- © 2013 Mineralogical Society of America
Nucleation is the seminal event in formation of new mineral phases. The subject has a long history, finding its modern roots in the work of Gibbs in the late 1800’s (Gibbs 1876, 1878) and continuing through the efforts of Stransky and others in the mid-20th century (Gutzow 1997). What emerged has come to be known as classical nucleation theory (CNT). Within this model, mineral nucleation comes about through ion-by-ion addition to a growing cluster (Fig. 1A), whose origin lies in the inherent thermodynamic fluctuations of the solution. In recent years, this well-accepted theory has been called into question by an increasing number of studies, both experimental and theoretical, reporting so-called “non-classical” pathways of nucleation involving aggregation of clusters or nanoparticles that exist in equilibrium with the free ions (Fig. 1B). Moreover, such studies often report the nuclei consist of an amorphous precursor, which transforms to the more stable crystalline products only after nucleation has long ended. Presumably, if these cluster-based and multi-stage processes dominate, it is because they offer a lower barrier pathway to nucleation than does the pathway through ion-by-ion addition. Based on the many years of surface science research since the development of CNT, we now know that energy landscapes are unlikely to be flat; i.e., the free energy of a cluster of ions should depend on its size and/or configuration, and composition, including that of the solvent. The resulting peaks and valleys in the energy landscape will naturally create cluster populations, which could give rise to unexpected pathways. In the context of a geological reservoir, consideration of mineral surfaces, organic films, and pores is crucial, because both substrates and pores can reduce the classical barrier to nucleation and, hence, may play a large role in modifying energy landscapes, re-directing pathways, and determining nucleation rates. The purpose of this chapter is to examine the impact of cluster-based pathways, surfaces, and complex energy landscapes on CNT, and then examine findings from in situ studies of nucleation on mineral substrates and organic films. We finish by discussing the implications for geologic CO2 sequestration.
THERMODYNAMIC DRIVERS OF NUCLEATION
Mineral nucleation occurs because the free energy of the initial solution phase is greater than the sum of the free energies of the crystalline phase plus the final solution phase (Gibbs 1876, 1878). In terms of solution activities (which are often well approximated by solution concentrations), an equivalent statement is that the actual activity product of the reactants, AP, exceeds the equilibrium activity product of those reactants, the latter being simply the equilibrium constant, Ksp. In cases where a single chemical component crystallizes, the driving force has been expressed not in terms of total free energy change during crystallization, but rather as the change in chemical potential of the crystallizing species, Δμ. This Δμ measures the free energy response to molecules transferring from one phase to the other. The larger Δμ becomes, the greater is the driving force for crystallization (Mullin 1992).
Not surprisingly, both the change in free energy and the change in chemical potential are directly related to the activity products (Johnson 1982). For the precipitation reaction:(1)
the activity product of the reactants, AP, and the value of Ksp are given by:(2) (3)
where the subscript “e” refers to the activity at equilibrium. In turn, the free energy of solution per molecule, ΔGs, and the change in chemical potential are:(4) (5a) (5b)
where kB is Boltzmann’s constant and T is the absolute temperature. Rather than use Δμ, most crystal growth analyses refer to the supersaturation, σ, which is related to Δμ by:(6) (7)
Readers may often see σ defined as AP/Ksp − 1 or even C/Ce − 1 where C and Ce are the equilibrium reactant concentrations. These are referred to as absolute supersaturations. They are valid approximations only at small values of σ and, in the case of the expression that utilizes concentrations, only useful when either the activity coefficients, χ = [M]/C and χe = [Me]/Ce for the reactants are close to unity, or when χ = χe. Nonetheless the absolute supersaturation is important when considering the kinetics of growth. This will become clear later in the chapter.
CLASSICAL NUCLEATION THEORY
Nucleation is one of the two major mechanisms of the first order phase transition, the process of generation of a new phase from an old phase whose free energy has become higher than that of the emerging new phase (Hohenberg and Halperin 1977; Chaikin and Lubensky 1995). Nucleation occurs via the formation of small embryos of the new phase inside the large volume of old phase. Another prominent feature of nucleation is metastability of old phase, i.e., the transformation requires passage over a free energy barrier (Kashchiev 1999). This is easily understood by considering the free energy changes associated with the formation of the nucleus. The statement that the free energy per molecule of the new phase is less than that of the solvated phase only applies to the bulk of the new phase. The surface is a different matter. Because the surface molecules are less well bound to their neighbors than are those in the bulk, their contribution to the free energy of the new phase is greater (Fig. 2A). The difference between the free energy per molecule of the bulk and that of the surface is referred to as the interfacial free energy. (It is sometimes called the surface free energy but, strictly speaking, this term should be reserved for surfaces in contact with vacuum.) The interfacial free energy is always a positive term and acts to destabilize the nucleus. As a consequence, at very small size when many of the molecules reside at the surface, the nucleus is unstable. Adding even one more molecule increases the free energy of the system. On average, such a nucleus will dissolve rather than grow. But once the nucleus gets large enough, the drop in free energy associated with formation of the bulk phase becomes sufficiently high that the surface free energy is unimportant and every addition of a molecule to the lattice lowers the free energy of the system (Fig. 2B). There is an intermediate size at which the free energy of the system is decreased whether the nucleus grows or dissolves, and it is known as the critical size. This phenomenon of size-dependent stability is referred to as the Gibbs-Thomson effect. Of course, if the supersaturation is high enough, the critical size can be reduced to less than one growth unit. Then the barrier vanishes and the old phase becomes thermodynamically unstable relative to the new so that an infinitesimal fluctuation of an order parameter, such as density, can lead to the appearance of the new phase. The rate of generation and growth of the new phase is then only limited by the rate of transport of mass or energy. This second process is referred to as spinodal decomposition, and the boundary between the regions of metastability and instability of the old phase is called spinodal line (Binder and Fratzl 2001; Kashchiev 2003).
Below the spinodal, the rate of nucleation is largely controlled by two important energetic barriers. The first is the excess free energy associated with the creation of the new interface as discussed above. This is an ensemble property that creates a thermodynamic barrier ΔGc due to the collective behavior of the ions in the solid and liquid phases. The second is an effective kinetic barrier EA arising from the individual reactions associated with transport of energy and mass, such as desolvation of solute ions, attachment to the forming nucleus, and structural rearrangements. Both barriers appear exponentially in the expression for the rate of nucleation J through (De Yoreo and Vekilov 2003):(8)
where A is a pre-factor that is determined by geometric factors and material-dependent parameters (e.g., density) and ΔGc is a decreasing function of the chemical potential Δμ = σ/kBT, where σ is the supersaturation, kB is Boltzmann’s constant and T is the absolute temperature.
