Chemistry B, 102, 2914–2922.
———(1998b) Molecular-dynamics simulation of MgO surfaces in liquid water
using a shell-model potential for water. Physical Review B, 58, 13901–13908.
———(2000) Modelling absorption and segregation of magnesium and cadmium
ions to calcite surfaces: introducing MgCO3 and CdCO3 potential models. Jour-
 DE LEEUW: MODELING DOLOMITE AND MAGNESIAN CALCITES
nal of Chemical Physics, 112, 4326–4333.
de Leeuw, N.H., Parker, S.C., and Harding, J.H. (1999) Molecular Dynamics simu-
lation of crystal dissolution from calcite steps. Physical Review B, 60, 13792–
13799.
Deer, W.A., Howie, R.A., and Zussman, J. (1992) An Introduction to the Rock-
Forming Minerals, 641 p. Longman Group, Harlow, U.K.
Dick, B.G. and Overhauser, A.W. (1958) Theory of the dielectric constants of alkali
halide crystals. Physical Review, 112, 90–103.
Duan, Z., Moller, N., and Weare, J.H. (1995) Equation of state for the NaCl-H2O-
CO2 system—prediction of phase-equilibria and volumetric properties.
Geochimica et Cosmochimica Acta, 59, 3273–3283.
Fenter, P., Geissbuehler, P., DiMasi, E., Srajer, G., Sorensen, L.B., and Sturchio,
N.C. (2000) Surface speciation of calcite observed in situ by high-resolution X-
ray reflectivity. Geochimica et Cosmochimica Acta, 64, 1221–1228.
Ferry, J.M. (1996) Three novel isograds in metamorphosed siliceous dolomites from
the Ballachulish aureole, Scotland. American Mineralogist, 81, 485–494.
———(2000) Patterns of mineral occurrence in metamorphic rocks, American
Mineralogist, 85, 1573–1588.
———(2001) Calcite inclusions in forsterite. American Mineralogist, 86, 773–779.
Fiquet, G., Guyot, F., and Itie, J.P. (1994) High-pressure X-ray-diffraction study of
carbonates–MgCO3, CaMg(CO3)2, and CaCO3. American Mineralogist, 79, 15–
23.
Fisler, D.K. and Cygan, R.T. (1999) Diffusion of Ca and Mg in calcite. American
Mineralogist, 84, 1392–1399.
Fisler, D.K., Gale, J.D., and Cygan, R.T. (2000) A shell model for the simulation of
rhombohedral carbonate minerals and their point defects. American Mineralo-
gist, 85, 217–224.
Forester, T.R. and Smith, W. (1995) W. DL_POLY user manual. CCLRC, Daresbury
Laboratory, Daresbury, Warrington, U.K.
Gale, J.D. (1997) GULP: a computer program for the symmetry-adapted simulation
of solids. Journal of the Chemical Society Faraday Transactions, 93, 629–637.
Gratz, A.J., Hillner, P.E., and Hansma, P.K. (1993) Step dynamics and spiral growth
on calcite. Geochimica et Cosmochimica Acta, 57, 491–495.
Hillner, P.E., Manne, S., Hansma, P.K., and Gratz, A.J. (1993) Atomic Force Micro-
scope: A new tool for imaging crystal growth processes. Faraday Discussions,
95, 191–197.
Hoover, W.G. (1985) Canonical dynamics: equilibrium phase-space distributions.
Physical Review A, 31, 1695–1697.
Ito, T. (1998) Factors controlling the transformation of natural ikaite from Shiowakka,
Japan. Geochemical Journal, 32, 267–273.
Katz, A. and Matthew, A. (1977) The dolomitization of CaCO3: An experimental
study at 252–295∞C. Geochimica et Cosmochimica Acta, 41, 297–308.
Kessels, L.A., Sibley, D.F., and Nordeng, S.H. (2000) Nanotopography of synthetic
and natural dolomite crystals. Sedimentology, 47, 173–186.
Krauskopf, K.B. and Bird, D.K. (1995) Introduction to Geochemistry. McGraw–
Hill Inc. New York, 79.
Letargo, C.M.R., Lamb, W.M., and Park, J.S. (1995) Comparison of calcite plus
dolomite thermometry and carbonate plus silicate equilibria —constraints on
the conditions of metamorphism of the Llano uplift, central Texas, USA. Ameri-
can Mineralogist, 80, 131–143.
Liang, Y., Baer, D.R., McCoy, J.M., Amonette, J.E., and LaFemina, J.P. (1996a)
Dissolution kinetics at the calcite-water interface. Geochimica et Cosmochimica
Acta, 60, 4883–4887.
Liang, Y., Lea, A.S.,
– Baer, D.R., and Engelhard, M.H. (1996b) Structure of the cleaved
CaCO3 (1014) surface in an aqueous environment. Surface Science, 351, 172–
182.
Lide, D.R. (2000) CRC Handbook of Chemistry and Physics. CRC Press, Boca
Raton.
Liebermann, O. (1967) Synthesis of dolomite. Nature, 213, 241–245.
