An important parameter for evaluating the amount of injected carbon dioxide that will be converted in a given time into solid carbonates is the rate of reaction (3) relative to the hydrodynamic parameters that characterize CO₂ leakage to the surface or to other reservoirs. Because reaction (3) is a combination of reactions (1) and (2), it is important to determine which of the two steps is kinetically limiting. Data from Daval et al. (2009a) are shown in Fig. 2. Modeling of the extent of carbonation of wollastonite (CaSiO₃) into calcite (CaCO₃) is not significantly different with and without consideration of the measured kinetics of precipitation of calcite instead of its equilibrium formation (i.e. at infinite rate). This observation demonstrates that what limits the kinetics of carbonation in such laboratory experiments is silicate dissolution (reaction (1)) and not the rate of carbonate formation (reaction (2)). This result can confidently be extended to all situations of calcite formation as a result of alteration of silicates at temperatures close to 100 °C, because wollastonite is presumably the most reactive Ca-bearing silicate under these conditions (cf. Fig. 7 in Daval et al., 2009a). At a temperature typical of the Earth's surface, where Si is incorporated in clays rather than in amorphous silica, this statement could be more questionable because the slow precipitation rate of these phases maintain aqueous concentrations close to equilibrium with respect to silicate dissolution, such that the weathering of primary silicates could be ultimately controlled by the rate of clay formation (see e.g., Maher et al., 2009, Zhu, 2009 for further details).

In cases where the carbonate formed is magnesite (MgCO₃), determination of the limiting step is not so clear and deserves more research. For instance, recent data on the precipitation kinetics of magnesite by Saldi (2009) suggest that at 100 °C, the rate constant of magnesite precipitation is 6 order of magnitude below that of calcite precipitation under similar conditions. However, the rate limiting step of the carbonation of Mg-rich silicates could still be the silicate dissolution reaction, at least below 120 °C. Moreover, at low temperatures, the carbonate formed would not be magnesite indeed, but rather hydrated magnesian carbonates such as hydromagnesite: Mg₅(CO₃)₄(OH)₂.4H₂O (Hanchen et al., 2008; Saldi et al., 2009, and references therein), with more rapid, yet unknown, precipitation rates.

A key issue for controlling conversion of CO₂ into solid carbonates is thus the rate of dissolution of silicate minerals. The reaction rate data in Fig. 2 for CaSiO₃ (wollastonite) at a grain size of about 0.2 mm show that full transformation of the silicate into carbonate occurs rapidly. This rapid transformation is due to two factors: wollastonite dissolves rapidly and Ca²⁺ is a carbonate-forming cation. In that respect, geological sites containing large amounts of wollastonite would be interesting for converting injected CO₂ into calcite, although wollastonite is not a common mineral. Other more common calcic minerals, e.g., anorthite, might also be favorable for carbonation, and more generally, minerals whose composition is dominated by magnesium and/or calcium would be interesting targets. This is the main idea behind projects involving CO₂ injection in basic and ultrabasic geological contexts (e.g., Gislason et al., 2010; Kelemen and Matter, 2008; Matter and Kelemen, 2009; McGrail et al., 2006; Schaef and McGrail, 2009; Schaef et al., 2010). The main premise of these strategies is that reaction (1) is an acid-basic neutralization of the mineral (base) by dissolved carbon dioxide (acid). Anorthite, CaAl₂Si₂O₈, an important component of basalts, could be an interesting subject in the perspective of geological storage of CO₂ in basalts. In this regard, a few recent studies have focused on water-CO₂-anorthite interactions (e.g., Berg and Banwart, 2000; Regnault et al., 2005; Sorai et al., 2007, Sorai and Sasaki, 2010). Indeed, basalts, although representing only ∼ 5% of the silicate rocks at Earth's surface contribute to more than 30% of reaction (1) (Dessert et al., 2003). In laboratory experiments, however, some promising minerals show unexpected slow carbonation rates at temperatures below 100 °C. Experimental studies of the dissolution of olivine (forsterite Mg₂SiO₄), even with small grains of 0.2 mm in characteristic sizes, have been highly disappointing because they yielded reaction rates being at least 4 orders of magnitude slower than that predicted by numerical simulation of the process (see Daval, 2009 and Daval et al., 2010b for details). Olivine, however, is one of the most thermodynamically favorable substrates for carbonation (Fig. 3), and is also one of the fastest dissolving minerals (Fig. 4). Therefore, this should make it an excellent substrate for conversion of CO₂ into carbonates. Slow experimental carbonation rates below 100 °C emphasize complexities whose origins will be discussed below. Of course, it is possible to increase carbonation rate by increasing temperature (Fig. 5) or by decreasing grain size (e.g., Dufaud et al., 2009, Garcia et al., 2010). Although achieved in some cases, temperatures will seldom reach 90 °C at CO₂ geological storage sites. Whereas temperature increase is an efficient but often impractical solution, increasing reactive surface area is an interesting option that deserves further research. Injection of CO₂ under pressures higher than hydrostatic and lithostatic values at the injection point, sufficient to create microcracks in the rocks, might help to increase fluid/rock contact surface area by creation of microfractures, and thus presumably increase reactive surface area as well. This main issue of coupled mineralogy and geomechanics thus needs to be investigated in detail (see analogous studies in carbonation of concrete (e.g., Corvisier et al., 2010; Fabbri et al., 2009) and limestone (e.g., Le Guen et al., 2007)). It should be noted, however, that the issue of inducing fracturing underground has proven difficult on a societal basis, at least under populated areas (Giardini, 2009). On the other hand, ex situ carbonation (i.e. in a chemical plant at Earth's surface, see e.g., Gerdemann et al., 2007) of materials susceptible to carbonation, with continuous grinding to reduce grain sizes, might be a promising approach (a French project supported by ANR CO₂ coordinated by Françoise Bodénan at BRGM in Orléans and by Florent Bourgeois at École de Chimie in Toulouse is an example of research efforts currently exploring this possibility). In this latter case, a condition of success is to dispose of materials favorable to carbonation, which are already present and occupying an important volume, and which cannot be disposed of in other manners. Mine waste materials would be interesting targets (see Wilson et al., 2006), especially if mining occurred in basic and ultrabasic rocks. Finally, even in the continental crust, ferromagnesian minerals such as biotite could be interesting targets.

