P1Freely available online through the publisher-supported open access option.


Natural hydromagnesite (Mg5(CO3)4(OH)2·4H2O) dissolution and precipitation experiments were performed in closed-system reactors as a function of temperature from 22.5 to 75°C and at 8.6 < pH < 10.7. The equilibrium constants for the reaction Mg5(CO3)4(OH)2·4H2O + 6H+ = 5Mg2+ + 4HCO3 + 6H2O were determined by bracketing the final fluid compositions obtained from the dissolution and precipitation experiments. The resulting constants were found to be 1033.7±0.9, 1030.5±0.5 and 1026.5±0.5 at 22.5, 50 and 75°C, respectively. Whereas dissolution rates were too fast to be determined from the experiments, precipitation rates were slower and quantified. The resulting BET surface area-normalized hydromagnesite precipitation rates increase by a factor of ~2 with pH decreasing from 10.7 to 8.6. Measured rates are approximately two orders of magnitude faster than corresponding forsterite dissolution rates, suggesting that the overall rates of the low-temperature carbonation of olivine are controlled by the relatively sluggish dissolution of the magnesium silicate mineral.


The precipitation rates of carbonate minerals are of current interest due to their potential role in carbon storage (e.g. Seifritz, 1990; Lackner et al., 1995; Xu et al., 2005; Marini, 2007; Flaathen et al., 2011). Carbon storage via carbonate mineral precipitation appears to be particularly favoured in basaltic and ultramafic rocks due to their high reactivity and the prevalence of divalent cations such as Ca2+ and Mg2+ (McGrail et al., 2006; Alfredsson et al., 2008; Oelkers et al., 2008; Matter et al., 2009, 2011; Schaef et al., 2010, 2011; Gislason et al., 2010; Gysi and Stefansson, 2012; Aradóttir et al., 2012; Oskierski et al., 2013; Gislason and Oelkers, 2014). Whereas anhydrous calcium carbonates such as calcite and aragonite precipitate readily at ambient temperatures (e.g. Burton and Walter, 1987; Zuddas and Mucci, 1998; Teng et al., 2000; Lakshtanov and Stipp, 2010; Rodriguez-Blanco et al., 2011; Ruiz-Agudo et al., 2013), the precipitation of the anhydrous Mg-carbonate mineral magnesite is apparently inhibited at temperatures of <~80°C (e.g. Saldi et al., 2009, 2012). At these lower temperatures, the precipitation of hydrous Mg carbonates, such as hydromagnesite and dypingite are favoured (Mavromatis et al., 2012; Shirokova et al., 2013). The goal of the present study was an improved understanding of the precipitation kinetics of hydrous magnesium carbonate minerals at temperatures below 80°C, in part to assess their potential to store carbon at ambient temperatures. Hydromagnesite growth rates were measured in closed-system reactors at 22.5–75°C. The results of these experiments are used here to evaluate the potential of hydrous magnesium carbonate minerals as a carbon-storage host.

Theoretical background

The standard state adopted in this study for thermodynamic calculations is that of unit activity for pure minerals and H2O at any temperature and pressure. For aqueous species other than H2O, the standard state is unit activity of the species in a hypothetical 1 molal solution referenced to infinite dilution at any temperature and pressure. Hydromagnesite dissolution can be described using:  
Taking account of this standard state, the law of mass action for reaction 1 is given by:  
where KHmg stands for the equilibrium constant of reaction 1 and ai represents the activity of the subscripted aqueous species. The chemical affinity (A) of reaction 1 can be expressed as:  
where R is the gas constant (= 8.3144 J/Kmol), T is absolute temperature, IAPHmg represents the aqueous ion activity product, and ΩHmg represents the saturation state of the aqueous phase with respect to hydromagnesite. All thermodynamic calculations in this study were performed using the PHREEQC for Windows (Version 2.18) computer code (Parkhurst and Appelo, 1999) together with its llnl database after adding the equilibrium constants for Mg2+ hydrolysis and the carbonic acid dissociation reported by Brown et al. (1996) and Millero et al. (2007), respectively.

