Divalent heavy metals and uranyl cations incorporated in calcite change its dissolution process

Due to the high capacity of impurities in its structure, calcite is regarded as one of the most attractive minerals to trap heavy metals (HMs) and radionuclides via substitution during coprecipitation/crystal growth. As a high-reactivity mineral, calcite may release HMs via dissolution. However, the influence of the incorporated HMs and radionuclides in calcite on its dissolution is unclear. Herein, we reported the dissolution behavior of the synthesized calcite incorporated with cadmium (Cd), cobalt (Co), nickel (Ni), zinc (Zn), and uranium (U). Our findings indicated that the HMs and U in calcite could significantly change the dissolution process of calcite. The results demonstrated that the incorporated HMs and U had both inhibiting and enhancing effects on the solubility of calcite, depending on the type of metals and their content. Furthermore, secondary minerals such as smithsonite (ZnCO3), Co-poor aragonite, and U-rich calcite precipitated during dissolution. Thus, the incorporation of metals into calcite can control the behavior of HMs/uranium, calcite, and even carbon dioxide.


Results
characterization of calcite incorporated with impurity metals. With the addition of HMs, the XRD patterns of all the Cd and Zn containing products were the same as the pure calcite reference pattern, indicating that no detectable secondary crystalline phases were present (Fig. S1a,b, Table 1). However, traces of aragonite were found in sample Co-10 ( Fig. S1c), while vaterite was present in U-08 and U-10 ( Fig. S1e). Meanwhile, both aragonite and vaterite occurred in Ni-0.4 and Ni-01 (Fig. S1d). To avoid the impact of aragonite and vaterite, we used samples without detectable secondary phases for further experiments.
As shown in Fig. 1a, pure calcite showed a typical rhombohedral morphology as a euhedral calcite crystal. With the incorporation of metals, the morphology changed to aggregates of semi-euhedral or anhedral (dumbbell) phases with small sizes (Fig. 1b-l). This trend is supported by the BET surface area data, which are positively correlated with the molar fraction of metals, with R 2 values of 0.65 (Ni), 0.89 (Co), 0.93 (U), and 0.96 (Cd, Zn) ( Table 1, Fig. S2). Usually, the incorporation of impurities will decrease the sizes of crystals 47,48 due to the inhibition of the crystal growth. Figure S3 showed the spatial distribution of metals in typical polished HM-calcite. Both of the line scan and elemental mapping results showed the presence of Cd/Zn-rich cores in Cd/Zn-calcite ( Fig. S3a-h). However, the Release of metals during the dissolution of calcite. Once calcite contacted the solution, dissolution occurred immediately, especially in an acidic solution. For example, when the initial solution pH ranged from 1.0 to 8.9, the proportion of pure calcite that dissolved within 20 min accounted for 83.8-97.4% of the total mass dissolved within 120 min (Fig. S4). However, the dissolution process proceeded until equilibrium was reached at approximately 2400 h with an initial pH of 5.0. (Fig. 2) 49 , the closed CO 2 in each tube was no less than 0.046 mmol, which can dissolve into the alkaline solution. However, this tiny CO 2 amount was not enough to increase the DIC up to 1.03 mmol at equilibrium. Therefore, the effect of environmental CO 2 on the experiment was not significant.
