The deleterious effect of secondary phases on olivine carbonation yield: Insight from time-resolved aqueous-fluid sampling and FIB-TEM characterization

Chemical Geology 357 (2013) 186–202

Contents lists available at ScienceDirect

Chemical Geology journal homepage: www.elsevier.com/locate/chemgeo

The deleterious effect of secondary phases on olivine carbonation yield: Insight from time-resolved aqueous-fluid sampling and FIB-TEM characterization Olivier Sissmann a,b,⁎, Damien Daval c, Fabrice Brunet d, François Guyot a,e, Anne Verlaguet f, Yves Pinquier b, Nathaniel Findling d, Isabelle Martinez a a

Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Univ. Paris Diderot, UMR 7154 CNRS, 1 rue Jussieu, F-75005 Paris, France Laboratoire de Géologie, UMR 8538 CNRS, École Normale Supérieure, 24 Rue Lhomond, 75005 Paris, France c Laboratoire d'Hydrologie et de Géochimie de Strasbourg, Université de Strasbourg/EOST CNRS UMR 7517, 1 Rue Blessig, 67084 Strasbourg, France d Institut de Sciences de la Terre, UMR 5275 CNRS, Université Joseph Fourier, 1381 rue de la Piscine, 38400 Saint Martin d'Hères, France e Institut de Minéralogie et de Physique des Milieux Condensés, UMR 7590 CNRS, Université Pierre et Marie Curie, 4 place Jussieu, 75005 Paris, France f ISTeP, UMR 7193 CNRS, Université Pierre et Marie Curie, 4 place Jussieu, 75005 Paris, France b

a r t i c l e

i n f o

Article history: Received 16 February 2013 Received in revised form 13 August 2013 Accepted 19 August 2013 Available online 5 September 2013 Editor: J. Fein Keywords: Olivine carbonation CO2 sequestration Dissolution kinetics Redox reactions Thermodynamic modeling

a b s t r a c t Geological storage of CO2 in mafic and ultramafic rocks relies on the dissolution of their silicate components, followed by the precipitation of carbonates, the overall process being commonly referred to as carbonation. To gain a better understanding of the rate- and yield-controlling factors of mineral carbonation, three batch experiments were conducted in CO2-saturated water, at (a) 90 °C, (b) 120 °C and (c) 170 °C, respectively (with citrate ligands added to the 90 °C experiment) in a Ti-reactor at pCO2 = 280 bar, using San Carlos olivine (Fo88) as a model mineral of (ultra)mafic environments. Those physicochemical conditions were purposely chosen so as to promote the dissolution rate of olivine. (a) At 90 °C, ~40 wt.% of olivine were dissolved within 3 weeks, a result which suggests that the passivation barrier evidenced in previous studies carried out in citrate-free media had been overcome. However, while saturation with respect to magnesite was overstepped in this experiment, the carbonation yield remained below ~1 wt.%, which we attributed to the slow kinetics of magnesite precipitation. (b) At 120 °C, the concentration of Mg and Si monitored by aqueous fluid sampling reached an apparent plateau for over 4 weeks, at conditions close to saturation with respect to amorphous silica, and with Fe concentration being below the detection limit. A resumption of the dissolution process was subsequently observed while the concentration of Fe suddenly increased in solution. The concentration plateau was attributed either to the formation of a protective Si–Fe(III) layer, or to the occurrence of Fe(III) phases clogging the porosity of a secondary silica layer. The resumption of olivine dissolution was ascribed to the increasingly reducing conditions that led to the breakdown of the protective phases. This late resumption led to a carbonation yield slightly above 2 wt.%. (c) At 170 °C, Fe was incorporated into a presumably permeable Fe-rich phyllosilicate layer, so that the dissolution process was likely less affected by such an interfacial layer. Nevertheless, carbonate minerals still formed in low quantity (~6 wt.%), while other amorphous phases represented the main sink for Mg cations. Overall, this study emphasizes that achieving carbonation reactions in reducing environments and circum-neutral aqueous solutions may represent a necessary requirement for making olivine carbonation a viable process in these temperature ranges. It also insists on the importance of transitory phases, which form both at the fluid-silicate interface and from the bulk solution. While they may not appear in overall carbonation equations or in thermodynamic databases, arguably they can drive the global kinetics of olivine dissolution and its transformation towards magnesite. © 2013 Elsevier B.V. All rights reserved.

As levels of atmospheric carbon dioxide (CO2) keep rising, the scientific community is seeking potential ways to store CO2 permanently. As such, the carbonation of metal-bearing silicates containing divalent

cations is considered as one of the most secure and sustainable options for sequestering CO2 over long time spans (Oelkers et al., 2008). Such a process relies on the dissolution of gaseous CO2 into water and its speciation into carbonate species, which, by dissolving silicates, releases divalent metal cations (M2+) in solution:

⁎ Corresponding author at: Institut de Physique du Globe de Paris, Sorbonne Paris Cité, Univ. Paris Diderot, UMR 7154 CNRS, 1 rue Jussieu, F-75005 Paris, France. E-mail address: [email protected] (O. Sissmann).

Mx Siy Oxþ2y−z ðOHÞ2z þ ðx−zÞH2 O þ 2xCO2 →xM þ ySiO2 :

1. Introduction

0009-2541/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.chemgeo.2013.08.031

2þ

þ 2xHCO3

−

ð1Þ

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

Depending on the saturation conditions of the aqueous medium, divalent cations M2+ can react with bicarbonate ions to form solid carbonate minerals as described by the reaction: 2þ

xM

−

þ 2xHCO3 →xMCO3 þ xH2 O þ xCO2 :

ð2Þ

Taken together, those two steps represent the carbonation process, described by the overall reaction (Oelkers et al., 2008): Mx Siy Oxþ2y−z ðOHÞ2z þ xCO2 →xMCO3 þ ySiO2 þ zH2 O:

ð3Þ

The efficiency of the carbonation process partly relies on the availability of cations released by the dissolution of the silicates. Rock formations rich in divalent cations and poor in silica are natural candidates for CO2 geological storage. In ultramafic rocks like peridotites and to a lesser extent, in mafic rocks like basalts, olivine ((Mg,Fe)2SiO4) is one of the most prominent minerals, and presents the advantage of being a very favorable substrate for reaction (3) from a thermodynamic standpoint (Guyot et al., 2011). In the specific case of Mg-rich olivine carbonation, magnesite (MgCO3), usually the most stable of Mg-carbonates, is the carbonate expected to form, since silicate dissolution will release mainly Mg to solution (and to a lesser extent, Fe). Though Fe-carbonates can potentially form as well, as separate phases, Fe is more likely to be incorporated within the Mg-carbonates if the conditions are reducing enough (e.g. Saldi et al., 2013). The carbonation of olivine is thus described by the reaction: ðMg; FeÞ2 SiO4 þ 2CO2 →2ðMg; FeÞCO3 þ SiO2 :

ð4Þ

Moreover, numerous studies dedicated to the measurement of the far-from-equilibrium dissolution kinetics of olivine showed that this mineral is among the fastest dissolving silicates (e.g. Blum and Lasaga, 1988; Pokrovsky and Schott, 2000; Rosso and Rimstidt, 2000; Golubev et al., 2005; Hanchen et al., 2006; Wolff-Boenisch et al., 2011), such that in spite of its lower abundance compared to e.g. plagioclase in basalts, olivine remains one of the main contributors to the release of cations (Gudbrandsson et al., 2011). Despite these observations, aqueous carbonation experiments of μm-sized olivine powders at temperature below 100 °C have shown negligible amount of carbonates formed after several weeks of experiment (e.g. Giammar et al., 2005; Daval et al., 2011; Wang and Giammar, 2013). In order to make carbonation a viable process for sequestering CO2, such a kinetic barrier needs to be understood in order to be, in turn, overcome. Schematically, olivine carbonation can be decomposed into successive steps: (1) species release by olivine dissolution; (2) transport of those dissolved species across a possible interfacial layer made of secondary phases (e.g.: amorphous silica); and (3) carbonate precipitation. Each of these steps, which may occur simultaneously, can potentially be rate-limiting. Regarding the competition between the processes (1) and (3) (Eqs. (1) and (2)), the comparison of magnesite precipitation rates with corresponding forsterite dissolution rates at conditions relevant to geologic CO2 sequestration led some authors (e.g. Saldi et al., 2012) to propose that the slow kinetics of magnesite precipitation may be rate-limiting with respect to the overall process. In qualitative agreement with this idea, previous studies (e.g. Gerdemann et al., 2007) showed that the optimal pH and alkalinity conditions for olivine carbonation rate represent a compromise between promoting the magnesite precipitation step, without lowering too much the dissolution rate of olivine. This observation emphasizes that increasing pH and alkalinity as a consequence of olivine dissolution is not necessarily detrimental to the overall carbonate rate. However, as detailed below, a growing number of studies pointed out that the formation of (transient or stable) secondary phases during the course of Mg-bearing silicate carbonation potentially played an even more important role in influencing the carbonation yields and rates.

187

A first straightforward effect is the limitation of carbonation yields due to the formation of silicate by-products incorporating Mg (e.g. Davis et al., 2009; Dufaud et al., 2009; King et al., 2010; Gysi and Stefansson, 2012). For instance, in their experiments carried out at T ≥ 200 °C, Dufaud et al. (2009) and King et al. (2010) showed that water–CO2–olivine interactions resulted in the formation of serpentine together with magnesite. Dufaud et al. (2009) acknowledged that depending on the reaction conditions, serpentine and magnesite minerals were produced in various proportions, drastically affecting the maximum extent of olivine carbonation. Although secondary Mg-bearing phases could negatively affect the carbonation yield, by-products of the carbonation reaction can also play a role in passivating the surface of the reactants, thereby decreasing the rate of the reaction. Such an effect has been widely documented for weathering reactions as a whole (e.g. Velbel, 1993; Nugent et al., 1998). Several studies emphasized that among the fundamental parameters which favor the protective nature of mineral surface coatings are (1) the molar volume ratio of product to reactant, a ratio larger than 1 leading to passivating properties (Velbel, 1993; Putnis, 2002; Putnis et al., 2005) and (2) the potential crystallographic relationships between coatings and underlying unaltered phases if both reactants and products are crystalline (e.g. epitaxial growth can dramatically affect the dissolution rate of the parent phase (Cubillas et al., 2005)). Regarding carbonation processes, the effect of the two main reaction products (carbonates and amorphous silica, see Eq. (3)) on the carbonation rates has been investigated as well. The corresponding studies revealed that the microstructure of the secondary assemblages and their overall evolution as the reaction proceeds were a supplementary factor which was worth considering. Some key observations are summarized hereafter. Dealing with carbonate minerals, their armoring ability depends both on the nature of the substrate and on the respective rates of carbonate nucleation and growth. For instance, Stockmann et al. (2011, 2013) did not measure any obvious decrease of basaltic glass and diopside dissolution rates at basic pH and T = 70 °C, in spite of the significant extent of CaCO3 precipitation. They concluded that these materials were not favorable substrates for the heterogeneous precipitation of calcite, such that most of the calcite actually formed by homogeneous nucleation, which did not directly affect the surface of the dissolving solids. In their experiments of wollastonite carbonation, Daval et al. (2009a,b) reported that when the reaction was initiated in ultrapure water equilibrated with elevated pCO2, a carbonation extent as high as 95% could be reached without observing any obvious negative feedback effect between carbonate precipitation and wollastonite dissolution. Conversely, when the reaction was initiated in a 0.4 M NaOH solution equilibrated with elevated pCO2, the extent of carbonation leveled to a maximum value of ~80%. The differences between the two series of experiments were ascribed to the relation between fluid composition and calcite grain size: the more alkaline the solution, the smaller the carbonate crystals, and the more compact the coatings. Furthermore, Daval et al. (2009b) observed that nanocrystals of calcite could form within silica rims embedding the wollastonite crystals, thereby decreasing their porosity and ultimately rendering the silica layers passivating. In the case of the transport properties of amorphous silica layers, opposite trends have been reported depending on the experimental conditions and the nature of the substrate. For instance, Daval et al. (2009a,b) showed that the formation of amorphous silica layers on the surface of wollastonite does not significantly affect its dissolution rate. Conversely, Park and Fan (2004) or Teir et al. (2007) proposed that silica layers formed on serpentine considerably slow down its dissolution kinetics. Likewise, a growing number of studies of olivine carbonation suggested that the formation of thin silica layers (b50 nm) could constitute a barrier to the transport of aqueous species (e.g. Bearat et al., 2006; Andreani et al., 2009; Daval et al., 2011; Wang and Giammar, 2013) and ultimately controls the rates of olivine carbonation. Importantly,