Even if the surface area problem could be solved, thanks to large porosities or to microfracturing, the effective dissolution rates of olivine and many other minerals will be in most cases still far too slow to make the rate of mineral carbonation competitive with the rate of hydrodynamic flow of CO₂ out of the system. We thus have to understand why olivine, as a typical example, converts into carbonates so slowly in comparison to wollastonite, especially in light of the observation that the far-from-equilibrium dissolution rates of olivine reported in the literature are only one order of magnitude slower than that of wollastonite (Fig. 4). The answer is found in the detailed examination of the mineral surfaces that underwent dissolution in batch experiments at similar conditions (90 °C and pCO₂ = 250 bar, duration: 2 days for wollastonite; 45 days for olivine) (Fig. 6). Whereas the silica layers shown in these pictures look different, they may originate from similar interfacial, nm-sized precursors. For both experiments, the bulk solution was found to be saturated with respect to SiO₂(am), such that the interfacial Si-rich layers usually observed to form at far-from-equilibrium conditions in mixed-flow reactor experiments for these minerals (see e.g., Casey et al., 1993 for wollastonite; Pokrovsky and Schott, 2000a,b for olivine) were stabilized. For wollastonite, it is striking that the silica-rich amorphous layer kept growing during the carbonation process while its external interface stopped dissolving. Thus, the growth of the silica layer at the expense of wollastonite was possible because this coating did not prevent the fluid from reaching the pristine interface of wollastonite, which thus could keep on dissolving. Significant porosity of the silica layer in the case of wollastonite (Fig. 7, see also further images and details in Daval et al., 2009b) was found and actually allows the reaction to proceed by transport of solutes through the layer, whereas this transport was severely limited for forsterite. In this latter case, the thin silica layer was non-porous, blocking dissolution and thus carbonation of forsterite.

The origin of such differences is so far unknown, and current research in our group is directed at ascertaining the origin of these surprising differences. Dissolution of nuclear waste glasses, where minor elements present in the initial material can strongly change the dissolution rate of the primary phase by modifying the structure of the secondary silica layer (see the case of Zr in Cailleteau et al., 2008), may be related to the dissolution process of silicate minerals. However, one major difference in the dissolution process of nuclear waste glasses and olivine is the much thinner silica layer on olivine, which makes its investigation highly challenging. Recent analytical developments in the surface analysis of minerals could help to resolve this problem (e.g., Daval et al., 2009b; Davis et al., 2009; Hellmann et al., 2003). Interesting studies could also be done on synthetic basaltic glasses with different minor elements in order to test the conjecture that alteration rates can be strongly modified by minor elements. In all cases, the issue will be to recognize, both in the laboratory and in the field, silica layers on different minerals and to determine how the chemistry of the surrounding fluid is manipulating their structure and properties, and in fine their alteration rates.