With the incorporation of HMs and uranium, the quantity of dissolved calcium changed, even at the early stage (first two hours) of dissolution (Table S2). After two hours of dissolution, the dissolved Ca concentrations were 0.13-0.15 mM and 0.15-0.07 mM for all the Cd-and Ni-incorporated calcite (Cd/Ni-calcite), which were less than the dissolved Ca concentration in the control (0.16 mM) (Fig. 2a,g and Table S2). On the contrary, more Ca was dissolved in Zn-calcite than in Cd/Ni-calcite (0.18-0.23 mM) ( Fig. 2c and Table S2). However, Co and U showed inhibition effects at low contents (0.14 and 0.15 mM for Co-2 and U-0.5) and an enhancement at a high content (0.17-0.22 mM for Co-06/08 and 0.16-0.31 mM for U-01/02/04) (Fig. 2e,i and Table S2). When calibrated with the BET surface area, the dissolution rates within the first 2 h for all heavy metal-incorporated calcites (2.90-31.70 μmol/m 2 ·h) were less than those for the pure calcite (37.36 μmol/m 2 ·h) (Fig. S6). This suggested that the dissolution rate was controlled both by the mineral surface area and the fraction of HMs. After 2880 h,  As expected, all the HMs and uranium released into the solution together with calcium, and the amounts released increased with time until reaching equilibrium, excepted for Co-calcite, Ni-0.04, and U-0.5/01/02, which did not reach equilibrium, even after 2880 h (Fig. 2f,h,j). Note that Cd-08 and Cd-10 displayed peak values of Cd at 960 h and 672 h (Fig. 2b), which is similar to the phenomenon observed for Cu-calcite and Mn-calcite 45 (Fig. 2j). The proportions of released Cd, Zn, Co, Ni and U were 0.01-0.02‰, 0.06-0.09‰, 1.11-4.03‰, 0.31-1.22‰ and 3.29-9.96% of the total loaded metals, respectively.
As demonstrated in Fig. S7, the molar ratios of total dissolved metals were 0.09-0.18‰ for Cd/Ca, 0.43-1.75‰ for Zn/Ca and 1.48-2.35% for Co/Ca, which were much less than those in the solids, except for Co-02, suggesting the nonstoichiometric release of Cd, Zn, Co and Ca. Contrarily, the molar ratios of released Ni/Ca and U/Ca were 1.17-3.93‰ and 7.23-14.63%, which are both higher than those in the solids (0.5-2.8‰ and 0.56-2.64%). These data demonstrated that all the studied HMs and U showed nonstoichiometric release behaviors during the dissolution of calcite. In particular, Ni and U tend to passively release into the solution, while Cd, Zn, and Co prefer to prevent dissolution. Figure 3 showed the total dissolved metals (Ca + M) and the corresponding solubility at 2880 h. Since the content of impurities is low in calcite, the addition of released HMs did not change the relationship between the total dissolved metals and the content of impurities (Fig. 3a) (Fig. 3b). As is known, the solubility of a material at equilibrium is independent of its surface area. Therefore, the relationship between the solubility of calcite and the molar fraction of HMs was fitted using linear equations, which provided an R 2 value of 0.96, 0.79, 0.69, 0.99 and 0.98 for Cd, Zn, Co, Ni, and U, respectively (Table S3). Except for Ni, the solubility of calcite is positively correlated with the molar fractions of HMs, which are the same as those of Mg-calcite [41][42][43][44] , Cu-calcite, and Mn-calcite 45 .
K sp and IAP are generally used to characterize the solubility of a material and are calculated using the ion concentration or activity. As shown in Fig. 3c, the data of all the IAP values were calculated using the ion activity of Ca 2+ and CO 2− 3 according to Eq. (8). In addition to U, the graphs of Cd, Zn, Co, and Ni showed a tendency similar to that shown in Fig. 3a and b. When we tried to calculate the IAP using Eq. (9), we obtained results . The data of Cu and Mn were taken from Zhang et al. 45 , and the data of Mg were obtained from Davis et al. 44  www.nature.com/scientificreports/ similar to those shown in Fig. 3c. Since the ion activity cannot be measured directly, we recommend using the total concentration of metals to express the solubility of impure calcite. This method should be more convenient in the field.

Morphology change and precipitation of secondary minerals during the dissolution of calcite.