188

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

the factors responsible for rendering silica layers passivating are apparently associated to the structure and chemistry of the parent substrate (Schott and Berner, 1983; Cailleteau et al., 2008; Daval et al., 2013) and remain poorly understood. Overall, carbonation reactions can be seen as a chain of successive processes with their own kinetics including the formation of passivating secondary phases, with silicate dissolution at one end and carbonate nucleation and growth at the other. As a result, the effect of olivine dissolution enhancement on Mg-carbonate yield is difficult to predict. In this study, we propose to investigate, using batch experiments, the effect of two main factors increasing olivine dissolution rate: temperature (e.g. Hanchen et al., 2006) and organic ligands added to the fluid (Grandstaff, 1986; Olsen and Rimstidt, 2008; Krevor and Lackner, 2009; Daval et al., 2010; Prigiobbe and Mazzotti, 2011; Bonfils et al., 2012). Our strategy consisted in an integrated approach which combines the time-resolved aqueous measurements of dissolved species, used as input data for aqueous speciation calculations using the thermodynamic code CHESS (van der Lee and De Windt, 2002), and the quantification of newly formed Mg carbonates at the end of the experiment. Solid products formed at the olivine/fluid interface have been characterized down to the nm-scale using transmission electron microscopy (TEM) on thin sections prepared by focused ion beam (FIB) milling. The results are discussed in light of the modification of the solubility and nature of the secondary phases formed at the olivine/water interface, and of their ultimate impact on the dissolution rate of olivine. 2. Materials and methods 2.1. Experimental setup 2.1.1. Starting materials This study was conducted on a batch of gem quality San Carlos olivine, with an average composition of (Mg0.88Fe0.12)2SiO4, as determined from 10 electron microprobe analyses (Table 1). Minor quantities of iron oxide impurities were removed manually under binoculars. Olivine crystals were crushed and sieved to recover the 100–300 μm grain size fraction, and subsequently cleaned in ethanol multiple times to remove fine particles. The powder was then rinsed with ultra-pure deionized water, and dried overnight at 50 °C. Scanning electron microscopy (SEM) confirmed that the olivine grain surfaces were devoid of finer particles. The specific surface area (SSA) of the sample was determined to be 0.041 m2·g−1, using a 3-point BET method with Kr as absorbent gas. 2.1.2. Experimental medium and conditions Overall, three experiments were carried out at pCO2 = 280 bars. Because citrate has been reported as being one of the most effective ligand to enhance olivine dissolution (Grandstaff, 1986), the first experiment was conducted at 90 °C for 22 days in a solution where 0.087 M of Na-citrate (Na3C6H5O7·2H2O) was added to ultra-pure water. We used a surface area to volume ratio (SA/V) of 656 m−1 (solid mass to volume of solution ratio of 16 g/L: 4.0010 g of olivine, 250 mL of solution). These experimental conditions (temperature, pCO2 and SA/V ratio) were purposefully close to those reported in Daval et al. (2011) carried out in pure water, for the sake of comparison. Two other experiments were conducted in ultra-pure water at 120 °C and 170 °C, for 55 and 48 days respectively. A SA/V ratio of 497 m−1 (12 g/L: 3.0032 g of olivine, 250 mL of solution) was used in the 120 °C experiment. A lower SA/V ratio of 249 m−1 (6 g/L: 1.521 g of olivine, 250 mL of Table 1 Olivine composition given by weight percentage of oxides as determined by electron probe microanalysis.

San Carlos olivine

SiO2

CaO

MnO

FeO

NiO

MgO

Total

40.65

0.08

0.11

11.19

0.35

47.75

100.13

solution) was used at 170 °C to better resolve the Mg and Si concentration increase in the first stages of the experiment. 2.1.3. Experimental equipment and protocols Carbonation experiments were conducted in a stirred batch reactor which allows regular sampling of internally filtered fluid at experimental conditions. The reaction cell (500 mL) is made of high-grade titanium, the surface of which has been passivated at 400 °C and covered with an inner Teflon jacket. The samples together with the aqueous medium were loaded into the reactor, which was then sealed with a lenticular Ti-gasket. In order to flush out O2 from the autoclave, CO2 was injected through the solution sampling tube and a CO2 circulation was maintained at 60 bars for 30 min. The autoclave was then loaded into a furnace, and the temperature was raised to its nominal value, with pCO2 increasing simultaneously. During the run, the aqueous solution inside the autoclave was continuously stirred with non-reactive Ti rotating pals, in order to avoid the formation of chemical gradients. The sampled fluid was internally filtered (2 μm pore size frit) and recovered into 50 mL syringes connected to the sampling tube of the apparatus by a PTFE 0.2 μm external filter. The procedure consisted of collecting 2 successive aliquots, the first one (1 mL) – used to flush the sampling line – being systematically discarded, while the second one (2 mL) was spared for analysis of the solution chemistry. The syringes were already pre-filled with a diluting volume of ultra-pure water, in order to prevent silica-phase supersaturation in the sampled fluid and subsequent precipitation at ambient conditions. After each sampling, additional CO2 was injected in the reaction cell via the sampling tube in order to both flush the sampling line and readjust pCO2. At the end of the experiment, the aqueous fluid was extracted still at high temperature from the reactor by the sampling tube to minimize the precipitation of secondary phases during the final quench phase. 2.2. Product characterization and analytical procedures 2.2.1. Fluid analyses The aqueous samples collected for each experiment were diluted 20 fold in ultrapure water and acidified with HNO3 2 vol.%. They were subsequently analyzed by inductively coupled plasma atomic emission spectrometry (ICP-AES) (Varian 720 ES) at ISTerre, (Grenoble) for total aqueous concentrations of Mg, Si, Fe, Mn and Ni. The analytical uncertainties were below 5%. 2.2.2. Carbonate identification and quantification Crystalline carbonates were identified by X-ray powder diffraction (XRD) using a Rigaku ultraX18HFCE Bragg–Brentano diffractometer equipped with a rotating copper anode. Carbonate phase quantification was achieved using a classical technique of CO2 extraction by carbonate decomposition in orthophosphoric acid (McRéa, 1950; Cornides and Kusabe, 1977). A precise mass of run products (mp) was placed in a Pyrex tube on a vacuum line. After a vacuum b10−5 mbar was obtained, the tube was isolated from the vacuum and the sample was immersed into orthophosphoric acid (H3PO4) and heated at 120 °C for 2 h, in order to dissolve the carbonates likely to have formed (here, MgCO3) and release CO2, along the following reaction: 3MgCO3 þ 2H3 PO4 →Mg3 ðPO4 Þ2 þ 3CO2 þ 3H2 O:

ð5Þ

The liberated gases were then released into the vacuum line and H2O was cryogenically separated from CO2 at −140 °C into liquid nitrogen. The resulting CO2 pressure was measured and converted to a molar quantity using internal standards. The obtained value was then converted into the mass of Mg-carbonate in the run products (mMgCarb) having reacted with orthophosphoric acid. The carbonation yield (expressed in wt.%) is subsequently determined as the ratio between the total mass of produced carbonate and the total mass of the sample (mMgCarb/mp). The detection

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

limit is estimated to 10−6 mol of CO2, i.e. 400 ppm of Mg-carbonate for a typical analysis of 200 mg of run products. 2.2.3. Water quantification in run products A mass of mp = 40 mg was introduced into a sealed quartz tube connected to a vacuum line, and subsequently heated at 50 °C for an hour to get rid of any adsorbed water. Then, the sample was heated to 1100 °C for an hour under a pressure of 1 mbar of pure O2 obtained by thermal breakdown of CuO. This procedure repeated on a sample that had previously been heated to 1100 °C for half an hour produced negligible amounts of water in agreement with our experimental blanks (about 0.5 μmol). The degassed CO2 and H2O were trapped in liquid nitrogen at −195 °C, and separated by heating the temperature up to −140 °C in order to keep only H2O, while incondensable gases and liberated CO2 were pumped out of the line. Water was then reduced to H2, using hot uranium (heated at 800 °C on a furnace), and measured using a calibrated mercury Toepler pump. The amount of water (expressed in wt.%) was calculated as the following ratio: mH2O/mp.

189

the database. The acid dissociation constants of citric acid and their temperature dependence were computed from Bénézeth et al. (1997). The formation constants of Mg, Na, and Fe citrate complexes are listed in Table 2, with detailed calculations reported in Appendix A1. Activity coefficients for aqueous species were calculated using the Davies equation, and the electrical balance was maintained by using pH as an adjustable parameter. Initial pH was calculated to be 4.96 for the experiment at 90 °C with Na-citrate, 3.14 and 3.31 for the experiments in pure water, at 120 °C and 170 °C, respectively. The aqueous magnesium and silica concentrations measured by ICP-AES served as input values to the geochemical code for computing, all along the run, the evolution of pH (Tables 3 to 5) and saturation indices (SI) of the solid phases, i.e. forsterite and magnesite (Mg end-members, taken respectively as a model for the experimental olivine and magnesite used in the present study) and amorphous silica. The Gibbs free energy (ΔGr) of dissolution reactions of various solid phases, including SiO2(am), forsterite, and magnesite, are reported in Tables 3 to 5, and were calculated from the saturation indices (SI) computed with CHESS, following:

2.2.4. Rietveld quantification The proportion of non-crystalline material (WNC) was determined by Rietveld refinement with corundum as internal standard (IS), as follows:

where R is the gas constant (J·mol−1·K−1) and T the temperature (K).

W NC ¼ 100ðW ISR −W IS Þ=W ISR

3. Results

ð6Þ

where WIS is the weight percentage of the corundum added to the sample (20 wt.%) whereas WISR is the weight percentage of corundum determined by Rietveld refinement of the XRD pattern. Practically, corundum was ground together with 100 mg of sample in ethanol using a McCrone micronizing mill. The resulting slurry was oven-dried at 40 °C before being prepared as a randomly oriented mount. A silicon plate was used as sample holder to minimize the background contribution. XRD patterns were recorded with a Bruker D5000 powder diffractometer equipped with a SolX Si(Li) solid state detector from Baltic Scientific Instruments using CuKα1 + 2 radiation and with a spinner. Intensities were recorded at 0.026° 2θ step intervals from 5 to 90° (10 s counting time per step). 2.2.5. Observation by electron microscopy The reaction products were placed on adhesive carbon tape, gold coated, and characterized by scanning electron microscopy (SEM) (Zeiss Ultra), operated at 15 kV for standard imaging and analyzed chemically using an energy dispersive X-ray spectrometer (50 mm2 detector; X-Max™ from Oxford Instrument). In addition, a low acceleration voltage (5 kV) was used to produce high resolution secondary electron (SE) images of olivine surface at a 3 mm working distance. Focused ion beam (FIB) milling was used to produce ultra-thin sample sections for analytical transmission electron microscopy (ATEM) on a Zeiss neon EesB40 FIB/FEG-SEM system (IEMN, Lille). The samples were carbon-coated beforehand in order to minimize the production of artifacts on the surface due to Pt and Ga beam interactions during the milling procedure (Lee et al., 2007). TEM analyses were carried out with a JEOL 2100F (FEG) operated at 200 kV (IMPMC, UPMC, Paris). 2.3. Thermodynamic calculations and kinetic modeling The CO2 fugacity was calculated based on pCO2 = 280 bars and using the software Thermosolver (Barnes and Koretsky, 2004) to be 140 bars for the experimental run at 90 °C, 170 bars at 120 °C and 208 bars at 170 °C Starting pH and dissolved CO2 concentrations were calculated with the geochemical code CHESS (Van der Lee and De Windt, 2002) using the LLNL EQ3/6 database. However, because citrate species and corresponding Mz+-citrate complexes are not incorporated into the database, a comprehensive effort was performed to implement them in

ΔGr ¼ R T SI ln 10

ð7Þ

3.1. Olivine carbonation at 90 °C with 0.1 M Na-citrate The evolution of aqueous Mg, Fe, Ni, Mn and Si concentrations in the fluid is plotted as a function of sampling time in Fig. 1a–b; the corresponding data are listed in Table 3. The evolution of aqueous Mg and Si concentrations in a citrate free solution at 90 °C (modified from Daval et al., 2011) is given in Fig. 1d for comparison. From Fig. 1a and b, it appears that the dissolution occurs following three steps. In the first step (t ≤ 1 day), the concentration of the dissolved species (Mg, Fe, Si, Mn, Ni) increases rapidly (several mmols per day) following concentration ratios which are consistent with stoichiometric olivine dissolution. In a second step (1 b t ≤ 5 days), [SiO2(aq)] starts decreasing until it reaches amorphous silica (SiO2(am)) saturation level while the aqueous concentration for all of the other elements keeps increasing in solution, with a slower release rate. Finally, in a last stage (t N 5 days), a constant but slow release of Mg, Fe, Mn and Ni is achieved (pseudo-plateau) whereas [SiO2(aq)] remains constant at the SiO2(am) saturation level. As shown in Fig. 1c, and Table 3, the aqueous concentrations of Mg, Fe, Ni and Mn are released congruently throughout the olivine dissolution process whereas silica starts departing from the congruent behavior after only one day (at the beginning of the second stage, see Table 3). The carbonate phase quantification by CO2 extraction indicates a small carbonation yield at the end of the experiment, of about 0.67 ± 0.04 wt.%. Based on a mass balance calculation on Mg and Table 2 Citrate complexation constants. T 2+

M = Mg

2+

M = Fe M = Na+

25 37 25 25

°C °C °C °C

ML

MHL

MLH2

ML2

ML2H

ML2-H

M2L

4.89 4.71 4.55 1.35

9.00 9.01 8.90 –

12.13 12.85 – –

– 5.44 6.88 –

– – 12.02 –

– – −1.06 –

– – – 1.67

Stability constants reported as log(K) and added to the EQ3/6 database, recalculated from ionic strength-dependant stability constants found in the literature. M stands for metal ions, H for hydrogen, and L for fully deprotonated citrate ligands (i.e. for Mg2+, ML = MgCitrate−, MHL = MgHCitrate, MLH2 = MgH2Citrate+ and ML2 = Mg(Citrate)4− 2 . Thus, for instance Mg2+ + Citrate3− → MgCitrate− is equivalent to M + L → ML). Constants for Mg complexes were calculated from Meyer (1974), Pearce (1980) and Covington and Danish (2009). Constants for Fe complexes were calculated from Konigsberger et al. (2000), and for Na complexes in Zelenina and Zelenin (2005).

