Towards efficient calcium extraction from steel slag and carbon dioxide utilisation via pressure-swing mineral carbonation

Abstract

Production of precipitated calcium carbonate (PCC) via carbon dioxide (CO₂) pressure-swing mineral carbonation is a potential way to utilise calcium-rich steel slag and carbon dioxide. Calcium supersaturation and slag surface passivation are two aspects of the calcium extraction step that strongly influence the choice of operating conditions necessary for rapid and complete calcium leaching, which are investigated in the present work. To investigate these aspects, slag dissolution characteristics were studied in a closed high-pressure batch-reactor taking care to eliminate gas–liquid mass transport limitations. This experimental design has two distinct advantages: (1) rapid CO₂ absorption necessary for dissolution under acidic conditions and to gain insights into calcium dissolution kinetics and (2) the closed system allowing measurement of the drop in reactor pressure, which along with elemental analysis of leachate is sufficient to determine the ionic species concentration and solution saturation state. The results provide evidence against surface passivation of residual slag by silica or calcium carbonate layers. Further, the experiments confirm high supersaturation with respect to calcite during the dissolution step, which we hypothesise to be a consequence of unfavourable calcite precipitation kinetics due to the low pH and high calcium to carbonate ion ratio. The results show scope for further enhancement in calcium solubility, up to the solubility limit of amorphous calcium carbonate, which can substantially reduce the water volume and CO₂ pressure required for dissolution. Pressure-swing to atmospheric pressure led to spontaneous co-precipitation of rhombohedral calcite and amorphous silica, also a paper-filler with similar optical properties to PCC, with impurities less than 1.5 wt%.

---

1 Introduction

The use of steel slag, the calcium-rich industrial waste from the steel industry, by mineral carbonation to sequester the carbon dioxide emitted by the process of steel making, has attracted considerable attention for carbon dioxide capture and better solid waste management.¹ Recently, the focus has been on producing precipitated calcium carbonate (PCC) by indirect mineral carbonation, a multi-step process which consists of selective calcium extraction and carbonate precipitation steps.² PCC is used as a filler material in writing and printing paper for better optical properties and has a high market value.³ The indirect route, in the case of steel slag, further yields Fe enriched residual slag, which can be recycled within the iron & steel industry⁴ and has a prospective use in production of cathode precursor materials.⁵

Slow calcium extraction rates from silicates and low extraction efficiency,⁶ especially in the case of coarser feedstock particle sizes,⁷ are serious impediments to commercialisation of the process. The calcium extraction rate and overall efficiency have been shown to increase with particle size reduction.^7,8 This behaviour, along with parabolic kinetics, led to the speculation that the reactive surface is passivated by silica and calcium carbonate layers.^1,7,9,10 To accelerate the calcium extraction, dissolution is typically carried out in mineral^11,12 or organic acids,^13–15 which are difficult to regenerate and require additional pH-swing, wherein the Ca-rich aqueous solution pH is increased by addition of alkaline media to facilitate CO₂ absorption and carbonate precipitation. Similarly, recycling and regeneration of the leaching agent (ammonium chloride solution) is a problem with spontaneous pH-swing methods.¹⁶ Alternatively, the carbon dioxide pressure-swing technique proposed by Iizuka et al.¹⁷ is promising for PCC production, as it does not involve the use of any acids or bases except carbon dioxide itself. Pressure-swing carbonation has been devised to take advantage of the increase in calcium carbonate solubility at high CO₂ partial pressure. The dissolution stage of the process involves leaching of Ca in aqueous solution under high-pressure of carbon dioxide. Post dissolution, Ca rich aqueous solution is filtered and precipitated at low partial pressure of CO₂. Studies on waste cement^17,18 showed potential for high Ca extraction rates and Ca extraction efficiency. The purity of calcium carbonate, however, was found to be 80% in unseeded precipitation; synthesis of high-purity (98 wt%) PCC required a large amount (2.5 wt%) of calcium carbonate seed material, an amount which was comparable to the expected precipitate weight. The nature of impurities and influence of dissolution conditions on carbonate characteristics have also not been reported. Further, in the case of Ca extraction from waste cement, the process required either high water volume (liquid-to-solid ratio of 350 ml g⁻¹) or large recycle of residual waste cement at a low liquid-to-solid ratio (L/S of 40 ml g⁻¹), which resulted in a significant process cost.¹⁸

In terms of the mechanisms involved, multi-oxide/silicate dissolution and carbonate precipitation under CO₂ pressure appear to be complex and not clearly understood. For instance, precipitation of carbonate and silica preferentially occurs on the surface of the slag¹ leading to speculation about surface passivation. However, such passivation has not been conclusively established.¹⁹ Iizuka et al.¹⁷ reported supersaturation of calcium during the dissolution step. However, reasons for the same are not clear. Calcium extraction beyond the calcite solubility limit can significantly reduce the water volume required for complete calcium extraction, an aspect which has not yet been explored. Also, some direct mineral carbonation experiments^1,20,21 showed better overall conversion at very low liquid-to-solid ratios, which was deduced as an enhancement in Ca dissolution efficiency. However, these studies were in contradiction to a few other studies^17,22 where better Ca dissolution efficiency was obtained at high L/S. To a large extent, the problem of not being able to understand the dissolution and precipitation mechanisms is due to limited data,⁹ inconsistency in methodology and incomplete evidence. In most high-pressure mineral dissolution and carbonation studies, researchers relied upon either inductively coupled plasma (ICP) analysis of filtered liquid samples or carbonate quantification of solid samples obtained after de-pressurisation of the system to interpret the kinetics. It must be noted that the precipitation of carbonate is spontaneous when the system is de-pressurised and the estimated carbonate content in the residues after filtration and drying does not therefore reflect the true process efficiency. Thus, it is impossible to resolve the true rate of dissolution from apparent rates without the knowledge of in situ carbonation. In addition to these shortcomings in existing data, there are specific complexities arising due to the nature of feedstock. For instance, in the case of mineral carbonation of steel slag, the influence of Mg and Fe which can potentially affect the carbonate morphology^23–25 and form secondary precipitates is yet to be explored.

This work attempts to address the above-mentioned gaps in experimentation and determine true calcium dissolution rates. High-pressure batch dissolution experiments have been carried out in a closed system after eliminating CO₂ mass transfer limitations (described in sections 2 and 3.2.1). Calcium leaching characteristics from steel slag under CO₂ pressure have been studied to probe surface passivation and its influence on dissolution kinetics. This study further attempts to understand the mechanism behind calcium supersaturation and the possibility of reducing the water requirement. To investigate these aspects, dissolution experiments were performed at moderate CO₂ pressures, temperatures and liquid-to-solid ratios (0.6–1.25 MPa, 25–60 °C, and 50–100 ml g⁻¹, respectively) using steel slag with fine (<75 μm) and coarse particle size fractions (90–125 μm, and 212–355 μm). A study has also been carried out on carbonate precipitation characteristics at atmospheric pressure by pressure-swing and PCC characterisation to determine the PCC purity and overall calcium conversion.