After dissolution, typical etch pits were present on the surfaces of pure calcite (Figs. 4a, S8a-f) and Cd/Ni-calcite (Figs. 4b,g, S8g-l). Unexpectedly, some secondary minerals precipitated on the surface of Zn/ Co/U-calcite (Figs. 4c-f,h,i, S9, S10). These secondary minerals were composed of lamelliform nanocrystals for Zn/U-calcite and prism nanocrystals for Co-calcite. Meanwhile, the secondary minerals formed earlier in the samples with a high content of impurities than in the one with a low content, i.e., new precipitates were first observed after 12 and 28 days for U-04 and U-02, respectively. As shown in Table 2 and Fig. S11, the EDS data demonstrated that the secondary minerals of Zn-10 and U-04 were rich in Zn and U, while Co-08 contained trace Co, which was distinguished with the calcite before dissolution. Furthermore, high resolution transmission electron microscopy (HRTEM) and selected area electron diffraction (SAED) were used to characterize the secondary minerals (Fig. 5). The HRTEM image of the secondary minerals of Zn-10 ( Fig. 5a) revealed the lattice fringes of the smithsonite (ZnCO 3 ) (104) plane (d = 2.77 Å) and (116) plane (d = 1.72 Å) (RRUFF 50 ), which was further validated by the SAED patterns (Fig. 5b). This observation agreed with the content of Zn being much higher than that of Ca, according to the EDS data. For the secondary mineral of Co-08, the HRTEM images show lattice spacings of 2.73 Å and 2.48 Å (Fig. 5c,e), which can be indexed to the (121) and (200)    www.nature.com/scientificreports/ peak at 205.4 cm −1 confirmed that aragonite was a secondary mineral in Co-08-120d (Fig. 6). Comparably, the HRTEM images of the secondary minerals of U-04 possessed lattice fringes with a d-spacing of 3.03 Å, which corresponded to the (104) reflection of calcite (RRUFF 52 ) (Fig. 5g), as confirmed by the SEAD pattern (Fig. 5h). Therefore, the secondary minerals observed in our experiment were Ca-poor smithsonite, U-rich calcite, and Co-poor aragonite.  [80][81][82][83][84][85][86][87][88][89][90][91] into the structure of calcite. In this study, the presence of aragonite and vaterite indicated the favoritism of these phases in the high concentration Co, Ni, and U system. The incorporated Co, Ni, and U in CaCO 3 could stabilize aragonite or vaterite phases, agreeing well with previous investigation 66,83 . Taking U as an example, a previous study reported that the U concentration in synthesized aragonite was higher than that in calcite, even though both of them grew in a solution containing the same concentration of uranium 90 . This phenomenon could be attributed  www.nature.com/scientificreports/ to the thermodynamic preference of U incorporation into aragonite 92 . Another reason was that aragonite has a higher surface area than calcite and can absorb more U on its surface. Aragonite was present only in Co-10, even though equal stoichiometric amounts of HMs were used for the synthesis of Cd-10, Zn-10, and Co-10. The exception of Co-10 demonstrated that the tolerance of the calcite structure for the impurity metals would be varied with its ionic radii, following the Goldschmidt's Rules. Traditionally, it is believed that the incorporated divalent cations would take the crystallographic site of Ca 2+53 . However, it is difficult to determine the coordination environment and, thus, the incorporation modes of the impurities until the development of extended X-ray absorption fine structure spectroscopy (EXAFS). According to the experimental EXAFS data and Ab initio molecular dynamics simulation, there are at least five models of incorporating impurities in calcite. (1) Isovalent substitution occurs for Ca 2+ with sixfold coordination, which is the same as the coordination of Ca 2+ in calcite, e.g., Ba 2+ , Cd 2+ , Co 2+ , Mn 2+ , Pb 2+ , Sr 2+ and Zn 2+64,65,69,93-95 . (2) Aliovalent substitution occurs for Ca 2+ with six-fold coordination, e.g., Nd 3+ , Sm 3+96 , Np 5+97 , U 4+98 and U 6+ (as UO 2+ 2 ) in natural calcite 84,87 . (3) Aliovalent substitution occurs for Ca 2+ with seven-fold coordination, e.g., Dy 3+ , Yb 3+96 , Am 3+99 and U 6+ (as UO 2+ 2 ) in synthesized calcite 80,81 . (4) 64 , nano hydrozincite (Zn 5 (CO 3 ) 2 (OH) 6 ) and sphalerite (ZnS) discovered in foraminiferal calcite shells 69 . Furthermore, the incorporation of monovalent cations, i.e., Na + and Li + , might be required to achieve a local charge balance in the calcite lattice 98,106 . In Zn-contaminated foraminiferal calcite shells, adsorption, isovalent substitution, and second nanophases coexisted with each other 69 . Coexisting multimechanisms are probably much more common in the natural system and high concentrations of impurities than in the other systems.

