Role of CaCO3° Neutral Pair in Calcium Carbonate Crystallization.

The molecular structure of the units that get incorporated into the nuclei of the crystalline phase and sustain their growth is a fundamental issue in the pathway from a supersaturated solution to the formation of crystals. Using a fluorescent dye we have recorded the variation of the pH value in time along a gel where CaCl2 and NaHCO3 counter-diffuse to crystallize CaCO3. The same pH-space-time distribution maps were also computationally obtained using a chemical speciation code (phreeqc). Using data arising from this model we investigated the space-time evolution of the activity of the single species (ions and ion pairs) involved in the crystallization process. Our combined results suggest that, whatever the pathway from solution to crystals, the neutral pair CaCO3° is a key species in the CaCO3 precipitation system.

The correct understanding of the nature (ionic or molecular) and chemical composition of the growth units building the crystals is important for a wide variety of crystallization processes ranging from biological mineralization to materials synthesis. Obtaining this knowledge from crystallization experiments is rather difficult since it is mandatory to find experimental evidence of these primary species. In this work we face such a challenge using the crystallization of calcium carbonate (CaCO3) as a model system. This case has been chosen because (i) the reactions between calcium ions and inorganic carbon may form many solution species that could either be the crystal building units or control their concentration; (ii) thermodynamic and kinetic data on the precipitation of CaCO3 is widely available; (iii) CaCO3 has important implications in industrial, technological, environmental, and biological processes and its nucleation and growth mechanisms are nowadays a matter of intense debate;[1−10] (iv) the precipitation of CaCO3 is a pH controlled process and can be monitored by pH probes.

This dependence on pH is the first key point for our work, because it opens a window to the otherwise elusive problem of identifying the solution species. The evolution of pH over time has certainly been recorded many times in CaCO3 precipitation studies. In commonly used precipitation methods, such as titration and direct mixing of a calcium ion solution with either a carbonated solution or CO2 gas, only a time series of volume-averaged pH values can be recorded. The information provided by these volume-averaged measurements does not inform about the space distribution of soluble species playing a role during the crystallization process.

Thus, the second key point of our work is using the counterdiffusion (CD) method instead of direct mixing. In CD experiments the distribution of aqueous species is unfolded along a one-dimensional space.[11−13] This allows the recording of thousands of pH values in different points and at different times to build a spatiotemporal map of the pH distribution. In CD crystallization of calcium carbonate two solutions, one containing hydrogen carbonate ions and the other calcium ions are allowed to counterdiffuse through a gel matrix or high-viscosity fluid.[14−16] The diffusion of the different chemical species leads to the development of species concentration gradients that evolve over time. The CD method has been traditionally used to optimize the size and quality of crystals for X-ray diffraction experiments (either from small molecules or macromolecules),[17,18] but their capabilities for space–time unfolding of experimentally measured values has seldom been explored.

pH is one of these unfolded values, and the one we use in our work. The last key ingredient of our study is the development of an experimental method to measure in situ the pH distribution in a crystallization experiment by using a pH sensitive fluorescent dye embedded in the gel. Using this pH imaging capability to track the evolution of a CaCO3 crystallization experiment by CD, we have produced space- and time-resolved maps of pH. These maps are made of thousands of measurements so they can be confidently compared with the output of transport/speciation simulations to derive conclusions and support hypotheses on the distribution of species in solution. Using this combined method we have derived the activity of all inorganic aqueous species revealing the key role of the neutral pair CaCO3° in the crystallization of CaCO3.

CD experiments have been carried out within U-tubes (Figure a). The two vertical branches were filled respectively with 0.5 M solutions of the two reagents, sodium hydrogen carbonate (NaHCO3) (right arm) and calcium chloride (CaCl2·2H2O) (left arm). The horizontal segment of the U-tube was filled with 3% (w/w) agarose gel containing the pH-fluorescent dye (8-hydroxypyrene-1,3,6-trisulfonic acid trisodium salt). The time evolution of the pH has been obtained from time-lapse series pictures following the procedure shown in Figure . Speciation, mass transport, and precipitation have been simulated using the Phreeqc code (v 3.1.4).[19−22] Species activities have been calculated using ion association with thermodynamic data from the Phreeqc.dat database. Mass transport was simulated using the multicomponent diffusion algorithm in Phreeqc. In counterdiffusion experiments, the ionic activity product increases slowly. In these conditions the first polymorph produced is always calcite, as confirmed in a previously reported experiment.[16] Therefore, only this crystalline phase was considered in the model. Agarose gel is not included in the solution chemistry, because although it is physically embedded into the crystals during their growth, it has no influence on chemical speciation. All details of the experiment (SI1) and the Phreeqc input file used in the simulation (SI2) are available in the Supporting Information.