190

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

Table 3 Summary of the results obtained for the 90 °C experiment in the presence of Na-citrate. No. sample

t (days)

[Mgaq] [SiO2aq] (mmol/L) (mmol/L)

[Niaq] (mmol/L)

[Mnaq] [Feaq] (mmol/L) (mmol/L)

Ratio [Mgaq]/ [SiO2aq]

Ratio [Mgaq]/ [Mnaq]

Ratio [Mgaq]/ [Niaq]

Ratio [Mgaq]/ [Feaq]

log rMg (mol/m2/s)

Calculated pH

ΔGr forsterite (kJ/mol)

ΔGr SiO2(am) (kJ/mol)

ΔGr magnesite (kJ/mol)

ΔGr nahcolite (kJ/mol)

A4-0 A4-1 A4-2 A4-3 A4-4 A4-5 A4-6 A4-7 A4-8 A4-9 A4-10 A4-11 A4-12 A4-13 A4-14 A4-15 A4-16

0 0.11 0.24 0.36 0.80 1.16 1.77 2.85 3.83 5.27 7.05 9.12 12.09 14.80 17.29 19.82 21.97

nd 3.01 7.15 15.54 31.29 39.23 50.29 61.53 66.77 73.11 78.27 82.19 87.25 90.40 91.53 94.39 96.33

nd 0.015 0.032 0.066 0.124 0.158 0.200 0.246 0.265 0.291 0.309 0.327 0.347 0.355 0.361 0.369 0.385

nd 0.004 0.010 0.021 0.041 0.051 0.066 0.080 0.087 0.095 0.102 0.108 0.115 0.117 0.119 0.122 0.125

– 1.8 1.8 2.0 2.6 4.9 6.9 9.3 10.6 12.1 13.4 14.4 15.7 16.5 17.2 17.8 18.1

– 680 724 745 764 766 766 767 766 767 765 764 762 776 767 771 768

– 202 226 237 252 248 252 250 252 251 253 251 252 255 254 256 250

– 9.6 10.0 10.2 10.5 10.5 10.5 10.6 10.6 10.6 10.6 10.6 10.6 10.7 10.6 10.7 10.6

nd nd −6.50 −6.19 −6.48 −6.68 −6.78 −7.03 −7.32 −7.41 −7.60 −7.79 −7.84 −8.01 −8.43 −8.04 −8.14

4.96 4.99 5.03 5.1 5.22 5.27 5.33 5.37 5.4 5.42 5.44 5.45 5.46 5.47 5.47 5.48 5.3

– −100.7 −91.0 −80.5 −68.8 −65.7 −60.7 −56.3 −54.5 −52.5 −51.0 −50.0 −48.7 −48.0 −47.9 −47.2 −50.8

– −3.71 −1.10 1.00 2.22 1.00 0.74 0.45 0.28 0.16 0.08 −0.01 −0.08 −0.13 −0.22 −0.23 −0.22

– −14.95 −11.38 −7.20 −1.93 0.22 2.87 5.21 6.18 7.24 8.03 8.58 9.24 9.62 9.75 10.06 8.31

−15.4 −15.2 −14.7 −13.9 −13.6 −13.2 −12.8 −12.7 −12.5 −12.4 −12.4 −12.3 −12.2 −12.2 −12.1 −12.1

nd 1.67 3.97 7.95 11.90 7.96 7.29 6.63 6.28 6.02 5.86 5.69 5.57 5.48 5.32 5.30 5.32

nd 0.31 0.72 1.53 2.98 3.74 4.77 5.83 6.31 6.90 7.39 7.78 8.26 8.46 8.60 8.85 9.09

The first two columns indicate the name of the sample and the time duration after which it was taken. The five following columns contain, respectively, the aqueous concentrations of Mg(aq), SiO2(aq), Ni(aq), Mn(aq) and Fe(aq) measured by ICP-AES (mmol·L−1), with uncertainties of +/−5%. The subsequent columns report the ratio of measured concentrations, with uncertainties of +/−11%. The next columns report respectively the dissolution rate based on Mg concentration, and calculated pH. Finally, the four following columns gather the values of Gibbs free energy of dissolution (kJ·mol−1) of forsterite, amorphous silica (SiO2(am)), magnesite and nahcolite at T = 90 °C and pCO2 = 280 bars. Those ΔGr values were converted from SI values provided by the CHESS code following Eq. (7). The stoichiometric element ratios of the dissolving olivine are: [Mgaq] / [SiO2aq] ≈ 1.8; [Mgaq] / [Mnaq] ≈ 764; [Mgaq] / [Niaq] ≈ 246; [Mgaq] / [Feaq] ≈ 8.

assuming that no other Mg solid-phase than MgCO3 was formed, from both CO2 extraction and final Mg concentration in the sampled solution (t = 22 days), it can be inferred that 39.8 ± 1.9 mol% of the starting olivine has been consumed after 22 days in this experiment. It is noteworthy that 82% of this final extent of reaction was reached within 5 days of reaction only.

The formation of Mg carbonates was confirmed by microscopic observations. SEM observations reveal that ovoid-shaped, Mg- and C-rich phases have precipitated on the olivine surface (Fig. 2a–c). Examination of FIB cross-section and electronic diffraction by TEM indicate that those ovoid grains are magnesite single-crystals (Fig. 2d). The odd morphology of the Mg-carbonates, usually rhomboedric, can

Table 4 Summary of the results obtained for the 120 °C experiment. No. sample

t (days)

[Mgaq] mmol/L

[SiO2aq] mmol/L

[Niaq] (mmol/L)

[Mnaq] (mmol/L)

[Fe2aq] (mmol/L)

A2-0 A2-1 A2-2 A2-3 A2-4 A2-5 A2-6 A2-7 A2-8 A2-9 A2-10 A2-11 A2-12 A2-13 A2-14 A2-15 A2-16 A2-17 A2-18 A2-19 A2-20 A2-21 A2-22 A2-23 A2-24 A2-25 A2-26 A2-27

0 0.1 0.72 1.26 1.79 2.37 3.05 3.85 4.84 6.18 7.81 9.28 11.85 14.04 16.02 20.01 22.11 26.20 29.73 34.86 35.92 36.94 38.30 39.96 42.11 44.91 54.07 54.73

nd 0.90 8.92 9.63 11.55 10.56 11.50 10.77 11.09 12.22 11.44 11.76 12.95 12.29 13.33 13.37 12.78 12.31 13.48 14.05 13.46 17.85 18.65 21.22 22.73 24.12 24.07 23.71

nd 0.48 4.86 5.22 6.24 5.72 6.25 5.96 6.01 6.62 6.28 6.35 6.93 6.61 7.10 7.13 6.84 6.56 6.96 7.18 7.24 8.35 10.22 12.26 11.34 11.16 11.71 11.43

nd 0.00518 0.03304 0.03474 0.04048 0.03702 0.04064 0.03854 0.04014 0.04295 0.04257 0.04191 0.04832 0.04590 0.04918 0.04897 0.04820 0.04781 0.05331 0.04684 0.05008 0.04715 0.03958 0.02363 0.01055 0.00922 0.00698 0.00502

nd 0.00188 0.01329 0.01383 0.01689 0.01480 0.01652 0.01588 0.01591 0.01763 0.01713 0.01698 0.01839 0.01795 0.01994 0.01941 0.01925 0.01818 0.01906 0.01885 0.01762 0.01086 0.00283 0.00165 0.00132 0.00106 0.00082 0.00060

nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd 0.00000 0.01881 0.06982 0.14063 0.09042 0.03400 0.01559 0.01212 0.00194 nd

Ratio [Mgaq]/ [SiO2aq]

Ratio [Niaq]/ [Mnaq]

Ratio [Mgaq]/ [Niaq]

1.89 1.83 1.85 1.85 1.84 1.84 1.81 1.85 1.85 1.82 1.85 1.87 1.86 1.88 1.88 1.87 1.88 1.94 1.96 1.86 2.14 1.82 1.73 2.00 2.16 2.06 2.07

2.8 2.5 2.5 2.4 2.5 2.5 2.4 2.5 2.4 2.5 2.5 2.6 2.6 2.5 2.5 2.5 2.6 2.8 2.5 2.8 4.3 14.0 14.3 8.0 8.7 8.5 8.4

175 270 277 285 285 283 280 276 284 269 281 268 268 271 273 265 258 253 300 269 379 471 898 2155 2617 3447 4723

Calculated pH

ΔGr forsterite (kJ/mol)

ΔGr SiO2(am) (kJ/mol)

ΔGr magnesite (kJ/mol)

ΔGr talc (kJ/mol)

3.14 3.58 4.43 4.45 4.52 4.49 4.52 4.5 4.51 4.55 4.52 4.53 4.57 4.55 4.58 4.58 4.56 4.55 4.58 4.6 4.59 4.69 4.7 4.75 4.77 4.79 4.79 4.79

– −107.3 −63.1 −62.0 −58.3 −60.0 −58.4 −59.5 −59.1 −57.5 −58.5 −58.0 −56.3 −57.2 −55.8 −55.7 −56.5 −57.2 −55.7 −55.0 −55.6 −51.0 −49.7 −47.3 −46.6 −45.8 −45.6 −45.9

– −9.14 −1.57 −1.34 −0.75 −1.04 −0.75 −0.90 −0.87 −0.76 −0.73 −0.69 −0.41 −0.56 −0.33 −0.32 −0.45 −0.59 −0.40 −0.29 −0.27 0.20 0.86 1.46 1.20 1.15 1.31 1.23

– −18.76 −0.41 0.16 1.54 0.87 1.51 1.02 1.24 1.96 1.47 1.67 2.39 2.00 2.61 2.63 2.29 2.02 2.69 2.99 2.68 4.75 5.06 5.97 6.45 6.87 6.86 6.75

– −113.13 −27.83 −25.15 −18.70 −21.87 −18.79 −20.88 −20.11 −17.48 −18.83 −18.07 −14.78 −16.56 −13.82 −13.70 −15.24 −16.62 −13.83 −12.52 −13.35 −5.25 −1.70 3.41 3.85 4.90 5.48 4.84

The first two columns indicate the name of the sample and the time duration after which it was taken. The five following columns contain, respectively, the aqueous concentrations of Mg(aq), SiO2(aq), Ni(aq), Mn(aq) and Fe(aq) measured by ICP-AES (mmol·L−1), with uncertainties of +/−5%. The subsequent columns report the ratio of measured concentrations, with uncertainties of +/−10.5%. The next column reports calculated pH. Finally, the four following columns gather the values of Gibbs free energy of dissolution (kJ·mol−1) of forsterite, amorphous silica (SiO2(am)), magnesite, and talc at T = 120 °C and pCO2 = 280 bars. Those ΔGr values were converted from SI values provided by the CHESS code following Eq. (7). The stoichiometric element ratios of the dissolving olivine are: [Mgaq] / [SiO2aq] ≈ 1.8; [Niaq] / [Mnaq] ≈ 3; [Mgaq] / [Niaq] ≈ 246.