While the exponential dependence of J on σ through the free energy barrier is a universal hallmark of nucleation that, in essence, distinguishes it from a simple chemical reaction, the exact forms of A and ΔGc are model dependent. The source of ΔGc is a positive excess free energy ΔGex of the solid phase that adds to the change in free energy for a simple chemical reaction, which is simply given by (dG/dn)Δn where Δn is the number of molecules passing from the solution to the solid phase. Note that for a supersaturated solution, (dG/dn) < 0. Thus without ΔGex there would be no barrier and precipitation would happen spontaneously at infinitesimal supersaturation without nucleation. As discussed above, in classical nucleation theory (CNT), ΔGex arises from the free energy of the interface between the mineral and the surrounding solvent and substrate (Fig. 1A). When the free energy landscape is flat—that is, the excess free energy is simply determined by the surface area times the interfacial free energy α, which is independent of size—then ΔGc is given by (De Yoreo and Vekilov 2003; Hu et al. 2012a):(9)
Equations (8) and (9) show that the magnitude of the interfacial free energy has a great influence over nucleation rates, because it appears as a cubic term in an exponential. Sohnel (1982) and Sohnel and Mullin (1978) showed that interfacial energies scale inversely with the logarithm of solubility. Consequently, for sparingly soluble crystals the interfacial energy is quite high and, hence, the thermodynamic barrier to homogeneous nucleation is formidable. In the case of calcite, based on the literature value of 109 mJ/m2 for the interfacial free energy (Sohnel 1982; Sohnel and Mullin 1978) the predicted classical barrier to homogeneous nucleation of a calcite rhomb ranges from 175 kBT to 93 kBT for CaCl2 and NaHCO3 concentrations between 10 mM and 29 mM—the latter marking the literature value for the solubility limit of amorphous calcium carbonate (Brecevic and Nielsen 1989) (ACC) (Fig. 3, bottom curve, σ = 4.8). (Note that, at 300 K, 1 kBT = 2.6 kJ/mol = 0.62 kcal/mol.)
Equation (9) also reveals the extreme level of supersaturation needed to reduce the classical free energy barrier for nucleation of ACC below that for calcite. Based on the scaling of interfacial free energy with solubility, the ratio of α for ACC to that of calcite is of order 0.75. Taking into account the differences in the parameter B for calcite and ACC, one finds that the free energy barrier to forming calcite will be less than the barrier to forming ACC until the solution concentration is increased to the point where the supersaturation relative to ACC exceeds ~65% of the supersaturation relative to calcite (Hu et al. 2012a). Because the solubility of ACC is so much higher than that of calcite, for equal mixtures of CaCl2 and NaHCO3, this condition can only be reached if the solution Ca2+ concentration is far in excess of 100 mM. Even if the solubility of ACC is considerably smaller than reported in the literature as seems likely based on recent titration studies (Gebauer et al. 2008) as well as in situ observations of nucleation in which bulk ACC formation was observed at less than half the literature value of the solubility (i.e., 13 mM vs. 28 mM final Ca2+ concentration) (Hu et al. 2012a), the required concentration for ACC to be favored thermodynamically is still extreme. For example, even if we take the equilibrium solubility to be as small as 10 mM (final Ca2+ concentration), a Ca2+ concentration of approximately 100 mM is required before there is a crossover in barriers. Even at that concentration, the classical barrier is still in excess of 53 kBT. The clear conclusion of this analysis is that, for a flat energy landscape, the concentrations required to achieve homogeneous nucleation of either phase are extreme and inconsistent with experimental observations of both calcite and ACC nucleation at significantly lower concentrations. Hence, either nucleation is heterogeneous or other pathways of precipitation that avoid this barrier must be at work.
The presence of a foreign surface can be used to exert even greater control over nucleation because, quite often, the interfacial energy between a crystal nucleus and a solid substrate is lower than that of the crystal in contact with the solution (Nielsen 1964; Abraham 1974; Chernov 1984; Mullin 1992; Mutaftschieve 1993). This is because the molecules in the crystal can form bonds with those in the substrate that are stronger than the bonds of solvation. Because the enthalpic contribution to the free energy comes primarily from chemical bonding, stronger bonds lead to a smaller interfacial free energy. This is likely to be a physical phenomenon of great importance in mineral environments, where a range of mineral-types and crystallographic faces are exposed. Clearly the strength of bonding at the interface is strongly dependent on the structure and chemistry of the substrate surface. If the atomic structure of the substrate surface closely matches a particular plane of the nucleating phase so that lattice strain is minimized and, in addition, the substrate presents a set of chemical functionalities that promote strong bonding to the nucleus, then the enthalpic contribution to the interfacial free energy becomes small and nucleation occurs preferentially on that crystal plane.
Because the thermodynamic barrier depends upon the cube of the interfacial energy, heterogeneous nucleation on surfaces (Fig. 4A) that reduce the interfacial energy can proceed at dramatically altered rates. In this case, α becomes an effective interfacial energy αhet that depends on the interfacial energies of the crystal-fluid, fluid-substrate, and crystal-substrate interfaces through:(10)
where h is a factor that depends on the aspect ratio of the nucleus (see SI of Hu et al. 2012a for details). As long as αcs < αls, the value of αhet will be reduced from that for the homogeneous nucleus. However, even if the effective αls equals αcs, that is, the interfacial energies for the crystal-substrate and fluid-substrate interfaces are equal, the barrier will already be reduced by a factor of 1.6 (Fig. 4B, αhet/αcl = 1.0) simply because a surface that would have been generated during homogeneous nucleation is now a crystal-substrate interface that carries no energy penalty. A further reduction in αhet by only 20% to 50% due to αcs < αls would lead to a decrease in the barrier by a factor of 3 to 13 (Fig. 4B, αhet/αcl = 0.8 and 0.5). Given that nucleation rate depends exponentially on this barrier, these large reductions mean that surfaces have the potential to completely alter the dynamics and pathways of calcite formation.
Based on Equation (10), it is worth pointing out that a substrate for which the interfacial energy with the surrounding solution is extremely high can drive nucleation to occur on the substrate rather than in the surrounding solution even on a crystal-substrate interface that is not particularly well matched, simply because covering a patch of the substrate with the crystalline nucleus hides it from the solution and thus reduces the overall free energy of the system.