Finally, even if both grain size issues and the problem of passivating layers are solved, another fundamental difficulty might be encountered. Dissolution rate laws are usually measured under far from equilibrium conditions–in practice, under an active fluid flow which is constantly renewing the water/solid interface (the so-called mixed-flow reactor set-ups, see, e.g., Posey-Dowty et al. (1986) for a technical description). This experimental design might be very different from the hydrodynamic reality at CO₂ injection sites where the flow will be on hydrological time scales rather than those typical of laboratory flow rates. Supporting this idea, a few field studies showed that fluid-rock interactions could occur over a broad range of degrees of undersaturation with respect to feldspars and other major rock forming minerals (e.g., Kampman et al., 2009; White et al., 2001; White et al., 2002). A consequence of having little renewal of the fluid in contact with minerals is that their dissolution rate might indeed be much slower than that measured under far from equilibrium conditions.

Overcoming this difference is one of the interests in performing batch experiments, which are obviously less precise than chemostats, but which allow integration of the real effect of different conditions of departures from equilibrium (see e.g., Daval et al., 2010c, for quantitative determination of kinetic parameters in batch reactors). In any case, accurate knowledge of the relations between mineral dissolution rate (R) and the Gibbs free energy of reaction (ΔG_r) is necessary to foresee the long-term evolution of fluids and rocks after CO₂ injection. The most widespread equation in geochemical codes that links R to ΔG_r was established in the framework of transition-state theory (Eyring, 1935a,b; Lasaga, 1981), and states that R is roughly independent of ΔG_r over a wide free energy range far-from-equilibrium (for instance, at 90 °C, the dissolution rate remains 95% of the far-from-equilibrium rate as long as ΔG_r < −10 kJ mol⁻¹), whereas it decreases linearly over a very narrow range of ΔG_r values as equilibrium is approached. The simplest TST equation, that is used in most of the long-term simulation attempts of CO₂ sequestration (e.g., Gherardi et al., 2007; Knauss et al., 2005; Xu et al., 2004, 2005) fails, however, to explain either sharp drops of dissolution rate close-to-equilibrium (see the comparison of the close-to-equilibrium data on feldspars weathering from Kampman et al. (2009) to the far-from-equilibrium dissolution rate of labradorite), or a decrease of mineral dissolution rate under conditions usually considered as to represent ‘far-from-equilibrium’ conditions (e.g., Beig and Luttge, 2006; Berger et al., 2002; Hellmann and Tisserand, 2006; Hellmann et al., 2010; or Taylor et al., 2000 for feldspars; Dixit and Carroll, 2007 or Daval et al., 2010a for pyroxenes; Cama et al., 2000 for clays). In that respect, the implementation in geochemical modeling of empirical kinetic rate laws that take into account the actual rate-affinity relations measured in recent studies was thus proposed to help bridge the gap between laboratory measurements and field estimations of weathering, at least partly (e.g., Daval et al., 2010a; Ganor et al., 2007; Maher et al., 2009; Zhu, 2009). Such results emphasize the need to increase our efforts to determine experimentally the effect of fluid composition on the dissolution rates of minerals, and eventually understand why the simplest TST expression fails to link far-from-equilibrium to close-to-equilibrium dissolution rates.

To date, the mechanistic reasons for these incorrect predictions of TST in the case of some minerals are not clear, and intensely debated. The existing explanations and/or alternative models can be divided into 4 main categories: (1) the actual interfacial phase which controls the dissolution of the primary silicates has an equilibrium constant different from that of the underlying phase (e.g., Bourcier et al., 1990; Grambow and Muller, 2001), such that the application of TST to the primary silicate may be meaningless; (2) dissolution is controlled by the spontaneous nucleation of etch pits at crystal defects (Lasaga and Luttge, 2003); (3) the silicate dissolution rate is controlled by the concentration of a precursor surface complex, and R – ΔG_r relations more sophisticated than TST can be derived within the framework of TST (Oelkers et al., 1994; Oelkers, 2001); (4) the abovementioned discrepancies may not be ascribed to ΔG_r directly, but to a combination of different processes such as the surface restructuring dynamics of primary silicates, diffusion of solutes across passivation barriers and others (Casey, 2008). Although reviewing in details all of the proposed mechanisms is beyond the scope of the present paper, the two first models will be briefly described because they heavily rely on mineralogical controls. Moreover, note that a thorough review of the 3^rd category not discussed hereafter can be found in Schott et al. (2009).