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

191

Table 5 Summary of the results obtained for the 170 °C experiment. No. sample

t (days)

[Mgaq] (mmol/L)

[SiO2aq] mmol/L

[Niaq] (mmol/L)

[Mnaq] (mmol/L)

[Feaq] (mmol/L)

A3-ini A3-0 A3-1 A3-2 A3-3 A3-4 A3-5 A3-6 A3-7 A3-8 A3-9 A3-10 A3-11 A3-12 A3-13 A3-14 A3-15 A3-16 A3-17 A3-18 A-19 A-20 A-21 A-22

0 0.05 0.10 0.23 0.36 0.51 0.63 0.76 0.98 1.77 2.26 2.82 3.96 4.85 5.84 7.04 8.92 11.13 12.96 15.96 22.00 33.00 40.00 48.00

nd 0.63 2.32 3.45 4.60 5.43 5.98 6.40 7.06 8.62 8.22 8.21 7.41 7.07 6.52 5.97 4.69 3.59 3.07 3.02 3.12 3.64 3.56 3.66

nd 0.34 1.24 1.90 2.51 3.12 3.69 4.03 4.85 7.09 8.42 9.74 11.14 11.86 12.20 12.77 12.96 12.59 12.44 12.34 13.33 12.83 13.88 13.93

nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd

nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd

nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd

Ratio [Mgaq]/ [SiO2aq] 1.89 1.87 1.82 1.83 1.74 1.62 1.59 1.46 1.22 0.98 0.84 0.67 0.60 0.53 0.47 0.36 0.29 0.25 0.24 0.23 0.28 0.26 0.26

Calculated pH

ΔGr forsterite (kJ/mol)

ΔGr SiO2(am) (kJ/mol)

ΔGr magnesite (kJ/mol)

ΔGr antigorite (kJ/mol)

ΔGr talc (kJ/mol)

3.31 3.74 4.21 4.36 4.46 4.52 4.56 4.58 4.62 4.69 4.68 4.68 4.64 4.62 4.59 4.56 4.47 4.37 4.31 4.31 4.32 4.38 4.37 4.38

– −92.3 −64.6 −56.2 −50.3 −46.7 −44.5 −43.1 −41.0 −36.2 −36.4 −35.8 −37.0 −37.5 −38.7 −40.0 −43.9 −48.5 −51.2 −51.5 −50.7 −48.2 −48.3 −47.8

– −13.12 −8.36 −6.78 −5.76 −4.96 −4.34 −4.01 −3.54 −1.93 −1.30 −0.76 −0.27 −0.04 0.07 0.24 0.29 0.18 0.14 0.11 0.39 0.25 0.54 0.56

– −14.88 −3.39 0.04 2.47 3.84 4.63 5.18 5.98 7.57 7.19 7.18 6.37 5.99 5.33 4.62 2.63 0.38 −0.96 1.10 −0.82 0.50 0.31 0.54

– −1640 −926 −708 −557 −464 −405 −367 −313 −182 −178 −161 −183 −193 −221 −250 −343 −455 −521 −529 −505 −447 −446 −434

– −104.9 −51.4 −34.8 −23.4 −16.1 −11.2 −8.3 −4.0 7.2 8.6 10.7 10.2 10.0 8.5 7.0 1.3 −5.9 −10.1 −10.6 −8.7 −5.3 −4.7 −3.9

The first two columns indicate the name of the sample and the time duration after which it was taken. The five following columns contain, respectively, the aqueous concentrations of Mg(aq), SiO2(aq), Ni(aq), Mn(aq) and Fe(aq) measured by ICP-AES (mmol·L−1), with uncertainties of +/−5%. The subsequent column reports the ratio of measured concentrations of Mg (aq) and SiO2(aq), with uncertainties of +/−10.5%. The next column reports respectively the calculated pH. Finally, the five following columns gather the values of Gibbs free energy of dissolution (kJ·mol−1) of forsterite, amorphous silica (SiO2(am)), magnesite, antigorite and talc at T = 170 °C and pCO2 = 280 bars. Those ΔGr values were converted from SI values provided by the CHESS code following Eq. (7). The stoichiometric element ratios of the dissolving olivine are: [Mgaq] / [SiO2(aq)] ≈ 1.8.

c [Mg aq]

100

12

.......... Congruent

[Fe aq]

10

[SiO2 aq]

SiO2 am

[Fe aq] (mmol/L)

[x] (mmol/L)

10

MgCO3 saturation

0.8

dissolution [Fe aq]

0.6

8

[Ni aq]

6 4

[Mn aq]

0.2

2

saturation

0

0 0

50

1 0

4

8

12

16

20

d 0.4

20

Modified from Daval et al, 2011

MgCO3 saturation

16

[x] (mmol/L)

[x] (mmol/L)

100

[Mg aq] (mmol/L)

24

b

0.3

[Ni aq]

0.2 0.1

4

8

12

t (days)

16

12 8

[Mn aq]

4

20

0

0 0

0.4

24

[Ni aq], [Mn aq] (mmol/L)

a

[Mg aq]

[SiO2 aq] 0

4

8

(SiO2)am saturation

12

16

20

24

t (days)

Fig. 1. (a), (b), (c) Olivine dissolution at 90 °C with citrate added to the solution monitored by time-resolved aqueous-fluid sampling. (a) Mg, Si and Fe concentrations of aliquots sampled as a function of time. Note the attainment of a quasi-steady state for Mg and Fe after 6 days of reaction, within experimental uncertainties. Amorphous silica saturation is temporarily overstepped. At the same time, the release of Mg and Fe begins slowing down. (b) Mn and Ni concentrations of the same aliquots. They follow the same trends as Mg. (c) Aqueous Fe, Mn and Ni concentrations plotted as a function of Mg concentration, showing a stoichiometric dissolution of the mineral for divalent cations, throughout the duration of the experiment. (d) (adapted from Daval et al., 2011) Macroscopic measurements of olivine dissolution rate at 90 °C in pure water, showing that a quasi-steady state is attained for Mg, well below magnesite saturation, while SiO2(aq) reaches equilibrium with respect to SiO2(am).

192

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

d

a

Olivine

006 018

Magnesite Magnesite

silica-free interface

2 µm 400 nm

b

Magnesite

c

Amorphous silica

Olivine 20 µm

4 µm

20 nm

Fig. 2. Electron microscopy analyses of reaction products from the carbonation of olivine, obtained at 90 °C with added citrate. (a), (b) Ovoid-shaped Mg-carbonates on the olivine surface, with secondary Si-phases. (b) Note the typical conical-shaped morphology of the etch pits formed on olivine (lower left). (c) Ovoid-shaped Mg-carbonates embedded in a silica-rich matrix, together with secondary silica phases. (d) TEM on a FIB thin section cut at the surface of one grain from the experiment. Analyses show no apparent silica layer at the carbonate/olivine interface. However, silica formed on the olivine surface which was exposed to the solution.

be explained by the presence of impurities and/or of organic ligands in the aqueous medium, which are known to modify the morphology of precipitated carbonate minerals (e.g. Montes-Hernandez et al., 2008). In addition, magnesite crystals are partially embedded in the interfacial silica layer covering the olivine surface (Fig. 2c). However, there is no apparent silica at the magnesite/olivine interface (Fig. 2d). Taken together, these observations suggest that interfacial silica deposition postdates Mg-carbonate crystallization onto the olivine surface. Speciation modeling with the CHESS code and estimations of olivine dissolution rate bring additional first-order information which must however be considered with great caution due to the uncertainties existing on the thermodynamic data of citrate speciation at high temperature (see Appendix A1). First, the release of divalent cations from olivine is dramatically slowed down after 5 days whereas the fluid remains highly undersaturated with respect to olivine (ΔGr = −51 kJ/mol calculated at the end of experiment). Second, the solution is undersaturated with respect to all Mg-carbonate included in the CHESS database except magnesite (ΔGr = 8.3 kJ/mol corresponding to a saturation index of SI = 1.20) and close to saturation with respect to SiO2(am) (ΔGr = −0.22 kJ/mol) at the end of the experiment. Moreover, the solution is not supersaturated with respect to any other Mg-bearing secondary silicate or oxide, or with respect to Na-carbonate (nahcolite), despite the large concentrations of Na in solution. Finally, it is calculated that Fe(aq) is mainly complexed by citrate ligands, which is in fair agreement with the observation that Fe is actually easily measurable in the sampled aliquots. 3.2. Olivine carbonation at 120 °C in pure water Compared to the previous experiment, the general evolution of [Mg] and [SiO2(aq)] is more complex and is characterized by a two-step release of these two species in the aqueous medium (Fig. 3a–b). The first step (t b 36 days) shows a very steep rise (t b 2 days) in the concentration of all elements (but Fe, whose concentration was below

the detection limit) before it progressively levels to a plateau for 2 b t b 36 days. Despite minor variations probably due to the uncertainties on the fluid sampling and analysis, the release of Mg cations dramatically slows down – if not stops – when the solution approaches saturation with respect to SiO2(am) (Fig. 3a and Table 4). The aqueous [Mg]/[Si] ratio remains around 2 (between 1.82 and 1.96), slightly above the stoichiometric ratio in the primary mineral (1.76). At t ≥ 36 days, [Mg] and [SiO2(aq)] increase abruptly again, suggesting the resumption of olivine dissolution. In contrast to the first dissolution stage, the evolutions of both major and minor elements are decoupled: while [Mg] and [SiO2(aq)] increase, the concentration of Mn and Ni steadily drops till t = 44 days. The [Fe] behavior is the most spectacular with the occurrence of a peak (t = 35 days) which seems to premise the resumption of olivine dissolution. Based on data from three sampled aliquots only, at t N 44 days, it seems that the aqueous Mg and Si concentrations level to a second plateau, with a [Mg]/[Si] ratio close to 2 (between 2.06 and 2.16). Similarly, Mn and Ni reach a second plateau at lower concentrations. Carbonate quantification leads to a carbonation yield of 2.2 wt.% at the end of the experiment. Assuming that (1) carbonate products are Mg-carbonates (Fig. 4a), and (2) no Mg-bearing phase but Mg-carbonate has formed, in fair agreement with SEM observations, mass balance calculation on Mg-content yields an overall olivine consumption of 12.8 mol% after 55 days. From the aqueous [Mg]/[Si] ratios, it can be inferred that a secondary silicon-bearing phase must have also precipitated. The absence of any new crystalline silicate phase in the XRD patterns suggests this phase to be amorphous. The olivine surface appears to be extensively covered by a Si-rich layer which is composed of silica spherules (see Fig. 4b–c). While widespread on the olivine surface, the Si-rich coating phase is absent from the carbonate surfaces. Consequently, this silica-rich phase most likely formed during the experiment rather than upon quench, otherwise it would have also covered the carbonates. Furthermore, detailed examination of a carbonate–olivine

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

c

25

9.5

20 8.5 15

[Mg aq]

SiO2 am saturation

5

0.15

[SiO2 aq]

0 0.10

[Feaq]

[Niaq]

0.05

[Mnaq]

0

4

8

12

16

20

24

28

32

36

40

44

48

Magnesite

6.5

5.5

4.5

3.5 -3.5

0.00 56

52

7.5

SiO2 am

b

Log ( a Mg2+ / H+2 )

MgCO3 saturation

10

[Feaq, Mnaq, Niaq] (mmol/L)

[Mgaq, SiO2 aq] (mmol/L)

a

193

t (days)

-2.5

-1.5

-0.5

Log ( aSiO2 )

Fig. 3. Olivine dissolution/carbonation at 120 °C monitored by time-resolved aqueous-fluid sampling. (a) Mg and Si concentrations of aliquots sampled as a function of time. (b) Minor and trace element concentrations as a function of time (note that Fe cannot be measured in solution before the 34th day of reaction; see text for interpretation). (c) Activity diagram of log(aSiO2(aq)) vs log(aMg2+/a2H+) showing the reaction pathway of the solution. The reaction successively exceeds the saturation with respect to magnesite, SiO2(am) and talc.

interface from a FIB cross-section under TEM shows the presence of interfacial silica phases over 100 nm thick (Fig. 5a–b). High resolution TEM images indicate that the crystalline olivine contacts a zone which contains a few amorphous domains (Fig. 5c), while the outermost fluid/solid interface consists of a fully amorphous silica layer (cf. Fig. 5d and f). In addition,

crystalline iron-rich phases appear to have formed within this amorphous layer (Fig. 5e–g). Speciation calculations reveal that fluid saturation with respect to magnesite is reached after two days (Fig. 3a). The solution remains highly undersaturated with respect to olivine throughout the duration

b Magnesite

a SiO2 (am) 1 µm

Magnesite

c

b

Magnesite

c 10 µm

SiO2 (am)

1 µm

Fig. 4. SEM observations of reaction products at 120 °C. (a) Magnesite precipitated on the surface of olivine. (b) and (c) High-resolution SEM observations (WD: 3 mm and acceleration voltage: 2 kV) of silica formed on the olivine surface. The absence of those “marbles” on the carbonates suggests that their presence does not result from quenching.

