Chemical Engineering Science 63 (2008) 1012–1028www.elsevier.com/locate/ces

Precipitation in the Mg-carbonate system—effects of temperatureand CO2 pressure

Markus Hänchen, Valentina Prigiobbe, Renato Baciocchi1, Marco Mazzotti∗

ETH Zurich, Institute of Process Engineering, Sonneggstrasse 3, CH-8092 Zurich, Switzerland

Received 28 June 2007; received in revised form 27 September 2007; accepted 29 September 2007Available online 17 October 2007

Abstract

The precipitation of different forms of magnesium carbonate has been studied at temperatures between 25 and 120 ◦C and at a partialpressure of CO2 between 1 and 100 bar. These conditions are relevant for mineral carbonation applications. Precipitation was triggered bythe supersaturation created by mixing Na2CO3 solutions in equilibrium with a CO2 atmosphere with MgCl2 solutions. Experiments weremonitored using attenuated total reflection Fourier transform infrared (ATR-FTIR) and Raman spectroscopy as well as a focused beam reflectancemeasurement (FBRM) probe and a turbidimeter. Solubility and supersaturation were calculated using the software package EQ3/6. Solidswere identified using X-ray diffraction (XRD) analysis and scanning electron microscope (SEM) images. At 25 ◦C and PCO2 = 1 bar, only thehydrated carbonate nesquehonite (MgCO3 · 3H2O) precipitates, as it has previously been observed. Solutions undersaturated with respect tonesquehonite did not form any precipitates in experiments lasting 16 h. Induction times increased with decreasing supersaturation with respectto nesquehonite. At 120 ◦C and PCO2 = 3 bar, hydromagnesite ((MgCO3)4 · Mg(OH)2 · 4H2O) was formed which transformed within 5–15 hinto magnesite (MgCO3). Solutions undersaturated with respect to brucite (Mg(OH)2) did not form any precipitates in experiments lasting 19 h.At 120 ◦C and PCO2 = 100 bar, direct formation of magnesite and, at elevated levels of supersaturation, the co-precipitation of magnesite andhydromagnesite has been observed. In the latter case, hydromagnesite transformed within a few hours into magnesite. Solutions undersaturatedwith respect to hydromagnesite did not form any precipitates in experiments lasting 20 h.� 2007 Elsevier Ltd. All rights reserved.

Keywords: Carbon dioxide capture and storage; Phase equilibria; Particulate formation; Kinetics; Nucleation; Raman spectroscopy; Precipitation

1. Introduction

Being natural minerals, the properties and the formation ofMg-carbonates have been studied for a very long time already.Nevertheless, open questions still remain both in regard to theformation of Mg-carbonates in natural systems as well as totheir production in the laboratory and in industrial processes.The complexity of the Mg-carbonate system, with numerousdifferent forms of hydrated or basic carbonates, is certainly onereason for the lack of a complete understanding of all the is-sues involved in the formation of Mg-carbonates. Industrial pro-duction processes for Mg-carbonate have only received limited

∗ Corresponding author. Tel.: +41 44 632 2456; fax: +41 44 632 1141.E-mail address: marco.mazzotti@ipe.mavt.ethz.ch (M. Mazzotti).

1 Permanent address: Department of Civil Engineering, University of RomeTor Vergata, Via del Politecnico 1, I-00133 Rome, Italy.

0009-2509/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.doi:10.1016/j.ces.2007.09.052

attention due to the low value of the resulting product, whichfor most uses can be substituted either by natural material orby other minerals like Ca-carbonate. Most often, no particularform of Mg-carbonate is required and, consequently, industrialprocesses often produce low purity mixtures of differentMg-carbonate forms.

However, efforts to counter global warming have broughtforward a new use for Mg-carbonate, for it to serve as a meansto store anthropogenically generated carbon dioxide in a solid,stable and long-lasting form in order to prevent its accumula-tion in the atmosphere. The process to achieve this, mineral car-bonation, can be part of a Carbon Dioxide Capture and Storage(CCS) system, which attempts to capture CO2 from industrialsources and to store it, so as to keep it as permanently as possi-ble isolated from the atmosphere. The most successful imple-mentation of the process of mineral carbonation involves theextraction of magnesium from Mg-bearing silicate rocks like

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1013

olivine or serpentine in aqueous solutions and the subsequentprecipitation of Mg-carbonate under a CO2 atmosphere (IPCC,2005; O’Connor et al., 2005).

In this work, besides a brief overview of various processesto produce Mg-carbonate under laboratory or industrial condi-tions, we study the operating conditions required to producedifferent forms of Mg-carbonate. Using various in situ mea-surement techniques like Raman spectroscopy, we presentdetailed experimental results of the formation of three differ-ent forms of Mg-carbonate: nesquehonite (MgCO3 · 3H2O),hydromagnesite ((MgCO3)4 · Mg(OH)2 · 4H2O) and mag-nesite (MgCO3), and discuss the data, interpreting themin the light of the observed kinetics and of the underlyingthermodynamics.

2. The Mg-carbonate system

Like with other carbonate systems, in the MgO–CO2–H2Osystem a number of hydrated and basic carbonates can beformed, all of which, however, are metastable compounds. A listof possible forms of Mg-carbonates is given in Table 1. Com-monly, all forms containing hydroxy groups (i.e., Mg(OH)2) arecalled basic carbonates since they form alkaline solutions upontheir dissolution.The stable phases in equilibrium with aqueoussolutions containing dissolved Mg-ions and carbon dioxide areeither brucite (Mg(OH)2, magnesium hydroxyde), at very lowpartial pressures of CO2, or magnesite (MgCO3) as illustratedfor the temperature range between 0 and 60 ◦C by Königsbergeret al. (1999), who based their calculations on calorimetricallymeasured thermodynamic properties and solubility data, usinga Pitzer model for the activity coefficients.

Using the software package EQ3/6 (Wolery, 1992) and adatabase employing a Debye–Hückel model for the activitycoefficient determination, it is possible to calculate thermody-namic properties over a wide range of temperature and pres-sure. The model includes two aqueous magnesium complexes,MgHCO+

3 (aq) and MgCO03(aq). The standard state used is one

molal, i.e., one mol per kg of water. The reference state that

Table 1List of selected minerals in the MgO–CO2–H2O system

Mineral Composition Activity product Q �G0f (kJ mol−1)

Brucite Mg(OH)2aMga

2OH

asolid−835.32

Magnesite MgCO3

aMgaCO2−3

asolid−1027.83

Nesquehonite MgCO3 · 3H2OaMgaCO2−

3a3

H2O

asolid−1723.95

Lansfordite MgCO3 · 5H2OaMgaCO2−

3a5

H2O

asolid−2199.20

Artinite MgCO3 · Mg(OH)2 · 3H2Oa2

MgaCO2−3

a2OHa3

H2O

asolid−2568.62

Hydromagnesite (MgCO3)4 · Mg(OH)2 · 4H2Oa5

Mga4CO2−

3a2

OHa4H2O

asolid−5864.66

In the expression of the activity product Q, ai is the activity of specified species and asolid is the activity of the relevant mineral. Values for �G0f are those

in the cmp database of the software package EQ3/6 (Wolery, 1992).

of infinite dilution. Phase diagrams based thereon are shownin Fig. 1; the solubilities of the different compounds are dis-played in Fig. 2. The results extend the model presented byKönigsberger et al. (1999) to higher temperature and pres-sure levels, i.e., more relevant for mineral carbonation. It alsoshows that magnesite is the most stable Mg-carbonate under allconditions.