The impact of heterogeneous surfaces on nucleation rates is expected to be enhanced when those surfaces contain steps, grooves or pores (Page and Sear 2006; Hedges and Whitelam 2012). The essential reason for the effect is that the presence of a corner increases the fraction of the forming nucleus in contact with the surface. The magnitude of the effect depends strongly on pore size and aspect ratio, and for a given driving force there is an optimum size. When the interaction between the surface and the crystal is favorable, the impact is dramatic. For example, Hedges and Whitelam (2012) show that, for conditions in which the bulk barrier to nucleation is 67 kBT and the interaction between the surface and the mineral reduces that barrier to 45 kBT placing an appropriately sized square pore in that surfaces reduces the barrier to only 20 kBT. When these barriers are translated to nucleation rates, the surface alone enhances the rate by 9 orders of magnitude over the bulk rate, a square groove increases it by another 7 orders of magnitude, and a square pore further increases it by a factor of 70, giving a total enhancement of nearly 18 orders of magnitude (Fig. 5)!
Deviations from a flat energy landscape: Cluster aggregation and size dependent α
In truth, the free energy landscape is unlikely to be flat (Fig. 6, long-dashes) at small sizes. ΔGex must approach zero at the size of a molecule (shown schematically in Fig. 6 by dotted curve) and probably exhibits local minima and maxima at very small cluster size as certain configurations expose more or less favorable coordination geometries for the surface ions (shown schematically in Fig. 6 solid and short-dashed curve). While these variations in ΔGex are easy to account for by expressing α as a function of size, they do little to change the basic physics of the nucleation process. Nonetheless, they can potentially have significant effects on the magnitude of the barrier (Figs. 7A and 7B) if the size at which that barrier is reached—i.e., the critical size—becomes comparable to the dimensions at which size effects begin to emerge (dotted curve in Fig. 6), or where local or global minima in the free energy landscape create a population of metastable or stable clusters, respectively (solid and short dashed curves in Fig. 6). These can aggregate to form a critical nucleus (Fig. 1B).
Unfortunately, not much is known about the size dependence of the interfacial free energy α. What little data do exist suggest a slight rise with decreasing size, followed by the beginning of a decrease in magnitude (Zhang et al. 2009), but those data do little to constrain the dependence in the region below 5 nm diameter, which is greater than the 1–3 nm critical size seen in Figure 2A. Theoretical treatments suggest that even a single formula unit already possesses much of the energetic features of the bulk (Gibbs et al. 2011). This suggests the fall-off in interfacial energy may not occur until diameters below 1 nm, though these simulations were performed for molecular solids and cannot be directly translated to ionic crystals like calcite. However, metadynamics simulations of equilibrium calcite structure suggest the energetic features of the bulk are still manifest below 2 nm (Harding 2011). Indirect evidence for a complex dependence on size comes from both cryoTEM (Pouget et al. 2009) and ultra-centrifugation data (Gebauer et al. 2008) that suggest there is indeed a population of sub-critical clusters (commonly referred to as pre-nucleation clusters) with a tight size distribution, which implies there is indeed a minimum in the free energy vs. size, as shown schematically in Figure 6 (solid and short-dashed curves). In fact, titration-based studies on the amount of calcium inferred to be bound in these clusters concluded that they occupy a global minimum, i.e., the free energy of the pre-nucleation clusters lies below that of the free ions (Fig. 6, short dashed curve) (Gebauer et al. 2008; Gebauer and Coelfen 2011).
Deviations from a flat landscape will change the dependence of ΔGc on σ (Eqn. 9). In the case of a size dependent α, the change can be complex and depends on the form of the size dependence. For nucleation by aggregation of clusters that occupy a local minimum in the free energy, the dependence becomes (Hu et al. 2012a):(11)
where C is a constant that depends on the shape factor, the cluster radius and the excess free energy of the cluster, and the plus or minus sign depends on whether the minimum in ΔG is local or global (for details, see Hu et al. 2012a). If it is a local minimum (Fig. 1D, solid green line), then the clusters are metastable, they carry excess free energy above the free ions, the plus sign applies, and the barrier is reduced. If it is a global minimum (Fig. 1D, dashed green line), then the clusters lie below the free ions, the minus sign applies and nucleation by cluster aggregation brings with it an extra energy cost. This is the case for pre-nucleation clusters, which were found to lie about 18 kJ/mol below the free ions (Gebauer and Coelfen 2011). Thus creation of a super-critical nucleus by aggregation of pre-nucleation clusters would bring with it a larger barrier than aggregation of free ions, regardless of whether the end product is an amorphous or crystalline phase.
One interesting implication of Equation (11) is that the impact of clusters—stable or metastable—is to effectively alter the supersaturation from σ to σ ± C, making it larger for metastable clusters and smaller for stable clusters. The result is that the existence of metastable clusters increases the probability of nucleating ACC rather than calcite, while the existence of stable clusters decreases that probability. This effect was further explored in a number of recent studies. Habraken et al. (2013) found that calcium phosphate nuclei formed by aggregation of nm-sized metastable complexes, which carried an excess free energy above the free ions, leading to a reduced barrier and appearance of the amorphous phases at anomalously low supersaturations. In the iron oxide system, Baumgartner et al. (2013) observed formation of the magnetite phase through aggregation of highly disordered ferrihydrite precursor particles. Similarly, Van Driessche et al. (2012) discovered that gypsum formed through a two-step process in which bassanite nanoparticles first formed at concentrations below the solubility limit for that phase and transformed into gypsum as they grew in size, or upon attachment to existing gypsum crystals. These latter cases harken to an effect predicted by Navrotsky (2004), who pointed out that, because enthalpies of formation of polymorphs can cross over as the particle size decreases into the nanoscale, one might observe an inversion of the relative solubilities. As a consequence, the less ordered phase, which is typically is more soluble at macroscopic size and has a lower interfacial energy, would become the thermodynamically favored phase. Because the disordered phase also typically exhibits the lower interfacial energy, considering Equation (2), it follows that, if this size effect on phase stability is indeed manifest, the less ordered phase would appear first at all supersaturations, forcing the system to follow a two-step pathway to formation of the final stable phase.
GISAXS MEASUREMENTS OF INTERFACE PRECIPITATION AND NUCLEATION RATES
GISAXS is a versatile extension of the well-known SAXS technique applied to interfaces between materials of different electron density (Levine et al. 1989). By suitable selection of experimental geometry it is possible to probe buried interfaces in solid materials (Naudon 1995), or at solid/liquid interfaces. The technique relies on the density differences at the interface to set up a grazing incidence geometry allowing total reflection of X-rays into the lower density medium (Fig. 8). Such scattering geometry allows scattering normal to the interface (z direction), and along the interface (xy direction or 2θ direction) to be probed independently, so that scattering information for all directions along and normal to the surface can be measured. Such data conveys detailed particle size and shape, as well as interparticle separation information unavailable with classical SAXS measurements, where all scattering directions in the detector plane are equivalent.
