CO₂ Absorption and Magnesium Carbonate Precipitation in MgCl₂–NH₃–NH₄Cl Solutions: Implications for Carbon Capture and Storage

Abstract

CO₂ absorption and carbonate precipitation are the two core processes controlling the reaction rate and path of CO₂ mineral sequestration. Whereas previous studies have focused on testing reactive crystallization and precipitation kinetics, much less attention has been paid to absorption, the key process determining the removal efficiency of CO₂. In this study, adopting a novel wetted wall column reactor, we systematically explore the rates and mechanisms of carbon transformation from CO₂ gas to carbonates in MgCl₂–NH₃–NH₄Cl solutions. We find that reactive diffusion in liquid film of the wetted wall column is the rate-limiting step of CO₂ absorption when proceeding chiefly through interactions between CO₂(aq) and NH₃(aq). We further quantified the reaction kinetic constant of the CO₂–NH₃ reaction. Our results indicate that higher initial concentration of NH₄Cl ($\ge 2\mathrm{mol}·{\mathrm{L}}^{-1}$) leads to the precipitation of roguinite [${\left({\mathrm{NH}}_4\right)}_2\mathrm{Mg}{\left({\mathrm{CO}}_3\right)}_2·4{\mathrm{H}}_2\mathrm{O}$], while nesquehonite appears to be the dominant Mg-carbonate without NH₄Cl addition. We also noticed dypingite formation via phase transformation in hot water. This study provides new insight into the reaction kinetics of CO₂ mineral carbonation that indicates the potential of this technique for future application to industrial-scale CO₂ sequestration.

1. Introduction

1.1. CO₂ Mineral Sequestration

Among the currently recognized strategies for Carbon Capture and Storage (CCS), sequestering atmospheric CO₂ in mineral forms, as first introduced by Seifritz [1], is the only known approach to permanent CO₂ storage. Mineral carbon sequestration mainly uses naturally abundant material resources like Mg-silicates [2,3] to react with atmospheric CO₂ to form carbonate minerals. Compared to other options (e.g., storing CO₂ in geological reservoirs like bedrock), mineral carbon sequestration has two major advantages: nearly unlimited raw material supplement, and relatively longer-term storage (>1 Myr, over geological timescales) [4]. However, at the Earth’s surface conditions, the rate of the silicate-CO₂ reaction, or carbonation, is quite low (~10⁻¹⁰ to 10⁻¹⁷ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$) [5]. To accelerate the reaction rate, several strategies have been tested, including increasing the ambient temperature and CO₂ pressure, enhancing the reactive surface areas of silicates [6,7,8], and removing the passivating layer on minerals [2,3,9,10]. Although those procedures show promise, they are often energy-intensive, making the carbonation process expensive and less practical [11].

The last decades have seen significant advances in the development of techniques aiming at economic and energy-efficient mineral carbon sequestration. Many groups have made efforts to accelerate silicate-CO₂ reaction rates. Notably, a pH-swing reaction path, which separated divalent ion leaching and carbonation process, was developed and proven to effectively enhance the mineral dissolution rate and optimize carbonate precipitation [12,13]. Extensive studies have focused on the mineral dissolution process. Among the many materials tested, magnesium-rich materials are thought to be good candidates for carbonation due to their natural abundance [14] and relatively higher unit mass binding efficiency (e.g., as compared to calcium). For example, physical activation (grinding [15,16,17] or heating [18,19]), chemical leaching (acid [20,21,22], recyclable ammonium salt [23,24] or non-acidic ionic solutions [25]), bioleaching [26,27] were applied to Mg rich minerals to dissolve and concentrate Mg. However, the other CO₂ absorption and carbonate precipitation steps have been less explored. Only a few studies have focused on testing the reactive absorption and precipitation using materials extracted from natural minerals. Soluble barium salt was used to absorb CO₂ via reactive crystallization [28,29,30], but was limited by the low natural abundance and toxicity. Ca(OH)₂ was also a possible absorbent [31,32], but Ca(OH)₂ solid is usually produced from CaCO₃ with CO₂ as a byproduct, negating its sequestering function. Reactive absorption of CO₂ in alkaline MgCl₂ solutions [33] or Mg(OH)₂ slurries successfully precipitated nesquehonite (MgCO₃·3H₂O), a common mineral, at Earth’s surface conditions [34,35]. Yet, the reaction kinetics of CO₂ absorption and carbonate precipitation in Mg solutions remains insufficiently understood. Specifically, few studies have investigated the CO₂ mass transfer processes on the gas-liquid interface, or how precipitated minerals vary under elevated ammonium nitrogen concentrations apart from temperature, duration, and Mg/Ca ratio [36,37,38].