Figure 1

Experimental procedure for pH measurements: (a) U-tube setup used for CaCO3 crystallization by CD method. (b) Optical setup: a narrow bandpass filter (pink) is used to select a wavelength of 450 nm (blue/purple in the figure) from the xenon lamp illumination. This radiation excites green fluorescence (green in the figure) in the pH sensitive ink. This radiation is selectively collected through a bandpass filter (brown in the figure) and imaged on a CCD. (c) Plot showing the wavelength filtering previously described; absorbance and emission curves are included for both the protonated and deprotonated ink along with their spectral position with respect to filters. (d) Raw fluorescence image as collected by the CCD. These images are quantitatively analyzed using a calibration curve (e) to produce the final pH profile (f). Plots (c) and (e) are available in the Supporting Information (SI3 and SI4).

Figure (left) displays a map showing the variation of pH in time (vertical axis) and space (horizontal axis) recorded during the crystallization experiment. The initial pH of the horizontal agarose-filled segment is 6.5 (red-orange in color scale). Once diffusion of ions starts, the pH of the right part of the agarose gel increases upon arrival of HCO3–/CO32– ions (color shifts quickly from red to dark blue, see label “a” in figure). At the left part of the gel tube the diffusion of Ca2+ in the opposite direction keeps the pH acidic, but progressively, the arriving HCO3–/CO32– ions start to increase the local pH. On the right part, the increment of pH creates an alkaline front that penetrates across the tube following a classical diffusion profile. Close to the center of the tube, a decrease of pH produced by chemical reactions between the incoming ions is observed (label “b”). Dashed lines with white arrows mark the evolution in time of the alkaline front at pH ∼8.

Figure 2

Experimental (left) and calculated (right) maps of pH–space–time. The color scale for pH is shown on top of the left map. The circle marks the position and the time where first nucleation occurs. Dashed line with white arrows marks the evolution in time of the alkaline front at pH ∼ 8. The horizontal dotted lines indicate the time for which the species activity is plotted in Figure 3.

Figure 3

Evolution in time of pH (black lines) and activities of CaHCO3+ (blue lines) and CaCO3° (orange lines) along the gel tube. For the sake of clarity only plots after 1 h (dotted lines), 8 h (dashed lines), and 14 h (continuous lines) are shown. The full time evolution is illustrated as a movie (SI6) in the Supporting Information. The thin horizontal dotted lines in the pH plot (top) represent the simulation input pH values in both reservoirs. The thin horizontal dotted line intercepting the log activity axis (bottom plot) at −5.26 represent the CaCO3° critical activity value for precipitation.

The location of the first forming crystals was observed by optical microscopy time-lapse images, but also from the pH versus space/time map, because as soon as the crystals reached a critical size of a few hundred nanometers they started to scatter light that, when recorded at the detector, shows as an apparent local decrease of pH in the map (label “c”). The circle in Figure (left) marks the position and the time where the first crystals are observed.

This complex experimental pH distribution in space and time is notably well reproduced by the simulation output (Figure , right). Apart from some minor differences like the pH minimum moving from left to right (label “d”) that is less prominent but present in the experiment or the apparent pH decrease at positions where crystals are growing (label “c”), the only difference between the experiment and the model is the time (vertical) scale. The multicomponent diffusion algorithm[21] implemented in Phreeqc was used in our model because the diffusion coefficients (unknown for most solution species) are computed ab initio and because ion transport is corrected by ensuring electrical neutrality over the solution volume. This ab initio computation of diffusion coefficients is based on tracer diffusion that is too fast for real solution diffusing through a gel. This overestimation of the diffusion coefficients produces a compression of the time dimension in the model output. The model assumes no nucleation barrier: calcium carbonate starts to precipitate as soon as it is supersaturated. This strategy was chosen to avoid the use of free parameters of kinetic nature in the model, and also contributes to the earlier nucleation events in the model. The fact that this complex pH landscape is notably reproduced by an speciation simulation with simple mass transport and chemical equilibrium equations suggests that the pH evolution and the onset of calcium carbonate precipitation is controlled by the distribution of species in solution. In order to explain these maps and to understand the precipitation behavior of CaCO3, one needs to know the spatial-temporal distribution of all solution species and their mutual interactions. The output of the simulation records each second at 100 points along the tube the concentration and activity of the following species: H+, OH–, Cl–, Na+, Ca2+, CO2, HCO3–, CO32–, CaHCO3+, CaCO3°, CaOH+, NaCO3–, NaHCO3, and NaOH. Maps of these species are available in the Supporting Information (SI5). Among them, CaCO3° and CaHCO3+ (along with pH) were found the most relevant for the problem at hand. Ca(HCO3)2, that forms at lower pHs (<6),[20] may form at concentrations much smaller than the major species and play no role in the overall speciation process. More complex species like polymers or clusters of the relevant pairs have not been considered in the calculations because their formation is not described by equilibrium chemical reactions, which is what the code computes.