Materials and methods

Natural hydromagnesite (Mg5(CO3)4(OH)2·4H2O) was collected as stromatolite precipitates from Lake Salda in southwest Turkey (cf. Shirokova et al., 2011) then hand-milled with an agate mortar and pestle. The resulting powder was cleaned with H2O2 to remove organics and oven dried at 50°C for 24 h. The particles were not cleaned ultrasonically due to the fine nature of these natural grains. Scanning Electron Microscope (SEM) images, obtained using a JEOL JSM-6360 LV microscope, of the resulting hydromagnesite are shown in Fig. 1. The hydromagnesite powder prepared consists of agglomerated crystals. The size of the individual hydromagnesite crystals ranges up to ~5 μm, the agglomerates range up to 50 μm. No other mineral phases were evident from SEM images nor were any detected using backscattered electron microscopy. The purity of this hydromagnesite was further verified by X-ray diffraction using an INEL CPS-120 diffractometer with CoKα radiation, λ = 1.78897 Å, and a graphite monochromator. X-ray diffraction was performed from 1 to 110°2θ at 0.09°2θ/min and at a step size of 0.029°2θ. This analysis revealed no phase other than hydromagnesite. The surface area of the hydromagnesite prepared was determined to be 8.5 m2/g ± 10% by multipoint krypton adsorption according to the BET method (Brunauer et al., 1938) using a Quantachrome Gas Sorption system.

Hydromagnesite dissolution and precipitation experiments were performed in closed-system 0.5 l high-density polypropylene reactors. Approximately 1 g of hydromagnesite powder and 500 g of an aqueous sodium carbonate buffer solution with ionic strength of 0.1 mol/kg were placed into each reactor. The reactors were subsequently sealed and placed in temperature-controlled shaking water baths. The initial solutions were composed of deionized H2O and Merck reagent-grade Na2CO3 and NaHCO3 at varying ratios (Table 1); the resulting aqueous solutions had a pH ranging from 7.7 to 11.2. The original hydromagnesite was allowed to dissolve in this fluid for about 5 days until a steady-state aqueous Mg concentration was attained and validated. After this time, ~3 g of a 1.81 mol/kg aqueous MgCl2 solution was added to each reactor to supersaturate the fluid with respect to hydromagnesite.

The fluid in each reactor was sampled regularly to monitor reaction progress. The reactive fluid Mg concentration was measured by flame atomic absorption spectroscopy using a Perkin Elmer Zeeman 5000 Atomic Absorption Spectrometer with an uncertainty of ±2% and a detection limit of 6 × 10−7 mol/kg. Alkalinity was determined by standard HCl titration using Schott TA 10plus with an uncertainty of ±2% and a detection limit of 5 × 10−5 eq/l. Fluid pH measurements were performed at 22.5°C immediately after sampling using a standard glass electrode, previously calibrated with 4.01, 6.86 and 9.18 NIST pH buffers.


Aqueous Mg concentrations as a function of time during a representative experiment are illustrated in Fig. 2. An approximately constant Mg concentration was obtained by dissolution within ~30 min. Following the addition of MgCl2 to the fluid phase, the aqueous Mg concentration decreased systematically with time. Note the near-constant aqueous Mg concentration attained at the end of the dissolution leg is distinct from that attained at the end of the precipitation leg of the experiment, reflecting the different fluid pH present during these legs.

The rate at which steady-state was attained during hydromagnesite dissolution was too rapid to obtain unambiguous dissolution rates from the temporal evolution of aqueous Mg concentrations (see Fig. 2). In contrast, the temporal evolution of aqueous Mg concentrations during precipitation was sufficiently slow to fit these data thereby retrieving hydromagnesite rates. Surface area-normalized hydromagnesite precipitation rates (rHmg) were fitted assuming they were consistent with surface growth in accord with:  
where kHmg refers to a rate constant, ΩHmg represents the saturation state of the aqueous phase with respect to the precipitating hydromagnesite, ν refers to the stoichiometric number of activated species in one mole of hydromagnesite (= 5; Gautier et al., 2014) and n stands for the reaction order (= 1; Gautier et al., 2014). Measured aqueous Mg concentrations were fitted following the method described by Harouiya et al. (2007) using equation 4 and a trial-and-error method by varying kHmg and comparing fluid concentrations calculated using an EXCEL spread sheet with measured aqueous Mg concentrations. The results of this fit for all experiments are shown in Fig. 3, where it can be seen that the calculated curves provide a close description of the temporal variation of the measured aqueous Mg concentrations. Note that in some cases the first few measured Mg concentrations during a precipitation leg plot below the curve. This observation may be due to the effects of hydromagnesite nucleation on aqueous fluid concentrations. Nucleation, which is favoured in highly supersaturated fluids, is typically faster than growth and more strongly dependent on fluid saturation state. As such, aqueous Mg concentrations in fluids experiencing hydromagnesite nucleation will decrease faster with time than Mg concentrations in fluids experiencing only growth.