In early studies of the MgO–CO2–H2O system, reported val-ues for the Gibbs free energy of magnesite differed noticeably(see Kittrick and Peryea, 1986, for a discussion). When tak-ing the upper end of the reported values, the result was thatmagnesite was the most stable Mg-carbonate compound onlyat elevated temperatures. One of the first authors publishing acomprehensive study of the MgO–CO2–H2O system, Langmuir(1965), therefore concluded magnesite to be unstable below60 ◦C at a partial pressure of CO2 of 1 bar. Numerous laterstudies have since disproved this assertion, showing magnesiteto be the most stable carbonate at all temperatures and par-tial pressures of CO2 (e.g., Christ and Hostetle, 1970; Kittrickand Peryea, 1986). The earlier conclusion about a greater sta-bility of the hydrated forms of Mg-carbonate, hydromagnesite,nesquehonite, lansfordite and artinite, at lower temperatures, isnow considered to be an artifact due to kinetic inhibitions inthe formation of magnesite.

3. Production of different forms of Mg-carbonate

Despite being the stable carbonate form, the formation ofmagnesite at ambient temperature is virtually impossible. Mostcommonly, only the mineral nesquehonite can be precipitatedfrom aqueous solutions at 25 ◦C and moderate partial pressureof CO2, i.e., close to ambient pressure or below (Dell andWeller, 1959; Davies and Bubela, 1973; Ming and Franklin,1985; Kloprogge et al., 2003; Zhang et al., 2006). At highertemperatures, i.e., above approximately 40 ◦C, various basicMg-carbonates are usually formed by precipitation, mostly inthe form of hydromagnesite (Dell and Weller, 1959; Daviesand Bubela, 1973; Fernandez et al., 2000; Zhang et al., 2006).

1014 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

-3 -2 -1 0 1 2

0

50

100

150

200

0

50

100

150

200

-3 -2 -1 0 1 2

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

T [

° C]

T [

° C]

0

50

100

150

200

0

50

100

150

200

T [

° C]

T [

° C]

Brucite

Magnesite

Brucite

(Hydromagnesite)

Brucite

(Artinite)

(Nesquehonite)

Brucite

(Nesquehonite)

Fig. 1. Phase diagrams for the system MgO–CO2–H2O. (a) Stable equilibrium of brucite with magnesite (nothing suppressed). (b) Magnesite was suppressed inthe calculations. Equilibrium between brucite and hydromagnesite. (c) Magnesite and hydromagnesite were suppressed in the calculations. Equilibria betweenbrucite, artinite and nesquehonite. (d) Magnesite, hydromagnesite and artinite were suppressed. Equilibrium between brucite and nesquehonite.

Fig. 2. Solubilities of different magnesium carbonates and magnesium hydroxide (brucite, Mg(OH)2) in solution with no foreign ions. From top to bottom,nesquehonite, hydromagnesite and magnesite, white surface is brucite.

In fact, the transformation from nesquehonite to hydromag-nesite has been widely reported at temperatures between50 and 100 ◦C, both for a suspension of particles and for

the dry compound, if at least some residual water vapor ispresent in the latter case (Dell and Weller, 1959; Davies andBubela, 1973).

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1015

The minimum reported temperature for the production of theanhydrous form, magnesite, is about 60–100 ◦C, its formationalso requiring the presence of an elevated CO2 pressure (Baronand Favre, 1958; Baron, 1958; Deelman, 2001; Giammar et al.,2005). Reportedly, the formation of magnesite is most often ei-ther preceded by the formation of hydromagnesite or observedonly in suspensions of solid hydromagnesite, i.e., with the an-hydrous carbonate being the result of a transformation fromhydromagnesite to magnesite (Sayles and Fyfe, 1973; Stevulaet al., 1978). This transition can be very slow, hence, below150 ◦C and at moderate pressures its duration is often in the or-der of days. In addition to higher temperatures, also high salin-ity, high CO2 pressure and low magnesium concentrations (forclosed systems) are known to accelerate the (trans)formation(Sayles and Fyfe, 1973). Zhang (2000) reported transforma-tion times between 2 h at 200 ◦C in a solution saturated withNaCl, and over 100 days at 110 ◦C and lower salinity in aclosed system. Magnesite formation without an apparent ini-tial or intermediate hydromagnesite presence has been reportedunder high CO2 pressure, for experiments coupling the disso-lution of Mg-silicates with carbonate precipitation. Using a mi-croreactor equipped with synchrotron X-ray diffraction (XRD)and Raman spectroscopy, magnesite formation was detected atPCO2 =150 bar and temperatures between 150 and 180 ◦C (Wolfet al., 2004). Production of magnesite had already been reportedearlier under similar conditions (O’Connor et al., 2002).

3.1. Mechanisms of Mg-carbonate formation

Mg-carbonate precipitation is strongly kinetically controlled,as the sequence of formation of the different compounds is gov-erned by kinetic constraints. The difficulty in precipitating theanhydrous Mg-carbonate (i.e., magnesite) is often attributed tothe highly hydrated character of Mg2+ ions in solution (Saylesand Fyfe, 1973; Deelman, 2001). As it was stated, hydratedforms of magnesium carbonate precipitate much more easilythan the anhydrous form, and therefore magnesite formationis often either preceded by the precipitation of hydromagne-site or of other forms of hydrated carbonates, or possibly evenby the initial formation of brucite. With increasing temperaturethis transformation and the direct precipitation of magnesitebecome fast enough to allow for the formation of magnesitewithout a discernable appearance of the hydrated phases.

The initial appearance of thermodynamically unstable phasesis a common phenomenon and often referred to as the Oswaldrule of stages (Mullin, 1993, p. 200). In the case of magnesiumcarbonates, the initial formation of highly hydrated compounds,such as nesquehonite, being replaced by less hydrated formssuch as hydromagnesite, can be partially explained by the factthat water molecules, being dipoles having strong electrostaticinteraction with all ions, form a barrier around the Mg-cationsand are easily incorporated into the carbonate structure.

The Mg-cation has a highly ionic character because it be-longs to the second group of the periodic table. In particular,magnesium possesses strongly bonded layers of water dipoles,Bol et al. (1970) distinguished six water molecules in primaryhydration and 12 water molecules in secondary hydration, and

only the secondary hydration numbers can be influenced bychanging the state variables of the system (e.g., the concentra-tion or the temperature) (Burgess, 1978).