For aqueous interfaces the technique is ideal for probing the evolution of precipitates from initial nucleation, through growth and evolution (Fig. 9), and into development of faceted crystallites (Fig. 10). The practical limits of the technique depend on range of the scattering vector (q, in dimensions of reciprocal length) available, which in turn is determined by the X-ray scattering setup, detector size, sample to detector distance, and incident beam divergence. Given highly brilliant X-radiation from a synchrotron undulator beamline, precipitates down to less than 1 nm can be detected, and growth followed up to sizes 2π/(lowest q value), typically on the order of 50–100 nm, but extendable to microns using long sample to detector distances with highly collimated X-rays. The technique is also fast, as scattering consists of a fraction of the incident beam (typically several %) and thus good counting statistics can be obtained in fractions of one second.
GISAXS combines elements of classical SAXS and grazing-incidence diffraction, hence for accurate analysis one would consider scattering not only involving particles at an interface, but also higher order scattering such as involving reflection from the surface first, then subsequent particle scattering, along with other scattering paths. For aqueous mineral interfaces with relatively low reflectivity, we might generally neglect the higher order scattering, a model known as the Born approximation. However in the case where reflectivity is high, e.g., at highly perfect surfaces below the critical angle, the second order scattering must be considered, and the distorted wave Born approximation (DWBA) is used in modeling to take account of several important scattering contributions (Rauscher et al. 1995; Lazzari 2002). In general the scattered intensity is related to the scattering power of the atoms in the precipitate, the difference in density between precipitate and contacting medium, and the number of scattering particles. One may speak of X-ray “contrast” which is proportional to the square of the electron density differences between precipitate and surrounding medium. Precipitates having low contrast are difficult to measure, even if abundant. The intensity is also modulated by the particular arrangement of particles in the precipitate, the structure factor or S(q), and the shape of the particles, the form factor P(q). Hence we have(13)
where N is the number of particles, Vp is the particle volume, V is the X-ray illuminated volume, ρ1 is the particle density and ρ2 is the matrix density. Data analysis is commonly first done with direct methods, i.e. methods that do not require use of specific model fitting or assumptions. Rather than process an entire 2D pattern, the image file is generally “sliced” to separate out the intensity change with q both vertically (z direction) and horizontally (lateral or xy direction). Such slice positions are shown in dashed lines in Figure 9. Analysis of these one dimensional patterns begins by subtracting the background scattering from the GISAXS pattern followed by proper scattering curve normalization using the incident beam intensity. For aqueous systems involving precipitation, it is important to consider the absorption of the X-rays in the liquid phase. If a high flow rate in a reaction cell is maintained, the liquid composition should remain stable, but if the solution is not continuously refreshed then a separate measurement of the solution absorption must be made to allow for solute loss to the interface. For cases of high dilution where the particle separation is relatively large (e.g. less than 1% interface coverage), inter-particle scattering effects are negligible, and we can work on the basis of analysis of a single averaged particle and neglect the structure factor. This allows use of the Guinier approximation, applicable to the smallest q region in the scattering curve. In this region the slope of the scattering curve plotted as log I vs. q2 is related to the size of this average particle, known as its radius of gyration (Rg). Rg is representative of the rms distance from the edges of the particle to its center of scattering, and for the Guinier region the slope of the scattering curve will be −Rg2/3.
In the high q limit of the GISAXS scattering, before approaching the wide angle X-ray scattering (WAXS) region (where interatomic distances within the particles become important) at still larger q, another useful approximation is due to Porod. Here if sharp interfaces are present between particles, the integrated surface area of the particles, S, can be estimated from(14)
from a log I versus log q plot. Other slopes are consistent with rough interfaces of fractal dimension D (slope = 1/q6–D), or other complex shapes. Figure 11 also shows other aspects of the scattering curve, e.g. the presence of knees indicative of particle shapes, e.g. elongate rods will have knees indicative of both the thickness and length. Additional data analysis can be done using the invariant function, Q, which is equivalent to the area under the weighted scattering curve, or(15)
taken over the full range of the SAXS data from small q limit to the beginning of the WAXS region. The invariant is proportional to the scattered volume and hence can be used to estimate precipitate volume. However it must be integrated well beyond the range where q shows any finite size features to be accurate.
The first application of the GISAXS method applied to aqueous geochemical interfaces has conducted by Jun et al. (2010) on precipitation of Fe oxide on quartz. Here the precipitate geometry, particle size and nucleation rate are seen to vary substantially as a function of saturation index or ionic strength. In addition, utilizing this method, the effect of environmentally abundant anions (Hu et al. 2012b) and organic coated mineral surfaces (Ray et al. 2012) on the nucleation and growth of Fe oxide were observed. If only nuclei contribute to the bulk of the scattering, and if there is little growth once nuclei form, one can use the invariant function as a proxy for nuclei density without use of a specific fitting model. Then changes in the invariant can be used to obtain nucleation rates. However if there is growth of nuclei in addition to new nuclei formation, then the various particle components must be fitted and the quantitative numbers of nuclei extracted. An additional consideration is the presence of homogeneous nucleation in solution, where the resulting particles can be collected by the interface and thus miscounted as heterogeneous nuclei. In synchrotron GISAXS experiments one can move the aligned sample in the z direction to allow the beam to transit only the solution, thus examining the scattering intensity from any homogeneous nuclei. Experiments may also be done in inverted mode so that gravity is used to frustrate homogeneous particle collection by the interface.
The first GISAXS studies of nucleation and growth of CaCO3 on high quality single crystal quartz substrates (Fernandez-Martinez et al. 2013) examined a range of solution saturations, each producing different nucleation rates (see next section). If all nuclei can be assumed to be the same phase, then the various nucleation rates can be directly related to precipitate/substrate interfacial energy. In principle similar analysis can be applied to any nucleation system to determine interfacial energies using the assumptions of classical nucleation theory. This is especially valuable as information on interfacial energies for mineral precipitation on contrasting mineral substrates is extremely limited.