Discussion
The spatial distribution of trace elements in biogenic calcite might be homogeneous or heterogeneous, which relates to the temperature, growth rate, or organic matter content 69,[107][108][109][110] . Due to the overgrowth of crystals, the heterogeneous distribution of impurities has been discovered in natural single crystals of low-temperature precipitated calcite 7,111 . In laboratory-synthesized calcite, the distribution of trace elements is not reported very often, even though heterogeneous distribution sometimes occurrs 66,112 . However, the HM-calcites in our experiment showed both heterogeneous (Cd, Zn, U) and homogeneous (Co, Ni, U) distributions. The homogeneous distribution of Co in Co-08 differed from the result of Katsikopoulos et al. 66 , probably due to the visible difference in the Co fractions in the solid phases, i.e., a Co/Ca molar ratio of 0.077 in Co-08 versus 1.67 in the previously published data. Unexpectedly, some U-04 particles showed heterogeneous distributions of uranium, while other particles showed homogeneous distributions. This phenomenon indicated the complexity of the distribution of impurities in HM-calcite, which required further investigations.
The influence of incorporated impurities on the dissolution of calcite. Since the natural water chemistry, e.g., seawater chemistry, is complex, numerous investigations have demonstrated that the dissolution of calcite is affected by the water constituents, including the major, minor, and trace constituents, regardless of whether the constituents are inorganic or organic [36][37][38] . Based on the inhibition dissolution of natural calcite, e.g., limestone and marble, Svensson and Dreybrodt suggested that Ca 2+ ions are adsorbed to impurities at the surface of the natural mineral to form a complex, which acts as an inhibitor 113 . Subsequently, Eisenlohr et al. suggested that aluminosilicate nano-complexes could incorporate in the calcite matrix. During dissolution, these impurities will release from the calcite matrix and then will be adsorbed irreversibly at the reacting surface, where they act as inhibitors 114 .
Except for Mg 2+ , the influence of the impurities incorporated metals in solids on the dissolution of calcite is unknown. According to the dissolution of magnesian calcite, the Mg 2+ in calcite enhanced the mineral solubility through magnesium incorporation and inhibited calcite growth [41][42][43][44] . On the one hand, high magnesian calcite compositions exist within the spinodal gap and are therefore unstable. On the other hand, calcite growth is inhibited through either step-blocking by Mg 2+ adsorption and slow dehydration or enhanced the mineral solubility associated with Mg 2+ incorporation [41][42][43][44] . Contrarily, Harstad and Stipp reported that naturally present trace metals (Fe 2+ , Mg 2+ , Mn 2+ , and Sr 2+ ) inhibited the dissolution rate of calcite based on AFM observations 40 . Recently, we demonstrated the enhanced dissolution of calcite once Cu 2+ and Mn 2+45 were incorporated. These results contradicted each other with regards to the role of Mg 2+ and Mn 2+ play in calcite dissolution.
Firstly, the calcite used by Harstad and Stipp contained many impurity metals. Therefore, the enhancement effect contributed by Mg 2+ and Mn 2+ might be neutralized by other inhibitors, such as Fe 2+ and Sr 2+ . Secondly, the relationship between the impurity molar fractions was inconsistent with the solubility/dissolution rate, as shown in Figs. 3 and S6. It has been observed that the surface area-normalized dissolution rates increase nonlinearly with decreasing the initial crystal size 115 . However, the normalized dissolution rates of our impuritycontaining calcite with small crystal sizes were lower than that of the large pure calcite, suggesting the inhibition of the dissolution rate by the impurities. This inconsistency indicates that the impurity-containing calcite required a longer time than pure calcite to reach equilibrium, as described previously. According to Fig. 2, the pure calcite reached equilibrium at approximately 2880 h, while Cd-calcite, Ni-calcite, and U-calcite still have an upward trend, suggesting that Cd/Ni/U-calcite did not reach equilibrium at that time. Third, the dissolution rate obtained from reactor experiments performed on crushed crystals (i.e., mineral powders) differed from that measured locally at the surface of polished or pristine crystals 116 . This phenomenon is ascribed to the differences in the surface reactivity distribution at the two scales of observation and different sample preparation methods, which could change the distribution of reactive sites 116 . In our samples, the BET surface area increased with the increase in the molar fraction of impurities, which partially due to the small crystal size. Compared to large crystals, the crystals with small sizes will provide more crystal edges, which can control the dissolution at the crystal scale. However, the contribution of the edges to the overall dissolution is crystal-size-and www.nature.com/scientificreports/ time-dependent, which is not well understood yet 116 . Furthermore, one should be very careful to hand these phenomena since the surface area-normalized dissolution rates increase non-linearly with decreasing initial crystal sizes 115 . Last but not least, the heterogeneous distribution of impurities in Cd/Zn/U-calcite should contribute more or less to the nonstoichiometric release of Cd, Zn, and U. We propose that the nonstoichiometric release of Cd, Zn during the weathering of Cd/Zn-bearing limestone is one of the reasons that Cd and Zn enrich the soil in the Karst area [31][32][33][34]117 .