This high hydration energy is due to the high ratio betweenthe electrical charge and the ion surface area of Mg2+. There-fore the reduction of water activity as in strongly saline envi-ronments could favor the dehydration by reducing the thicknessof the secondary hydration layer (Christ and Hostetle, 1970).

Since CO2−3 ions have a larger ion radius and are not com-

pletely surrounded by water, they are adsorbed on the crystalsurface easier and, therefore, a low Mg2+ to CO2−

3 ion ratio,ri = [Mg2+]/[CO2−

3 ], might help to keep the water moleculesaway from the crystal surface.

The hydrated compounds precipitated first can trans-form following two possible mechanisms: shrinkage (thatis, dehydration) and dissolution–precipitation solvent me-diated transformation. Two examples are brought herein:nesquehonite–hydromagnesite transition studied by Davies andBubela (1973), Lanas and Alvarez (2004), and Zhang et al.(2006), and hydromagnesite–magnesite transition studied byZhang (2000).

4. Experimental

4.1. Experimental methods

Two different experimental set-ups were used in this study,namely a stirred 300 ml titanium high pressure reactor forhigh temperature experiments (Fig. 3), and a 500 ml glassreactor for experiments at ambient pressure. A stirring rateof 300 rpm was used in all experiments. They have been de-scribed more in detail previously (Hänchen et al., 2006; Schöllet al., 2007). Precipitation of magnesium carbonate from theH2O–CO2–Na2CO3–MgCl2 system was investigated in exper-iments of three different types, i.e., under different operatingconditions:

• T = 25 ◦C and PCO2 = 1 bar (type 1);• T = 120 ◦C and PCO2 = 3 bar (type 2);• T = 120 ◦C and PCO2 = 100 bar (type 3).

In all cases an aqueous solution of Na2CO3 was prepared,heated up to the desired temperature and the gas phase flushedwith CO2. For the low temperature experiments a CO2 atmo-sphere in the reactor was maintained by feeding a small streamof CO2 through a dip tube into the reactor throughout the entireexperiment. For the high temperature experiments the reactorwas heated up, pressurized with CO2 and then kept at a con-stant pressure by feeding CO2 via a front pressure regulator.

Supersaturation with respect to Mg-carbonate was achievedby adding a solution of MgCl2, after making sure that dissolu-tion and speciation of CO2 had reached a state of equilibrium.This addition defines the start of the induction period, whoseend is defined by the onset of precipitation, i.e., the time offirst detection of solid particles. At the end of the precipita-tion experiment, the produced solids were collected, filtered,washed with de-ionized water and dried in an oven at 60 ◦C for

1016 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

Temperaturecontrolled oil

bath

Stirred titanium

autoclave

MgCl2 solution

CO2 buffer tank

Front pressure

regulator

Tmax = 200°C

Pmax = 200 bar

HPLC

pumpRaman probe /

Turbidity meter

Fig. 3. Scheme of the experimental set-up used for the high-temperature experiments.

about 12 h, and then characterized using SEM/EDX (Brukeraxs, model D8 Advance) and XRD (Zeiss, model LEO 1530).Simple bulk chemical analysis of the solid produced was usedwith a few samples only. It agreed well with the species iden-tification achieved based on the very distinct crystal shape andthe experimental XRD data.

4.2. Monitoring

The solutions of both the experiments at ambient pressureand at 100 bar (type 1 and type 3) were monitored using Ramanspectroscopy (Mettler-Toledo, model RA 400). Infrared spec-troscopy (Attenuated Total Reflectance Fourier TransformedInfrared spectroscopy, ATR-FTIR, Mettler Toledo, ReactIR4000, K6Conduit) was used alternatively during part of theexperiments at ambient pressure (type 1). With IR and Ramanspectroscopy, it is possible to detect and measure the con-centration of dissolved molecular species in solution, e.g.,HCO−

3 and CO2−3 ions, but not of mono-atomic species such as

Mg2+ ions, since IR and Raman spectroscopy measure prop-erties associated to the bonds between atoms in the molecule.Raman spectroscopy, in addition, allows also to detect andidentify suspended solid compounds in solution, a task whichcannot be achieved using IR spectroscopy only. The onlinemonitoring has been used among other tasks for verifying theachievement of a state of equilibrium before the addition ofthe MgCl2 solution. The onset of precipitation was measureddifferently during each type of experiment. During the exper-iments of type 1 a Focussed Beam Reflectance Measurementinstrument (FBRM, Lasentec, model 600L) for measuring insitu the chord length distribution (CLD) was used. During theexperiments of type 2 a turbidity meter (Mettler Toledo, modelFSC402) was used to monitor the turbidity variations. Dur-ing the experiments of type 3 Raman spectroscopy was used.Commonly, FBRM is used for detecting the onset of precip-itation, in our case we have tested the sensitivity of both theturbidity meter and Raman spectroscopy during the ambientpressure experiments. We observed that the induction times

obtained from all three techniques differed only very slightly.Figs. 4 and 5 illustrate two precipitation experiments of type 1and 3. Fig. 4(a) shows a Raman spectrum of the experimentalsolution exhibiting the CO2−

3 , HCO−3 , CO2(aq) and nesque-

honite absorption bands. Fig. 4(b) shows a Raman spectrumof the experimental solution with the HCO−

3 , hydromagnesiteand magnesite absorption bands. The position of the peaks forthe different compounds matches those of the characteristicRaman bands reported in the literature (Edwards et al., 2005).The peak position differs only slightly between suspensionsand dry solids, which were additionally identified using XRD.An algorithm for peak deconvolution of overlapped Ramanpeaks was applied. The peaks were described using Gaussianfunctions; the resulting separated peaks were then integratedover time. In Fig. 5(a) and (b) the evolution of the integratedpeaks of these species during the precipitation experimentsis shown; also plotted is the number of particles as mea-sured by FBRM. Concentrations are related to the integratedpeak area for CO2−

3 , HCO−3 , and CO2(aq) and the precipi-

tated solid nesquehonite (MgCO3 · 3H2O), hydromagnesite((MgCO3)4 · Mg(OH)2 · 4H2O), and magnesite (MgCO3).

As the peak areas of the species CO2−3 , HCO−

3 , and CO2(aq)reach a plateau, thus indicating the achievement of equilibrium,a solution of MgCl2 is added (start of the induction period forMg-carbonate precipitation). The end of the induction periodcan easily be detected both from the peak area curve of pre-cipitated solids and from the curve indicating the number ofparticles that start growing rapidly as the precipitation starts.At the same time, the CO2−

3 , HCO−3 , and CO2(aq) curves start

to decrease due to a reduction of the solution measurement vol-ume seen by the Raman probe caused by the increased solidconcentration in the suspension.