2 Methodology of the study

The present methodology involves the use of a closed system to study the existence and extent of calcium supersaturation in the liquid and surface passivation of slag particles by calcium carbonate precipitation. The rationale behind the choice of a closed system is to monitor the CO₂ pressure change with time, in addition to the elemental concentration in the aqueous solution by ICP analysis. As shown below, based on these measurements in a closed system, wherein the CO₂ moles, gas phase volume, and temperature are constant (constant NVT), it is possible to determine the state of the solution (pH, supersaturation). We base this on a calculation, from the experimental data, of the moles of CO₂ fixed as bicarbonate and carbonate per unit mole change of M²⁺ ions in solution (ΔC_s/Δ[M]). Relationships between ΔC_s/Δ[M] and the solution state are obtained using an overall carbon balance (equation given below) and the condition that the aqueous solution is saturated with CO₂.‡where [CO₂]_tot is the total moles of CO₂ per unit liquid volume (V_L), P_CO2 is the CO₂ partial pressure, [CO₂]_aq is the concentration of undissociated CO₂ absorbed in the liquid (H₂CO₃ is usually negligible under these conditions), V_G is the volume of gas, and R is the universal gas constant.

Under CO₂ saturated conditions (using Henry's law [CO₂]_aq = k_H × P_CO2), the change in the concentration of CO₂ fixed as carbonate and bicarbonate in aqueous solution is, from eqn (1),

Table 1 shows the reactions involved. Based on the reaction chemistry and charge balance equation, depending on the solution state, different conditions for ΔC_s/Δ[M] exist (as shown below). Based on the saturation state of the solution, these conditions are categorised into two cases.

Table 1 Reaction chemistry of the slag–CO₂–H₂O system

Step	Reactions
[M^2+] = [Ca^2+] + [Mg^2+] + [Fe^2+]; PCC: precipitated calcium carbonate.
CO2 absorption & speciation
Slag dissolution	C3S + H2O → C2S + Ca(OH)2
	C2S + 4H^+ → 2Ca^2+ + H4SiO4
	MgO + 2H^+ → Mg^2+ + H2O
	FeO + 2H^+ → Fe^2+ + H2O
Overall dissolution	C2S + 4H2CO3 → 2Ca^2+ + 4HCO3^− + H4SiO4
PCC precipitation	Ca^2+ + CO3^2− → CaCO3
Charge balance	2M^2+ + H^+ + MHCO3^+ − OH^− − HCO3^− − 2CO3^2− = 0

---

Case I: ionic activity product (IAP) less than solubility product (K_sp) for CaCO₃

During the dissolution step (under acidic and circumneutral pH conditions, where [CO₃²⁻] ≪ [HCO₃⁻]),

ΔC_s/Δ[M] ≈ 2 (3)

On the other hand, when dissolution takes place in a mildly basic environment (8 < pH < 10) where [CO₃²⁻] and [HCO₃⁻] are comparable,

1 < ΔC_s/Δ[M] < 2 (4)

In the case of dissolution under very basic conditions (pH > 10) where [CO₃²⁻] ≫ [HCO₃⁻],

ΔC_s/Δ[M] ≈ 1 (5)

Case II: IAP greater than K_sp for CaCO₃

Here, three possibilities exist—(a) the solution may remain supersaturated with respect to CaCO₃, or (b) the supersaturation may be discharged as a precipitate that remains suspended and does not show up in the liquid samples, or (c) the supersaturation is discharged as a colloidal precipitate that passes through the sampling filter and shows up in the liquid sample. These cases may be distinguished by defining , the moles of CO₂ consumed to dissolve Ca²⁺ in the solution,§eqn (6)

(a) In the case of acidic pH (eqn (3)) and true solution saturation with respect to CaCO₃, we have,

(b) In the case where supersaturation is discharged as a CaCO₃ precipitate and the precipitate is not passed into the liquid sample (precipitate size > filter pore size),

Δ[Ca] ≤ 0 (8)

(c) In the case where the CaCO₃ precipitate exists in the colloidal form, which passes through the sampling filter and dissolves upon acidification of the sample,¶ if the calcium dissolution rate from the slag is negligible,

Δ[Ca] = 0 (9)

On the other hand, when the calcium dissolution and CaCO₃ precipitation rates are comparable,

Thus, by measuring the ΔP_CO2 under CO₂ saturated conditions, it is possible to distinguish the Ca dissolution reaction from the carbonate precipitation reaction, and a true supersaturated solution from a colloidal precipitate.

Though the above methodology is limited by the sensitivity of pressure loss measurement and ability to maintain the system at constant temperature, the concept is useful to gain insights into various regimes of dissolution and precipitation. A detailed discussion on such limitations and numerical methods to overcome them is presented in the Results and discussion section 4.2.

3 Materials and methods

3.1 Materials

Basic oxygen furnace slag was kindly provided by M/s JSW Ltd, India, from their steel-making plant. The as-received slag was washed in acetone to remove surface impurities. The washed slag was dried and crushed in a Herzog pulverizing mill (HSM100) for 60 seconds. The crushed sample was carefully sieved to separate 212–355, 150–212, 90–125 and <75 μm fractions using standard test sieves, and the fractions were stored in airtight containers. The chemical and mineral composition of the slag was characterized by X-ray fluorescence (XRF) and X-ray diffraction (XRD). The slag was found to be rich in elements such as Ca (30.7%), Fe (16.9%), Si (8.0%), and Mg (5.5%) which were mainly concentrated in C₂S (36.2 wt%), C₃S (9.1 wt%), the RO phase containing FeO and MgO (20.4 wt%), and dicalcium ferrite (23.5 wt%) mineral phases. CO₂ gas (99.99% purity) was obtained from BOC. Deionized water from a Milli-Q® dispenser was used in all experiments. 0.01 M nitric acid prepared from 65% concentrate (analytical reagent grade) was used to measure the alkalinity of slag leachate samples by potentiometric titration. Standard buffers (pH 4 and 7) received from Horiba Scientific along with a LAQUAtwin pH meter were used for pH meter calibration. Multi-element standard solution 6 for ICP (TraceCERT® grade) obtained from Sigma-Aldrich was used to prepare standard calibration solutions for elemental analysis.

3.2 High-pressure experiments

All high-pressure experiments were performed in a 600 ml bench top Parr autoclave (4568) equipped with a Parr 4848 reaction controller (shown in Fig. 1). The reactor temperature is maintained using an electrically heated mantle and circulating cooling water. The reactor stirrer has two 4-blade impellers and has a maximum stirring speed of 800 rpm in water. Details of the setup with inlet and outlet ports are shown in Fig. 1.

Fig. 1 Schematic of the high-pressure autoclave and modified stirrer which acts as a self-inducing impeller.