GISAXS: from scattered intensity to interfacial energy
The use of time-resolved in situ GISAXS experiments using a flow-cell allows keeping a constant supersaturation in the solution, i.e., allows working in an open system and probing the thermodynamics of carbonate mineral nucleation. Experiments are performed measuring scattering patterns at different time points while the mineral substrate of interest is being exposed to a saturated carbonate solution. By placing the mineral substrate in total reflection conditions, the particles nucleated on the surface scatter the incident radiation onto a 2D detector, generating a scattering pattern that contains 3D information about the precipitate. This is illustrated here with the example of CaCO3 particles nucleated onto a quartz (100) substrate (Fernandez-Martinez et al. 2013) With time, an increase in the scattered intensity is observed at intermediate q values, which is associated with the formation of CaCO3 particles on the quartz substrate. As mentioned above, the average vertical and lateral particle sizes of the nucleated particles can be obtained from the distribution of the scattering intensity along the momentum transfer directions perpendicular and parallel to the plane of the substrate, respectively. Figure 12a shows a horizontal integration of the intensity at different times during the nucleation experiment.
In this particular case, the scattering intensity is for that of an isotropic system, and is proportional to:(16)
where P(q,R,σd) is the form factor and q is the scattering vector (Renaud et al. 2003). The intensity can be modeled using a log-normal polydisperse model of non-interacting spherical particles (Guinier 1994) with:(17)
where n(R,σd) is the log-normal distribution, which is used to represent the observed size polydispersity of the nucleated particles, with R being the mean radius of the particle and σd its standard deviation. V and Δρ are the particle volume and the electronic density difference (or contrast) between the particle and the surrounding solution. The low-q part of the curves shows a power law behavior associated with the formation of aggregates. A fit to the data in the high-q region using a log-normal distribution of polydisperse spherical particles yields a mean particle size (radius) of ~2 nm, with typical standard deviation values of ~0.09 nm.
Analyzing the development of particle scattering at the initial stages of the experiments under different supersaturation conditions allows us to obtain thermodynamic information of the CaCO3-quartz system. According to Equations (16) and (17), three factors affect the total level of scattered intensity, namely the contrast (Δρ), the square of the particle volume (V) and the number of particles (n). In the case of CaCO3 nucleation of quartz (100), the size of these primary particles remains constant during these initial stages, which suggest that nucleation, instead of growth, is the dominant precipitation mechanism contributing to the increase of the GISAXS intensity. This allows using the time-evolution of the GISAXS invariant (Eqn. 15) as a proxy for the nucleation rate, under the assumption that the nature of the CaCO3 polymorph that is being formed remains unchanged, and hence that V and Δρ are constant. Plots of the so-called Lorentz-corrected GISAXS intensity (f(q) = I(q)·q2) for two different time points are shown in Figure 12a. These curves have been calculated from the fitted function to the experimental data, which allows extrapolating the data at high-q values and obtaining a converged value for the invariant at each time point. The invariant is equivalent to the subtended area underneath the Lorentz-corrected GISAXS intensity curves. Plots with the time-evolution of the invariant for a set of experiments at different solution supersaturation values are shown in Figure 12b, together with linear fits to the data. Following the observation that the size of the primary particles is constant throughout the early stages of the experiment, the slope of each of these curves is then equivalent to the rate of heterogeneous nucleation of CaCO3 on quartz.
In the case of systems showing nucleation and growth (and in the absence of Ostwald ripening processes), fitting polydisperse size distributions to the data is required. Nucleation rates can then be extracted then by integrating the derived particle number distributions at each time point, and making a linear regression to these data.
The effective interfacial energies can be extracted from these data with the help of classical nucleation theory discussed earlier. A plot of the natural logarithm of J against 1/σ2 is shown in Figure 12c. A linear regression of this plot yields a value for the slope that can be interpreted in terms of an interfacial free energy αhet by using Equation (10). The fact that the obtained αhet value is smaller than αwater-calcite = 109 mJ/m2 (the interfacial energy that governs CaCO3 homogeneous nucleation) implies that the thermodynamic driving force for heterogeneous nucleation of CaCO3 on quartz is higher than for homogenous nucleation. A comparison between the height of the molar free energy barriers for nucleation (ΔGc) for both processes gives a ratio ΔGc-hete/ΔGc-homo × 100 = 2.5% for calcite and an estimated ratio of ΔGc-hete/ΔGc-homo × 100 = 15% lower if ACC is the nucleated polymorph. This implies that heterogeneous nucleation will be more significant over homogeneous nucleation given that kinetically-limiting factors for both processes are unimportant.
AFM OBSERVATIONS OF NUCLEATION AND GROWTH OF NEWLY FORMED PRECIPITATES
Atomic force microscopy (AFM) is a powerful tool for qualitatively and quantitatively observing in situ surface reactions. It also provides a means to observe in situ nucleation at solution-substrate interfaces in real-time while allowing for changing bulk aqueous conditions, but it cannot detect nuclei in solution (Jun et al. 2005). The technique has sub Angstrom vertical resolution, but the lateral resolution is limited to ~40–60 nm features at solution-solid interfaces depending on the specific instrument. If the nuclei are smaller than the lateral resolution, one cannot make an assessment of shape. In addition, newly formed nuclei are often very soft and can be easily wiped out by interactions with the AFM tip. Therefore, careful operations are needed.
The AFM operates a piezo feedback loop based on the force relationship between the tip and sample, which leads to deflection of the cantilever. AFM height mode images provide vertical and lateral dimensional information and phase mode images (see inset images in Fig. 15 below) provide variations in composition, adhesion, friction, viscoelasticity, and other properties (Babcock and Prater 2004). Surface morphology changes by dissolution of pre-existing phases and nucleation and growth of new phases at a solution-solid interface can be observed using an AFM equipped with a fluid cell (Fig. 13). For example, metal carbonate surfaces can be reacted with solutions introduced into the fluid cell by a syringe pump and effluent from the fluid cell can be collected and analyzed off-line for aqueous metal concentrations using inductively coupled plasma mass spectrometry (ICP-MS) or atomic absorption spectroscopy. The setup in Figure 13 allows for real-time molecular scale observations of film growth while controlling factors such as ionic strength, pH, and saturation ratio.