5. Experimental results

Experiments were carried out over a range of sodium andchloride concentrations. Due to our experimental protocol,the initial magnesium and chloride concentrations are related,

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1017

800 900 1000 1100 1200 1300 1400 15002000

3000

4000

5000

6000

7000

8000

9000

10000

11000

raman shift [1/cm]

Inte

nsity

HCO3

-

CO3

2-

Nesquehonite

CO2(aq)

850 900 950 1000 1050 1100 1150 1200 1250200

250

300

350

400

450

500

550

600

650

700

raman shift [1/cm]

Inte

nsity

HCO3

-

Magnesite

Hydromagnesite

Fig. 4. Raman spectrum of the experimental solution; band position of selected species with the associated Raman shift values: magnesite, 1099 cm−1;nesquehonite, 1100 cm−1; hydromagnesite, 1119 cm−1. The peak position of aqueous species is to a certain extent affected by temperature: (a) Experiment type 1(1 bar, 25 ◦ C) HCO−

3 , 1017 cm−1; CO2−3 , 1065 cm−1; CO2(aq), 1364 cm−1. (b) Experiment type 3 (100 bar, 120 ◦ C) HCO−

3 , 1003 cm−1; CO2−3 , 1032 cm−1;

CO2(aq), 1351 cm−1.

i.e., [Cl] = 2[Mg]. The solubility of Na-carbonates and MgCl2limit the maximum sodium and chloride concentration achiev-able experimentally, although by varying the ratio betweenthe MgCl2 and the Na-carbonate solutions, these limits can beshifted to a certain extent. The range of possible experimentscan best be illustrated by plotting the initial Mg concentra-tion obtainable in our system over the range of Cl and Naconcentrations together with the solubility of various Mg-compounds.

The supersaturation ratio, S, for all compounds is calculatedusing the activity product, Q, as defined in Table 1:

S =(

Q

Ksp

)1/n

, (1)

where Ksp is the corresponding equilibrium constant or solubil-ity product and n is the total number of aqueous species formedupon dissolution of the compound, including water, i.e., n = 5for nesquehonite.

5.1. Low temperature experiments

For the experiments at 25 ◦C and PCO2 = 1 bar the solu-bilities of artinite, nesquehonite, hydromagnesite and magne-site (from top to bottom) are shown in Fig. 6 together withthe range of possible initial Mg-concentrations. It can beseen that over the plotted range of Na and Cl concentrationall solutions are undersaturated with respect to artinite (aswell as with respect to brucite, not shown here) and always

1018 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

0

1

2

3

4

5

6

7

x 105

Are

a p

ea

k [

I cm

-1]

# p

art

icle

sHCO3

-

CO3

CO2

Nesquehonite

#particles

MgCl2addition

0 100 200 300 400 500 600

0

1

2

3

4

5

6

7

8

9x 104

Are

a p

ea

k [

I cm

-1]

time [min]

0 100 200 300 400 500 800600 700

time [min]

MgCl2 addition

Magnesite

HCO3

Hydromagnesite

2-

-

Fig. 5. Monitoring of a precipitation experiment with Raman spectroscopy and FBRM. Evolution of concentration (i.e., integrated peak area of the Ramansignal) of CO2−

3 , HCO−3 , CO2(aq) and MgCO3 · 3H2O, as well as of the number of particles measured with FBRM (experiment type 1), and integrated peak

area of the Raman signal for magnesite and hydromagnesite (experiment type 3). (a) Experiment type 1. (b) Experiment type 3.

supersaturated with respect to hydromagnesite and magnesite.Except for high sodium and chloride (i.e., MgCl2) concen-trations, the solutions are also supersaturated with respect tonesquehonite.

The solution compositions of the precipitation experimentscarried out are shown in Fig. 7 where contour lines indicateconditions of constant supersaturation ratio of the initial solu-tion with respect to nesquehonite. The main property varyingalong these iso-supersaturation lines is the ratio between Mg2+and CO2−

3 ion concentrations, ri , as indicated by a second setof contour lines. Operating conditions along with supersatura-tion ratio and Mg2+ to CO2−

3 ratio are listed in Table 2.

All solids from the low temperature experiments exhibitthe typical morphology of nesquehonite crystals (well-formedneedles) as shown in Fig. 8(a). Such identification is con-firmed by the very good match between the nesquehonite pat-tern obtained from a database and our X-ray measurements(see Fig. 8(b)).

The experiments can be grouped into three sets. Two sets atapproximately constant ion ratio, i.e., ri � 5 and 25, and oneset at constant supersaturation ratio, i.e., SNes � 1.11.

Plotting the induction time over the supersaturation ratio Swith respect to nesquehonite for set one, at ri � 5, and includ-ing one experiment at ri � 6.5, one can clearly see that the

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1019

Fig. 6. Solubility, expressed as Mg-concentration, of artinite, nesquehonite, hydromagnesite and magnesite (from top to bottom) over a range of Na- andCl-concentration at 25 ◦C and PCO2 = 1 bar. White wireframe surface shows Mg-concentration of solutions obtainable by mixing Na2CO3 solutions withMgCl2 solutions, i.e., possible initial conditions using our experimental protocol.

. . . . . . . .

15

Fig. 7. Contour line representing supersaturation ratio, S, with respect to nesquehonite (0.94–1.21) and Mg2+ to CO2−3 ratio, ri , (5–25) over the experimental

range of Na- and Cl-concentrations at the possible initial conditions using our experimental protocol, i.e., projected on the wireframe surface in Fig. 6. Dashedlines represent pH contour lines (7.4–7.6). Symbols indicate solution composition of the precipitation experiments in Table 2.

induction time increases as SNes approaches one whereas al-most instantaneous precipitation occurs as the supersaturationis increased to SNes = 1.15 (Fig. 9(a)). A very similar behavior

is observed at ri � 25 (Fig. 9(b)). The uncertainty in the mea-sured induction time can be inferred from the different timesmeasured when repeating experiments at SNes = 1.12 and 1.14.

1020 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

Table 2Experimental conditions and results: 25 ◦C, PCO2 = 1 bar

cNa (mol/kg) cCl (mol/kg) SNes (–) SLan (–) SHM (–) SMag (–) ri (Mg2+/CO2−3 ) (–) tind

a (min) texp (min) Exp. no

1.2 0.04 0.64 0.76 0.66 7.9 0.4 NA 960 35.12 0.12 0.86 0.91 1.14 18.5 0.4 NA 960 2

1.5 0.24 1.11 1.12 1.65 31.3 4.4 80 450 201.3 0.18 1.03 1.06 1.45 25.7 4.4 685 1100 211.4 0.22 1.08 1.1 1.58 29.1 4.7 387 1084 221.44 0.31 1.15 1.15 1.76 34.4 6.5 1 300 19

1.3 0.28 1.11 1.12 1.66 31.3 7.4 NA 648 251.3 0.28 1.11 1.12 1.66 31.3 7.4 180 290 261.25 0.29 1.11 1.12 1.65 31.1 8.5 77 1030 271.2 0.31 1.12 1.12 1.66 31.4 10.1 271 790 241.1 0.34 1.12 1.13 1.66 31.3 14.2 303 1130 281.05 0.37 1.12 1.13 1.67 31.5 17.4 281 764 231.05 0.37 1.12 1.13 1.67 31.5 17.4 274 940 36

1.02 0.46 1.16 1.16 1.77 33.7 24.3 10 1120 111.04 0.48 1.16 1.16 1.77 34.3 24.5 NA 1070 170.97 0.41 1.12 1.13 1.66 31.3 24.5 228 870 140.97 0.41 1.12 1.13 1.66 31.3 24.5 300 1470 151.01 0.45 1.14 1.14 1.72 33.2 24.5 126 395 161.01 0.45 1.14 1.14 1.72 33.2 24.5 0.01 275 340.88 0.36 1.07 1.1 1.55 28.2 25.4 600 1120 10

aNA means that no precipitation was observed.