To fill the knowledge gap, we systematically investigated the reaction kinetics and mass transfer process in CO₂–MgCl₂–NH₃–NH₄Cl solutions, as a case study to understand CO₂ absorption and precipitation in Mg solutions. Our work combined theoretical modeling of solution chemistry and laboratory experiments. We carefully designed and optimized the solution recipes via geochemical modeling. We built a novel wetted wall column device and conducted a series of controlled experiments of CO₂ absorption and precipitation. We calculated reaction kinetic parameters and discussed the potential controlling factors. We studied how the precipitation products change under different conditions, and developed original methods to obtain industrially useful Mg carbonates. Our work contributes to a better understanding of the CO₂ absorption and precipitation processes in industrially practical systems (e.g., CO₂–Mg solutions), and provides valuable information for the design of absorption devices and optimizing absorbent composition.

1.2. Chemistry and Kinetics of the CO₂–MgCl₂–NH₃–NH₄Cl System

A brief introduction to the aqueous phase reactions after gaseous CO₂ diffusing into the solution, together with the associated kinetic equations, is given below. After entering the aqueous phase, CO₂ reacts with water, hydroxide ions, and ammonia simultaneously [39,40,41]. This process can be expressed by the following chemical reactions:

$${\mathrm{CO}}_2\left(\mathrm{aq}\right)+{\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{HCO}}_3^{-}+{\mathrm{H}}^{+}$$

(1)

$${\mathrm{CO}}_2\left(\mathrm{aq}\right)+{\mathrm{OH}}^{-}\leftrightarrow {\mathrm{HCO}}_3^{-}$$

(2)

$${\mathrm{CO}}_2\left(\mathrm{aq}\right)+2{\mathrm{NH}}_3\left(\mathrm{aq}\right)\leftrightarrow {\mathrm{NH}}_2{\mathrm{COO}}^{-}+{\mathrm{NH}}_4^{+}$$

(3)

Equation (1) combines the hydration of CO₂ with rapid dissociation to ${\mathrm{HCO}}_3^{-}$ and ${\mathrm{H}}^{+}$, and its reaction rate can be written as:

$$r_{{\mathrm{CO}}_2-{\mathrm{H}}_2\mathrm{O}}=k_{{\mathrm{H}}_2\mathrm{O}}^{\text{'}}{C}_{{\mathrm{CO}}_2}$$

(4)

where $r_{{\mathrm{CO}}_2-{\mathrm{H}}_2\mathrm{O}}$ is the rate of reactions between CO₂ and water, $k_{{\mathrm{H}}_2\mathrm{O}}^{\text{'}}$ is the rate constant, and $C_{{\mathrm{CO}}_2}$ represent the concentration of CO₂. The temperature (T, in Kelvin) dependence of rate constant ($k_{{\mathrm{H}}_2\mathrm{O}}$, in ${\mathrm{s}}^{-1}$) has been determined previously [42] as:

$$\log \left(k_{{\mathrm{H}}_2\mathrm{O}}^{\text{'}}\right)=329.850-110.541\log \left(T\right)-\left(\mathrm{17,265.4}/T\right)$$

(5)

Under experimental conditions (T = 298.15 K), $k_{{\mathrm{H}}_2\mathrm{O}}^{\text{'}}$ is 0.026 ${\mathrm{s}}^{-1}$. Similarly, the reaction rate constant of Equation (2), given in ${\mathrm{m}}^3·{\mathrm{mol}}^{-1}·{\mathrm{s}}^{-1}$, is

$$\log \left(k_{{\mathrm{OH}}^{-}}\right)=10.635-\left(2895/T\right)$$

(6)

which is 8.41 ${\mathrm{m}}^3·{\mathrm{mol}}^{-1}·{\mathrm{s}}^{-1}$ at 298.15 K [42].