194

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

a

b

Olivine surface

Magnesite

e

covered with SiO2am

g

f

b

d

c Olivine

Magnesite

1 µm

c

d

Olivine

Assemblage of olivine and and SiO2 (am)

Assemblage of olivine and and SiO2 (am)

SiO2(am)

10 nm

e

5 nm

f

Magnesite

g Fe-rich phase

Fe-rich phase

Fe-rich phase

Olivine SiO2(am)

SiO2(am)

50 nm

50 nm

SiO2(am)

200 nm

Fig. 5. HRTEM on FIB thin sections of the carbonate/silicate interface formed at 120 °C. Rectangles indicate the zoomed areas. (a) SEM image of the location of the FIB thin section (indicated by the red line). (b) TEM view of the phases formed at the interface between the olivine and the carbonate. (c) HRTEM of the innermost interface. Olivine lattice fringes are easily identified, and a few amorphous domains are occasionally evidenced. (d) HRTEM of the outermost interface with the carbonate, showing the presence of amorphous silica. (e), (f) nm-sized discontinuous band of Fe-rich phases located within the SiO2(am) layer. (g) SiO2(am) layer containing Fe-rich phase exposed to the fluid.

of the experiment (ΔGr b −45 kJ/mol at the end of the run). With respect to pure, anhydrous amorphous silica, the solution is slightly undersaturated for 2 b t b 36 (−1.0 b ΔGr b −0.3 kJ/mol). For t N 42 days, the solution is supersaturated with respect to all SiO2 polymorphs included in the EQ3/6 database, and slightly supersaturated with respect to talc (see activity diagram, Fig. 3c). If ever talc-like phases did form, we were unable to identify them by either SEM or TEM analyses. 3.3. Olivine carbonation at 170 °C in pure water In this experiment, magnesite saturation is reached before that of amorphous silica (Fig. 6a). After a couple of days, the aqueous [Mg]/[Si] ratio is found to depart from the stoichiometric ratio in olivine. [Mg] reaches a maximum after 1.8 days while the fluid is greatly supersaturated with respect to magnesite (ΔGr(mag) = 7.6 kJ/mol; i.e. SI = 0.89) but undersaturated with respect to any other Mg-carbonate. [Mg] subsequently steadily decreases for 2 ≤ t ≤ 13 days (Fig. 6a). It eventually levels to a plateau at t N 13 days: the solution then

appears to be at equilibrium with respect to magnesite (− 0.96 b ΔGr(mag) b 1.10 kJ/mol (SI = 0.13) for 13 b t b 48 days) and remains undersaturated with respect to any other Mg-bearing phases available in the EQ3/6 database. Amorphous silica saturation is reached after about 5 days (ΔGr = 0.07 after 5.84 days, Table 5). There is no direct evidence on whether or not olivine keeps dissolving when [Mg] and [SiO2(aq)] have reached steady-state concentrations. From a thermodynamic standpoint, the dissolution of olivine remains energetically favorable since the solution remains very undersaturated with respect to forsterite (ΔGr(ol) = −48 kJ/mol after 48 days). The present experiment strongly contrasts with the previous ones by the amount of olivine which has dissolved. Rietveld analysis on XRD pattern (Fig. 6b) indicates that only about half of the initial olivine mass is recovered as unreacted (about 55 wt.%). This value contrasts with the relatively low carbonation yield (6 wt.% and 5.6 wt.% according to Rietveld refinement and carbon analysis, respectively). This can be explained by the presence of amorphous phases representing about 39 wt.% of the recovered run products, quantified using corundum as an internal standard.

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

a 14 12

(SiO2) am saturation

[SiO2aq]

[x] (mmol/L)

10 8 6

[Mgaq]

4 MgCO3 saturation

2 0 0

10

20

b

40

o

5000

experimental fit

o

4500 o

baseline

o

4000 3500

c

o

o

3000

o c o

c

c m

m

m o m

o

500

o

o o

o

1000

m

o

m

o

1500

o

2000

c

o c

2500 o

Intensity (counts)

30

t (days)

0 5

15

25

35

45

55

65

75

85

2θ(°) Fig. 6. (a) Olivine dissolution/carbonation at 170 °C monitored by time-resolved aqueousfluid sampling. Magnesite saturation is overstepped in the first stage of the experiment. The concentration of Mg eventually stabilizes at a level close to that of magnesite saturation from day 12 to the end of the experiment. The solution reaches saturation with respect to SiO2(am) on day 8, and the Si concentration is subsequently quasi-stable until the end of the experiment. (b) XRD pattern of the reaction products and Rietveld refinement of the XRD pattern. Mineral phases are identified by letters (o for olivine, m for magnesite and c for corundum, the internal standard).

Optically, the reaction products display a reddish color consistent with the presence of ferric iron phases. SEM imaging shows that olivine grains are sparsely covered with euhedral magnesite crystals (Fig. 7a–b). In addition, some complex assemblages of Fe–Mg–Si-rich phases (2–10 μm across) are observed, either as small aggregates or as surface layers covering the olivine and carbonates (Fig. 7g–j). By their abundance, they could represent the amorphous phase inferred from the quantitative Rietveld analysis. At the submicronic scale, TEM investigations reveal that a continuous porous and fibrous Fe-rich layer (50 nm-thick) has developed on the surface of the primary mineral that was in contact with the solution, as well as beneath the Mg-carbonate (Fig. 7c–f). Both stoichiometry and habitus of this phase suggest that it could be a serpentine-type mineral. Water determination indicates that the reaction products are weakly hydrated, as water represent 0.39 wt.% of the recovered sample powder; i.e. 1.02 wt.% of the amorphous solid products. 4. Discussion 4.1. Rate-limiting processes of olivine carbonation at 90 °C with 0.1 M Na-citrate The experiments by Daval et al. (2011) in a citrate-free medium (Fig. 1d) have been performed with otherwise identical experimental

195

parameters (T, SA/V ratio, pCO2, and starting materials). They will therefore be taken here as reference to discuss the intrinsic effects of citrate ligands on olivine carbonation. In their study, Daval et al. (2011) showed that dissolution of olivine appears to stop once the fluid approaches equilibrium with SiO2(am), as aqueous [Mg]/[Si] stabilizes slightly above the stoichiometric ratio. TEM observations of a FIB thin section revealed that a 30 nm-thick amorphous silica layer formed at the olivine/fluid interface, and was suggested to passivate the mineral surface. With the addition of Na-citrate (0.1 M), the main observations are that (1) olivine consumption increases by more than an order of magnitude (N39 mol% vs. 3 mol% in citrate-free solution), (2) the fluid is supersaturated with respect to magnesite, leading to magnesite precipitation, (3) transient SiO2(am) oversaturation is achieved and precipitation of silica-rich products occurs at the olivine surface while the latter is already partly covered with carbonate grains, (4) despite the increase of olivine dissolution rate, the carbonation yield remains fairly small (0.67 wt.%). 4.1.1. Initial stages (0 b t ≤ 5 days) The passivation barrier, evidenced by Daval et al. (2011) in a citratefree solution, is overcome (Fig. 1a) in our experiment since amorphous silica saturation is overstepped in the early stage (t ≤ 1 day). Further olivine dissolution in the citrate aqueous medium likely led to amorphous silica precipitation since aqueous silica concentration decreases (1 b t ≤ 5 days) to reach amorphous silica saturation level. Another evidence of the absence of olivine surface passivation in the early stage of the dissolution reaction is given by TEM microstructural characterization. Silica precipitation occurred onto the olivine surface while Mg-carbonates had already formed. Therefore, silica precipitation did not prevent the fluid from reaching Mg-carbonate supersaturation, as expected if precipitated silica had been passivating. Kinetically speaking, the passivation process can occur when the rate of formation of the passivating layer exceeds that of the host solid dissolution. Undoubtedly, the dissolution rate of olivine has been increased by the presence of Na-citrate. During the first 5 days of the run, the dissolution rate remained greater than 1.7 · 10−7 mol/m2/s (see Table 3), which is about one order of magnitude faster than those reported in Daval et al. (2011) in a citrate-free medium where pH was even lower. This is also consistent with the well-documented effect of organic ligands on the dissolution rate of olivine (e.g. Grandstaff, 1986; Olsen and Rimstidt, 2008; Daval et al., 2010; Prigiobbe and Mazzotti, 2011). Moreover, it cannot be excluded that citrate ligands may also have lowered the kinetics of formation of interfacial Si-rich layer. Such explanation would be supported by the role played by some elements (e.g. iron) on the rapid formation of passivating silica layers, as suggested by Schott and Berner (1983) for Fe-rich pyroxenes. In their study, Schott and Berner (1983) showed that the formation of a so-called “leached-layer” consisting of Si and Fe was responsible for dramatically decreasing the dissolution rate of the Fe-rich pyroxene bronzite, whereas such an effect was not observed for Fe-free pyroxenes. Since in our experiment Fe was complexed by citrate in solution, we suggest that the reactants that may be required for rapidly forming the passivating interfacial layer (e.g. Fe(H2O)++ 6 ) were not available, so that olivine dissolution could proceed unhindered. Such a proposition is further explored and expanded in Section 4.2. 4.1.2. Quasi steady-state stage (5 b t b 22 days) Concomitantly to silica precipitation in the early stages of olivine dissolution (see above), the release rate of all measured elements but Si has dramatically slowed down while remaining congruent (Fig. 1a–c) until it reached a quasi steady-state regime. Given the minuscule carbonation yield (0.67 wt.%), the observed quasi steady-state regime cannot be ascribed to the precipitation of (Fe, Ni)-rich magnesites. Moreover, we verified that none of the parameters that impact olivine dissolution rate (namely: olivine surface area, pH, and concentration of freecitrate ligands) vary significantly enough to account for the observed

196

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

a

b

Fe-rich phase

Magnesite

Magnesite

Olivine

b 100 nm

20 nm

c

e

d

Olivine Gold Platinum

100 nm

f

TEM

50 nm

Si

Mg

Fe

Olivine Phyllo

Magnesite Mg/ Si rich secondary phase

Mg/ Si rich secondary phase

Mg/ Si rich secondary phase

Mg/ Si rich secondary phase

olivine

4 µm

g

4 µm

olivine

h

40 µm

i

50 µm

j

Fig. 7. Electron microscopy observations of the reaction products obtained at 170 °C. (a), (b) SEM images of magnesite and Fe-rich phases formed on olivine grains. (c), (d) TEM images of the carbonate/olivine interface, showing a 50 nm-thick phyllosilicate layer instead of an amorphous silica-rich layer. (e) EDS spectrum of a chemical point analysis in the phyllosilicate layer. (f) EDS maps of the interface, illustrating that the phyllosilicate is Fe-rich. (g)–(j) SEM images of Mg-rich secondary phases formed among the reaction products.

decrease of the dissolution rate by ~2 orders of magnitude. In addition the experiment took place in a Gibbs free energy range usually referred to as far-from-equilibrium, which is characterized by a ‘rate plateau’ (see e.g. Burch et al., 1993; Hellmann and Tisserand, 2006; Arvidson and Lüttge, 2010; Hellmann et al., 2010 for details), such that presumably, ΔGr(ol) did not impact the dissolution rate. Besides, mass balance calculations indicate this SiO2(am) represents about 14.9 wt.% of the run products, most of which having precipitated at the olivine surface. Therefore, we propose that the heterogeneous precipitation of amorphous silica (SiO2(am)) from the bulk solution is responsible for a limitation of the species transport from the olivine surface to the solution.

In addition, Mg concentration does not drop down though magnesite has apparently precipitated (Fig. 1a). This can be explained by the very slow nucleation and growth kinetics of magnesite at this temperature (Giammar et al., 2005). To conclude, at 90 °C, increasing the dissolution rate of olivine by the means of adding citrate ligands only marginally increases the yield of carbonate precipitation. Most of the Mg which was released in the fluid was actually complexed by citrate ligands (about 70% of aqueous Mg was complexed by citrate according to our speciation calculations) and, although magnesite saturation was achieved, magnesite precipitation (nucleation and/or growth) occurred to be one of the carbonation

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

limiting factor. The conclusions of this section echo the recent suggestions by Bonfils et al. (2012) about the weak to non-existent beneficial effect that various organic ligands have on increasing carbonation yields. In addition, the formation of secondary coatings dramatically slows down the transport of aqueous species from and to the surface of olivine, so that the supply of Mg appears to ultimately limit the carbonation yields. 4.2. Rate-limiting processes of olivine carbonation at 120 °C 4.2.1. Evidence for a transient passivation stage (0 b t b 36 days) For 0 b t b 36 days, the concentrations of all elements but Fe, which will be treated separately (Section 4.2.2), increase and level to a plateau while the solution approaches saturation with respect to amorphous silica (Fig. 3). As illustrated in Table 4, all divalent cations (Mg, Mn, Ni) are released almost congruently, with a slight retention of SiO2; aqueous [Mg]/[Si] ratio is comprised between 1.82 and 1.96 for all aliquots sampled for t b 36 days, whereas a strictly congruent process would correspond to a stoichiometric ratio of 1.76. Silica retention is confirmed by the presence of amorphous Si-rich phases on the surface of olivine as observed by TEM (Fig. 5). From a kinetic standpoint, considering the error bars on the data, the dissolution rate can be calculated only during the very first stages of the process. For t b 1 day, the calculated dissolution rate of olivine is 1.7 × 10−7 mol/m2/s. For the sake of example, this value falls within the interval of the minimum and maximum far-from-equilibrium rates, defined by extrapolating the rate laws of Rosso and Rimstidt (2000) and Hänchen et al. (2006), at the T and pH conditions of the experiment (see details in Daval et al., 2011): 7.5 × 10−8 mol/m2/s and 1.3 × 10−6 mol/m2/s. If ever the rate of olivine dissolution is not nil between 2 and 36 days, it remains far too low to be measured by our analytical techniques. In addition, one could show that the evolution of all parameters discussed in the previous section (ΔGr(ol), surface area, pH) would marginally affect the dissolution rate of olivine, such that none of them could account for such a dramatic drop of olivine dissolution rate. The observation, furthermore, that the dissolution rate drop coincides with the approach of SiO2(am) saturation, favors the idea of a passivation process of olivine surface by an interfacial amorphous silica-like phase, similar to what has been proposed at 90 °C in similar experimental conditions (Daval et al., 2011). The facts that (1) the fluid composition remained supersaturated with respect to magnesite (0.87 b ΔGr(magnesite) b 2.99 kJ/mol; i.e. 0.11 b SI b 0.40) and that (2) the total carbonation yield did not exceed 2.2 wt.% confirm the very low nucleation and growth kinetics of magnesite at this temperature (Hanchen et al., 2008; Saldi et al., 2009). 4.2.2. Resumption of olivine dissolution (36 b t b 46 days) Until t = 34 days, iron concentration remained below the detection limit whereas olivine has been shown to dissolve almost congruently with respect to all the other species (Table 4). This unexpected feature likely indicates that the starting solution was oxidizing despite our attempt to lower the O2(aq) concentration by flushing CO2 at the beginning of the experiment (see Section 2.3). Fe2+ ions released from olivine were presumably rapidly oxidized, and as Fe3+ is much less soluble, it precipitated as Fe(III)-bearing phases, such as oxides and/or hydroxides, or as Fe(III)-bearing amorphous silica. A drastic change in iron behavior occurred at 34 ≤ t b 42 days, as indicated by the [Fe] peak which coincided with a strong Mg and Si concentration increase. Because this increase took place according to an aqueous [Mg]/[Si] ratio close to that of olivine stoichiometry (between 1.82 and 2.16), it can be confidently concluded that olivine dissolution resumed. The most straightforward explanation for this resumption is the loss of the passivation property of the interfacial amorphous silica layer. It must be noted that the [Fe] increase slightly predates the olivine dissolution resumption and Fe must therefore have been supplied by another Fe-bearing phase. Iron oxides and/or hydroxides from the silica layer, or the (Fe-bearing) silica