10 20 30 40 50 600

2000

4000

6000

8000

10000

Lin

(counts

)

2-Theta-Scale

Sup 36

Nesquehonite

Fig. 8. SEM image and X-ray diffraction pattern for the precipitated solids in the Na-system at 25 ◦C, PCO2 = 1 bar. Red stars indicate position and heightof a reference XRD spectra. Except for a few peaks significantly higher in the reference data, a very good agreement is found. (a) SEM. (b) XRD.

Induction times for the third set, at constant SNes, only exhibita weak trend and, given the limitations of the thermodynamicmodel used, therefore, do not allow for conclusions to be madeabout the influence of the Mg2+ and CO2−

3 ratio on precipitationkinetics. Similarly, differences between set one and two, i.e.,high and low Mg2+ and CO2−

3 ratio, are too small to draw anyconclusions in that respect.

However, it is quite evident that at ambient temperaturesupersaturation with respect to nesquehonite is governingprecipitation kinetics. Therefore, the most likely reaction

path is the direct nucleation and growth of nesquehonite.In fact, nesquehonite was identified using Raman spec-troscopy as the only solid Mg-compound detected over thecourse of the whole experiment. Moreover, experiments con-ducted at MgCl2 concentrations corresponding to undersat-uration with respect to nesquehonite (not shown in Fig. 7)and lasting more than 16 h did not yield any Mg-carbonateprecipitation.

The formation of nesquehonite at low temperature is wellknown from literature (see Section 3). Although any solution

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1021

1 1.05 1.1 1.150

200

400

600

SMg-carbonate

Ind

uctio

n t

ime

[m

in]

0

200

400

600

Ind

uctio

n t

ime

[m

in]

1 1.05 1.1 1.15

SNesquehonite

Fig. 9. Induction time for precipitation experiments at 25 ◦C in the Na-system at approximately constant Mg2+ to CO2−3 ratio, ri , plotted over supersaturation

ratio, S, with respect to nesquehonite. (a) ri � 5 (Exp. no.: 21, 22, 20 and 19). (b) ri � 25 (Exp. no.: 10, 15, 23, 16, 34, 11).

010

2030

4050

6070

1060

1080

1100

1120

1140

1160

800

1000

1200

1400

1600

1800

2000

Time [min]

Inte

nsity [-]

Raman shift [cm-1]

Fig. 10. Raman spectra of the transformation of nesquehonite to hydromagnesite at 72 ◦C. Peaks at Raman shifts of 1100 and at 1119 cm−1 representnesquehonite and hydromagnesite, respectively.

precipitating nesquehonite is also supersaturated with respect tomagnesite as well as hydromagnesite, their nucleation kineticsas well as the kinetics of transformation from nesquehonite intohydromagnesite or magnesite are too slow to allow for theirformation at ambient temperature.

5.2. Transformation of nesquehonite

As described in Section 3, the transformation of nesquehoniteinto hydromagnesite has been observed at temperatures aboveabout 50 ◦C. We have studied this transformation for temper-atures ranging from 50 to 72 ◦C. Online in situ Raman spec-troscopy as described in Section 4.2 was used to monitor thedisappearance of nesquehonite and the formation of hydromag-nesite. The Raman spectra thus obtained are shown for oneexperiment in Fig. 10. Experiments were carried out at atmo-spheric CO2 concentration.

Table 3Transition experiments from nesquehonite into hydromagnesite at atmosphericCO2 concentration for several different temperatures

T (◦C) ttrans (min) tind (min) Exp. no. (source material) Exp. no.

50 289 108 19 t1155 51 3 23 t760 138 35 19 t1065 14 2 23 t672 13 4 34 t4

Up to 60 ◦C, the transformation is preceded by an induc-tion period with no discernable change in the Raman signal,followed by the simultaneous disappearance of nesquehoniteand the appearance of hydromagnesite during a time periodthat depends on the operating conditions and is reported inTable 3. Above 60 ◦C, the induction time is close to zero. As

1022 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

Fig. 11. Solubility, expressed as Mg-concentration, of nesquehonite, brucite, artinite, hydromagnesite and magnesite (from top to bottom) over a range of Na-and Cl-concentration at 120 ◦C and PCO2 = 3 bar. White wireframe surface shows Mg-concentration of solutions obtainable by mixing Na2CO3 solutions withMgCl2 solutions, i.e., possible initial conditions using our experimental protocol.

reported in Table 3, increasing the temperature accelerates thetransformation. It should be noted that the nesquehonite crys-tals used as seeds had been previously formed in differentexperiments.

5.3. High temperature, low pressure experiments

For the experiments at 120 ◦C and PCO2 = 3 bar the solu-bilities of nesquehonite, brucite, artinite, hydromagnesite andmagnesite (from top to bottom) are shown in Fig. 11 togetherwith the range of Mg-concentrations attainable using our ex-perimental procedure. The sharp increase of the solubilities ofhydromagnesite and magnesite at ratios of c(Cl) to c(Na) �1is due to a significant decrease in the pH as the balance be-tween chloride and metal cations is shifted towards a surplusof the former. The change in pH shifts the balance betweenCO2−

3 and HCO−3 ions in the direction of the latter, thereby

leading to increasing magnesite and hydromagnesite solubilityas expressed via the Mg-concentration in Fig. 11. The solubil-ity of the other carbonates is much less affected by the phe-nomenon since their high solubility brings with it high magne-sium concentrations, which largely compensate the increasedchloride concentrations. It can be seen that except for highsodium and chloride concentrations the solutions are, over theplotted range of Na and Cl concentration, undersaturated withrespect to nesquehonite and always supersaturated with respectto artinite, hydromagnesite and magnesite. Except for low chlo-ride concentrations, the solutions are also supersaturated withrespect to brucite.