This technique has been previously used to investigate thin film growth on carbonate substrates. Lea et al. (2001, 2003) reported that by introducing 150 mM [CO32−] and 10−5.7 M Mn2+ at pH = 8.9 Mn0.5Ca0.5CO3 nucleates and grows on the (101̄4) surface of CaCO3. In particular, the shape of the new Mn0.5Ca0.5CO3 precipitate grows parallel to the [22̄1] direction of the CaCO3 substrate. Astilleros et al. (2002) also observed in situ precipitate film growth on the (101̄4) CaCO3 surface by adding Ca2+, Mn2+, and [CO32−] at pH = 10.2 and showed that, for low [Mn2+](aq), a MnxCa1−xCO3 film initiates at steps, whereas at high [Mn2+](aq) 2D nuclei form on terraces. Jun et al. (2005) also reported the heterogeneous nucleation and growth of a Mn(III) oxide thin film on the (101̄4) surface of MnCO3 at circumneutral pH and 1 atm O2. Heteroepitaxial nucleation and oriented growth of Mn(III) oxide islands on the carbonate substrates MnCO3, MgCO3, and CaCO3 was also observed (Fig. 14). These islands grow on MnCO3 terraces with a rhombohedral 2D shape. On highly stepped surfaces, the islands lack a euhedral form and additional aqueous Mn2+ created rounded precipitate islands rather than rhombohedral shaped islands. Interestingly, the Mn(III) oxide islands were observed on MgCO3, but not on CaCO3 due to lattice mismatch between the new precipitate film and CaCO3 substrate. While these studies provide good qualitative descriptions of nucleation and growth processes, statistically improved quantitative analyses are required to obtain the nucleation rates for larger surface areas.
Ex situ AFM observations of GISAXS samples
AFM also can be used to observe surface morphological changes ex situ. Figure 15 shows ex situ AFM observations of CaCO3 nucleation on the (100) surface of quartz as function of saturation (defined as the natural log of the ion activity product divided by the calcite solubility product, or ln IAP/Ksp,). The measurements were made within 5 hours of nucleation completion, and the images were shown here in height mode. The solution conditions were similar to analogous GISAXS experiments described above. For AFM imaging water was removed, first by washing with pure ethanol to remove remaining dissolved metal ions and solution, and then with drying by application of a high purity nitrogen gas jet. As predicted from classical nucleation theory, the nucleus size at lower saturation was the largest observed, and the size decreased with saturation. The observed sizes from AFM measurements can be somewhat smaller than from GISAXS observations due to particle dehydration. A few larger particles were observed that were formed by homogeneous nucleation and collected by the interface. However, their particle sizes were much larger than the 10 nm range exhibited by the heterogeneously nucleated particles, as also expected from classical nucleation theory. Morphology and preliminary Raman spectroscopy indicate that these can particles may be vaterite. Further investigation is needed to confirm whether there is a phase transformation after the nucleation process, perhaps from amorphous calcium carbonate (ACC) to vaterite.
CALCIUM CARBONATE NUCLEATION ON ORGANIC FILMS
The recognition that organic films could have a large impact on nucleation came from investigations in the field of biomineralization where an organic matrix composed of proteins and/or polysaccharides is often found in association with highly organized mineral elements of hard tissues ranging from the carbonate shells of marine organisms (Addadi et al. 2006) to the phosphate-based bones of humans and other mammals (Weiner et al. 2005). Early work on control over calcium carbonate nucleation by organic films focused on either Langmuir monolayers at an air-water interface, polydiacetylene films on mica, or organothiol self-assembled monolayers (SAMs) on noble metal surfaces. Following the landmark paper by Aizenberg et al. (1999a), showing that SAMs could be used to control the location and orientation of calcite crystals with high specificity and fidelity, many studies, both experimental and theoretical, explored the source of structural control in this system (Aizenberg 1999b; Han and Aizenberg 2003; Travaille et al. 2003; Duffy and Harding 2004; Lee et al 2007). Even small modifications in SAM composition or structure, such as changing the endgroup functionality (Aizenberg 1999b) or hydrocarbon chain length (Han and Aizenberg 2003), or altering the noble metal substrate (Aizenberg et al. (1999a) were shown to alter the preferred orientation of the final suite of crystals.
More recently, in situ AFM studies of calcium phosphate nucleation on collagen demonstrated that the effective interfacial energy was greatly decreased from that expected for bulk solution (Habraken et al. 2013), while in situ optical studies of calcium carbonate nucleation came to a similar conclusion in the case of organothiol SAMs on gold for both odd and even parity hydrocarbon chain length (Hu et al. 2012a), as well as a range of different headgroup chemistries Hamm et al. 2013). In the latter case, a direct correlation between crystal-SAM binding energy and interfacial energy was established. Similar studies using polysaccharide films revealed a systematic relationship between the barrier to nucleation and the surface charge of the film (Giuffre et al. 2013). The latter two studies provided further insights into the terms within the effective interfacial energy αhet (Eqn. 4) that are responsible for the reduced barriers. As we show below, these studies reveal the powerful impact surfaces can have on controlling the pathway and enhancing the rate of nucleation.
In situ studies of calcium carbonate nucleation on organic films have been based on the use of a quartz cuvette modified to allow continuous flow of solution and placed in an inverted optical microscope (Hu et al. 2012a; Nielsen et al. 2012) (Fig. 16A). Calcium and carbonate buffers from two syringe pumps are mixed at a T-junction some distance from the inlet to the cell and a substrate coated with the organic film is placed upside down within the cell. Images are recorded using a CCD camera and the number of crystals forming on the substrate (Fig. 16B–D) is measured as a function of time (Fig. 17A). Once a necessary incubation time has passed, the number of crystals grows linearly with time until the spacing between them becomes too small to allow for independent nucleation events, at which point the increase in number density begins to level off. The nucleation rate is then defined by the slope of the linear region in the plot of number vs. time. These measurements are repeated over a range of supersaturation, giving the dependence of nucleation rate on supersaturation.(18)
Hu et al. (2012a) measured J for a range of supersaturations on SAMs of 16-mercaptohexadecanoic acid (MHA) and 11-mercaptoundecanoic acid (MUA) to determine if indeed the prediction of CNT was obtained and whether these surfaces led to a reduction in interfacial free energy. Qualitatively similar behavior was observed for both the even chain-length (MHA, C-16) and odd chain-length (MUA, C-11) films, but the nucleation rate was greater on MHA. In addition, in accord with previous reports, nucleation occurred on distinct crystallographic planes for the two different SAMs (Travaille et al. 2002; Han and Aizenberg 2003). In contrast, under identical conditions, nuclei on SAM-free gold films were few in number and exhibited random orientations with sporadic occurrences of vaterite.