The first idea is that the dissolution rate of primary phases would drop because the fluid would reach equilibrium not with the original mineral but with secondary phases coating the surface, potentially including silica layers that contain minor elements, which would determine the solid/fluid exchanges (see for e.g., Grambow and Muller (2001) for glasses and Oelkers (2001) for minerals; a theoretical review of some of these models is also available in Gin et al., 2008). Detailed knowledge of the structures and compositions of these secondary phases, as well as their observation in the field, will thus be another key for identifying appropriate field sites for efficient conversion of CO₂ into solid carbonates. Although such a model may be consistent with our own observations on olivine, note that wollastonite carbonation cannot be explained in this framework because this mineral keeps on dissolving whereas saturation is reached with respect to SiO₂(am) which coats its surface (Daval et al., 2010b). Thus, although interesting and as emphasized above, this model would need further development to be consistent for all silicates.

The second idea was proposed by Lasaga et al. in the early 1990s (e.g., Burch et al., 1993; Nagy et al., 1991; Taylor et al., 2000), and further explored and developed by Luttge et al. (e.g., Arvidson and Luttge, 2010; Beig and Luttge, 2006; Lasaga and Luttge, 2001; 2003; Lüttge, 2006). Basically, the observation of a dramatic drop in dissolution rate under conditions classically considered to be “far-from-equilibrium” (ΔG_r << 0) led these authors to propose that the driving process of dissolution should be the spontaneous nucleation of etch pits at crystal defects (screw dislocations) when ΔG_r < ΔG_r^crit and where ΔG_r^crit is the critical free energy required for opening an etch pit at a given crystal defect. Above ΔG_r^crit, it is predicted that deep etch pits should not open up spontaneously, and this, in turn, would explain the slow dissolution rates observed in the range ΔG_r^crit < ΔG_r < 0. In spite of meaningful theoretical developments of this model by Lasaga and Luttge (2001, 2003) and convincing experimental evidence (e.g., Arvidson and Luttge, 2010; Beig and Luttge, 2006), it is worth noting that the parameters controlling the numerical value of ΔG_r^crit are poorly constrained, and further research could aim at determining (1) the effective surface energies of silicate minerals and (2) the exact Burgers vectors of the dislocations from which etch pits actually originate.

To sum up this section about mineralogical controls on carbonation kinetics, it appears that the main obstacle to rapid carbonation of silicates at temperatures typical of the majority of envisaged CO₂ storage sites is the relatively small reactive (i.e. pristine) surface area actually present in most geological settings. However, a detailed knowledge of the petrophysics of the site (actual reactive surface areas) and of involved mechanisms (passivating layers, role of other phases, e.g., Zuddas and Pachana, 2010, influence of crystal defects on R – ΔG_r relations) might help to identify minerals and environments favorable to that process, especially in rocks containing significant basic and ultrabasic components. In cases where conversion of carbon dioxide into solid carbonates will be only marginal, a detailed knowledge of the fluid/mineral interface both in the laboratory and at the injection site, will still be the basis of our understanding of the environmental mineralogy of carbon dioxide geological storage.

At the Earth's surface, life drives two important CO₂ sinks, as described in the first section-the so-called biological pump and some CO₂ conversion by biomineralization into solid carbonates which eventually are stored for million years in sedimentary rocks. Can processes similar to these two occur at CO₂ injection sites?

Let us first examine the formation of carbonates, which on Earth is mostly biological, according to the following reaction:

In surface ocean waters, reaction (7) is usually thermodynamically favorable (unlike silica biomineralization by diatoms, which occurs in waters thermodynamically undersaturated with respect to solid silica), but it does not occur for kinetic reasons. Biology nevertheless allows CaCO₃ precipitation to occur using highly complex mechanisms that are crudely summarized as a concentration of calcium and carbonate ions in order to overcome the kinetic barrier (similar to diatoms which concentrate silicon for overcoming both thermodynamic and kinetic barriers for silica precipitation). This concentration phenomenon occurs in intracellular vesicles such as the coccolithosomes of the coccolithophores (e.g., Young and Henriksen, 2003). It should be noted that, in principle, the formation of carbonates according to reaction (7) is adverse to CO₂ storage because out of the two molecules of bicarbonate (which represents one form of carbon dioxide storage, the so-called ionic sequestration, which has the lifetime of the corresponding water mass), one carbon atom is liberated as carbon dioxide, dissolved in water and in equilibrium with atmospheric CO₂. This is possibly why carbonate-forming organisms are in danger when atmospheric carbon dioxide increases by virtue of a Le Chatelier principle on reaction (7) or simply because their carbonate components might eventually dissolve. However, because one of the two carbon atoms present at the beginning of reaction (7) ends up in solid carbonates, which, if sedimentation conditions are favorable, constitute a very stable storage of CO₂ in the solid Earth, the process subtracts carbon dioxide from the atmosphere/hydrosphere. This is somewhat of a paradox because the formation of carbonates by life liberates some CO₂ but it is also the mechanism by which carbon dioxide is converted into a stable solid form that is separated from the hydrosphere/atmosphere system. The liberation of CO₂ in reaction (7) is strongly decreased by coupling with the organic biological cycle as shown in reaction (6). The degree of coupling between reactions (7) and (4) on the large scale is one of the most difficult global geochemical mass balances to measure, but it is believed to be high. One might argue that organisms such as mussels, oysters, and snails obviously have no direct relationship with photosynthesis, which is true, but the amount of carbonate precipitation they represent is very small when compared to coccolithophores, foraminiferas, and corals, which either are photosynthetic (coccolithophores) or often coupled to photosynthetic symbionts (foramniferas and corals). It should be noted that even in case of perfect coupling of reactions (7) and (4), and if reaction (6) fully describes carbonate biomineralization, an increase of CO₂ concentration (i.e. ocean acidification) will still be adverse to carbonate biomineralizing organisms if dissolution of CaCO₃ occurs. Such a scenario would damage the carbonate shells and skeletons and favor an ecological transition (a major biological crisis and possible extinction event) toward non-carbonate producing organisms such as diatoms.

How can these concepts be transferred to the subterranean context of a CO₂ injection site? Not easily, obviously. Of course, photosynthesis will not occur deep in the subsurface. But other mechanisms can lead deep biosphere microorganisms to induce carbonate precipitation. Although biomineralization of carbonates has not yet been documented in the subterranean biosphere, one microorganism, Sporosarcina pasteurii, also often called Bacillus pasteurii, isolated from wastewaters, has proven to be very efficient for carbonate biomineralization under underground conditions (Ferris et al., 2004; Mitchell and Ferris, 2005). This microorganism performs urea degradation, causing a pH increase which favors the precipitation of carbonates, according to the following urea hydrolysis (8) and carbonate precipitation (9) reactions:

Several laboratory experiments (Dupraz et al., 2009a,b) have shown that injection of urea and of B. pasteurii in a synthetic aquifer water (inspired by the Paris Basin Dogger aquifer) leads to an effective pH increase and to the formation of CaCO₃ (Fig. 8). Of course, this has no operational value for CO₂ injection and should only be considered as a useful model since urea degradation produces its own inorganic carbon, as shown by reaction (8). The next step in the context of possible application to carbon dioxide geological storage would be to find subterranean species with metabolisms related to that of B. pasteurii, able to degrade organic matter into ammonia and into partially or totally denitrogenized organic molecules according to a reaction of the type:

In that case, the hydrolysis of the amine would be associated with a pH increase, allowing conversion of injected CO₂ into carbonate resulting in calcite precipitation. This process would imply that there is sufficient pre-existing organic matter in the injection reservoir or to co-inject wastewaters with the carbon dioxide, which can be envisaged since CO₂ injection will never occur in potential water resource aquifers.