For the reaction (Equation (3)) between ${\mathrm{CO}}_2\left(\mathrm{aq}\right)$ and ${\mathrm{NH}}_3\left(\mathrm{aq}\right)$, a multi-step zwitterion mechanism is often used [40]: A CO₂ molecule reacts with an ammonia molecule to form a zwitterion, then gets deprotonated by bases (e.g., NH₃, OH⁻, and H₂O) in solution to form carbamate. For the carbamate, a hydrolytic reaction may occur to convert carbamate to bicarbonate. The detailed reaction paths of Equation (3) are:

$${\mathrm{CO}}_2\left(\mathrm{aq}\right)+{\mathrm{NH}}_3\stackrel{k_2/k_{-1}}{\leftrightarrow } {}_{}^{\oplus }\mathrm{N}{\mathrm{H}}_3{\mathrm{COO}}^{\ominus }$$

(7)

$${}_{}^{\oplus }\mathrm{N}{\mathrm{H}}_3{\mathrm{COO}}^{\ominus }+\mathrm{Base} \stackrel{{\mathrm{k}}_{\mathrm{b}}}{\leftrightarrow }{\mathrm{NH}}_2{\mathrm{COO}}^{-}+{\mathrm{BaseH}}^{+}$$

(8)

$${\mathrm{NH}}_2{\mathrm{COO}}^{-}+{\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{NH}}_3+{\mathrm{HCO}}_3^{-}.$$

(9)

Combining Equations (3) and (7)–(9), the rate expression for Equation (3) can be obtained as:

$$r_{{\mathrm{CO}}_2-{\mathrm{NH}}_3}=\frac{C_{{\mathrm{CO}}_2}{C}_{{\mathrm{NH}}_3}}{\left(\frac{1}{k_2}\right)+\left(\frac{k_{-1}}{k_2}\right)\left(\frac{1}{{{\displaystyle \sum }}^{\text{}}\mathrm{b}{k}_{\mathrm{b}}{C}_{\mathrm{B}}}\right)},$$

(10)

where b represents any base in the solution and $C_{\mathrm{B}}$ is their concentrations.$k_2$ and $k_{-1}$ are the forward and backward rate constant of Equation (7), respectively. In this case, a base can be a water molecule, hydroxide ions, or ammonia. Replacing the base using the specific molecules, Equation (10) can be rewritten as:

$$r_{{\mathrm{CO}}_2-{\mathrm{NH}}_3} = \frac{C_{{\mathrm{CO}}_2}{C}_{{\mathrm{NH}}_3}}{\left(\frac{1}{k_2}\right)+\left(\frac{k_{-1}}{k_2}\right)\left(\frac{1}{{{\displaystyle \sum }}^{\text{}}{k}_{b,{\mathrm{H}}_2\mathrm{O}}{C}_{{\mathrm{H}}_2\mathrm{O}}+k_{\mathrm{b},{\mathrm{OH}}^{-}}{k}_{{\mathrm{OH}}^{-}}+k_{\mathrm{b},{\mathrm{NH}}_3}{C}_{{\mathrm{NH}}_3}}\right)}$$

(11)

Note that in this work, ammonia is introduced as a base to bind hydrogen ions and buffer pH change.

$${\mathrm{NH}}_3+{\mathrm{H}}^{+}\leftrightarrow {\mathrm{NH}}_4^{+}$$

(12)

The bicarbonate resultant from Equations (1), (2), and (9) transforms into carbonate and carbonic acid through

$${\mathrm{HCO}}_3^{-}\leftrightarrow {\mathrm{CO}}_3^{2-}+{\mathrm{H}}^{+}$$

(13)

$${\mathrm{HCO}}_3^{-}+{\mathrm{H}}^{+}\leftrightarrow {\mathrm{H}}_2{\mathrm{CO}}_3$$

(14)

When ${\mathrm{Mg}}^{2+}$ is present, various forms of magnesium carbonates can precipitate from saturated solution [43,44,45]. The anhydrous form of Mg-carbonate, magnesite (${\mathrm{MgCO}}_3$), is usually difficult to precipitate under ambient conditions, but relatively easier under elevated CO₂ pressure and temperature (>90 °C), likely caused by the high hydration energy of Mg²⁺ ions in aqueous solution [46]. A highly hydrated Mg-carbonate, lansfordite (${\mathrm{MgCO}}_3·5{\mathrm{H}}_2\mathrm{O}$), appears near the freezing point of water (4~9 °C). Other hydrated Mg-carbonates, such as nesquehonite (${\mathrm{MgCO}}_3·3{\mathrm{H}}_2\mathrm{O}$), hydromagnesite (${\mathrm{Mg}}_5{\left({\mathrm{CO}}_3\right)}_4{\left(\mathrm{OH}\right)}_2·4{\mathrm{H}}_2\mathrm{O}$), and dypingite (${\mathrm{Mg}}_5{\left({\mathrm{CO}}_3\right)}_4{\left(\mathrm{OH}\right)}_2·5{\mathrm{H}}_2\mathrm{O}$), can form within wider ranges of temperature and pressure. Nesquehonite is a common magnesium carbonate precipitated under experimental conditions at room temperature and CO₂ pressure due to its kinetic advancement [47]. However, the natural occurrence of nesquehonite in large deposition is rather rare compared with other magnesium carbonates, such as hydromagnesite [48], magnesite [49,50], etc. Fresh nesquehonite can be seen under low temperature in a natural environment (e.g., [51,52,53]). Although it is not abundant, field and experimental evidence shows that nesquehonite could act as a precursor of more abundant hydromagnesite or magnesite [43,50,54], and dypingite is a transitory phase during this transformation [36,55]. Temperature, fluid pH, and CO₂ partial pressure are important parameters of phase transformation [55,56]. A generalized reaction equation representing the precipitation of these Mg-carbonates can be written as:

$$\left(x+y\right){\mathrm{Mg}}^{2+}+x{\mathrm{CO}}_3^{2-}+2y{\mathrm{OH}}^{-}+z{\mathrm{H}}_2\mathrm{O}\leftrightarrow x{\mathrm{MgCO}}_3·y\mathrm{Mg}{\left(\mathrm{OH}\right)}_2·z{\mathrm{H}}_2\mathrm{O}\left(\mathrm{s}\right).$$

(15)

In solutions with more complex compositions, magnesium can form a crystalline double carbonate with the alkali metals in the form of ${\mathrm{R}}_2{\mathrm{CO}}_3·{\mathrm{MgCO}}_3·4{\mathrm{H}}_2\mathrm{O}$, where R represents ${\mathrm{K}}^{+}$, ${\mathrm{Rb}}^{+}$, ${\mathrm{Cs}}^{+}$, and ${\mathrm{NH}}_4^{+}$.

2. Materials and Methods

2.1. Reagents

We chose magnesium chloride hexahydrate (MgCl₂·6H₂O) as the magnesium source material, and ammonia as the base to adjust the solution pH. Analytical graded aqueous NH₃ solution (28 wt %, produced by Nanjing Chemical Reagent Co., Ltd., Nanjing, China), was used to adjust the solution pH. Distilled deionized water produced by a High-tech^TM ultra-pure water system (resistivity >18.0 $\mathrm{M}\mathsf{\Omega }·\mathrm{cm}$) was used to wash any precipitated solid product and to dissolve analytical graded MgCl₂$·$6H₂O and NH₄Cl solids (produced by Nanjing Chemical Reagent Co., Ltd.) to make the desired solutions. Diluted HCl (0.1 $\mathrm{mol}·{\mathrm{L}}^{-1}$) solution was also used to clean the liquid system after each run. The CO₂-N₂ mixture with a specified CO₂ concentration was made by mixing pure (>99.99%) CO₂ and N₂ from gas cylinders, with their proportions regulated by flow rate controllers.

2.2. Absorption Experiments

To investigate CO₂ absorption and Mg-carbonate precipitation kinetics in detail, we adopted a wetted wall column device that was originally designed for gas-liquid reaction, and updated the device to make it suitable for the study of triple phase reaction kinetics (Figure 1). A detailed description and the testing results of the device are reported in Zhu et al. (2016) [57]. We briefly introduce the key components and functions here. The core of the column (Figure 1a) was made of a hollow stainless tube, allowing homogeneous distribution of liquid film on its outer surface, forming a uniform gas-liquid contact layer with a film thickness of 0.29 mm and a total area ($a$) of 3364 mm². The mole amount of CO₂ aborted within a certain time interval (~1 s) was calculated by measuring the difference of CO₂ flux between gas inlet ($n_i^{in}$) and outlet ($n_i^{out}$) based on the ideal gas law ($PV=n\mathrm{R}T$). A crystallizer was attached to the wetted wall column, allowing precipitation of solids and solution sampling, while solution pH was kept constant by a proportional integral (PI) control sequence managing NH₃ addition. The pH in both the wetted wall column and the crystallizer was kept constant by circulating the solution through two peristaltic pumps (E in Figure 1b).

A normal experimental run starts with purging the reaction chamber with N₂ and rinsing the liquid tubing using deionized water and the reaction solution. Subsequently, the peristaltic pumps were started up after filling the glass reactor with 660 mL of the reaction solution. Formation of homogenous liquid film was confirmed by a visual check. Next, the reaction chamber was bypassed and CO₂ was introduced into the gas line. The volumetric fraction of CO₂ was determined by a non-dispersive infrared (NDIR) gas analyzer (15.0% for all experiments). Once the solution reached set temperature, the CO₂–N₂ mixture was fed into the reaction chamber, and the pH control sequence was launched. Real-time experimental status was monitored and recorded, including all parameters of interests such as room temperature, solution pH, solution temperature, base adding rate, solution transmittance, reaction chamber temperature, and the CO₂ volumetric fraction at the gas outlet. An individual experiment runs end when the solution transmittance reaches a stable reading. Afterwards, diluted hydrochloric acid was used to wash the entire liquid line and the reaction chamber, to eliminate potential crystal seed contamination. Solution samples from experiments C, E, F, and G (

Table 1. Chemistry of starting solutions for CO₂ absorption and Mg-carbonate precipitation experiments. Reaction temperature was kept at 25 °C.