the precipitation of secondary minerals during the dissolution of calcite. In principle, the chemical weathering of rocks involves the coupled dissolution of the primary minerals and precipitation of more stable secondary minerals. The precipitation of secondary minerals is significant for the formation of soil. For example, feldspar, one of the most abundant rock-forming minerals, chemically reacts with water and water-soluble compounds to form clay, such as halloysite and kaolinite 118,119 . In this case, the principal components of the primary mineral (feldspar) (Al and Si) are inherited by the secondary minerals (halloysite and kaolinite).
The trace elements, such as the rare earth elements (REEs), might be enriched or depleted in the altered products depending on the soil and groundwater pH 120 . In South China, the REEs adsorbed on the secondary clay minerals develop world-class heavy rare earth element (HREE) deposits, known as regolith-hosted HREE deposits or ion-adsorption REE deposits, which provide most of the HREE produced in the world 121,122 . Another interesting example is the fate of trace arsenic (As) during the dissolution of As-containing pyrite; As might be adsorbed by the secondary goethite or might be incorporated into the secondary jarosite via substitution for sulfate 123 . Therefore, the minor/trace elements in minerals are hard to form independent secondary minerals during dissolution/weathering. During the dissolution of (Pb x Ca 1−x ) 5 138 and fluorite (CaF 2 ) 139,140 , via a coupled dissolution-precipitation mechanism. Interestingly, the capture of aqueous Mg on the calcite surface was recognized when calcite dissolved in seawater, resulting in an enriched Mg surface 141 . With the growth of the secondary minerals covering the calcite surface, the dissolution of calcite was inhibited accordingly. Note that almost half of the components of these secondary minerals are liquid sources, which combine with dissolved calcium or carbonate from calcite.
Recently, hydrated Mg-carbonate phases have been identified as the secondary precipitates that form during the dissolution of dolomite and magnesite in pure water 130 . For siderite dissolution, the Raman spectra proved the precipitation of iron oxyhydroxide (goethite or ferrihydrite) 130,142 . When siderite dissolved under extreme conditions, a complex assemblage of Fe II -Fe III -iron oxides forms, which was dominated by wüstite (FeO) but also contained some iron(III) observed as hematite and possibly magnetite or defective wüstite 143 . Except for the hydroxyl group (OH − ), which is the water source, the other elements of these secondary minerals came from the parent dolomite and magnesite. Similarly, the secondary minerals in our experiment, i.e., Ca-poor smithsonite, U-rich calcite, and Co-poor aragonite, inherited all the components of the original impurity-bearing calcite. A possible pathway for the formation of Ca-poor smithsonite and U-rich calcite could be expressed as Eqs. (1) and (2). The dissolved Zn/Ca molar ratio of Zn-calcite was much less than that of the solids, suggesting that more Ca 2+ was released than Zn 2+ (Fig. S7). This phenomenon was consistent with Eq. (1). In contrast, the dissolved U/Ca molar ratio was higher than that in the solid, indicating that more UO 2+ 2 ions were released, which provided the possibility of Eq. (2).