Solution compositions of the precipitation experiments areshown in Fig. 12 as red symbols, where contour lines indi-cate conditions of constant supersaturation ratio of the initialsolution with respect to brucite (lower left corner) and nesque-honite (upper right corner). One experiment was carried out ata lower CO2 pressure and could therefore not be representedon the graph. The operating conditions for all experiments at120 ◦C and PCO2 � 3 bar along with supersaturation and Mg2+to CO2−

3 ion ratio are listed in Table 4. All but one experimentwere carried out at a Mg2+ to CO2−

3 ratio between three andsix. As it can be seen, most experiments were carried out un-der conditions of supersaturation with respect to brucite, withsome of them also being supersaturated with respect to nesque-honite (incl. the experiment at a lower pressure, not indicatedin Fig. 7).

In the experimental set-up used for the experiments at 120 ◦Cthe MgCl2 solution was added via an HPLC pump. The additiontime was 2 min. Observed induction times at 120 ◦C were con-siderably shorter than at ambient temperature; indeed, exceptfor two experiments, precipitation started before the addition ofthe MgCl2 solution was completed. In one experiment wherethe solution was highly undersaturated with respect to bruciteno precipitation was observed over the entire experimental du-ration of 19 h. For the experiment close to supersaturation withrespect to brucite, precipitation started about 1 min after the ad-dition of the MgCl2 solution was complete. In all other exper-iments, precipitation started before the addition of the MgCl2had finished, making it impossible to measure an inductiontime. The evolution of the supersaturation during the MgCl2

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1023

ci(Cl) [m]

ci(N

a)

[m]

0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4

0.2

0.25

0.3

0.35

0.4

0.45

0.5

0.55

0.6

0.65

Fig. 12. Contour line representing supersaturation ratio, S, with respect to brucite (0.7–1.3) and nesquehonite, S, (0.91–1.03) over the experimental range ofNa- and Cl-concentrations at 120 ◦C and PCO2 � 3 bar. Dashed lines represent pH contour lines (6.85–7.5). Symbols indicate solution composition of theprecipitation experiments.

Table 4Experimental conditions and results: 120 ◦C, PCO2 = 3 bar

PCO2 (bar) Ptot (bar) cNa (mol/l) cCl (mol/l) SBru (–) SNes (–) SHM (–) SMag (–) ri (Mg2+/CO2−3 ) (–) tind

a (min) texp (min) Exp. no.

2.8 4.8 0.2 0.02 0.72 0.51 2.71 32 4.2 NA 1140 392.7 4.7 0.2 0.4 1.45 0.75 5.25 285.5 322 0b 50 383.1 5.1 0.3 0.04 0.98 0.62 3.7 51 2.5 1 170 433 5 0.4 0.08 1.36 0.75 5.15 84 2.8 0b 140 402 4 0.5 0.3 2.89 1.05 9.39 196 5.7 0b 900 373 5 0.66 0.38 2.41 1.05 9.1 199 5.6 0b 310 413.1 5.1 0.66 0.38 2.41 1.05 9.1 199 5.6 0b 980 42

aNA means that no precipitation was observed.bPrecipitation occurred before addition of MgCl2 was fully completed, i.e., before targeted supersaturation was achieved.

solution addition is illustrated in Fig. 13 where the solubili-ties of several Mg-carbonates expressed as Mg-concentration,at constant sodium concentration, are plotted over MgCl2, i.e.,chloride concentration. Plotted as well is the Mg-concentrationin solution as generated by the addition of MgCl2. It can clearlybe seen that with the increasing addition of MgCl2 solution,supersaturation, first with respect to magnesite, and then sub-sequently with respect to hydromagnesite, brucite and in thesecond example also with respect to nesquehonite is created.

The duration of the experiments varied between a few hoursand more than 15 h. The solids collected in the four experi-ments lasting up to about 5 h exhibit the crystal morphologyof an agglomerate of very thin platelets (Fig. 14(a)), typical

of hydromagnesite, as confirmed by XRD (Fig. 14(b)). For theexperiment lasting 5 h, the longest of this set, it can be seenthat the major peak of magnesite at 32.7◦ is already present inthe XRD spectra. Solids collected from experiments runningfor more than 15 h, on the contrary presented the structure ofmassive hexagonal prisms (Fig. 15(a)) typical of magnesite.The formation of highly crystalline magnesite was confirmedby XRD (Fig. 15(b)), with the XRD pattern exhibiting a perfectmatch to the characteristic peaks.

These results suggest that at 120 ◦C and PCO2 = 3 bar super-saturation with respect to brucite determines initial precipitationkinetics. We observed in fact no precipitation in solutions un-dersaturated with respect to brucite, fast but not instantaneous

1024 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

100

10-2

10-4

10-6

c(M

g)

[m]

100

10-2

10-4

10-6

c(M

g)

[m]

Nesquehonite

Brucite

Artinite

Hydromagnesite

Magnesite

0 0.1 0.2 0.3 0.4

c(Cl) [m]

0 0.1 0.2 0.3 0.4

c(Cl) [m]

Nesquehonite

Brucite

Artinite

Hydromagnesite

Magnesite

Fig. 13. Solubility, expressed as Mg-concentration, of nesquehonite, brucite, artinite, hydromagnesite and magnesite (from top to bottom) at constant c(Na)over Cl-concentration. Dotted line: trajectory of Mg-concentration in solution as MgCl2 solution is added assuming no precipitation, red symbol denotes finalMg-concentration after completion of MgCl2 solution addition for individual experiment. (a) c(Na) = 0.4 m, Sup 40. (b) c(Na) = 0.5 m, Sup 37.

10 20 30 40 50 600

500

1000

1500

2000

2500

3000

3500Lin

(counts

)

2-Theta-Scale

Sup 41

Hydromagnesite

Fig. 14. Typical SEM image and X-ray diffraction pattern for the precipitated solids at 120 ◦C, PCO2 = 3 bar for experiments lasting up to 5 h. Red symbolsmark the position and height for a reference hydromagnesite spectra. Both peak position and height show a very good match. (a) SEM. (b) XRD.

precipitation at solutions close to the solubility of brucite andvery fast precipitation as the Mg-concentration was increasedabove the solubility of brucite during the addition of MgCl2 so-lution. However, no brucite was collected in our experiments,hence we conjecture that brucite might only act as a precursorfor the nucleation of hydromagnesite or might transform intohydromagnesite during the experiment (in at most 1 h, the du-ration of our shortest experiment).

Compared to low temperature experiments, the formationkinetics of hydromagnesite is greatly increased at 120 ◦C thusallowing for its formation within less than 1 h. Similarly, thekinetics of magnesite formation becomes fast enough to trans-form hydromagnesite to magnesite in a time-span of hours.As mentioned in Section 3, this transformation might hap-pen via two different reaction mechanisms, which are notnecessarily alternative to each other: dehydration of hydro-magnesite (favored at high concentration of magnesium andhigh ionic strength), and hydromagnesite dissolution followed

by magnesite precipitation (favored at low concentration ofmagnesium).