3.2.1 Mass transfer characterisation of the carbonation reactor. CO₂ absorption in water was studied to estimate the gas–liquid mass transfer characteristics in the reactor. The autoclave was filled with 250 ml of water and the air present in the reactor was purged out by performing three pressurising–depressurising cycles with CO₂. After air removal, the reactor was pressurized to a desired pressure of 8 bar. After achieving the desired initial reactor pressure, the inlet valve was closed and stirring was commenced at 800 rpm. Changes in CO₂ pressure with time were noted. It was observed that it takes about 25 minutes before water is saturated with CO₂; the estimated k_La was found to be an order of magnitude lower than desired to avoid significant gas–liquid mass transport limitations. To overcome the problem, the reactor stirrer was fitted with three thin plastic tubes with one end open to the gas space (shown in Fig. 1). This would act like a self-inducing impeller to improve the gas–liquid mass transfer (also see section 4.1). An absorption experiment under the same operating conditions with the aforesaid modification improved the k_La value significantly and reduced the saturation time to three minutes. This arrangement was adopted in all subsequent experiments. Under these conditions, the liquid could be assumed to be saturated with CO₂ at all times.

3.2.2 Pressure-swing carbonation of slag – calcium extraction step. All dissolution experiments were performed in 250 ml of water. The slag was loaded into the reactor and the slurry was stirred for two minutes to avoid lump formation which was otherwise observed. The liquid was sampled for elemental analysis before pressurizing the reactor. The reactor was then closed and rapidly pressurized to the desired initial pressure. The reaction mixture was stirred at 800 rpm and changes in pressure with time were recorded. About 2 ml of the supernatant from the reaction slurry was sampled at 3, 10, 20, 60, 90 and 120 minutes through a high-pressure sampling valve after temporarily stopping (for about 30 seconds) the stirrer to avoid flow of thick slurry into the sampling line. The slurry was immediately filtered through a 0.2 μm syringe filter and acidified using 1 M nitric acid for ICP-OES analysis. Also, potentiometric titration with 0.01 M HNO₃ was carried out using a part of the unacidified liquid sample for alkalinity measurement. Sample alkalinity is used to cross-validate the total divalent metal concentration measured using ICP-OES. Dissolution experiments were also repeated without intermittent liquid sampling to ensure a completely closed system for CO₂ pressure loss measurement. Pressure change measurements with and without liquid sampling matched when the pressure loss due to liquid sampling was neglected. At the end of the experiment, the slag residue was filtered and dried overnight at 100 °C.

3.2.3 Precipitation kinetics of CaCO₃. At the end of the dissolution step, the leachate from the reactor obtained under CO₂ pressure was filtered using a 0.2 μm filter. The filtered solution was stirred in a glass beaker at a stirring speed of 500 rpm. Precipitation was carried out for up to three hours by monitoring the progress by pH measurement and sampling for ICP-OES analysis. The calcium carbonate precipitate was filtered using Whatman filter paper and left for overnight drying in a hot air oven maintained at 100 °C.

3.2.4 Characterisation. Elemental concentrations (Ca, Mg, Fe and Si) in the liquid samples were determined using a PerkinElmer Avio 200 inductively coupled plasma optical emission spectrometer (ICP-OES). Scanning electron microscopy (SEM) with energy dispersive spectroscopy (EDS) was used to study the morphology and chemical composition of PCC and residual steel slag. The calcium carbonate content in the PCC precipitate was determined based on the weight loss observed using thermogravimetric analysis (TGA) in the temperature range of 600–900 °C. Fourier transform infrared spectroscopy (FT-IR) was used for chemical analysis of PCC, mainly to determine the Si compounds.

4 Results and discussion

4.1 Elimination of gas–liquid mass transfer limitations

To understand the rate of CO₂ absorption in the autoclave, absorption experiments were performed in a closed system with 250 ml of distilled water and a gas volume of 350 ml of at 25 °C. As described in the experimental section, the experiment was stopped when the CO₂ pressure remained constant. Since, the absorption is performed in a constant NVT and variable pressure system, the following equation is used to determine the mass transfer coefficient (k_La),where [CO₂]* is the saturation concentration of CO₂ (at the gas–liquid interface) and [CO₂]_aq is the concentration of CO₂ in the bulk liquid.

Using eqn (1), Henry's law ([CO₂]* = k_H × P_CO2), and neglecting [HCO₃⁻] + [CO₃²⁻] which is ≪[CO₂]_aq under CO₂ pressure greater than 1 bar in distilled water, the following equation was arrived at.

Integrating the above equation, we obtain

χ(T, P_CO2, [CO₂]_total) = −k_La × t + C (13)

where C is a constant, and

Using eqn (13), the mass transfer coefficient (k_La) is estimated as the slope of the linear fit as illustrated in Fig. 2B.

Fig. 2 (A) Drop in the reactor pressure during absorption in distilled water without (○) and with (△) stirrer modification and (B) the corresponding k_La estimation.

Absorption experiments showed that the time required to reach constant pressure (CO₂ saturation time) with the existing reactor configuration was about 25 min (Fig. 2A). Since this time was comparable to the estimated dissolution time of 1–2 hours,^1,17 it was anticipated that external mass transfer limitations would influence the reaction rate. To overcome the problem, as described in section 3.2.1, the reactor stirrer was modified to act like a self-inducing impeller. As demonstrated in Fig. 2, the mass transfer coefficient improved by an order of magnitude from 1.45 ± 0.07 ×10⁻³ s⁻¹ to 1.50 ± 0.01 × 10⁻² s⁻¹, reducing the saturation time to three minutes. The pressure data in slag carbonation experiments (to be discussed later) showed that the fall in pressure in the post saturation phase was much more gradual than that in the initial saturation phase. Thus, it can be stated that the rate of CO₂ absorption is much faster than those of slag dissolution reactions and does not influence the dissolution kinetics.

4.2 Temperature sensitivity of pressure loss measurement

In a closed system, the reactor pressure is expected to be influenced by fluctuation in reactor temperature, mainly due to changes in CO₂ solubility, and in turn influence the pressure loss measurements. An analysis has therefore been carried out to understand the influence of temperature fluctuations on the observed reactor pressure.