ID	Base	pH 1	${\mathit{C}}_{\mathbf{M}\mathbf{g}}^0$ 2	${\mathit{C}}_{\mathbf{N}{\mathbf{H}}_4\mathbf{C}\mathbf{l}}^0$	Duration	Mg-Carbonate Phase 3	tind 4
			$\mathbf{m}\mathbf{o}\mathbf{l}·{\mathbf{L}}^{-1}$	$\mathbf{m}\mathbf{o}\mathbf{l}·{\mathbf{L}}^{-1}$	s		s
A	NH3	9.00	0.00	0.0	16,532	N/A	-
B	NH3	9.00	0.05	0.0	22,728	Nes	12,792
C	NH3	9.00	0.10	0.0	19,117	Nes	9302
D	NH3	9.00	0.20	0.0	22,416	Nes	7899
E	NH3	8.76	0.10	0.0	20,000	Nes	11,222
F	NH3	9.16	0.10	0.0	18,380	Nes	8036
G	NH3	9.74	0.10	0.0	18,839	Nes	4185
H	NH3	9.20	0.20	2.0	28,800	Nes + Rog	-
I	NH3	9.63	0.20	6.7	19,740	Rog	-
X	NaOH	9.00	0.00	0.0	16,353	N/A	-

¹ pH indicates the solution pH value maintained by the pH controller; ² $C_{\mathrm{Mg}}^0$ and $C_{{\mathrm{NH}}_4\mathrm{Cl}}^0$ represent initial Mg and NH₄Cl concentrations; ³ “Nes” and “Rog” represent nesquehonite and roguinite [${\left({\mathrm{NH}}_4\right)}_2\mathrm{Mg}{\left({\mathrm{CO}}_3\right)}_2·4{\mathrm{H}}_2\mathrm{O}$] respectively; ⁴ t_ind represents the time lag of nesquehonite precipitation.

) were collected and filtered at certain time intervals (20–30 min). The filtrate was acidized immediately to avoid nitrogen loss due to vaporization. Total ammonium nitrogen was measured using the distillation and titration method. The total carbon (TC, in mole) absorbed into the solution was calculated using the equation below:

$$\mathrm{TC}\left(t\right)={\displaystyle \sum }_0^{i=t}\left(n_i^{in}-n_i^{out}\right)$$

(16)

where $t$ is the sampling time, and $n_i^{in}$ and $n_i^{out}$ represent the amount of CO₂ absorbed (mol) during the time interval (~1 s) in and out of the wetted wall column, respectively. Thus, the experimental absorption rate (${\phi }_{{\mathrm{CO}}_2}$) is defined as

$${\phi }_{{\mathrm{CO}}_2}=\frac{\mathrm{TC}\left(t_2\right)-\mathrm{TC}\left(t_1\right)}{\left(t_2-t_1\right)a}$$

(17)

Note that for experiments H and I, a uniform liquid film is difficult to maintain, thus the CO₂ absorption rate data of these experiments are unavailable. A pre-established CO₂ absorption baseline, which uses NaOH solution (pH = 9.00, 25 °C, experiment X) as the absorbent [57], was adopted to run parallel experiments for comparison to the previous experiments run in NH₃ solutions, for the purpose of revealing the role of NH₃ in absorption.

2.3. Mg-Carbonate Precipitation and Phase Transformation Experiments

The solid products produced after each absorption experiment were filtered and air-dried, followed by X-ray powder diffraction (XRD) characterization (Rigaku, DMX-IIIA, Tokyo, Japan). Mg concentration in the solution sample from experiment C was analyzed by an inductively coupled plasma atomic emission spectrometer (ICP-AES, Thermo Scientific, iCAP 6000, Waltham, MA, USA). The amount of Mg precipitated was then calculated by subtracting the remaining Mg in solution from the total initial Mg in system. If the precipitated Mg-carbonate was nesquehonite, the amount of carbon (mol) removed from the solution was determined as equivalent to the amount of precipitated Mg, and the dissolved inorganic carbon (DIC) was calculated by subtracting the precipitated carbon from the TC absorbed. The obtained data were then used to estimate the saturation state of nesquehonite.