5.4. High temperature, high pressure experiments

To illustrate the experiments at 120 ◦C and PCO2 of 100 bar,the solubilities of magnesite (bottom) and hydromagnesite areshown in Fig. 16 together with the range of possible initial Mg-concentrations (uppermost surface). It can be seen that over theplotted range of sodium and chloride concentrations, except forvery low Na and Cl values, the initial solutions applied duringthe experiments are supersaturated with respect to hydromag-nesite and magnesite. They are, however, undersaturated withrespect to brucite, artinite, nesquehonite, whose solubility isnot shown here for the sake of simplicity.

More in detail, solution compositions of the precipitation ex-periments are shown in Fig. 17. In this graph, contour linesindicate conditions of constant supersaturation of the initial

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1025

10 20 30 40 50 600

2000

4000

6000

8000

10000

12000

14000

Lin

(counts

)

2-Theta-Scale

Sup 42

Magnesite

Fig. 15. SEM image and X-ray diffraction pattern for the precipitated solids at 120 ◦C, PCO2 = 3 bar for experiments lasting more than 16 h. Red symbolsmark the position of characteristic magnesite peaks, indicating an almost perfectly crystalline magnesite. (a) SEM. (b) XRD.

Fig. 16. Solubility, expressed as Mg-concentration, for magnesite and hydromagnesite over a range of Na- and Cl-concentration at 120 ◦C and PCO2 = 100 bar.White wireframe surface shows Mg-concentration of solutions obtainable by mixing Na2CO3 solutions with MgCl2 solutions, i.e., possible initial conditionsusing our experimental protocol (from bottom to top).

solution with respect to hydromagnesite (0.7–2.7) and to mag-nesite (10–40). Detailed operating conditions along with super-saturation ratio and Mg2+ to CO2−

3 ratio are reported in Table 5.The solution composition and the suspension were monitoredusing online Raman spectroscopy. The supersaturation ratio val-ues given for each experiment in Fig. 17 and in Table 5 are dif-ferent, since the simulated PCO2 was equal to 100 bar whereasits measured values were slightly higher.

Three different types of outcomes have been observed. In so-lutions which are only very slightly supersaturated with respect

to hydromagnesite (h3), no precipitation occurred during thewhole experiment, of duration 20 h. At slightly higher supersat-uration (h11), precipitation of a solid compound, identified insitu with Raman spectroscopy as magnesite, occurs after a shortinduction time. Increasing the supersaturation even further re-duces the induction time to almost zero, i.e., instantaneous pre-cipitation (h5, h10). At even higher supersaturation levels (h8,h13, h19), the co-precipitation of two different solids was ob-served. They were identified as hydromagnesite and magnesitefrom the comparison of the measured Raman spectra with the

1026 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

ci(Cl) [m]

ci(N

a)

[m]

6

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1

Fig. 17. Contour line representing supersaturation ratio, S, with respect to hydromagnesite (dashed line, 0.7–3.5) and magnesite (15–75) over the experimentalrange of Na- and Cl-concentrations at 120 ◦C and PCO2 = 100 bar. Straight solid lines represent pH contour lines (6–6.37). Symbols indicate solutioncomposition of the precipitation experiments.

Table 5Experimental conditions and results: 120 ◦C, PCO2 = 100 bar, Na-system

T (◦C) fCO2 (bar) Ptot (bar) cNa (mol/l) cCl (mol/l) SHM (–) SMag (–) ri (Mg2+/CO2−3 ) (–) Exp. no.

119 83 106 0.6 0.02 1.1 11.5 6.9 h3119 82 104 0.5 0.1 1.6 21 94 h11119 81 102 1.0 0.06 2 32 8 h10120 85 108 0.6 0.12 2.2 33 100 h5123 84 106 0.7 0.4 2.8 48 129 h8119 83 106 1.0 0.7 3.4 65 105 h19119 82 103 1.0 0.7 3.4 66 102 h13

characteristic Raman bands (see Section 4.2). Hydromagnesitedisappears again after about 2 h and only magnesite remains inthe solution, see Figs. 5(b) and 18. The final product in all ex-periments was magnesite, as confirmed by the very good matchbetween the magnesite pattern obtained from a database andthe X-ray measurements, as well as by the particle shape seenin SEM micrographs (as illustrated for a low pressure experi-ment in Fig. 15). These results indicate that under the experi-mental conditions of 100 ◦C and PCO2 =100 bar solutions haveto be supersaturated with respect to hydromagnesite to allowfor unseeded precipitation of magnesium carbonates within 20h. This is in agreement with Giammar et al. (2005), which at

90 ◦C and PCO2 � 90 bar reported significant precipitation onlyin experiments supersaturated with respect to hydromagnesite.At the same time, the formation of hydromagnesite in measur-able quantities is only observed at elevated values of supersat-uration and is only a temporary phenomenon since in all casesmagnesite is the final product.

6. Conclusions

The magnesium carbonate system exhibits a complex behav-ior due to the large number of possible phases and to the strongkinetic inhibitions that are dependent on temperature, partial

M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028 1027

Fig. 18. Raman spectra of the transformation of hydromagnesite into magnesite for experiment h13 at 119 ◦C and Ptot = 103 bar. Peaks at Raman shifts of1099 and at 1119 cm−1 represent magnesite and hydromagnesite, respectively.

pressure of CO2 and possibly other parameters. Depending onthe operating conditions, usually only one compound is con-trolling the formation of a solid phase, with the fastest reactiondetermining the observed overall rate of precipitation. If forgiven conditions it is known which phase is rate-controlling,precipitation kinetics depends essentially on the supersatura-tion with respect to this species, i.e., thermodynamics can beused to predict both the appearance of solid phases and theprecipitation kinetics. The formation of the anhydrous form ofmagnesium carbonate, i.e., magnesite, has been demonstratedat 120 ◦C and a PCO2 of 3 bar. However, it required the forma-tion of hydromagnesite as an intermediate substance and is tooslow a process for industrial purposes, in particular mineral car-bonation. Moreover, to produce magnesite under these condi-tions, solutions have to be supersaturated not only with respectto magnesite but also brucite and hydromagnesite, which re-quires e.g., much higher magnesium concentrations than super-saturation with respect to magnesite would need. Compared toexperiments at T =120 ◦C and PCO2 =3 bar, formation kineticsof magnesite is greatly enhanced at 100 bar thus allowing for itsimmediate formation. Precipitation was only observed in solu-tions supersaturated with respect to hydromagnesite (but under-saturated with respect to brucite). At sufficient supersaturation,hydromagnesite co-precipitates with magnesite but transformsthen into magnesite within a few hours, i.e., faster than at lowerpressures. These results show that precipitation and formation

of various Mg-carbonates can be understood and described us-ing thermodynamic models if the kinetic inhibitions have beenadequately characterized. They provide a useful basis for thedesign of operating conditions to precipitate Mg-carbonates inthe context of mineral carbonation, e.g., using natural MgO-bearing silicates as substrates.