A total CO₂ balance in the slag carbonation system with a solid phase gives,

In pure water, under CO₂ pressure, the concentrations of [HCO₃⁻] and [CO₃²⁻] are insignificant compared to [CO₂]_aq. On the other hand, in the case of slag carbonation, the concentration of these species is strongly dependent on the concentration of metal ions (for charge balance) and is expected to show negligible influence with small temperature changes. Hence, eqn (15) can be used for further analysis under saturated conditions, where [CO₂]_aq can be described by Henry's law.

where mol L⁻¹ bar⁻¹.²⁶

Differentiating the above equation with respect to P and T, we obtain dn_CO2 = 0 and

Eqn (16) is rearranged as follows to estimate the changes in reactor pressure due to a small magnitude of change in reactor temperature,

Under typical experimental conditions of V_G = 0.35 L, V_L = 0.25 L, T = 298.5 K and P_(CO2) = 800 kPa, a change in reaction temperature of 1 °C influences the pressure measurement by ∼10 kPa (Fig. 3A). Though this temperature change induced error in pressure is insignificant when compared to the total reactor pressure, it is large enough to introduce noise into reaction induced pressure loss estimates. Ideally, temperature fluctuation less than 0.1 °C is desired to contain errors in pressure measurement less than 1 kPa. However, the reactor used for the current study was electrically heated and the PID controller often resulted in fluctuations of about 1–2 °C due to the reaction exothermicity and changes in cooling water temperature. As the observed accuracy in the pressure loss measurement was lower than the desired value, recorded pressure was corrected based on the temperature changes using eqn (17). Changes in reactor pressure observed with a change in reactor temperature in a closed system were experimentally determined under typical experimental conditions of V_G = 0.35 L, V_L = 0.25 L, T = 298.15 K and P_(CO2) = 885 kPa and compared with the predictions of eqn (17). As shown in Fig. 3A, corrections in pressure predicted by eqn (17) are in excellent match with experimental observations. Further, as shown in Fig. 3B and C, corrections using eqn (17) significantly reduced the noise in pressure loss measurements, which allowed to accurately determine the relationship between ΔC_s and Δ[M].

Fig. 3 (A) Sensitivity of pressure with temperature change during CO₂ absorption in distilled water at 25 °C; (B) typical temperature fluctuation during slag carbonation experiment, and (C) fluctuations in the corresponding pressure loss measurement during slag dissolution and the numerically corrected response using eqn (17).

4.3 Calcium leaching characteristics from slag—parametric study

A parametric study was conducted to understand the influence of various operating parameters such as the liquid-to-solid ratio, particle size, temperature and carbon dioxide pressure on the dissolution characteristics of slag. Based on these results, an attempt has been made to identify the optimal conditions for mineral carbonation.

4.3.1 Effect of the liquid-to-solid ratio. The influence of the liquid-to-solid ratio (L/S) on the extent of calcium dissolution was studied in the range of 50–100 ml g⁻¹ using <75 μm particles at 25 °C and an initial CO₂ pressure of 12.4 bar. In these experiments, the slag weight was varied while keeping the water volume constant to obtain the desired L/S ratio. This procedure was chosen over other options to vary the liquid volume since the gas–liquid interfacial area was nearly constant in this procedure, and hence the gas–liquid mass transfer rate. As shown in Fig. 4A, the rate of CO₂ pressure loss was found to increase with the increase in solid loading suggesting that the rate of CO₂ absorption does not limit the dissolution. As shown in Fig. 4B, increasing the solid loading (low L/S ratio) resulted in higher initial rates of Ca release and higher final concentration of dissolved calcium. The initial calcium release rates are proportional to the mass of slag (also surface area). However, as the dissolution proceeded, the Ca release rate per unit mass of slag at L/S of 50 ml g⁻¹ decelerated faster compared to that at L/S of 100 ml g⁻¹, understandably due to the lower CO₂ pressure in the reactor and higher alkalinity of the solution which reduced the rate of leaching. Consequently, % Ca dissolved from slag reduced from 57.1% to 40.2% (after 120 min of dissolution) as the L/S ratio was lowered from 100 to 50 ml g⁻¹ (refer to Fig. S1 in the ESI†).

Fig. 4 Effect of the liquid-to-solid ratio (T = 25 °C, PS = <75 μm, P_CO2,initial = 12.4 bar) on (A) pressure loss during the dissolution step, (B) Ca leaching characteristics, and (C) Mg leaching characteristics.

The current experimental observations on the influence of L/S on calcium dissolution rates are in contrast with previously reported work. For instance, Huijgen et al.¹ reported that the overall calcium carbonate yield monotonically increased with the decrease in the liquid-to-solid ratio. It was hypothesised that ionic strength improved the dissolution of Ca. Also, some studies^20,21 reported the presence of an optimal L/S ratio for unstirred systems and rather weakly stirred systems.²⁷ These studies hypothesised that too much water hindered CO₂ permeability into the material. Together, these hypotheses that the dissolution step accelerates when L/S is reduced contradict current experimental results and those reported by a few others.¹⁷ Further, all these studies determined the extent of carbonation by measuring the weight gain of the carbonated residue which did not provide direct insights into Ca dissolution characteristics. Possible reasons for these contradicting observations are i) varying liquid volume reduced the liquid surface-area-to-volume ratio which in turn reduced gas–liquid absorption rates and ii) incomplete carbonation precipitation at a higher L/S ratio (due to low supersaturation) led to low overall efficiency. Under CO₂ saturated conditions, it is clear from the current experimental observations that higher L/S improves the calcium extraction efficiency.

Similar to Ca leaching behaviour, the initial rate of Mg dissolution was observed to proportionally increase with increasing mass of slag owing to the higher surface area (Fig. 4C). However, the deceleration in dissolution of Mg with progression of leaching at a low L/S ratio is comparatively lower than that of Ca. This resulted in a marginal increase in the Mg/Ca ratio from 0.11 to 0.14 when L/S was varied from 100 to 50 ml g⁻¹.

The influence of L/S on Si dissolution characteristics is very similar to that on Ca dissolution behaviour. As shown in Fig. 5A, a stoichiometric relationship exists between the temporal evolution of dissolved Si and Ca. This can be explained by the mineralogy of the slag where extractable Ca and Si are present in dicalcium silicate and tricalcium silicate phases (calcium present in the dicalcium ferrite phase is not susceptible for leaching²⁸). These phases belong to the orthosilicate group containing no bridging oxygen (Si–O–Si) and dissolution is expected to occur in two-steps where the breaking of Ca–O bonds is followed by liberation of SiO₄ tetrahedra without any need to break Si–O bonds.⁹ The overall dissolution step from the calcium silicate phase present in the slag can be considered as follows with soluble silica (present as silicic acid) as a stable intermediate which is further expected to undergo polymerization to form silica particles.

Fig. 5 (A) Relative dissolution of calcium and silicon; (B) effect of L/S on Si release.

In all three cases, the Si concentration reached a plateau within the two hour period of dissolution (Fig. 5B). And the maximum Si concentration (18 mmol L⁻¹) was significantly higher than its expected solubility (1.9 mmol L⁻¹) under prevailing reaction conditions. Given the extremely high apparent supersaturation, it is highly likely that silica exists as colloidal particles of size less than 0.2 μm (filter pore size) instead of a true solution. At the prevailing pH (4–6), nanocolloid silica demonstrates uncharacteristic stability with a lifetime ranging from a few hours to days.²⁹ This behaviour is expected due to strong repulsion among nanocolloid silica particles due to their negative surface charge. However, the increase in ionic strength³⁰ and the presence of divalent cations³¹ could accelerate the growth of particles, possibly due to reduction or cancellation of the negative surface charge.³² As observed by Iler,³³ a critical calcium ion concentration is necessary for coagulation of colloidal silica where the critical concentration is higher for smaller colloidal silica particles. Thus, silica precipitation is induced with further Ca dissolution as shown in Fig. 5B. As the slag loading was increased, understandably, the rate of Si precipitation increased because of higher Ca concentration in the solution.