A second issue is the interaction of injected CO₂ with the production of organic matter discussed in the first section of this paper. At the Earth's surface, some CO₂ is converted into refractory organic matter, escaping the almost fully closed (to about 999 per mil) cycle of photosynthesis and respiration, thus constituting a second stable form of carbon dioxide storage (the so called kerogen reservoir, which has about one fifth of the mass of solid carbonates, a proportion that has been maintained over most of the Earth's history, e.g., Schidlowski, 1988). Can similar processes occur underground at CO₂ injection sites? As already mentioned, no photosynthesis occurs in the absence of light; however, some other modes of primary production are adapted to underground conditions. Among those actually documented, methanogenesis is probably the most important (Fig. 9). Other metabolisms also exist (e.g., acetogenesis) that also mostly rely on H₂ naturally present in the subsurface. Some CO₂ might thus be pumped by thermodynamically favorable reactions with autochthonous or co-injected H₂. The fate of the organic matter thus synthesized (in first instance methane and acetic acid) from CO₂ and H₂ in the deep subsurface is absolutely unknown because the knowledge of the biological cycle of organic molecules in the deep subsurface is only in its infancy. It is thus hard to speculate whether these processes could operate at CO₂ injection sites for a durable conversion of carbon dioxide into stable organic matter in response to large CO₂ overconcentrations. Also, the extent of biomass that could be stimulated by carbon dioxide injection and would thus constitute another form of CO₂ storage is unknown. Geotechnical engineers contending with the development of massive biofilms in deep geothermal and oil fields installations know that biomass formation is not always negligible and that biocides need to be injected for avoiding these types of underground blooms (e.g., Fardeau et al., 2009, and references therein). The consequences of CO₂ injection on autochthonous or co-injected microorganisms are thus largely unknown and need to be studied as a high priority. Indeed, methanogenesis, although considered by some as a way to backtransform CO₂ into a useful CH₄ resource in former gas reservoirs, could also be seen as a very adverse effect of the interaction of injected carbon dioxide with the deep biosphere. This is because methane has a stronger greenhouse gas potential than carbon dioxide, and even small leakages would be counterproductive with respect to greenhouse gas balance. Incubation experiments with cores drilled from the deep subsurface with special care to maintain the original (very small) biomass and to avoid contaminations are probably the most appropriate way to progress in this so far very poorly constrained research area. One point of knowledge we have in this ocean of unknowns is provided by the microbiology of oil fields (Ollivier and Magot, 2005) and by that of deep aquifers (e.g., Fardeau et al., 2009). We know from these studies that sulfate-reducing bacteria are prominent in these media (Fardeau et al., 2009; Ollivier and Magot, 2005). In ocean waters, methanogenesis and sulfate reduction are indeed sometimes intimately related; microbial consortia (or looser connections) of methanotrophic and sulfate-reducing organisms have been evidenced (e.g., Boetius et al., 2000), which transform methane according to the following reaction:

In this process, carbonate ions are produced which eventually lead to the formation of calcium carbonates (e.g., Aloisi et al., 2000). The whole biological transformation of CO₂ can thus be represented by the following reaction:

Whether such metabolic phenomena can occur underground, nobody knows. A potential subsurface biological process based on this reaction is shown in Fig. 9. It would store injected carbon dioxide as solid carbonates but would, however, be a potential problem because of the generation of H₂S (a corrosive and strong greenhouse gas that could escape unless used by other microbes as it occurs on the ocean floor). Such a biological phenomenon is related to well-documented so-called souring in oil reservoirs (Ollivier and Magot, 2005). In some geological environments poor in sulfates or where iron-reducing metabolism would have been favored (e.g., in the presence of Geobacter or Shewanella species), the problem might be solved and could lead to durable conversion of the injected CO₂ into solid carbonates, FeCO₃ siderite in that case, according to the following reaction (see Fig. 9):

Such microbial processes and environments need to be actively searched for in the deep subsurface. Interestingly, it is worth noting that some siderite and ankerite precipitation are expected at some sites such as Frio (Xu et al., 2004)—far surpassing storage of injected CO₂ in calcite. The role of ankerite has also been emphasized in McGrail et al., 2006. Identification of possible biomineralization processes requires field investigations of the mineralogy of the new carbonates formed from the injected carbon dioxide.

The purpose of this paper has been to show that by careful examination of the natural processes of carbon dioxide transfer from Earth's fluid envelopes to the solid Earth (i.e. precipitation of solid carbonates and formation of refractory organic matter), one might find innovative processes in the field of CO₂ geological storage. We have emphasized two important environmental mineralogy aspects: (1) detailed understanding of the surface of silicates during their dissolution is a key to discovering and implementing rapid underground or ex situ carbonation of minerals; (2) understanding of carbonate biomineralization by deep biosphere microorganisms opens possibilities for subterranean biological conversion of carbon dioxide into stable carbonates. However, the deep biosphere also has the potential to complicate CO₂ geological storage by generating some strong greenhouse gases such as CH₄ and H₂S. Subterranean microorganisms, therefore, need to be better mapped and understood. In any case, comparison of laboratory (bio)mineralogy studies with field observations (at CO₂ injection pilot sites, in enhanced oil and gas recovery systems, and in natural CO₂ fields) is a formidable opportunity (and challenge) for environmental mineralogy.