197

layer itself, are the most likely sources since Fe release affected the olivine surface passivation properties. The fact that [Fe] decreased after reaching a peak value around 0.15 mmol/L indicates that iron was removed from the solution by a precipitation process. This precipitation also impacted Ni and Mn, the concentration of which dropped simultaneously with that of Fe. Fe(II)-bearing phases such as oxides (e.g. magnetite) or iron carbonates are potential host phases for these divalent cations. The important observation that iron became suddenly soluble suggests that appropriate reducing conditions were reached for modifying iron speciation and redox in solution. To test this assumption, we followed an approach similar to Saldi et al. (2013) to model the sequence of secondary Fe-bearing phases expected to be formed during this experiment (see Appendix B). The main results can be summarized as follows: (i) Early in the experiment, when the system is under oxidizing conditions, Fe(III) oxides/hydroxides (such as goethite, FeOOH) are stable and form, consuming O2(aq); (ii) as the reaction proceeds, the solution gets further reducing, and magnetite (a mixed Fe(II)/Fe(III) oxide) forms at the expense of goethite, which ultimately disappears; (iii) Fe becomes suddenly soluble and starts to increase in the solution. Interestingly, if goethite is considered as to represent a proxy for the passivating phase, this sequence qualitatively mimics our observations, with the increase of Fe(aq) together with the breakdown of the passivating phase. Two additional comments arise from this model: (1) At that stage of our study, it is difficult to unambiguously assess whether the passivating layer formed at t b 36 days was either a Fe3+-Si rich layer, or a mixture of SiO2(am) and goethite filling up its porosity. It is noteworthy that the FIB thin section that we obtained integrates the evolution of interfacial phases throughout the carbonation experiment from t = 0 to t = 54 days. Accordingly, the Fe-rich crystallites observed by TEM within the amorphous silica layer could be either residues from the Fe3+ phases formed during the first dissolution step under oxidizing conditions (supposedly, goethite filling the porosity of amorphous silica), or Fe2+/Fe3+-solids that appeared later during the more reducing dissolution step (i.e. through the destabilization of the Fe3+-Si rich layer). To unravel this question, a thin section of the interface at t b 36 days would have been required. However, previous results on the passivating ability of surface layers suggested that passivation of Fe-rich silicate surfaces rather originates from Fe(III)-rich silica layers than from Fe(III) oxides (e.g. Schott and Berner, 1983; Santelli et al., 2001). In addition, Daval et al. (2011) reported that, at 90 °C and under oxic conditions, olivine was passivated by a silica layer. However, their TEM investigations carried out in a similar way as to those reported in the present study failed at identifying separate Fe-oxides within the amorphous silica rich layer, which suggests that the passivating ability of the silica layer did not result from a mixture of Fe-oxides and SiO2(am). This may be a supplementary argument to support that, at 120 °C, the passivating layer is an Fe3+-rich silica phase. (2) Importantly, the evolution from oxidizing to reducing conditions requires a continuous supply of Fe(II) to the solution. Therefore, we suggest that for 2 b t b 36 days, olivine was still dissolving, with a rate which was out of reach of our analytical techniques. 4.2.3. Towards a new steady-state (t N 46 days)? For t N 46 days, both aqueous Mg and Si concentrations apparently level to second plateaus, with an aqueous [Mg]/[Si] ratio slightly above 2, which presumably indicates that olivine dissolution slowed down again. The experiment having been terminated at t = 54 days, the attainment of steady state conditions associated with olivine surface passivation cannot be safely inferred on such a short time basis. The aqueous silica concentration stabilized above SiO2(am) saturation and the solution was even supersaturated with respect to several other

198

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

secondary phases such as talc which, however, were not observed directly (Table 4). In the absence of obvious secondary phases controlling the solution composition and visible in the experimental products, it is difficult to speculate about the nature of the passivating phase, if any. However, this second stage of olivine dissolution depicts again the general trend of a fast dissolution followed by a strong slowdown of species release into the solution whereas olivine saturation has not yet been achieved. Even if it could not be characterized here, the role of precipitated secondary phases as olivine dissolution moderator is very likely to have been prominent. 4.3. Limited Mg-carbonate precipitation at 170 °C From a kinetic standpoint, olivine dissolution rates can be calculated only when the process is not obviously affected by secondary precipitation (i.e., as long as the aqueous [Mg]/[Si] ratio remains close to stoichiometry, which corresponds to t b 0.6 day). During this time interval, the measured dissolution rate of olivine decreases from 3.1 × 10−7 to 1.8 × 10−7 mol/m2/s. For the sake of example, the minimum and maximum estimations based on the rate laws of Rosso and Rimstidt (2000) and Hänchen et al. (2006) extrapolated at 170 °C for the corresponding calculated pH range, respectively yield: 1.3 × 10−5 to 4.7 × 10−6 mol/m2/s and 5.7 × 10−7 to 1.9 × 10−7 mol/m2/s. Therefore, the measured rates are about 1.5 orders of magnitude slower than those determined from the study of Hänchen et al. (2006), but are in close agreement with the rates extrapolated from the data of Rosso and Rimstidt (2000) (see similar conclusions in Daval et al. (2011) and Saldi et al. (2013) for olivine dissolution at 90 °C and 150 °C, respectively). As temperature increases, magnesite saturation is achieved before amorphous silica saturation (Fig. 8). It therefore suggests that the crystallization sequence is reversed with magnesite precipitating prior to amorphous silica. Obviously, the slow rate of magnesite precipitation tends to shift that inversion point towards high temperatures since magnesite supersaturation can be maintained at least at the timescale of the experiments (see 120 °C experiment). Here, we show that at 170 °C, magnesite does precipitate before amorphous silica saturation is achieved. Carbonation occurs therefore before any interfacial silica

layer has stabilized at the olivine–solution interface. The passivation effects described in the two previous experiments is therefore expected to be limited or at best avoided. However, the carbonation yield of the 170 °C experiment remains unexpectedly low, around 6 wt.%. 4.3.1. Carbonation still controlled by secondary phases The low carbonation yield of ca. 6 wt.% may also appear inconsistent with the large proportion of olivine dissolved in this experiment of ca. 51 wt.% deduced from Rietveld data (Appendix A2). This apparent discrepancy can be explained by the presence of about 39 wt.% of Mg-Si rich amorphous material in the experimental product. This phase cannot have formed during quench at the end of the experiment, as there is a very small difference (less than 3%) in [Mg] and [SiO2(aq)] measured in the fluid before (last aliquot) and after quench. Overall, after 48 days of experiments, olivine reaction with CO2 saturated aqueous solution at 170 °C produced more amorphous magnesium silicates than carbonates. We have shown that at 90 and 120 °C, carbonation was limited by interfacial secondary products (e.g. amorphous silica) which isolate olivine surfaces from the aqueous solution. At 170 °C, olivine extensively dissolves at least in the first stage, but the massive formation of secondary amorphous Mg-silicate(s) sequesters magnesium which was expected to be incorporated in Mg-carbonates. Considering that after 12 days, silica concentration reached about 12.5 mmol/L, more than 22 mmol/L of Mg must have been released in the solution by stoichiometric dissolution of olivine. After 12 days, being the volume of the remaining solution about 200 mL, and since [Mg] levels at about 3.5 mmol/L due to magnesite saturation, in addition, a total of ca. 3.7 mmol Mg must have precipitated. This would correspond to ca. 18.4 wt.% of magnesite (and ca. 21 wt.% of dissolved olivine), which is far above the amount of carbonate measured in the run product recovered after 48 days (~ 6 wt.%). Several possibilities can account for these observations: (1) either some Mg was already included in the Mg-silicate amorphous material precipitating early in the t b 12 days interval together with magnesite or (2) magnesium carbonates formed during the first 12 days and further reacted with silica to produce some amorphous Mg-silicates (decarbonation). A stoichiometry of 2.9MgO·SiO2 (with

a) 90 C,pCO2 = 280bars

b) 120 C,pCO2 = 280bars 20

20

MgCO3 saturation

15

[Mgaq] (mmol/L)

[Mgaq] (mmol/L)

15

10

10

5

5

0

0 0

2

4

6

[SiO2 aq](mmol/L)

8

0

2

4

6

8

[SiO2 aq](mmol/L)

Fig. 8. Equilibrium reaction pathways of olivine carbonation calculated with the CHESS code at 90 °C (a) and 120 °C (b), neglecting the passivating ability of secondary phases (a). At 90 °C, dissolution is congruent until the fluid reaches equilibrium with respect to SiO2(am) (vertical solid line). Then, ideally, as dissolution proceeds, the fluid should reach equilibrium with respect to magnesite (horizontal solid line). (b) At 120 °C, the solubility of silica increases, while that of the carbonate decreases; therefore, the reaction pathways change. Equilibrium with respect to magnesite is expected to be reached before equilibrium with respect to SiO2(am).

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

minor FeO and H2O) can be inferred for the amorphous Mg-silicate phase from both mass balance calculation (see Appendix A2) and EDS analyses. From its relatively high Mg-content, this amorphous compound represents a potential sink for Mg. Therefore, even though, its possible coating role with respect to olivine surface could not be inferred here, the formation of this amorphous compound will compete with that of magnesium carbonate and will hence have a deleterious effect on olivine carbonation yield. 4.3.2. Iron incorporated at the olivine interface in a phyllosilicate In the 120 °C experiment, we have emphasized the prominent role played by iron on the olivine dissolution dynamics. Iron was found to precipitate at the olivine surface in the Fe3+ form as inferred from the Fe concentrations, which were below the detection limit in the aqueous solution. Following the same reasoning, iron is believed to be present as ferric iron in this 170 °C experiment. Iron-rich poorly-crystallized phyllosilicate is actually observed by TEM at the interface between carbonate and olivine, and could be the Fe3+-bearing phase. This observation supports the notion already pointed out in the 120 °C experiment, that at least part of the iron from the olivine precipitates, locally at the olivine surface, in the form of a Fe-bearing (silica-rich) phase. At a temperature of 170 °C however, a Fe-bearing serpentine-like phase is apparently stabilized over iron oxide, as has already been described by King et al. (2010). This observation is also consistent with the experimental work by Malvoisin et al. (2012) at higher temperature, in the 250–450 °C range, on San Carlos olivine hydration, which shows that Fe is favorably partitioned into serpentine at the expense of iron oxide (magnetite) as temperature decreases. Finally, the amorphous silica layer, which has been shown to passivate the olivine surface at 90 °C and 120 °C, is not observed here. However, possible deceleration or even cancelation of the olivine dissolution rate in relation to secondary phase precipitation onto its surface cannot be excluded from the present dataset; this is evidenced through the observation that, according to [SiO2(aq)], only 12 days are required to dissolve 24 mol% of olivine, whereas olivine consumption measured after 48 days represents 51 mol% only (Appendix A2). 5. Conclusions As the release in large quantities of divalent cations during the dissolution of a silicate is a prerequisite to the efficiency of the carbonation reaction, it is necessary to understand the processes which limit the availability of those cations. In the temperature range 90–170 °C, we emphasized the critical role played by secondary by-products in decreasing and/or controlling the dissolution rate of olivine, thereby dramatically impacting the carbonation yields and rates. The observed nm- to μm-thick interfacial Si-rich layers are consistent with a large amount of studies dedicated to the characterization of olivine surface during dissolution/carbonation processes at temperature between 20 and 150 °C (e.g. Schott and Pokrovsky, 2000; Bearat et al., 2006; Andreani et al., 2009; Davis et al., 2009; King et al., 2010, 2011; Daval et al., 2011; Hellmann et al., 2012). At T = 90 °C, adding citrate ligands to the solution prevents the formation of passivating interfacial Si-rich layer in the early stages of the reaction, either by promoting the intrinsic dissolution rate of olivine which in turn exceeds that of the passivating layer formation, or by a combined effect consisting in increasing the dissolution rate of olivine and concomitantly impeding the formation of the Si-rich layer. However, olivine dissolution was ultimately impacted by the precipitation of SiO2(am) from the bulk solution (olivine dissolution rate was dramatically decelerated). Being Fe3+ a critical contributor to the rapid formation of passivating silica layers (e.g. Schott and Berner, 1983; Santelli et al., 2001), its unavailability in the experiment carried out at 90 °C (Fe was most likely complexed by citrate ligands in solution) might explain why the olivine dissolution proceeded unhindered in the very early stages of the reaction. The role of Fe3+ in the passivation process