From the current investigation, it may thus be concluded that higher L/S is desirable for maximum utility of calcium in the slag. Further, slower polymerisation of silica at high L/S allows easy filtration as polymerised silica from orthosilicates is gelatinous and makes filtration difficult.³⁴

4.3.2 Effect of CO₂ pressure. The influence of the carbon dioxide pressure on Ca dissolution characteristics was studied by varying the initial reactor pressure (CO₂ partial pressure) from 6 to 12.4 bar using <75 μm particles at 25 °C and L/S of 100 ml g⁻¹. As shown in Fig. 6, an increase in the initial CO₂ pressure resulted in a higher extent of Ca dissolution at the end of 120 min. In both the regimes where CO₂ was undersaturated (<3 min) and saturated (>3 min), increasing pressure showed a positive influence on the rate of Ca dissolution. The influence of pressure on the Ca dissolution rate is estimated to be higher in the first three minutes of dissolution compared to that in the CO₂ saturated regime. This can be attributed to the rapid rate of acidification of slurry under higher CO₂ pressure. On the other hand, in the CO₂ saturated regime, enhancement at higher CO₂ pressure can be attributed to a smaller change in pH as compared to lower CO₂ pressure. CO₂ pressure has a similar influence on Mg leaching. An increase in pressure led to a marginal increase in the net CO₂ pressure loss indicating a higher rate of the overall elemental dissolution (Fig. 6). Overall, the CO₂ partial pressure has a positive and noticeable influence on mineral dissolution kinetics.

Fig. 6 Effect of CO₂ pressure (T = 25 °C, PS = <75 μm, L/S = 100 ml g⁻¹) on (A) Ca leaching characteristics, (B) Mg leaching characteristics, and (C) pressure loss during the dissolution step.

4.3.3 Effect of particle size. The effect of particle size on Ca release and Mg release was studied using <75 μm, 90–125 μm and 212–355 μm at L/S of 100 ml g⁻¹, 25 °C and an initial CO₂ pressure of 12.41 bar. As shown in Fig. 7, the Ca and Mg elemental release showed a parabolic trend for both fine and coarse particles in the first 10 min of dissolution. However, beyond 10 min of dissolution, the rate of dissolution was observed to be constant for coarser particles while the finer particles continued to exhibit a parabolic concentration profile. In fact, coarser particles (90–125 μm) continued to exhibit the linear concentration profile even when the reaction time was extended up to 240 min. The difference, however, does not appear to originate from the non-homogeneity of the sample or changes in reaction chemistry as ratios of release of elements (Ca/Si and Ca/Mg) remained the same for both cases. It is likely that the parabolic trend prevalent in the finer particle-size fraction (<75 μm) is due to the presence of ultra-fine particles and the broader particle-size distribution in the sample,³⁵ wherein a high initial rate corresponds to rapid dissolution of ultra-fine particles and the progressive change to the coarser particle-size distribution reduces the dissolution rate.

Fig. 7 Effect of particle size (T = 25 °C, L/S = 100 ml g⁻¹, P_CO2,initial = 12.4 bar) on (A) Ca leaching characteristics, (B) Mg leaching characteristics, and (C) pressure loss during dissolution step.

Surface passivation by the Si rich layer or precipitated calcium carbonate layer has been widely suspected as the reason for the reduction in the overall Ca leaching/carbonation efficiency when the particle size is increased.^1,7 Extended dissolution experiments with 90–125 μm particles for 240 min showed that the Ca extraction efficiency can reach as high as that for finer particles without showing any sign of surface passivation. These results suggest that high Ca extraction efficiency from coarser particles can be achieved, which can bring down the need for energy-intensive particle-size reduction.

4.3.4 Effect of temperature. The effect of temperature on dissolution kinetics was studied at L/S of 100 ml g⁻¹ and an initial CO₂ pressure of 12.4 bar using <75 μm particles. As shown in Fig. 8A, the overall extent of calcium dissolution was observed to decrease with the increase in temperature. During the initial stage of dissolution (at 3 min), the Ca extraction efficiency is higher at 25 °C. This observation is consistent with the finding that the rate of CO₂ absorption in water decreases with the increase in temperature.³⁶ After 3 minutes of dissolution, the net rate of Ca dissolution marginally increased when the temperature was increased from 25 to 40 °C. It is to be noted that [CO₂] in the solution is lower at 40 °C owing to lower solubility and a negative influence on Ca dissolution kinetics is expected based on observations at lower CO₂ partial pressure. Thus, a net marginal increase in the Ca dissolution rate reflects a considerable influence of temperature which overwhelmed the effect of lower CO₂ concentration. This is a consequence of the positive influence of temperature on mass transfer or surface reaction rates. However, the increase in temperature strongly influenced the CaCO₃ solubility in the solution and led to precipitation after 60 min when the dissolution was carried out at 60 °C.

Fig. 8 Effect of temperature (PS = <75 μm, L/S = 100 ml g⁻¹, P_CO2,initial = 12.4 bar) on (A) Ca leaching characteristics, (B) Mg leaching characteristics, and (C) pressure loss during the dissolution step.

As shown in Fig. 8B, the temperature has only a marginal influence on dissolution of Mg. It appears that the rate of Mg dissolution increases with the increase in temperature, however, due to very small changes and low concentration it could not be clearly resolved. The drop in reaction pressure, which is shown in Fig. 8C, was found to be higher at 25 °C as compared to those at higher temperatures; this is due to the higher overall extraction of calcium at low temperature. Thus, in view of complete extraction of Ca from slag, lower temperature appears to be optimal for mineral carbonation through the pressure-swing route.

4.4 Supersaturation of calcite

Iizuka et al.¹⁷ reported up to three times supersaturation of calcium in dissolution experiments with waste cement. As shown in Table 2, similar observations were made in the current study where all the fine particle experiments, depending on the L/S ratio, showed an ionic activity product (IAP) 4–10 times higher than the solubility product of calcite. Since such an observation has implications to obtain high Ca dissolution at lower CO₂ (reaction) pressure than thermodynamically predicted, it is worth exploring the reasons for the same. Further, understanding reasons for the observed remarkable solution stability for long time periods can pave the way for complete extraction of calcium without surface passivation posing a limitation as posited by many researchers. Also, even if the apparent supersaturation is due to the colloidal nature of precipitated calcium carbonate formed in solution (which passed through a syringe filter during sampling), it is worth understanding the reason behind nucleation in bulk solution so that surface precipitation can be avoided.