Notation

ai activity of ion or solid compound i, dimension-less

cCl total chloride concentration in solution,mol kg−1

cNa total sodium concentration in solution, mol kg−1

fCO2 CO2 fugacity, barKsp solubility product (equilibrium constant), di-

mensionlessn total number of aqueous species formed upon

dissolution, dimensionlessPCO2 CO2 partial pressure, barPtot total pressure, barQ activity product, dimensionlessri ratio between the molality of Mg2+ and CO2−

3ions, dimensionless

SHM supersaturation ratio with respect to hydromag-nesite, dimensionless

1028 M. Hänchen et al. / Chemical Engineering Science 63 (2008) 1012–1028

SLan supersaturation ratio with respect to lansfordite,dimensionless

SMag supersaturation ratio with respect to magnesite,dimensionless

SNes supersaturation ratio with respect to nesque-honite, dimensionless

texp total duration of the experiment, mintind induction time until onset of nucleation, minttrans duration of the phase transformation, minT temperature, ◦C

Acknowledgements

Financial support from the Fondation Claude & Giuliana,Basel, Switzerland, is gratefully acknowledged.

References

Baron, G., 1958. Precipitation de la giobertite et de la dolomie a partir dessolutions de chlorures de magnesium et de calcium. Comptes RendusHebdomadaires Des Seances De L Academie Des Sciences 247 (19),1606–1608.

Baron, G., Favre, J., 1958. État actuel des recherches en direction de lasynthèse de la dolomie. Revue de l’Institut Française du Pétrole 13,1067–1085.

Christ, C.L., Hostetle, P.B., 1970. Studies in system MgO–SiO2–CO2–H2O.2. Activity-product constant of magnesite. American Journal of Science268 (5), 439–453.

Davies, P.J., Bubela, B., 1973. Transformation of nesquehonite intohydromagnesite. Chemical Geology 12 (4), 289–300.

Deelman, J.C., 2001. Breaking Ostwald’s rule. Chemie Der Erde-Geochemistry 61 (3), 224–235.

Dell, R.M., Weller, S.W., 1959. The thermal decomposition of nesquehoniteMgCO3 · 3H2O and magnesium ammonium carbonate MgCO3 ·(NH4)2CO3 · 4H2O. Transactions of the Faraday Society 55 (12),2203–2220.

Edwards, H.G.M., Villar, S.E.J., Jehlicka, J., Munshi, T., 2005. FT-Raman spectroscopic study of calcium-rich and magnesium-rich carbonateminerals. Spectrochimica Acta Part A—Molecular and BiomolecularSpectroscopy 61 (10), 2273–2280.

Fernandez, A.I., Chimenos, J.M., Segarra, M., Fernandez, M.A., Espiell, F.,2000. Procedure to obtain hydromagnesite from a MgO-containing residue.Kinetic study. Industrial & Engineering Chemistry Research 39 (10),3653–3658.

Giammar, D.E., Bruant, R.G., Peters, C.A., 2005. Forsterite dissolution andmagnesite precipitation at conditions relevant for deep saline aquiferstorage and sequestration of carbon dioxide. Chemical Geology 217 (3–4),257–276.

Hänchen, M., Prigiobbe, V., Storti, G., Seward, T.M., Mazzotti, M., 2006.Dissolution kinetics of fosteritic olivine at 90–150 ◦C including effects

of the presence of CO2. Geochimica et Cosmochimica Acta 70 (17),4403–4416.

IPCC, 2005. IPCC Special Report on Carbon Dioxide Capture and Storage.Cambridge University Press, Cambridge, UK, New York.

Kittrick, J.A., Peryea, F.J., 1986. Determination of the Gibbs free-energy offormation of magnesite by solubility methods. Soil Science Society ofAmerica Journal 50 (1), 243–247.

Kloprogge, J.T., Martens, W.N., Nothdurft, L., Duong, L.V., Webb, G.E.,2003. Low temperature synthesis and characterization of nesquehonite.Journal of Materials Science Letters 22 (11), 825–829.

Königsberger, E., Königsberger, L.C., Gamsjager, H., 1999. Low-temperaturethermodynamic model for the system Na2CO3–MgCO3–CaCO3–H2O.Geochimica et Cosmochimica Acta 63 (19–20), 3105–3119.

Lanas, J., Alvarez, J.I., 2004. Dolomitic lime: thermal decomposition ofnesquehonite. Thermochimica Acta 421 (1–2), 123–132.

Langmuir, D., 1965. Stability of carbonates in the system MgO–CO2–H2O.Journal of Geology 73, 730–754.

Ming, D.W., Franklin, W.T., 1985. Synthesis and characterization oflansfordite and nesquehonite. Soil Science Society of America Journal 49(5), 1303–1308.

Mullin, J.W., 1993. Crystallization. third ed. Butterworth-Heinemann, Oxford.(j. W. Mullin 24 cm ill. andere Ausgabe: Crystallisation).

O’Connor, W.K., Dahlin, D.C., Rush, G.E., Dahlin, C.L., Collins, W.K.,2002. Carbon dioxide sequestration by direct mineral carbonation: processmineralogy of feed and products. Minerals & Metallurgical Processing 19(2), 95–101.

O’Connor, W.K., Dahlin, C.L., Rush, G.E., Gerdemann, S.J., Penner,L.R., Nilsen, D.N., 2005. Final report aqueous mineral carbonation:mineral availability, pretreatment, reaction parametrics, and process studies.Technical Report DOE/ARC-TR-04-002, Albany Research Center.

Sayles, F.L., Fyfe, W.S., 1973. Crystallization of magnesite from aqueous-solution. Geochimica et Cosmochimica Acta 37 (1), 87–99.

Schöll, J., Lindenberg, C., Vicum, L., Brozio, J., Mazzotti, M., 2007.Precipitation of � L-glutamic acid: determination of growth kinetics.Faraday Discussions 136, 247–264.

Stevula, L., Petrovic, J., Kubranova, M., 1978. Formation of syntheticmagnesite by carbonatization of hydromagnesite under hydrothermalconditions. Chemicke Zvesti 32 (4), 441–443.

Wolery, T.J., 1992. EQ3/6, a software package for geochemical modeling ofaqueous systems: package overview and installation guide (version 7.0).Technical Report UCRL-MA-110662 PT I, Lawrence Livermore NationalLaboratory, September 14.

Wolf, G.H., Chizmeshya, A.V.G., Diefenbacher, J., McKelvy, M.J., 2004. Insitu observation of CO2 sequestration reactions using a novel microreactionsystem. Environmental Science & Technology 38 (3), 932–936.

Zhang, L., 2000. Kinetics and mechanisms of formation of magnesitefrom hydromagnesite in brine. Technical Report SAN099-19465, SandiaNational Laboratories, Albuquerque, NM 87185-0750, USA.

Zhang, Z.P., Zheng, Y.J., Ni, Y.W., Liu, Z.M., Chen, J.P., Liang, X.M.,2006. Temperature- and pH-dependent morphology and FT-IR analysis ofmagnesium carbonate hydrates. Journal of Physical Chemistry B 110 (26),12969–12973.