Figure shows profiles of these species at times well before nucleation (1 h), right before nucleation (8 h), and after nucleation (14 h). These times are marked as white horizontal dotted lines in Figure (right). The full time evolution is illustrated as a movie (SI6) in the Supporting Information.

One hour after starting the experiment, the pH profile is dominated by the presence of Ca2+ and HCO3– because the activity of each of them is negligible in all places where the activity of the other is high. The activity of CO32– at that time is very low compared to that of HCO3–. The minimum of pH close to the cationic reservoir (left in the plot) is due to the hydrolysis of Ca2+, which drives the pH value close to 5.8 (lower dotted horizontal line) according to the equilibrium[23]Close to the anionic reservoir, the diffusion of HCO3– raises the initial pH of the gel (6.5) up to 8.3 (the upper dotted line).

As the counterdiffusion process continues diffusing Ca2+ to the right and CO32–/HCO3– to the left, new species in solution start forming (Figure , bottom) according to equilibria[24]

As shown in the plot (bottom of Figure ) as well as in movie SI6, the activities of both pairs increase continuously. The first solid CaCO3 is formed when the activity of CaCO3° reaches a critical value (thin horizontal dotted line in the bottom plot). After this event, the CaCO3° activity, in the region where crystals are nucleating and growing (marked by the vertical blue rectangle), remains constant at the critical value level (the flat top of the orange curve), indicating that this species is being produced and consumed at the same rate. The maximum activity of CaHCO3+ is to the left of the crystallization region and keeps increasing after solid CaCO3 is formed instead of remaining constant (Figure ). In the same region, after the first precipitation, Ca2+ and CO32– activities diverge over time while the ionic product is kept constant, as expected (Figure ). The evolution of the activities of both pairs in a scenario rich in Ca2+ and HCO3– ions indicates that neutral CaCO3° is consumed in the process of calcium carbonate crystallization while charged CaHCO3+ acts as an alternative reservoir contributing ions to form CaCO3° in addition to the Ca2+ and CO32– supplied by counterdiffusion. Note that the overall equilibrium 5, that delivers neutral pairs on demand, is a combination of equilibria , 3, and 4. The released protons yield the pH minimum that can be observed at 14 h (Figure , solid black line) at the crystallization region. This release of protons also keeps increasing the concentration of CaHCO3+ according to[24]

Figure 4

Time evolution of activity for the ions Ca2+, CO32– and the pairs CaHCO3+, CaCO3° (bottom). The ionic activity product a(Ca2+) a(CO32–) is shown in the top panel.

The relevance of ion pairs in the dissolution and precipitation processes of CaCO3 was already theoretically suggested. Plummer and Busenberg[24] showed that the experimental values of solubility equilibrium constants of CaCO3 polymorphs are consistent with an aqueous model including CaCO3° and CaHCO3+. Gómez-Morales et al.[25] found that, by maintaining the ionic activity product constant, the maximum nucleation rate (minimum induction time) occurred in solutions with initial stoichiometric Ca2+/CO32– concentrations, whereas maximum growth rate occurred in the experiments with excess of Ca2+; the activities of CaCO3° in solution and at the crystal surface respectively were found to be maximized in these conditions, so they proposed that the activities of CaCO3° in solution and on the crystal surface are rate limiting for nucleation and growth, respectively. Ruiz-Agudo et al.[26] also suggested that the CaCO3° should play an important role in crystal growth of calcium carbonate on the basis of its smaller degree of hydration that would facilitate the incorporation in the structure by making it faster. Additionally, it has been experimentally shown that the Ca2+/CO32– molar ratio in solution, affecting the concentrations of CaCO3° and CaHCO3+, strongly influences the growth rate of calcite.[27,28] Our results indicate that among the two pairs only CaCO3° is consumed in equilibrium during the precipitation process. Whether CaCO3° or a larger neutral cluster formed by aggregation of these discrete species is the relevant growth unit cannot be derived from our findings, since clustering is not considered in the calculations.