Table 2 Estimation of saturation indices for CaCO3 polymorphs

Experimental conditions	[Ca]total	[Mg]total	P CO2	pH	[Ca^2+]	[CO3^2−]		IAP	Ω calcite	Ω aragonite	Ω vaterite	Ω ACC
	mM	mM	bar	—	mM	mM	—	M^2	—	—	—	—
IAP: ionic activity product; Ω: saturation index (IAP/Ksp); ACC: amorphous calcium carbonate.a Precipitation was noticed during dissolution; Ksp values for various polymorphs reported in ref. 37 were used; the Davies equation^37 was used for estimation of the activity coefficient for ionic species.
L/S = 100 ml g^−1, 120 min	43.7	5.1	8.05	5.71	31.6	4.5 × 10^−3	7022	1.7 × 10^−8	4.5	3.8	1.4	0.04
L/S = 71.4 ml g^−1, 120 min	51.5	6.7	7.85	5.78	36.0	6.5 × 10^−3	5538	2.6 × 10^−8	6.8	5.7	2.2	0.07
L/S = 50 ml g^−1, 120 min	61.5	8.6	7.38	5.88	41.2	9.9 × 10^−3	4162	4.2 × 10^−8	10.9	9.2	3.5	0.11
T = 60 °C, 60 min^a	31.4	4.5	9.31	5.82	22.9	6.3 × 10^−3	3635	1.8 × 10^−8	9.4	7.1	3.0	0.15

---

Firstly, to determine whether the apparent supersaturation is a true solution or a colloidal suspension, the moles of CO₂ consumed per mole of metal ions released into the solution are estimated. Theoretical formulation based on which solution state could be interpreted is presented in section 2. Since the observed supersaturation levels are as high as 4–10 times ([Ca]_total − [Ca]_saturated > 20 mM), the sensitivity of pressure loss measurement was found to be suitable to distinguish a true solution from a saturated colloidal suspension. For all the datasets where supersaturation was evident (Ω_calcite > 1) and Ca concentration monotonously increased, an excellent prediction of total metal ion concentration could be made from the pressure loss data when the ratio of CO₂ moles consumed per moles of metal ions was assumed to be 2 (Fig. 9). This shows that the solution is a true solution and disproves formation of any measurable colloidal precipitation or formation of a non-recoverable precipitate. Further, in the case of the coarse particle size (90–125 μm) experiment, where rapid filtration was possible to avoid further CaCO₃ precipitation on residual slag after reactor de-pressurisation, the thermogravimetric analysis (TGA) on the slag residue obtained after 240 min dissolution (Ω_calcite = 4.5) showed no weight loss corresponding to carbonate decomposition (Fig. S2 in the ESI†). This corroborates that there is no precipitation of calcium carbonate on the slag surface.

Fig. 9 Total concentration of metal ions in super saturated solutions – measured (solid symbols) and predicted (open symbols) assuming true solution (Δ[M] = ΔC_s/2).

High supersaturation compared to calcite appears to be a consequence of calcite precipitation being kinetically unfavourable compared to other less-stable polymorphs such as amorphous calcium carbonate (ACC) or vaterite which subsequently transform to calcite. As the more stable calcite grows, the solution becomes undersaturated with respect to less-stable polymorphs leading to their dissolution and complete transformation to calcite. Multiple studies^38,39 agree that amorphous calcium carbonate is the preferred nucleating polymorph when the supersaturation is sufficient to produce ACC. However, in the current study all the solutions, including where precipitation occurred, were undersaturated with respect to ACC (Table 2). Under circumstances where ACC cannot be produced, several studies^38,40,41 agree that the first formed solid is metastable vaterite with the exception of the study of Clarkson et al.⁴² who reported it to be calcite when there are no calcite inhibiting species. Vaterite also appears to be the kinetically preferred polymorph because of its very high [Ca²⁺]/[CO₃²⁻],⁴¹ a parameter which significantly affects the calcite growth rate.⁴³ Also, condensation of supersaturated silica (>10 mM) on carbonate particles especially under acidic conditions can potentially stabilize the growth and transformation of carbonate particles to calcite.⁴⁴ Though it is difficult to precisely conclude the reason for the remarkable stability of supersaturated solution, low pH and very high [Ca²⁺]/[CO₃²⁻] appear to be important factors.

4.5 Surface passivation

Several previous studies^1,7 reported passivation of the residual slag surface due to precipitation of silica and calcium carbonate. In the current study calcium carbonate precipitated during the dissolution step only at high temperature and very high supersaturation. Accordingly, as shown in Fig. 10, no calcium carbonate precipitate was observed on the surface where supersaturated aqueous solution was determined to be a true solution. Thus, it can be concluded that calcium carbonate does not passivate the surface under current experimental conditions. As reported in section 4.3.1, Si releases as monomeric units of silica and does not polymerise even at high supersaturation. For instance, in the case of coarse particle dissolution, Si was found to remain stable in solution for a dissolution time of 240 minutes. Overall, about 57% of the calcium could be extracted from fine and coarse particles in 120 minutes and 240 minutes, respectively, without any sign of surface passivation.

Fig. 10 SEM image of the slag residue after 240 minutes of leaching at 25 °C (PS: 212–355 μm).

In the case where precipitation occurred (60 °C), as shown in Fig. 11, micron-sized calcite crystals were observed as agglomerates along with silica, with the slag surface still being exposed. In contrast, Huijgen et al.¹ reported that the precipitation occurred exclusively on the surface and no carbonate crystals were observed in the bulk. The study also reported the presence of a silica rich layer on the surface of residual slag. Based on the current experimental results, it appears that calcite and silica co-precipitated as agglomerates in the bulk rather than precipitating as individual layers on the slag surface. This may be because of the acidic slurry pH, under which conditions, the calcium silicate surface and calcite nuclei are generally expected to be positively charged and more so because of their high Ca²⁺/CO₃²⁻ ratio (>4000).⁴⁵ Thus, calcite is expected to co-precipitate with negatively charged silica. Smaller calcite crystals which appear to partly cover the slag surface might have precipitated after depressurisation when the saturation index increases multi-fold along with slurry pH.

Fig. 11 SEM image of the slag residue after 120 minutes of leaching at 60 °C (PS: <75 μm).

Based on these results, it can be concluded that surface passivation is not significant with steel slag as the feedstock under current experimental conditions.

4.6 Precipitation

Unseeded precipitation of CaCO₃ was carried out under atmospheric CO₂ pressure using the filtrate (through 0.2 μm filter) from dissolution experiments, using <75 μm particles at 25 °C, an initial pressure of 12.41 bar and varying L/S in the range of 50–100 ml g⁻¹. A reduction in CO₂ pressure from 7.38–8.05 bar to atmospheric conditions increased the supersaturation by more than an order of magnitude (even after considering that the solution is not completely free of dissolved CO₂). After 2–5 minutes of stirring, the clear solution turned turbid indicating the onset of precipitation. During this induction period, a rapid increase in pH to 7.5 was noticed (Fig. 12A). This probably indicates degassing of excess dissolved CO₂. After 60 minutes of precipitation, the solution pH reached 8.45 and turbidity was completely lost with a suspension of calcium carbonate particles in a clear solution. Beyond this point, the solution pH continued to increase at a slower rate until it stabilized in the range of 8.5–8.6.