Hydromagnesite precipitation-rate constants, consistent with the regression curves shown in Fig. 3, are provided as a function of pH in Fig. 4. (1) Hydromagnesite precipitation-rate constants decrease by a factor of ~2 with increasing pH from 8.6 to 10.7 at 22.5°C. A similar decrease in precipitation rates with pH was observed for the case of magnesite at 100 to 200°C by Saldi et al. (2012). (2) Hydromagnesite precipitation-rate constants increase slightly with increasing temperature from 22.5 to 75°C.

The final steady-state aqueous Mg concentrations during the dissolution and precipitation leg of each experiment were used together with corresponding pH and alkalinity values to determine the final IAP of hydromagnesite. The resulting IAP values are illustrated as a function of pH in Fig. 5. As these values were obtained from both dissolution and precipitation experiments, they bracket the equilibrium constant of the hydromagnesite hydrolysis reaction 1. The equilibrium IAP values shown in Fig. 5 were determined by the requirement that the equilibrium IAP be greater than those measured for dissolution and lower than those measured from precipitation. In each case, the equilibrium IAP was chosen as the largest highest IAP value determined from dissolution and the lowest determined from precipitation. There is some overlap among the measured IAP values for the 75°C experiments, possibly due to uncertainties in the thermodynamic database adopted in this study. The logarithms of equilibrium constants for reaction 1 generated by equating KHmg to this equilibrium IAP (see equation 3) are 33.7±0.9, 30.5±0.5 and 26.5±0.5, for 22.5, 50 and 75°C, respectively.


The role of hydromagnesite precipitation rates in mineral carbonation

The carbonation of Mg-rich olivine dissolution has been studied widely (e.g. Béarat et al., 2006; Andreani et al., 2009; Prigiobbe et al., 2009; King et al., 2010; Daval et al., 2011). Olivine is commonly thought to be the best source of the divalent metal cations required to create carbonate minerals due to its fast dissolution rates and worldwide abundance. The carbonation of Mg-enriched fluids due to forsterite dissolution at low temperatures, where magnesite precipitation is kinetically not favoured (e.g. Saldi et al., 2009), can proceed via:  
Forsterite dissolution has been shown to follow a relatively simple dissolution mechanism (Pokrovsky and Schott, 2000; Oelkers, 2001); dissolution rates of forsterite have been reported by Wogelius and Walther (1991), Pokrovsky and Schott (2000), Oelkers (2001) and Hänchen et al. (2006) as a function of reactive fluid pH. At steady state, the stoichiometry of reaction 5 requires:  
where rFo and sFo refer to the surface area-normalized dissolution rate and surface area of forsterite, respectively, and sHmg stands for the surface area of hydromagnesite. A comparison of surface area-normalized forsterite dissolution rates and hydromagnesite precipitation rates is shown in Fig. 6. Forsterite dissolution rates after normalization to its Mg content are ~2 orders of magnitude slower than corresponding hydromagnesite precipitation rates. This observation indicates that, in contrast to the carbonation of forsterite to form magnesite (Saldi et al., 2009), the limiting step of the carbonation process to form hydromagnesite from forsterite is the relatively slow dissolution of forsterite. As forsterite dissolution rates are not increased by changing saturation state at far-from-equilibrium conditions (cf. Oelkers et al., 2001) or substantially by the addition of organic ligands (Declercq et al., 2013), it seems likely that the most efficient way of accelerating this overall process would be by increasing forsterite reactive surface areas (e.g. by increased grinding leading to reduced particle sizes).