199

was directly inferred from the reaction conducted at 120 °C, where increasingly reducing conditions resulted in the breakdown of the passivating Fe3+/Si-rich surface layer and in the resumption of olivine dissolution. At 170 °C, Fe was (at least partly) incorporated into a porous Fe-rich phyllosilicate layer, so that the dissolution process was most likely less affected by this interfacial layer. An important lesson of this study is that passivation is a process which can be transient, and prone to be greatly influenced by external forcings. From a general standpoint, any secondary phase which would represent a sink for Fe3+ cations and/or SiO2(aq) will impede the long-term stabilization of a passivating Si-rich layer (see similar conclusions in Frugier et al. (2008) with respect to nuclear glasses, and references therein). From a carbonation yield point of view, however, overcoming the passivation barrier does not guarantee high carbonation yields (Table 6). Although the addition of citrate may overpower the deleterious effect of Fe, the carbonation yield remained marginal because citrate ligands complexed Mg as well, thereby leading to minuscule carbonation yields. Similarly, recent experiments conducted on olivine with 0.1 M oxalate at close temperature conditions (120 °C) (Bonfils et al., 2012) yielded no carbonation at all. Although citrate and oxalate are considered as two of the most effective ligands for enhancing olivine dissolution (Grandstaff, 1986), it would therefore appear that their actual benefit to enhance carbonation yields is close to nil. At 120 °C, despite a temporary resumption of olivine dissolution, the role of precipitated secondary phases as olivine dissolution moderator is very likely to have been prominent anyway. Finally, at 170 °C, the carbonation yields were ultimately limited by the formation of amorphous Mg-silicates, competing with the growth of Mg-carbonates. Though those last results seem to be at odds with previous studies conducted at close P/T conditions, considered as optimum for carbonation (e.g. Gerdemann et al., 2007, where T = 185 °C and pCO2 = 150 bars), possible explanations might be sought in the pH of the bulk solutions (acidic in our case, alkaline in the case of Gerdemann et al., 2007), which impact the respective solubilities of the Mg-phases. Overall, carrying out the carbonation reactions in reducing environments and in circum-neutral aqueous solutions may represent a necessary requirement for making olivine carbonation a viable process in the T-range 120–170 °C. A final concluding remark is that this study emphasized the importance of transitory phases as a whole, which form both at the dissolution interface and from the bulk solution. Although they do not necessarily appear in global carbonation equations or in themodynamical databases, they are likely to dramatically drive the global kinetics of olivine dissolution and its subsequent transformation towards magnesite.

Acknowledgments The authors wish to thank Carine Chaduteau for help with carbonate quantification, David Troadec from IEMN, Lille for preparation of the FIB thin sections, Delphine Tisserand, Lionel Rosseto and Sarah Bureau from ISTerre, Grenoble for help with the ICP-AES analyses, and Jean-Michel Guigner from IMPMC for help with TEM analyses. O. Sissmann thanks the ANR CO2FIX (ANR-08-PCO2-003-03) and the IPGP/Ademe/Schlumberger/Total CO2 geological storage program for funding, and the members of its respective committees for stimulating discussions. Pierre Agrinier and Benedicte Menez are warmly acknowledged for continuous support and stimulating exchanges. Finally, the

Table 6 Summary of olivine dissolution and carbonate yields (in wt.%) for each experiment.

wt.% dissolved olivine wt.% formed carbonates

90 °C

120 °C

170 °C

39.8 0.67

12.8 2.2

51.4 6

200

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

thorough reviews and constructive comments by two anonymous reviewers and the editor (J. Fein) were also much appreciated and helped improve the manuscript. This is IPGP contribution n° 3430.

In the present section, we propose a rough estimation of the stoichiometry of the amorphous phase that is formed together with magnesite in the experiment carried out at 170 °C through a series of mass balance calculations. Corresponding data can be found in Table A-1.

Appendix A1. Calculations of the stability constants for M-citrate complexes In the vast majority of the papers we reviewed for calculating the stability constants reported in our study, the stability constant were determined for a given, fixed ionic strength (I). Our treatment of the data consisted in extrapolating such constants for I = 0. The overall reaction which was taken into account can be written as follows: ðA1Þ

pM þ qL þ rH⇆Mp Lq Hr

with L, H and M standing for the fully deprotonated citrate ligand, hydrogen and metal ions, respectively, where charges were omitted for the sake of simplicity. The corresponding I-dependent stability constant (referred as to: βMp Lq Hr ðT Þ ) is reported in the literature following:

βMp Lq Hr ðT Þ ¼

aMp Lq Hr aM p aL q aH r

¼

A2.1. Mass of the final product If we consider m0, the mass of initial olivine, then mf is the mass of final reaction products, such as:

m f ¼ m0 −mMgO fluid −mSiO2

þ mCO2 mag þ mH2 O

ðA6Þ

where mMgO and mSiO2 represent respectively the masses of dissolved MgO and SiO2, and mCO2 and mH2O the masses of CO2 and H2O incorporated in the solids. Sampling the fluid (s times) removes matter from the system (Si and Mg). The total amount of Si removed from the system during

ðA2Þ

γMp Lq Hr γM p γL q γH r

sampling is ∑ ½Sii V i , where s is the number of aliquots sampled i¼1

throughout the experiment, [Si]i the Si concentration in each aliquot of volume Vi. The same holds for Mg. Consequently, the final mass of the solid can be written as follows:

m f ¼ m0 − ½Mg f  V f þ βMp Lq Hr ðT Þ

! pffiffi
 2 −A I pffiffi þ AbI ln γ j ¼ z j 1þ I

! ½Mg i V i

s X

!

½Sii V i  MSiO2  MMgO − ½Si f  V f þ !i¼1
 MCO2 þ 0:4%  m f þ 6%  m f  MMgCO3

ðA7Þ

where Vf, [Mg]f and [Si]f stand for the final volume and [Mg] and [Si] in the remaining solution. ðA4Þ A2.2. Amorphous phase composition

where zj is the electric charge of the jth species, and A and b are constants with values of 1.17 and 0.3, respectively. Therefore, by combining Eqs. (A3) and (A4), K Mp Lq Hr ðT Þ values can be calculated from the following expression:
 2 2 2 2 zMp Lq Hr −p:zM −q:zL −rzH !! pffiffi −A I pffiffi þ AbI  βMp Lq Hr ðT Þ: 1þ I

s X i¼1

ðA3Þ

where γj stands for the activity coefficient for the jth species. Because of the range of ionic strengths investigated in the studies we reviewed (0 b I b 1), we decided to relate the activity coefficients to I by using the Davies equation:

K Mp Lq Hr ðT Þ ¼ exp

fluid

s

⌊Mp Lq Hr ⌋ : ½M p ½Lq ½Hr

Therefore, the corresponding stability constant, K Mp Lq Hr ðT Þ, expressed with the activities (aj) of the dissolved species can be related to βMp Lq Hr ðT Þ following: K Mp Lq Hr ðT Þ ¼

Appendix A2. Estimation of the chemical composition of the amorphous phase formed in the experiment conducted at 170 °C

ðA5Þ

In this calculation, the number of moles of the various compounds is symbolized as follows, n0, nr, nam and nmag, for the initial olivine, the residual olivine, the amorphous product and magnesite, respectively. The composition of the amorphous product is unknown and can be expressed as: xMgO·yFeO·zH2O·SiO2. Therefore the molar mass of this phase is Mam ¼ xM MgO þ MSiO2 þ yM FeO þ zMH2 O :

ðA8Þ

A2.3. Water content of the amorphous product The original values of βMp Lq Hr ðT Þ for a given ionic strength were taken from Meyer (1974), Pearce (1980), Covington and Danish (2009) for Mg-citrate complexes; Konigsberger et al. (2000) for Fe(II)-citrate complexes, Zelenina and Zelenin (2005) for Na-citrate complexes. Note that for Na and Fe(II)-citrate complexes, because we were not able to find data at two different temperatures, we provided the EQ3/6 database with the stability constant determined at one temperature only. Consequently, the speciation calculations at 90 °C were performed with the values of K Mp Lq Hr ðT ¼ 25BC Þ, and must be considered with great caution.

If one assumes that all the water recovered in the solid product (mH2O) is hosted by the amorphous phase then:

z¼

mH2 O mam



Mam MH2 O

where mam is the mass of the amorphous product.

ðA9Þ

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

201

A2.4. Mass balance assuming stoichiometric dissolution

Appendix B. Supplementary data

The mass balances for all of the major oxides incorporated in the solid phases (see Eq. (A8)) are detailed hereafter:

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.chemgeo.2013.08.031.

− Si mass balance: n0 ¼ nr þ nam þ ½Si f  V f þ

References

s X

½Sii V i

ðA10Þ

i¼1

with: nr ¼

55%  m f : M oliv

ðA11Þ

Therefore, the amount of the amorphous phase (nam) can be calculated using Eqs. (A10) and (A11) following: nam ¼ n0 −

s 55%  m f X − ½Sii V i −½Si f  V f Moliv i¼1

ðA12Þ

− Mg mass balance:

1:76ðn0 −nr Þ ¼ nmag þ ½Mg  V f þ

s X

½Mg i V i þ xnam :

ðA13Þ

i¼1

Therefore: 1:76ðn0 −nr Þ−nmag −½Mg   V f −

s X

½Mg i V i

i¼1

x¼

nam

¼ 2:99:

ðA14Þ

− Fe mass balance:

0:24ðn0 −nr Þ ¼ 0:54: nam

y¼

ðA15Þ

− H mass balance: mH

2 am According to Eq. (A9) z ¼ mam ñM × mnam ¼ 0:14 with H O O

2

mH

2O

¼ 0:4%  m f

ðA16Þ

The yield of olivine consumption is: 1−

nr ¼ 51:4%: n0

ðA17Þ

Andreani, M., Luquot, L., Gouze, P., Godard, M., Hoise, E., Gibert, B., 2009. Experimental study of carbon sequestration reactions controlled by the percolation of CO2-rich brine through peridotites. Environ. Sci. Technol. 43 (4), 1226–1231. Arvidson, R.S., Luttge, A., 2010. Mineral dissolution kinetics as a function of distance from equilibrium—new experimental results. Chem. Geol. 269 (1–2), 79–88. Barnes, C.S., Koretsky, M.D., 2004. Engineering and Chemical Thermodynamics. John Wiley and Sons (553 pp.). Béarat, H., McKelvy, M., Chizmeshya, A., Gormley, D., Nunez, R., Carpenter, R., Squires, K., Wolf, G., 2006. Carbon sequestration via aqueous olivine mineral carbonation: role of passivating layer formation. Environ. Sci. Technol. 40 (15), 4802–4808. Benezeth, P., Palmer, D.A., Wesolowski, D.J., 1997. Dissociation quotients for citric acid in aqueous sodium chloride media to 150 degrees C. J. Solut. Chem. 26 (1), 63–84. Bonfils, B., Julcour, C., Guyot, F., Bodenan, F., Chiquet, P., Bourgeois, F., 2012. Comprehensive analysis of direct aqueous mineral carbonation using dissolution enhancing organic additives. Int. J. Greenhouse Gas Control 9, 334–346. Burch, T.E., Nagy, K.L., Lasaga, A.C., 1993. Free energy dependence of albite dissolution kinetics at 80 °C and pH 8.8. Chem. Geol. 105, 137–162. Cailleteau, C., Angeli, F., Devreux, F., Gin, S., Jestin, J., Jollivet, P., Spalla, O., 2008. Insight into silicate-glass corrosion mechanisms. Nat. Mater. 7 (12), 978–983. Cornides, I., Kusakabe, M., 1977. Preparation of carbon-dioxide from magnesite for isotopic analysis. Fresenius Z. Anal. Chem. 287 (4–5), 310–311. Covington, A.K., Danish, E.Y., 2009. Measurement of magnesium stability constants of biologically relevant ligands by simultaneous use of pH and ion-selective electrodes. J. Solut. Chem. 38 (11), 1449–1462. Cubillas, P., Kohler, S., Prieto, M., Causserand, C., Oelkers, E.H., 2005. How do mineral coatings affect dissolution rates? An experimental study of coupled CaCO3 dissolutionCdCO3 precipitation. Geochim. Cosmochim. Acta 69 (23), 5459–5476. Daval, D., Martinez, I., Corvisier, J., Findling, N., Goffé, B., Guyot, F., 2009a. Carbonation of Ca-bearing silicates, the case of wollastonite: experimental investigations and kinetic modeling. Chem. Geol. 265 (1–2), 63–78. Daval, D., Martinez, I., Guigner, J.‐M., Hellmann, R., Corvisier, J., Findling, N., Dominici, C., Goffe, B., Guyot, F., 2009b. Mechanism of wollastonite carbonation deduced from micro- to nanometer length scale observations. Am. Mineral. 94 (11–12), 1707–1726. Daval, D., Testemale, D., Recham, N., Tarascon, J.-M., Siebert, J., Martinez, I., Guyot, F., 2010. Fayalite (Fe2SiO4) dissolution kinetics determined by X-ray absorption spectroscopy. Chem. Geol. 275 (3–4), 161–175. Daval, D., Sissmann, O., Menguy, N., Saldi, G.D., Guyot, F., Martinez, I., Corvisier, J., Garcia, B., Machouk, I., Knauss, K.G., Hellmann, R., 2011. Influence of amorphous silica layer formation on the dissolution rate of olivine at 90 degrees C and elevated pCO2. Chem. Geol. 284 (1–2), 193–209. Daval, D., Hellmann, R., Saldi, G.D., Wirth, R., Knauss, K.G., 2013. Linking nm-scale measurements of the anisotropy of silicate surface reactivity to macroscopic dissolution rate laws: new insights based on diopside. Geochim. Cosmochim. Acta 107, 121–134. Davis, M.C., Brouwer, W.J., Wesolowski, D.J., Anovitz, L.M., Lipton, A.S., Mueller, K.T., 2009. Magnesium silicate dissolution investigated by Si-29 MAS, H-1-Si-29 CPMAS, Mg-25 QCPMG, and H-1-Mg-25 CP QCPMG NMR. Phys. Chem. Chem. Phys. 11 (32), 7013–7021. Dufaud, F., Martinez, I., Shilobreeva, S., 2009. Experimental study of Mg-rich silicates carbonation at 400 and 500 degrees C and 1 kbar. Chem. Geol. 265 (1–2), 79–87. Frugier, P., Gin, S., Minet, Y., Chave, T., Bonin, B., Godon, N., Lartigue, J.E., Jollivet, P., Ayral, A., de Windt, L., Santarini, G., 2008. SON68 nuclear glass dissolution kinetics: current state of knowledge and basis of the new GRAAL model. J. Nucl. Mater. 380 (1-3), 8–21. Gerdemann, S.J., O'Connor, W.K., Dahlin, D.C., Penner, L.R., Rush, H., 2007. Ex situ aqueous mineral carbonation. Environ. Sci. Technol. 41 (7), 2587–2593. Giammar, D.E., Bruant, R.G., Peters, C.A., 2005. Forsterite dissolution and magnesite precipitation at conditions relevant for deep saline aquifer storage and sequestration of carbon dioxide. Chem. Geol. 217, 257–276. Golubev, S.V., Pokrovsky, O.S., Schott, J., 2005. Experimental determination of the effect of dissolved CO2 on the dissolution kinetics of Mg and Ca silicates at 25 °C. Chem. Geol. 217, 227–238.