Fig. 12 (A) Change in pH during the precipitation step; change in concentration of (B) Ca and Si, and (C) Mg and Fe during the precipitation step.

As shown in Fig. 12B, the calcium concentration reached a plateau after rapidly decreasing for the first 30 minutes. The silicon concentration, however, continued to decrease at the same rate for 120 minutes. These results suggest that while very high calcium recovery as carbonate is possible, the amount of silicon in PCC increases exponentially with calcium recovery. Mg remained in the mother liquor and only a small fraction precipitated along with PCC (Fig. 12C). The combined maximum concentration of Mg and Fe as impurities (determined based on elemental concentrations in the mother liquor) was found to be less than 1.5 wt% as magnesium carbonate and Fe(III) hydroxide. Silicon is the major secondary constituent with a maximum concentration of 16.6 wt% as SiO₂, as estimated based on the Si concentration change in the mother liquor. The calcium carbonate content determined by thermal decomposition of PCC samples up to 900 °C showed a weight loss of 36%, which was equivalent to 81.5 wt% purity. This is in close agreement with the elemental analysis where other constituents were estimated to be ∼18 wt%. Si is known to co-precipitate as silica along with calcium carbonate without formation of silicate.⁴⁶ As shown in Fig. 13, the morphology of precipitated calcium carbonate under all experimental conditions was found to be rhombohedral indicating the calcite polymorph. Additionally, a globular amorphous phase rich in silicon was also observed in SEM-EDS analysis (Fig. 13B). FT-IR analysis was performed to investigate the chemical form of Si in the PCC sample. The FT-IR spectra shown in Fig. 14, show peaks corresponding to symmetric (794 cm⁻¹), anti-symmetric (1082 cm⁻¹), and bending vibrations (475 cm⁻¹) of the Si–O–Si group. Additionally, peaks corresponding to anti-symmetric (1418 cm⁻¹), out-of-plane bending (871 cm⁻¹), and planar bending (713 cm⁻¹) vibrations of the CO₃ group were noticed. No shoulders or peaks corresponding to Si–O–Ca which were expected at 950 cm⁻¹ (ref. 47) were noticed. This result suggests the absence of silicate and existence of Si as silica. Silica is also a paper-filler with similar optical and ink adsorption properties to PCC.⁴⁸ These observations suggest that further purification of PCC is not required for its use as a filler in writing and printing paper.

Fig. 13 SEM micrographs of PCC produced at L/S = 100 ml g⁻¹, 60 min of precipitation – (A) 3000× magnification and (B) 10 000× magnification with EDS spectra; (C) L/S = 71 : 4 ml g⁻¹, 120 min of precipitation; (D) L/S = 50 ml g⁻¹, 120 min of precipitation.

Fig. 14 FT-IR spectra of the PCC sample.

4.7 Outlook for pressure-swing carbonation

In the current study, about 57% of calcium from steel slag could be extracted in 120 minutes under mild operating conditions of 12.4 bar and 25 °C. Under these conditions, the quantity of PCC generated including silica is 535 kg per tonne of steel slag. Considering the complete extraction of calcium from silicates with extended dissolution time (assuming dicalcium ferrite is inert), the maximum potential for PCC generation is as high as 720 kg per tonne of steel slag. The corresponding potential for CO₂ sequestration is 260 kg per tonne of steel slag.

In the current study, very high solubility of calcium (2465 mg L⁻¹) could be obtained at an initial CO₂ pressure of 12.4 bar (actual pressure after CO₂ absorption was <10 bar) and L/S of 50 ml g⁻¹ while a previous study⁴⁹ reported a maximum concentration of 1200 mg L⁻¹ under similar conditions for waste cement. Rapid CO₂ absorption allowed dissolution under acidic conditions, which in turn facilitated high supersaturation and potential to reduce the water volume. The remarkable stability of supersaturated solution suggests that calcium extraction up to the solubility limit of amorphous calcium carbonate (ACC) may be achieved by further process intensification. Steel slag is found to be very reactive as calcium is mainly present in orthosilicates, a class of silicates which dissolve by breaking weaker Ca–O bonds (instead of stronger Si–O–Si bonds) and dissolve as fast as calcium oxide in aqueous solutions.^34,50 Further, congruent dissolution of Si from orthosilicates does not leave leached layers which can potentially passivate the reactive surface. Overall, pressure-swing carbonation appears to be very attractive for steel slag and CO₂ utilisation.

5 Conclusions

In this study we have investigated the pressure-swing route as a potential route for production of PCC using steel slag to capture CO₂. Simple stirrer modification allowed us to increase the gas–liquid mass transfer coefficient and perform studies on mineral carbonation under CO₂ saturated conditions, which in turn permitted us to perform carbon balance and determine that the solution is truly supersaturated during the dissolution stage. Precipitation of calcite is not kinetically favoured possibly due to the low slurry pH and high calcium to carbonate ion ratio. Precipitation of silica or calcium carbonate on the residual slag surface which could lead to surface passivation was not observed in this study. Thus, the overall recovery of Ca as calcium carbonate is an order of magnitude higher than the equilibrium estimate based on calcite solubility. Within the investigated parametric space, dissolution at low temperature, high-pressure and high liquid-to-solid ratios are favourable conditions for PCC production. Silica which has similar optical and ink adsorption characteristics to PCC co-precipitated with calcite, wherein further purification may not be required for its use as a filler in the paper industry. Under all conditions investigated, rhombohedral calcite was the product. The overall extraction efficiency obtained after 120 minutes of dissolution was 57% leading to a production of 535 kg of PCC per tonne of steel slag. Further extension of dissolution time can enhance the PCC production and CO₂ sequestration to 720 kg and 260 kg per tonne of steel slag, respectively.

This study shows significant scope for process-intensification of pressure-swing carbonation, especially to reduce the water requirement. Further investigations on the dissolution step to achieve high and stable calcium concentration in the leachate, up to the solubility limit of amorphous calcium carbonate, are necessary. Also, investigations to obtain various morphologies of PCC and to control the particle size distribution are necessary for taking the process a step closer to commercialisation.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The authors thank the JSW Foundation, India for funding the project. The authors also thank Dr. Suman Majumdar of the JSW Foundation for providing the steel slag sample and for suggestions during collegial discussions.