Comparison with hydromagnesite hydrolysis constants available in the literature

As experiments reported in this study bracket equilibrated hydromagnesite from both undersaturated and supersaturated conditions, the results presented above constrain strongly the equilibrium constant of reaction 1. There have been few reports of hydromagnesite solubility measurements in the literature. Data present in the llnl database of PHREEQC originate from SUPCRT92 (Johnson et al., 1992), which takes account of the thermodynamic properties of hydromagnesite reported by Helgeson et al. (1978), which themselves are based on the calorimetric measurements of Robie and Hemingway (1972, 1973). The equilibrium constants for reaction 1 as calculated using PHREEQC and its llnl database as a function of temperature are plotted in Fig. 7. Gautier et al. (2014) reported the hydromagnesite hydrolysis constants based on dissolution experiments performed on natural abiotic hydromagnesite. Measured equilibrium constants on natural biologically induced hydromagnesite, as determined in the present study, are 1–3 orders of magnitude greater than those generated from the Robie and Hemingway (1972, 1973) calorimetric data. Although these differences may be within the uncertainty of the calculations (c.f. Oelkers et al., 2009), such differences may also result from the presence of small hydromagnesite grains. Hydrolysis constants generated in this study are less than an order of magnitude different from those reported by Gautier et al. (2014), confirming these previous results, and suggesting a similar stability of biotic vs. abiotic hydromagnesite.

The variation of hydromagnesite precipitation rates as a function of temperature

The hydromagnesite precipitation rates illustrated in Fig. 4 suggest that there is little effect of increasing temperature on rates from 22.5 to 75°C. This atypical observation can stem from a decrease in the number of active sites on the hydromagnesite surface in the higher-temperature experiments. A potential mechanism for the decrease in reactive surface area with increasing temperature is the agglomeration of hydromagnesite particles during precipitation. Particle agglomeration decreases rates by decreasing the hydromagnesite surface area in contact with the reactive fluid. Indeed, SEM images taken after the 22.5 and 75°C experiments can be compared in Fig. 1, where it can be seen that the hydromagnesite recovered following the higher-temperature experiments are larger, potentially reflecting the increased agglomeration of these grains. Note also that Pokrovsky et al. (2009) and Schott et al. (2009) reported that carbonate mineral dissolution rates even decreased with increasing temperature from 100 to 150°C. They attributed this atypical behaviour to changes in the surface speciation of carbonate minerals with increasing temperature.

A comparison between hydromagnesite precipitation-rate constants generated in the present study and those of Gautier et al. (2014) is also provided in Fig. 4. Rate constants generated in the present study on biogenic hydromagnesite are ~20 times faster than corresponding rate constants reported by Gautier et al. (2014) on abiotic hydromagnesite. This difference suggests that biogenic hydromagnesite has far more active sites per unit surface area than its abiotic counterpart. Comparison of photomicrographs presented by Gautier et al. (2014) and those from the present study shows the abiotic hydromagnesite to consist of euhedral and larger crystals. Note that rates at 75°C reported for the biotic and abiotic hydromagnesite are similar in value, but as mentioned above, rates generated at this temperature in the present study may have been reduced due to agglomeration.


The results summarized above confirm the rapid dissolution and precipitation kinetics of hydromagnesite. Hydromagnesite dissolution rates were so rapid that they could not be quantified accurately from the closed-system reactor experiments performed in this study. Due to its rapid reaction rates, hydromagnesite should equilibrate quickly with low- and ambient-temperature natural fluids. The fast precipitation rates of hydromagnesite suggest that it could be an efficient product phase for the ex situ and in situ sequestration of CO2.


The authors thank Vasilis Mavromatis for insightful discussions and encouragement, Carol Causserand for her help with wet-chemical analysis, Thierry Aigouy for technical support in obtaining SEM images and Michell Thibaut for X-ray analysis. This study was supported by the Centre National de la Recherche Scientifique (CNRS), the European Commission (through Marie Cuire ITN project MINSC, 290040; and PROCOPE – DAAD 55923335/Egide 28469SD).

This paper is published as part of a special issue in Mineralogical Magazine, Vol. 78(6), 2014 entitled ‘Mineral–fluid interactions: scaling, surface reactivity and natural systems’.