If non-nesquehonite Mg-carbonate was precipitated (e.g., roguinite in this study), solid phase transformation experiments were then conducted to obtain nesquehonite or more stable basic magnesium carbonate hydrate due to their diverse industrial applications. Specifically, 5 g of air-dried solid product from experiment I (Table 1) were placed in a 100 mL flask and further washed using 50 mL distilled deionized water. The slurry was then set in a water bath and stirred for 1 h under various temperature settings (20 °C, 50 °C, and 80 °C). After washing, the mixture was filtered and characterized using XRD.

2.4. Theoretical Modeling of Solution Chemistry

Our previous studies concerning nesquehonite nucleation [58,59] demonstrated the importance of precise control of solution chemistry in avoiding Mg(OH)₂ precipitation (e.g., brucite). In the present work, we constrained solution chemistry based upon the results of equilibrium modeling. Assuming fast ionic reactions, sauturation index (SI, the logarithm of saturation degree ($\mathsf{\Omega }=\mathrm{IAP}/{\mathrm{K}}_{sp}$) to the base 10, where IAP and ${\mathrm{K}}_{sp}$ represent the ionic activity product and solubility product respectively) of both nesquehonite and brucite under various total carbon and ammonium nitrogen concentration were calculated using the PHREEQC program (a computer program for aqueous geochemical calculations). pH was allowed to drift, while Mg concentration was set at 0.10 $\mathrm{mol}·{\mathrm{L}}^{-1}$ during modeling. Such equilibrium treatment of bulk solution chemistry would introduce ~8% uncertainty according to our previous test [57]. To avoid the formation of brucite, a common co-product at high pH, and to form pure nesquehonite, the solution should be supersaturated with respect to nesquehonite and unsaturated to brucite. During CO₂ absorption, ammonia was added to adjust pH, because the solution pH was also a key factor determining precipitated minerals and would decrease with the accumulation of carbon species. As shown in Figure 2 (shaded area), once the concentrations of ammonia and carbon are carefully controlled at any given Mg concentration, pure nesquehonite can be obtained with no precipitation of brucite. Such a state can be maintained throughout the whole process of CO₂ absorption and Mg-carbonate precipitation. For reference,

Table 2. Equilibrium constants of various ionic reactions in bulk solution at 25 °C and their sources.

Equation	pK 1	Source
(1)	5.86	[60]
(2)	−8.03	Calculated 2
(3)	−3.29	[61,62]
(12)	−9.24	[63]
(13)	10.32	[64]
(14)	−6.37	[64]
(15)	−5.34	[65,66,67,68,69] 3

¹ pK is defined as –log(K); ² Calculated from the equilibrium constants of Equation (1) and pure water (pKw = 14); ³ Average value of nesquehonite solubility product in the literature.

shows the equilibrium constants of several reactions and their sources. For the other Mg-carbonate, roguinite, thermodynamic data are unavailable.

3. Results

In this section, we report our results from the experiments of CO₂ absorption under various initial Mg concentrations and pH values, and of Mg-carbonate precipitation under different ammonium concentrations.

3.1. CO₂ Absorption under Changing Mg Initial Concentration

Figure 3a shows how the total carbon absorbed (TC) evolves over time under variable initial Mg concentrations from 0.00 to 0.20 $\mathrm{mol}·{\mathrm{L}}^{-1}$. The overall trends of TC evolution with time are similar despite the varying Mg concentrations. The averaged initial absorption rate (for the first two minutes of absorption, Figure 3b) of those experiments is 8.38(74) × 10⁻⁴ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$, close to that of NaOH absorption baseline (experiment X, 7.70 × 10⁻⁴ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$). The instantaneous CO₂ absorption rate increases over time in the NH₃ system but not in the NaOH system. The averaged final absorption rate (in a span of four hours) between experiment A–E is 1.28(4) × 10⁻³ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$, which is almost double that of experiment X (also 7.70 × 10⁻⁴ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$). The increased CO₂ absorption rates over time indicate kinetic accelerations of the CO₂–NH₃–MgCl₂ system. The normalized deviations in both initial and final absorption rates over experiments A–E, 8.8% and 3.5%, respectively, are insignificant, suggesting that changing initial Mg concentration has a limited effect on the CO₂ absorption. The initial Mg concentration does seem to affect the timing of nesquehonite’s first occurrence. As shown in Table 1, no Mg-carbonate appears when Mg concentration is zero, and a higher initial Mg concentration shortens the time lag of nesquehonite precipitation (from 3.6 h at 0.05 $\mathrm{mol}·{\mathrm{L}}^{-1}$ to 2.2 h at 0.2 $\mathrm{mol}·{\mathrm{L}}^{-1}$). Another interesting observation is that experiments A–E all have similar, smooth, and continuous TC-time curves, suggesting that precipitation does not seem to lead to abrupt change in the absorption process.