In accordance with the prediction of Eqn. 12, plots of ln(J) vs. σ −2 were found to be linear (Fig. 17B). From these data Hu et al. (2012a) obtained an αhet of 72 mJ/m2 for MHA and 81 mJ/m2 for MUA, both of which were substantially smaller than previous estimates of 109 mJ/m2 for α in the case of (homogeneous) calcite nucleation in bulk solution (Sohnel and Mullin 1978; Sohnel 1982). In the middle of the concentration range of the study, these differences in interfacial energy corresponded to free energy barriers for nucleation on MHA, MUA and in bulk solution of 19 kBT, 27 kBT and 105 kBT, respectively. All other factors being equal, these differences alone would correspond to relative nucleation rates JMHA:JMUA:Jsol of 1:3.4×10−4:4.5×10−38. The results of these studies showed that calcite nucleation on canonical carboxyl-terminated SAMs proceeded as expected from CNT and that both the enhancement of nucleation on the SAMs relative to bulk solution and the advantage of the SAM with even parity over that with odd parity could be explained in purely classical terms through differences in interfacial energy. This conclusion in no way pre-supposes whether the fundamental units that comprise the nucleus are individual ions or pre-nucleation clusters, because, as Equation (11) indicates, the form of the rate equations is the same in either case and, in fact, for the range of supersaturations used in this study, the value of C never exceeded 10% of σ.
In a subsequent study, Hamm et al. (2013) performed the same experiments for a series of SAMs with distinct headgroup chemistries including COOH, SH and PO4. As in the study of Hu et al. (2012a), the data were described well by the classical expression, despite the fact that the supersaturations used in this study straddled the solubility limit of ACC as given in the literature. (Though as discussed above, there is good evidence that this limit is substantially lower than previously reported.) Figure 18A, which shows the dependence of ΔGc on supersaturation for each type of surface in comparison to the barrier to homogeneous nucleation of a calcite rhomb estimated from the literature value for α in pure solution, demonstrates that: 1) the thermodynamic barrier to nucleation onto the SAM surfaces is significantly lower than estimates for homogeneous nucleation; and 2) substrate functional group chemistry and alkanethiol chain length have significant effects on the barrier. For example, comparisons of substrates with the same chain length but different head groups show that ΔGc is lower for C16-COOH than C16-SH and that C11-SH gives a value that is lower than for C11-PO4. The influence of chain length is seen by the lower ΔGc value for C16-SH as compared to C11-SH.
Extraction of the interfacial energy from the barriers showed that αhet ranges from 81 to 95 mJ/m2 for these SAMs, which is considerably less than the bulk value of 109 mJ/m2. Considering the expression for αhet in terms of the individual interfacial energies given by Equation (10), Hamm et al. (2013), reasoned that the crystal-liquid interfacial energy should be nearly the same for all films, as should the SAM-liquid interfacial energy given the pkA of the films and the pH of the experiments. The only term that should differ significantly from one film to the other is the interfacial energy between the SAM and the crystal αcs with smaller values leading to lower barriers. Moreover, αcs should be largely controlled by the binding free energy between the crystal and the SAM. Using dynamic force spectroscopy (DFS) with gold-coated AFM probes, Hamm et al. (2013) measured the free energy of binding between the SAMs and the surfaces of calcite single crystals. The results, shown in Figure 18B, reveal a direct, linear correlation between these two quantities. A large crystal-SAM binding free energy leads to a small crystal-SAM interfacial energy, resulting in a small barrier to nucleation.
Using the same in situ optical technique, Giuffre et al. (2013) looked at nucleation on polysaccharide films deposited by electrodeposition onto gold substrates. They considered five different films including de-N-sulfated Heparin containing amine, acetylamine, carboxyl, and sulfate groups; LG and HG sodium alginates containing carboxyl groups with different proportions of guluronic acid monomer content, hyaluronate containing acetylamine and carboxyl groups, and chitosan containing amine and acetylamine groups. As in the studies of Hu et al. (2012a) and Hamm et al. (2013), the nucleation rates were found to vary with supersaturation according to the dependence expected classically and given by Equation (12). Extracting the barrier, Giuffre et al. (2013) found a systematic variation with polysaccharide charge density, with the most negatively charged film giving the highest barrier (Fig. 19A). Moreover, when the data from Hamm et al. were put on the same plot, they found this correlation extended smoothly from the polysaccharide films to the SAMs. Considering, once again, the terms in Equation (10), Giuffre et al. (2013) reasoned that, as in the study of Hamm et al. (2013), the magnitude of the crystal-liquid interfacial energy was the same in all cases. However, because the polysaccharides have very different charge densities, the film-liquid interfacial energy might be very different for each polysaccharide. Taking the film-air interfacial energy as a surrogate for the film-liquid interfacial energy—with larger values of one corresponding to smaller values of the other—they found that, indeed, the nucleation barrier scaled linearly with the film-air interfacial energy as expected (Fig. 19B).
There are three major conclusions of these studies of calcite nucleation on organic films: First, they demonstrate the dependence of nucleation rate on supersaturation is in accordance with the predictions of CNT without consideration of “non-classical” behavior. Second, the film-to-film differences are controlled by the effective interfacial energy αhet, also as expected from CNT. Third, both the substrate-liquid and substrate-crystal interfacial energies can be manipulated to alter αhet in a systematic manner.
IMPLICATIONS OF NUCLEATION INFORMATION ON GEOLOGIC CO2 SEQUESTRATION
Nucleation information is pertinent to both mineral trapping in geologic carbon sequestration (GCS) reservoirs, and geochemical processes occurring in caprock formations, especially changes in porosity and flow due to precipitation. For GCS applications requiring effective mineral trapping of carbonate, it is essential to understand the factors affecting both the volume and specific location of carbonate produced from exposure of porous media to carbonate rich brines. If nucleation occurs preferentially in the throats of pores near an injection site, then brine infiltration of the reservoir may be restricted by rapidly decreasing permeability, limiting the effectiveness of the effort (Armstrong and Ajo-Franklin 2011). On the other hand, if nucleation is limited, leaking of the emplaced fluid into caprock is a concern. In a typical scenario, the degree of saturation increases after brine implacement, due to the dissolution by the brine of minerals having Ca2+, Mg2+ and Fe2+. Once saturation with respect to a crystalline phase is exceeded, the nucleation rate is then determined in part by the interfacial energy between the precipitate and the substrate mineral. A low interfacial energy supports easier nucleation, and hence certain minerals are more likely to support early and high-density nucleation. However at very high saturations, for example when the reservoir fluid already contains high cation concentrations, homogeneous nucleation becomes important, allowing nucleation to become significant within the fluid. Nuclei formed in this way may travel further into the reservoir, leading to reduced porosity and flow, depending on permeability. We are thus concerned about where precipitation occurs and to what extent, and require the ability to model the process from knowledge of the fluid composition, types of minerals in the reservoir, and the likely hydrological networks in the reservoir. Basic information supplied by GISAXS, optical and AFM nucleation studies includes: a) estimation of precipitate/mineral surface interfacial energies, b) locations for high nucleation density and topology of growth following nucleation, and c) effects of solution composition on nucleation and growth rates. In actual GCS field situations, these considerations are all likely to have influence, perhaps strongly coupled to local mineralogical, biological, and geochemical characteristics in the reservoir. Hence they must be considered together to effectively understand the precipitation process on all appropriate length scales.