Table A-1 Data used for mass balance calculations during the 170 °C experiment. S

S

∑ ½Sii V i (mol)

∑ ½Mgi V i (mol)

3.49E−04

5.98E−04

i¼1

[Mg] f (mol)

[Si] f (mol)

Vf H2O (L)

Moliv (g/mol)

Mmag (g/mol)

m0 (g)

mf (g)

n0 (mol)

nr (mol)

mH2O (g)

nmag (mol)

3.66E−03

1.39E−02

0.181

148.26

84.3

1.531

1.35

1.03E−02

5.012 E−03

5.4 E−03

0.962 E−03

i¼1

The two first columns represent respectively the sum of the molar quantity of Si and Mg sampled with each aliquots throughout the experiment. The two following columns are the Mg and SiO2(aq) concentrations measured at the end of the experiment by ICP-AES. The next column is the volume of fluid remaining at the end of the experiment. The two following columns are respectively the molar mass of the San Carlos olivine and magnesite. The two next columns are respectively the initial mass of olivine at the beginning of the experiment, and the final mass of run products at the end of the experiment. The two following columns are respectively the initial molar quantity of olivine, and the final molar quantity of olivine. The next column is the mass of water in the run product, and the last column is the molar quantity of magnesite formed during the experiment.

202

O. Sissmann et al. / Chemical Geology 357 (2013) 186–202

Grandstaff, D.E., 1986. The dissolution rate of forsteritic olivine from Hawaiian beach sand. In: Colman, S.M., Dethier, D.P. (Eds.), Rates of Chemical Weathering of Rocks and Minerals. Academic Press, Orlando, FL, pp. 41–57. Gudbrandsson, S., Wolff-Boenisch, D., Gislason, S.R., Oelkers, E.H., 2011. An experimental study of crystalline basalt dissolution from 2 ≤ pH ≤ 11 and temperatures from 5 to 75 °C. Geochim. Cosmochim. Acta 75, 5496–5509. Guyot, F., Daval, D., Dupraz, S., Martinez, I., Menez, B., Sissmann, O., 2011. CO2 geological storage: the environmental mineralogy perspective. C.R. Geosci. 343 (2-3), 246–259. Gysi, A.P., Stefansson, A., 2012. Mineralogical aspects of CO2 sequestration during hydrothermal basalt alteration—an experimental study at 75 to 250 degrees C and elevated pCO(2). Chem. Geol. 306, 146–159. Hanchen, M., Prigiobbe, V., Storti, G., Seward, T.M., M., M., 2006. Dissolution kinetics of fosteritic olivine at 90–150 °C. Geochim. Cosmochim. Acta 70, 4403–4416. Hanchen, M., Prigiobbe, V., Baciocchi, R., Mazzotti, M., 2008. Precipitation in the Mg-carbonate system—effects of temperature and CO2 pressure. Chem. Eng. Sci. 63 (4), 1012–1028. Hellmann, R., Tisserand, D., 2006. Dissolution kinetics as a function of the Gibbs free energy of reaction: an experimental study based on albite feldspar. Geochim. Cosmochim. Acta 70, 364–383. Hellmann, R., Daval, D., Tisserand, D., 2010. The dependence of albite feldspar dissolution kinetics on fluid saturation state at acid and basic pH: progress towards a universal relation. C.R. Geosci. 342 (7–8), 676–684. Hellmann, R., Wirth, R., Daval, D., Barnes, J.-P., Penisson, J.-M., Tisserand, D., Epicier, T., Florin, B., Hervig, R.L., 2012. Unifying natural and laboratory chemical weathering with interfacial dissolution–reprecipitation: a study based on the nanometer-scale chemistry of fluid–silicate interfaces. Chem. Geol. 294–295 (0), 203–216. King, H.E., Plumper, O., Putnis, A., 2010. Effect of secondary phase formation on the carbonation of olivine. Environ. Sci. Technol. 44 (16), 6503–6509. King, H.E., Plumper, O., Geisler, T., Putnis, A., 2011. Experimental investigations into the silicification of olivine: implications for the reaction mechanism and acid neutralization. Am. Mineral. 96 (10), 1503–1511. Konigsberger, L.C., Konigsberger, E., May, P.M., Hefter, G.T., 2000. Complexation of iron(III) and iron(II) by citrate, implications for iron speciation in blood plasma. J. Inorg. Biochem. 78 (3), 175–184. Krevor, S.C., Lackner, K.S., 2009. Enhancing process kinetics for mineral carbon sequestration. Greenhouse Gas Control Technologies, 9, 1(1), pp. 4867–4871. Lee, M.R., Brown, D.J., Smith, C.L., Hodson, M.E., Mackenzie, M., Hellmann, R., 2007. Characterization of mineral surfaces using FIB and TEM: a case study of naturally weathered alkali feldspars. Am. Mineral. 92 (8–9), 1383–1394. Malvoisin, B., Brunet, F., Carlut, J., Roumejon, S., Cannat, M., 2012. Serpentinization of oceanic peridotites: 2. Kinetics and processes of San Carlos olivine hydrothermal alteration. J. Geophys. Res. Solid Earth 117. McCrea, J.M., 1950. On the isotopic chemistry of carbonates and a paleotemperature scale. J. Chem. Phys. 18 (6), 849–857. Meyer, J.L., 1974. Formation constants for interaction of citrate with calcium and magnesium ions. Anal. Biochem. 62, 295–300. Montes-Hernandez, G., Fernandez-Martinez, A., Charlet, L., Renard, F., Scheinost, A., Bueno, M., 2008. Synthesis of a Se–O/calcite composite using hydrothermal carbonation of Ca(OH)(2) coupled to a complex selenocystine fragmentation. Cryst. Growth Des. 8 (7), 2497–2504. Nugent, M.A., Brantley, S.L., Pantano, C.G., Maurice, P.A., 1998. The influence of natural mineral coatings on feldspar weathering. Nature 395 (6702), 588–591.

Oelkers, E.H., Gislason, S.R., Matter, J., 2008. Mineral carbonation of CO2. Elements 4 (5), 333–337. Olsen, A.A., Rimstidt, J.D., 2008. Oxalate-promoted forsterite dissolution at low pH. Geochim. Cosmochim. Acta 72 (7), 1758–1766. Park, A.-H.A., Fan, L.-S., 2004. CO2 mineral sequestration: physically activated dissolution of serpentine and pH swing process. Chem. Eng. Sci. 59, 5241–5247. Pearce, K.N., 1980. Formation constants for magnesium and calcium citrate complexes. Aust. J. Chem. 33, 1511–1517. Pokrovsky, O.S., Schott, J., 2000. Kinetics and mechanism of forsterite dissolution at 25 °C and pH from 1 to 12. Geochim. Cosmochim. Acta 64 (19), 3313–3325. Prigiobbe, V., Mazzotti, M., 2011. Dissolution of olivine in the presence of oxalate, citrate, and CO2 at 90 degrees C and 120 degrees C. Chem. Eng. Sci. 66 (24), 6544–6554. Putnis, A., 2002. Mineral replacement reactions: from macroscopic observations to microscopic mechanisms. 18th General Meeting of the International-MineralogicalAssociation. Mineralogical Society, Edinburgh, Scotland, pp. 689–708. Putnis, C.V., Tsukamoto, K., Nishimura, Y., 2005. Direct observations of pseudomorphism: compositional and textural evolution at a fluid–solid interface. Am. Mineral. 90 (11–12), 1909–1912. Rosso, J.J., Rimstidt, D.J., 2000. A high resolution study of forsterite dissolution rates. Geochim. Cosmochim. Acta 64 (5), 797–811. Saldi, G.D., Jordan, G., Schott, J., Oelkers, E.H., 2009. Magnesite growth rates as a function of temperature and saturation state. Geochim. Cosmochim. Acta 73 (19), 5646–5657. Saldi, G.D., Schott, J., Pokrovsky, O.S., Gautier, Q., Oelkers, E.H., 2012. An experimental study of magnesite precipitation rates at neutral to alkaline conditions and 100–200 degrees C as a function of pH, aqueous solution composition and chemical affinity. Geochim. Cosmochim. Acta 83, 93–109. Saldi, G.D., Daval, D., Morvan, G., Knauss, K.G., 2013. The role of Fe and redox conditions in olivine carbonation rates: an experimental study of the rate limiting reactions at 90 and 150 °C in open and closed systems. Geochim. Cosmochim. Acta 118, 157–183. Santelli, C.M., Welch, S.A., Westrich, H.R., Banfield, J.F., 2001. The effect of Fe-oxidizing bacteria on Fe-silicate mineral dissolution. Chem. Geol. 180 (1–4), 99–115. Schott, J., Berner, R.A., 1983. X-ray photoelectron studies of the mechanism of iron silicate dissolution during weathering. Geochim. Cosmochim. Acta 47 (12), 2233–2240. Stockmann, G.J., Wolff-Boenisch, D., Gislason, S.R., Oelkers, E.H., 2011. Do carbonate precipitates affect dissolution kinetics? 1: basaltic glass. Chem. Geol. 284 (3–4), 306–316. Stockmann, G.J., Wolff-Boenisch, D., Gislason, S.R., Oelkers, E.H., 2013. Do carbonate precipitates affect dissolution kinetics?: 2: diopside. Chem. Geol. 337–338 (0), 56–66. Teir, S., Revitzer, H., Eloneva, S., Fogelholm, C.J., Zevenhoven, R., 2007. Dissolution of natural serpentinite in mineral and organic acids. Int. J. Miner. Process. 83 (1–2), 36–46. Van der Lee, J., De Windt, L., 2002. CHESS tutorial and cookbook. Updated for version 3.0., Manual Nr. LHM/RD/02/13. (Paris, 116 pp.). Velbel, M.A., 1993. Formation of protective surface layers during silicate-mineral weathering under well-leached, oxidizing conditions. Am. Mineral. 78 (3–4), 405–414. Wang, F., Giammar, D.E., 2013. Forsterite dissolution in saline water at elevated temperature and high CO2 pressure. Environ. Sci. Technol. 47 (1), 168–173. Wolff-Boenisch, D., Wenau, S., Gislason, S.R., Oelkers, E.H., 2011. Dissolution of basalts and peridotite in seawater, in the presence of ligands, and CO2: implications for mineral sequestration of carbon dioxide. Geochim. Cosmochim. Acta 75, 5510–5525. Zelenina, T.E., Zelenin, O.Y., 2005. Complexation of citric and tartaric acids with Na and K ions in aqueous solution. Russ. J. Coord. Chem. 31 (4), 235–242.