As a typical and well known polymorphic transition, the hydrothermal transformation of aragonite to calcite provides a perfect example of interface-coupled dissolution-reprecipitation 144 . The reverse transformation of calcite to aragonite is also relatively common in nature 145 , though it is usually simulated in the laboratory under high-temperature/high-pressure conditions [146][147][148] . Remarkably, Huang et al. observed the complete transition of high Mg-calcite contents to low Mg-aragonite contents at room temperature/pressure within days 149 . Probably, the formation of Co-poor aragonite in our experiment occurs by a similar mechanism because aragonite is the most stable polymorph of calcium carbonate in the presence of a significant amount of Co 2+ ions. As mentioned previously, the dissolved Co 2+ content gradually increased with increasing the dissolution time, and the final released Co 2+ content observed for Co-08 was as high as 16.13 μM after 2880 h (Fig. 2f). Meanwhile, Co-poor aragonite precipitated as a secondary mineral via the dissolution-precipitation process (Figs. 4e,f, S10a-h). The proposed mechanism is shown in Eq. (3). However, there is no direct evidence at present, thus requiring further study.
(1)  154,155 and precipitates as abiogenic and biogenic minerals. In contaminated soils and sediments, bioavailable carbonate-bounded HMs mainly coprecipitate with calcite 156,157 . Moreover, primary carbonate dissolution and secondary carbonate precipitation often happen in soil 158 . Therefore, our results provided further evidence that calcite could control the migration of HMs, especially for Cd, Zn, and Ni. The residual calcite has a higher HM concentration than the primary calcite, which could contribute to the enrichment of Cd and Zn in the soil weathered from limestone [31][32][33][34]117 . However, calcite-bound U tends to release during dissolution, suggesting that U might be more sensitive than other metals to soil and ocean acidification. Although the general content of U in natural calcite is 0.1-10 mg/kg and occasionally up to 360 mg/kg 84 , the dissolution of limestone could rein in the uranium concentrations in river water 29 . Meanwhile, the change in the calcite solubility indicates that the stability of a carbon sink in soil and ocean might be reconsidered once HMs were incorporated. Last but not least, the precipitation of smithsonite, Co-poor aragonite, U-rich calcite, and the previously reported rhodochrosite (MnCO 3 ) and malachite (Cu 2 (OH) 2 CO 3 ) indicated a new mineralization pathway of these minerals, i.e., secondary minerals formed after the dissolution of minerals doped with impurity metals.

conclusion
In this study, four divalent heavy metal (Cd 2+ , Co 2+ , Ni 2+ , and Zn 2+ ) and uranyl ( UO 2+ 2 )-containing calcite were successfully synthesized with changing the M/Ca molar ratio from 0.05 to 12.36%. With the incorporation of impurities, the calcite crystal sizes decreased with increasing the BET surface area. According to the batch dissolution data, the BET surface area calibrated dissolution rates at the early stage decreased with the increase in the impurity molar fraction, indicating an inhibition effect on the dissolution rate. However, both inhibition and promotion effects were observed on the calcite solubility, depending on the type of impurities and their content. The dissolution of impure calcite is incongruent, especially for Cd-, Ni-, and U-containing calcite. The retention ability of Cd and Ni in calcite is remarkably better than that of U, suggesting the potential risk of uranium release due to U-rich calcite dissolution. Meanwhile, the HRTEM and SAED results demonstrated that partial zinc and uranium were immobilized in the reprecipitated phase as smithsonite and U-rich calcite, respectively. Furthermore, the discovery of Co-poor aragonite as a secondary mineral during the dissolution of Co-calcite provided a new type of calcite-aragonite transition. In conclusion, our results demonstrated that the dissolution process of calcite could be significantly changed by incorporating divalent heavy metals and uranyl cations, which has potential environment and climate implications.