Precipitate/mineral interfacial energies
Significant differences in interfacial energies can dictate where nucleation is likely to first occur, and will affect the density of nuclei. Both factors influence further growth (trapping) of carbonate precipitates. The observations for quartz indicate that it supports nucleation at ambient conditions, but comparisons with other phases, though in progress, are not yet available. With increasing temperature, retrograde solubility of CaCO3 will increase saturation, so nucleation and growth are expected to be robust in reservoirs dominated by quartz, e.g. sandstones. However the effect of pressure and temperature on interfacial energies is not yet known, and requires investigation. Another factor is the wettability of mineral surfaces, which are known to become more hydrophobic after contact with scCO2 (Chiquet et al. 2007; Espinoza and Santamarina 2010). Additional hydrophobicity will alter the thermodynamics of nucleation by affecting the difference between the αls and αcs terms in the nucleation rate equation (see Eqns. 9 and 10). αls will increase while a change in αcs, ostensibly due to lattice mismatch between substrate and nucleating crystal, is difficult to predict. Lattice mismatch strain assumes coherency between precipitate and substrate, and coherency may be reduced in a hydrophobic interface. However, assuming nil change in αcs, the net effect would be a lowering of the effective interface energy for nucleation, αhet, increasing the driving force for heterogeneous nucleation. It is important to note that both precipitate/substrate interfacial energies and relevant solution/substrate interfacial energies are not yet known for most of the cases expected in GCS. This affects not only modeling of nucleation processes, but also determination of thermodynamic and kinetic factors affecting dissolution (Lasaga and Blum 1986).
Location and topology of high-density nucleation and subsequent growth
There is evidence from simulations (Hedges and Whitelam 2012) that confined pore spaces can enhance nucleation rates, leading to rapid cutoff of pore flow and reduction in permeability after nucleation. This may be further enhanced if the nucleation density is favored by interfacial energy. However the combined effect of dimensionality, roughness of surfaces, and interfacial energies on nucleation in pore throats has not been treated. Particular nucleation environments may favor increased precipitate growth, or conversely, lead to much reduced growth, given varied topological factors. For example, nucleation in narrow pore throats may produce limited growth due to rapid reduction in solution flow, while nucleation in wide pores may lead to extensive growth. Growth may also be affected by the shape of nuclei and the phase nucleated, e.g. if it is an amorphous carbonate phase, or if it has variable composition, such as might be expected in the presence of significant Mg levels.
A further example of how nucleation may be affected by interface and mineral types is afforded by the work of Shao et al. (2010), which reported the evolution of nanoscale amorphous silica precipitation on phlogopite surfaces reacted with scCO2 in 1 M NaCl under 102 atm of CO2 at 95 °C, conditions relevant to GCS. The actual nanophase formed depends on temperature, pressure and salinity, but amorphous silica is overall dominant. Images depicting a range of reaction times are shown in Figure 20, where parts A, B, C, E and F refer to 5, 8, 22, 43 and 159 h reaction times, and D is an enlargement from a small area of B. All are AFM images shown here in height mode, with insets in the lower left corners collected in phase mode, which can be used to identify the nanoparticle phase. Insets in right upper corners indicate schematic particle positions, and the height cross sections below the images correspond to the white dotted lines in height images. Arrows indicate the position of particles on the edge of dissolution pits. The dotted curved line in image D indicates the evolution of a single dissolution pit. An interesting aspect is that the nanoparticles prefer to form at the edges of dissolution pits rather than on flat surfaces. If the particles grow continuously, they may close over the dissolution pit, locally passivating the reaction. Dissolution pits might be considered analogous to pore structures in GCS sites, and must be considered when there are simultaneous dissolution and nucleation processes. As the location of nucleation is important owing to the impacts on CO2 permeability changes in porous media, the mineralogical constituents of caprocks and formation rocks in GCS sites are also crucial to understand. Mineralogical distributions in GCS sites are quite diverse and often their heterogeneities create more complex interactions with CO2 and brines (Landrot et al. 2012). Consequently, nucleation of secondary precipitates can form with totally different kinetics and extent as a function of exposed reactive mineral (for examples, quartz, mica, or feldspar). These differences can then further affect pore-scale microstructure and reactive fluid transport. Hence, a better understanding of the location, morphology, and their mineralogy of the pre-existing mineral substrates in GCS would be needed.
Effects of solution composition on nucleation and growth rates
Solution composition is expected to mitigate nucleation in several ways apart from degree of supersaturation and effects associated with scCO2 exposure. An important characteristic environmental factor is high salinity (Shao et al. 2011a), which can cause an increase in αls, as shown by Mizele et al. (1985) in amorphous silica surface in 1 M NaCl solutions. This factor is relevant in especially relevant in sedimentary reservoir rocks where salt domes are coexistant, and which can have salinities up to 2 M NaCl in the pore-space water (Kharaka et al. 2006). Another important aspect is the presence of organic species that may contaminate interfaces, affect near interface water structure, or bind ions which result in slowed growth, e.g., by chelation (Yang et al. 2011, 2013). Because supercritical CO2 is well known as a good organic solvent, the injection of supercritical CO2 can also increase concentrations of organic species in brine (Shao et al. 2011b). Recent work on peptoid effects on calcite growth show that some of these organic species can enhance growth at low concentrations, but retard growth as concentration increases (Chen et al. 2011). Natural organics may have similar effects (Ray et al. 2012), although the more important considerations may be increased hydrophobicity of pore space surfaces by nonpolar hydrocarbons, or the complete blockage of hydrological channels by larger concentrations of organics such as remaining after natural oil recovery (Butkus and Grasso 1998). These considerations require knowledge of both fluid composition and the response of mineral surfaces and confined spaces to dissolved and particulate organic matter.
J.J.D., G.A.W, Y.-S.J. were supported as part of the Center for Nanoscale Control of Geologic CO2, an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Award Number DE-AC02- 05CH11231. A.F.-M. was supported as part of Le Centre national de la recherche scientifique (CNRS) at Institut des Sciences de la Terre, Universite Joseph-Fourier Grenoble I and CNRS.