MacInnis, I.N. and Brantley, S.L. (1992) The role of dislocations and surface mor-
phology in calcite dissolution. Geochimica et Cosmochimica Acta, 56, 1113–
1126.
Martinez, I., Zhang, J.Z., and Reeder, R.J. (1996) In situ X-ray diffraction of arago-
689
nite and dolomite at high pressure and high temperature: Evidence for dolomite
breakdown to aragonite and magnesite. American Mineralogist, 81, 611–624.
Nordeng, S.H. and Sibley, D.F. (1994) Dolomite stoichiometry and Ostwald step
rule. Geochimica et Cosmochimica Acta, 58, 191–196.
Nosé, S. (1984) A unified formulation of the constant temperature molecular-dy-
namics methods. Journal of Chemical Physics, 81, 511–519.
Oliver, P.M., Parker, S.C., Egdell, R.G., and Jones, F.H. (1996) Computer simula-
tion of the surface structures of WO3. Journal of the Chemical Society Faraday
Transactions, 92, 2049–2056.
Paquette, J., Vali, H., and Mountjoy, E. (1999) Novel TEM approaches to imaging
of microstructures in carbonates: Clues to growth mechanisms in calcite and
dolomite. American Mineralogist, 84, 1939–1949.
Park, N.-S., Kim, M.-W., Langford, S.C., and Dickinson, J.T. (1996) Atomic layer
wear of single-crystal calcite in aqueous solution scanning force microscopy.
Journal of Applied Physics, 80, 2680–2686.
Parker, S.C. and Price, G.D. (1989) Computer modelling of phase transitions in
minerals. Advances in Solid State Chemistry, 1, 295–327.
Pavese, A., Catti, M., Parker, S.C., and Wall, A. (1996) Modelling of the thermal
dependence of structural and elastic properties of calcite, CaCO3. Physics and
Chemistry of Minerals, 23, 89–93.
Pokrovsky, O.S., Schott, J., and Thomas, F. (1999a) Processes at the magnesium-
bearing carbonates/solution interface. I. A surface speciation model for magne-
site. Geochimica et Cosmochimica Acta, 63, 863–880.
———(1999b) Dolomite surface speciation and reactivity in aquatic systems.
Geochimica et Cosmochimica Acta, 63, 3133–3143.
Reeder, R.J. (1990) Crystal chemistry of the rhombohedral carbonates. In Reeder,
R.J., Ed., Carbonates: Mineralogy and Chemistry, 11, 1–48. Reviews in Miner-
alogy, Mineralogical Society of America, Washington, D.C.
Redfern, S.A.T. and Angel, R.J. (1999) High-pressure behaviour and equation of
state of calcite, CaCO3. Contributions to Mineralogy and Petrology, 134, 102–
106.
Ross, N.L. (1997) The equation of state and high-pressure behavior of magnesite.
American Mineralogist, 82, 682–688.
Ross, N.L. and Reeder, R.J. (1992) High-pressure structural study of dolomite and
ankerite. American Mineralogist, 77, 412–421.
Schubel, K.A., Elbert, D.C., and Veblen, D.R. (2000) Incommensurate c-domain
superstructures in calcian dolomite from the Latemar buildup, Dolomites, North-
ern Italy. American Mineralogist, 85, 858–862.
Sibley, D.F., Nordeng, S.H., and Borkowski, M.L. (1994) Dolomitization kinetics
in hydrothermal bombs and natural settings. Journal of Sedimentary Research
Section A-Sedimentary Petrological Processes, 64, 630–637.
Tasker, P.W. (1979) The surface energies, surface tensions and surface structure of
the alkali halides. Philosophical Magazine A, 39, 119–136.
Titiloye, J.O., de Leeuw, N.H., and Parker, S.C. (1998) Atomistic simulation of the
differences between calcite and dolomite surfaces. Geochimica et Cosmochimica
Acta, 62, 2637–2641.
Verlet, L. (1967) Computer ‘experiments’ on classical fluids. I. Thermodynamical
properties of Lennard-Jones molecules. Physical Review, 159, 98–103.
Watson, G.W., Kelsey, E.T., de Leeuw, N.H., Harris, D.J., and Parker, S.C. (1996)
Atomistic simulation of dislocations, surfaces and interfaces in MgO. Journal
of the Chemical Society Faraday Transactions, 92, 433–438.
Wenk, H.R. (1993) Partially disordered dolomite—microstructural characterization
of Abu-Dhabi Sabkha carbonates. American Mineralogist, 78, 769–774.
Williams, Q., Collerson, B., and Knittle, E. (1992) Vibrational-spectra of magnes-
ite (MgCO3) and calcite-III at high-pressures. American Mineralogist, 77, 1158–
1165. –
Wright, K., Cygan, R.T., and Slater, B. (2001) Structure of the (1014) surfaces of
calcite, dolomite and magnesite under wet and dry conditions. Physical Chem-
istry and Chemical Physics, 3, 839–844.
MANUSCRIPT RECEIVED AUGUST 17, 2001
MANUSCRIPT ACCEPTED DECEMBER 24, 2001
MANUSCRIPT HANDLED BY MICHAEL E. FLEET