Our model, based on equilibrium calculations, logically implies that the ionic activity product of Ca2+ and CO32– must be constant when the activity of the neutral pair is constant, and this is what we observed (Figure ). Therefore, we cannot conclusively determine whether the calcium/carbonate ions or the neutral pair are the units whose aggregations feeds calcite nucleation and growth, but both the previous reports on this subject[25,29] and the evolution of the Ca2+ and CO32– concentration after nucleation do support the hypothesis of CaCO3° being the most likely candidate. The activities of Ca2+ and CO32– diverge after the first precipitate forms. At the first nucleation, their activities are very similar and both increase with time, but after this event, the activity of Ca2+ keeps increasing while that of CO32– starts to decrease quickly. The ratio Ca2+/CO32 changes from around 1 a few hours before nucleation to 100 a few hours after nucleation and more than 1000 1 day after.

In this strongly nonstoichiometric situation, an expression for supersaturation as a function of the activity of the CaCO3° instead of the lattice ions could be more realistic to define the rate equations in calcium carbonate precipitation.

From a chemical point of view, calcium binding to HCO3– through monodentate interaction or to CO32 through a bidentate interaction modifies the coordination of the hydrated calcium. The strongest bidentate interaction of carbonate explains why this anion, rather than bicarbonate, is the one entering the elemental building brick even though bicarbonate is the prevailing anion in solution.[30]

Besides the interest of the plausible role of CaCO3° as the calcite growth unit, these results can also shed light on the onset of CaCO3 crystallization process (nucleation). In the classical view, the nucleation is a stochastic process in which fluctuations induce the formation of clusters that evolve by accretion of discrete units. In the case of calcium carbonate, our results suggest that instead of calcium and carbonate ions, these units can be CaCO3° following the reaction

Gebauer et al.,[3] by using titration and ultracentrifugation techniques, demonstrated the formation of (CaCO3) prenucleation clusters. According to their model the nucleation follows a nonclassical pathway and CaCO3 forms by aggregation of these clusters. Our results suggest the formation of these neutral clusters could take place by aggregation of CaCO3° as an intermediate step toward the formation of solid CaCO3.

Wallace et al.,[5] by using molecular dynamics simulations, reported that previously formed clusters adopt chain, ring, and branched structures. They predict the formation of nanoscale droplets in concentrated solutions by liquid–liquid phase separation.[4] These droplets coalesce and solidify to form the solid phase, amorphous calcium carbonate.

Neither clustering processes nor liquid–liquid separation are considered in our calculations. However, the identification of CaCO3° as a key species involved in CaCO3 precipitation may reconcile both views, the nonclassical where large neutral clusters (CaCO3) form and the classical where both solid nuclei and CaCO3° in solution do form from the same ions.

The fact that (CaCO3) or CaCO3° can be the key species in CaCO3 crystallization has important implications in the understanding of precipitation phenomena. According to eq , and considering x = 1, the supersaturation β of the solution with respect to the most stable polymorph, calcite, should be expressed aswith Ksp (calcite) = 10–8.48,,[24] instead of the usual expression in terms of the ionic activity product (IAP) and solubility product (Ksp). Letting β = 1 in eq , the equilibrium CaCO3° activity can be computed to be 10–5.[26] This is the critical activity value that should be overtaken to produce solid CaCO3 (horizontal dotted line in Figure ). In macroscopic equilibrium experiments both approaches are obviously equivalent, but the finding of (CaCO3) as a relevant species for CaCO3 crystallization is expected to have a deeper impact in situations where the interactions at the molecular scale are relevant as in molecular dynamics studies; the distribution of solution species is unsteady, as in counterdiffusion experiments, where the calcium to carbonate ratio changes over time and along space; and the ionic strength and type of electrolyte added change or additives that can modify the speciation equilibria are present.

In these situations a critical reexamination considering the neutral cluster (CaCO3) species as the primary building unit for CaCO3 crystallization can bring new insights into unexpected or unexplained precipitation behaviors and open new possibilities for controlling the nucleation and growth of calcium carbonate.