Notes and references

1. W. J. J. Huijgen, G.-J. Witkamp and R. N. J. Comans, Environ. Sci. Technol., 2005, 39, 9676–9682.
2. A. Sanna, M. Uibu, G. Caramanna, R. Kuusik and M. M. Maroto-Valer, Chem. Soc. Rev., 2014, 43, 8049–8080.
3. S. Teir, S. Eloneva and R. Zevenhoven, Energy Convers. Manage., 2005, 46, 2954–2979.
4. H. Shen and E. Forssberg, Waste Manage., 2003, 23, 933–949.
5. Z. J. Wu, B. F. Jiang, W. M. Liu, F. B. Cao, X. R. Wu and L. S. Li, Ind. Eng. Chem. Res., 2011, 50, 13778–13788.
6. K. S. Lackner, D. P. Butt and C. H. Wendt, Energy Convers. Manage., 1997, 38, S259–S264.
7. S. N. Lekakh, C. H. Rawlins, D. G. C. Robertson, V. L. Richards and K. D. Peaslee, Metall. Mater. Trans. B, 2008, 39, 125–134.
8. R. M. Santos, D. François, G. Mertens, J. Elsen, T. V. Gerven and T. Van Gerven, Appl. Therm. Eng., 2013, 57, 154–163.
9. L. Marini, Geological Sequestration of Carbon Dioxide - Thermodynamics, Kinetics, and Reaction Path Modeling, Elsevier, 2006, vol. 11, pp. 27–51.
10. R. M. Santos, M. Bodor, P. N. Dragomir, A. G. Vraciu, M. Vlad, T. V. Gerven and T. Van Gerven, Miner. Eng., 2014, 59, 71–81.
11. J. Mulopo, M. Mashego and J. N. Zvimba, Water Sci. Technol., 2012, 65, 2236.
12. S. Van der Zee and F. Zeman, J. Cleaner Prod., 2017, 174, 718–727.
13. S. Eloneva, S. Teir, J. Salminen, C.-J. Fogelholm and R. Zevenhoven, Energy, 2008, 33, 1461–1467.
14. J. Baldyga, M. Henczka and K. Sokolnicka, Chem. Eng. Res. Des., 2011, 89, 1841–1854.
15. H. Zhao, Y. Park, D. H. Lee and A.-H. A. Park, Phys. Chem. Chem. Phys., 2013, 15, 15185–15192.
16. S. Kodama, T. Nishimoto, N. Yamamoto, K. Yogo and K. Yamada, Energy, 2008, 33, 776–784.
17. A. Iizuka, M. Fujii, A. Yamasaki and Y. Yanagisawa, Ind. Eng. Chem. Res., 2004, 43, 7880–7887.
18. Y. Katsuyama, A. Yamasaki, A. Iizuka, M. Fujii, K. Kumagai and Y. Yanagisawa, Environ. Prog., 2005, 24, 162–170.
19. D. Daval, I. Martinez, J. Corvisier, N. Findling, B. Goffé and F. Guyot, Chem. Geol., 2009, 265, 63–78.
20. M. F. Bertos, X. Li, S. J. R. Simons, C. D. Hills and P. J. Carey, Green Chem., 2004, 6, 428–436.
21. N. L. Ukwattage, P. G. Ranjith and S. H. Wang, Energy, 2013, 52, 230–236.
22. N. L. Ukwattage, P. G. Ranjith and X. Li, Measurement, 2017, 97, 15–22.
23. J. D. Rodriguez-Blanco, S. Shaw, P. Bots, T. Roncal-Herrero and L. G. Benning, J. Alloys Compd., 2012, 536, S477–S479.
24. M. M. Reddy and G. H. Nancollas, J. Cryst. Growth, 1976, 35, 33–38.
25. T. Chen, A. Neville and M. Yuan, Chem. Eng. Sci., 2006, 61, 5318–5327.
26. R. Sander, Atmos. Chem. Phys., 2015, 15, 4399–4981.
27. R. R. Tamilselvi Dananjayan, P. Kandasamy and R. Andimuthu, J. Cleaner Prod., 2015, 112, 4173–4182.
28. M. Bodor, R. M. Santos, L. Kriskova, J. Elsen, M. Vlad and T. Van Gerven, Eur. J. Mineral., 2013, 25, 533–549.
29. C. F. Conrad, G. A. Icopini, H. Yasuhara, J. Z. Bandstra, S. L. Brantley and P. J. Heaney, Geochim. Cosmochim. Acta, 2007, 71, 531–542.
30. E. A. Gorrepati, P. Wongthahan, S. Raha and H. S. Fogler, Langmuir, 2010, 26, 10467–10474.
31. D. A. Crerar, E. V. Axtmann and R. C. Axtmann, Geochim. Cosmochim. Acta, 1981, 45, 1259–1266.
32. D. J. Belton, O. Deschaume and C. C. Perry, FEBS J., 2012, 279, 1710–1720.
33. R. K. Iler, J. Colloid Interface Sci., 1975, 53, 476–488.
34. B. Terry, Hydrometallurgy, 1983, 10, 135–150.
35. G. R. Holdren and R. A. Berner, Geochim. Cosmochim. Acta, 1979, 43, 1161–1171.
36. K. Lívanský, Folia Microbiol., 1982, 27, 55–59.
37. L. Brečević and A. E. Nielsen, J. Cryst. Growth, 1989, 98, 504–510.
38. T. Ogino, T. Suzuki and K. Sawada, Geochim. Cosmochim. Acta, 1987, 51, 2757–2767.
39. J. Rieger, T. Frechen, G. Cox, W. Heckmann, C. Schmidt and J. Thieme, Faraday Discuss., 2007, 136, 265.
40. N. Spanos and P. G. Koutsoukos, J. Phys. Chem. B, 1998, 102, 6679–6684.
41. L. Z. Lakshtanov and S. L. S. Stipp, Geochim. Cosmochim. Acta, 2010, 74, 2655–2664.
42. J. R. Clarkson, T. J. Price and C. J. Adams, J. Chem. Soc., Faraday Trans., 1992, 88, 243.
43. K. Larsen, K. Bechgaard and S. L. S. Stipp, Geochim. Cosmochim. Acta, 2010, 74, 2099–2109.
44. M. Kellermeier, F. Glaab, R. Klein, E. Melero-García, W. Kunz and J. M. García-Ruiz, Nanoscale, 2013, 5, 7054.
45. E. Chibowski, L. Hotysz and A. Szcze's, Colloids Surf., A, 2003, 222, 41–54.
46. Y. Kitano, M. Okumura and M. Idogaki, Geochem. J., 1979, 13, 253–260.
47. G. S. Pappas, P. Liatsi, I. A. Kartsonakis, I. Danilidis and G. Kordas, J. Non-Cryst. Solids, 2008, 354, 755–760.
48. R. Gill, Paper Chemistry, Springer Netherlands, Dordrecht, 1st edn, 1995, ch. Fillers, pp. 1–8.
49. Y. Abe, A. Iizuka, H. Nagasawa, A. Yamasaki and Y. Yanagisawa, Chem. Eng. Res. Des., 2013, 91, 933–941.
50. W. H. Casey, J. Colloid Interface Sci., 1991, 146, 586–589.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c8re00167g

‡ Experimental design to achieve rapid CO₂ saturation is presented in section 3.2.1.

§ In the current study, the reaction slurry is undersaturated with respect to MgCO₃ and Fe²⁺ leaching is negligible at the operating CO₂ pressure.

¶ Standard practice to store the sample for ICP analysis.

---