Methods
Solid preparation and characterization. The coprecipitation method followed the method used by our previous investigation 45 and was similar to the method used by Katsikopoulos et al. 66 . First, a 0.1 mol/L heavy metal and uranyl solution (M = Cd 2+ , Co 2+ , Ni 2+ , Zn 2+ , UO 2+ 2 ) and a 1 mol/L CaCl 2 solution were mixed with different molar ratios (Table 1). Afterward, a Na 2 CO 3 solution was delivered at a constant rate using a peristaltic pump under continuously stirring. The solid-solution precipitates were allowed to age and incubate for three days, and then, the solutions were centrifuged and dried at 80 °C. All the experiments were performed at room temperature and atmospheric pressure. In the following text, the calcite doped with impurities will be recorded as HM-calcite, e.g., Cd-calcite means calcite containing Cd.
The mineral identity of the precipitates was confirmed by X-ray powder diffraction (XRD) using a Bruker D8 Advance powder diffractometer with Ni-filtered Cu-Kα radiation (λ = 1.5406 Å, 40 kV and 40 mA). The samples were scanned from 10° to 80° (2θ), a range that covers the characteristic peaks of calcium carbonate, using a step size of 0.01° (2θ) and a scan rate of 3.0° (2θ) per minute. Then, 0.05 g of each precipitate was dissolved in concentrated HNO 3 and used to analyze the cation contents, which were determined by inductively coupled plasmaoptical emission spectrometry (ICP-OES) (VARIAN VISTA PRO) after dilution. The Brunauer-Emmett-Teller (BET) surface area of the precipitates was determined using a Micromeritics ASAP 2020 M surface area analyzer, in which the samples were dried and degassed at room temperature for 24 h using N 2 gas. Scanning electron microscopy (SEM) (ZEISS Gemini 500) was performed to obtain images of the precipitates and to provide an approximate size and morphology of the crystalline samples.
Dissolution experiments and characterization. The procedure of dissolution experiment is the same as described in our previous investigation 45 . In brief, an initial dissolving solution (pH = 5.00 ± 0.05) was prepared using deionized water (18.2 MΩ·cm) and hydrochloric acid (AR). Each 0.125 g of calcite powder was put into a 15 mL polypropylene tube containing 10 mL of the initial reaction solution. Afterward, all the tubes were capped and shaken using an overhead shaker in an incubator chamber at 25 °C. At a specified time, the reaction solution was sampled and filtered by a 0.22 μm pore filter and stored in a clean polyethylene tube. The aqueous pH was measured immediately using a pH meter (Sartorius PB-10). Cations were analyzed using ICP-OES, and the dissolved inorganic carbon (DIC) was measured using a total organic carbon (TOC) analyzer (TOC-VCPH, Shimadzu, Japan). All the dissolution experiments were run in duplicate, and an average was obtained for each metal concentration. Meanwhile, the solids were sampled from the bottles, dried, and preserved for characteri- Ca (1−x) Co x CO 3(s) (Calcite) ⇋ Ca y Co z (CO 3 ) (y+z)(s) Co -poor Aragonite + 1 − x − y Ca 2+ (aq) x > z, y > z, z y < x 1 − x Scientific RepoRtS | (2020) 10:16864 | https://doi.org/10.1038/s41598-020-73555-6 www.nature.com/scientificreports/ zation. The morphology of calcite after dissolution was observed by SEM and a high-resolution transmission electron microscopy (HRTEM) system operated at 200 kV (FEI Talos F200S).
calculation of solubility and chemical species. The dissolution of pure calcite and impurity-containing calcite can be expressed as Eqs. (4) and (5).
where x is the molar fraction of MCO 3 in the solid-solution.
The solubility was calculated using the dissolved metal (Ca + M) and stoichiometric carbonate. The corresponding equilibrium solubility product expression is showing as Eqs. (6) and (7).
Usually, the ion activity product (IAP) is also used to describe the solubility of a solid 41 , as shown in Eqs. (8) and (9).
where α i is the activity of species i.
However, Davis et al. used the same IAP expression for Mg-calcite that was used for pure calcite 44 .
In this investigation, the species activity and saturation index (Table S1) were calculated using Visual MINTEQ. For comparison, IAP was expressed in the same way as it was expressed by Davis et al. 44 .

Data availability
The datasets generated during and/or analyzed during the current study, whether they were included in this published article and its Supplementary Information files or not, are available from the corresponding author upon reasonable request. www.nature.com/scientificreports/