3.2. CO₂ Absorption under Changing Solution pH

Figure 4a shows the TC-time curves for experiments C, E, F and G under the same initial Mg concentration ($C_{Mg}^0$ = 0.10 $\mathrm{mol}·{\mathrm{L}}^{-1}$) but different pH values (from 8.76 to 9.74). Different from Figure 3a, the TC-time curves in Figure 4a separate from each other more significantly, pointing to a stronger pH effect on CO₂ absorption kinetics.

Although these experiments share similar initial CO₂ absorption rates (averaged over the first two minutes ~8(1) × 10⁻⁴ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$, Figure 4b), those with higher pH lead to a faster four hour-averaged absorption rate (0.97 to 3.76 × 10⁻³ $\mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{s}}^{-1}$). The delay for nesquehonite precipitation also shortens with increasing pH, from 3.1 h at pH 8.76 to 1.2 h at pH 9.74 (Table 1).

3.3. Mg-Carbonate Precipitated under Changing Ammonium Concentration and Its Purification

In solutions without initial NH₄Cl addition (experiments A–G), nesquehonite is the main precipitate, as revealed by XRD (Figure 5 and Table 1). However, with higher initial NH₄Cl concentration (>2 $\mathrm{mol}·{\mathrm{L}}^{-1}$), roguinite starts to form (Figure 5a). Pure roguinite appeared when initial NH₄Cl reached 6.7 mol L⁻¹, but transformed to nesquehonite at 20–50 °C (Figure 5b) and dypingite at 80 °C when washing at such a temperature.

4. Discussion

In this section, we first simulate the temporal evolution of solution components under constant pH conditions to detail solution chemistry. We next build a model to link solution chemistry to CO₂ absorption rate, and use the model to discuss the CO₂ absorption kinetics quantitatively. Finally, we discuss Mg-carbonate precipitation kinetics and the mineral forms of the products.

4.1. Modeling of Solution Chemistry and Temporal Evolution

To characterize the dynamic nature of the reaction, the temporal evolution of primary and secondary species during experiments is simulated using the PHREEQC program. Several parameters are imported: (1) DIC concentration values, which are calculated by subtracting precipitated carbon from TC before passed to PHREEQC program. Since Mg/C ratio in nesquehonite structure is 1:1, precipitated carbon can be calculated through subtracting measured Mg concentrations from initio Mg concentration; (2) Concentrations of Cl, which are set to double the initial Mg concentration; (3) Measured Mg concentrations in solution; (4) Measured pH values; (5) With the knowledge of DIC, Mg, and Cl concentrations and pH, the concentration of total ammonium nitrogen (TN) can be calculated through charge balance. Finally, solution speciation equations are solved automatically by PHREEQC.

Figure 6 shows the modeled temporal evolution curves of the solution species, with pH maintained at 9.00 and the initial Mg concentration set as 0.10 $\mathrm{mol}·{\mathrm{L}}^{-1}$, a scenario where nesquehonite appears (experiment C). Before precipitation, the major cation in the solution is ${\mathrm{Mg}}^{2+}$, followed by ${\mathrm{NH}}_4^{+}$ (the dominant contributor to the total ammonium nitrogen) (Figure 6b,c). The main carbon species is ${\mathrm{HCO}}_3^{-}$ followed by ${\mathrm{MgCO}}_3^0$, ${\mathrm{MgHCO}}_3^{+}$, and ${\mathrm{CO}}_3^{2-}$. The main anion is ${\mathrm{Cl}}^{-}$. Nesquehonite reaches saturation rapidly, but its precipitation does not occur until ~2.5 h after the beginning of the reaction (Figure 6d). Although the modeled results also suggest that the solution became oversaturated rapidly after absorption, such a long time for the appearance of crystal indicates slow nucleation. Upon precipitation, the ${\mathrm{Mg}}^{2+}$ concentration in the solution decreases rapidly, together with ${\mathrm{HCO}}_3^{-}$, ${\mathrm{MgCO}}_3^0$ and ${\mathrm{CO}}_3^{2-}$, and ${\mathrm{NH}}_4^{+}$ becomes the dominant cation. At this stage, addition of ammonia is required to neutralize the ${\mathrm{H}}^{+}$ produced by the removal of ${\mathrm{CO}}_3^{2-}$ (Equation (13)), to maintain the pH. As the precipitation continues, the concentrations of ${\mathrm{HCO}}_3^{-}$ and ${\mathrm{CO}}_3^{2-}$ gradually recover but not the Mg-related species (Figure 6a,b). Throughout the whole course of the experiments, brucite remains undersaturated (Figure 6d), indicating a successful control of the solution composition. Based on the discussion above, the overall reaction in solution for experiments A–G appears to be:

$${\mathrm{NH}}_3+{\mathrm{Mg}}^{2+}+{\mathrm{HCO}}_3^{-}+3{\mathrm{H}}_2\mathrm{O}\leftrightarrow {\mathrm{NH}}_4^{+}+{\mathrm{MgCO}}_3·3{\mathrm{H}}_2\mathrm{O}\downarrow$$

(18)

4.2. Gas-Solution Interface Kinetics

4.2.1. Reaction Kinetics between CO₂ and NH₃

To characterize the reaction kinetics at the gas-solution interface, here we consider the physical and chemical processes of CO₂ mass transfer from gas phase to solution. After transporting CO₂ into solution, three reactions can promote the transformation of molecular CO₂(aq) to ${\mathrm{HCO}}_3^{-}$ (Equations (1)–(3)). Since the concentration of water is a constant, and the solution pH is fixed, the reaction rates of Equations (1) and (2) should be constant, i.e., CO₂ absorption rate should be stable in a pH-controlled NaOH solution (experiment X, Figure 3a and Figure 4a). In contrast, CO₂ absorption rates in NH₃ solutions exhibit an increasing trend even with a fixed solution pH. The enhanced CO₂ absorption rates in NaOH versus NH₃ solutions can be explained by the promotion of reaction kinetics, as shown by Equation (3), likely as the result of an increasing concentration of NH₃(aq) (Figure 6c).

The reaction mechanism of Equation (3) discussed in Section 1.2 yielded a complex kinetic expression (Equation (11)), with up to five rate constants. To reduce the computational complexity, the reaction is simplified to a second-order reaction. Consequently, Equation (11) is simplified to:

$$r_{{\mathrm{CO}}_2-{\mathrm{NH}}_3}=k_{\mathrm{app}}{C}_{{\mathrm{CO}}_2}{C}_{{\mathrm{NH}}_3},$$

(19)

where $k_{\mathrm{app}}$ is the apparent kinetic constant of Equation (3).

To link the CO₂ absorption rate to reaction kinetics, a “two-film” model is adopted to describe the mass transfer process. Figure 7 illustrates the concept of this model. In essence, the diffusion film on the gas side and the solution side determine the efficiency of mass transfer. In this framework, the rate of CO₂ absorption (${\phi }_{{\mathrm{CO}}_2}$) can be written as:

$${\phi }_{{\mathrm{CO}}_2}=k_{\mathrm{ov}}\Delta P_{{\mathrm{CO}}_2}\approx k_{\mathrm{ov}}{P}_{{\mathrm{CO}}_2}^{\mathrm{bulk}}$$

(20)

and

$$\frac{1}{k_{ov}} = \frac{1}{k_g} + \frac{1}{k_l},$$

(21)

where $\Delta P_{{\mathrm{CO}}_2}$ is the pressure gradient of CO₂ from bulk gas phase to bulk solution, and $k_{ov}$ is the overall mass transfer coefficient. $k_{ov}$ consisted of a gas-side ($k_g$) and solution-side ($k_l$) mass transfer coefficient, and their reciprocal forms represent mass transfer resistance at the gas side and solution side, respectively (Equation (21)). The magnitude of $k_g$ is determined by fluid properties of the gas phase, which has been calibrated by previous studies [57]. In the two-film model, an enhancement factor (E) is used to describe the mass transfer accelerated by a chemical reaction on the solution side (i.e., the solution-side mass transfer coefficient with reactions ($k_l$) versus physical mass transfer coefficient with no chemical reaction ($k_l^0$)). Methods to calculate $k_l^0$ can be found elsewhere [40].

Table 3. Solution chemistry and CO₂ absorption kinetics data for experiments C and E–G.

ID	Time	pH	TN	${\mathit{C}}_{\mathbf{N}{\mathbf{H}}_3\left(\mathbf{a}\mathbf{q}\right)}$	DIC	${\mathit{C}}_{\mathit{M}\mathit{g}}^{}$	${\mathit{k}}_{\mathit{o}\mathit{v}}$	${\mathit{k}}_{\mathit{l}}$	${\mathbf{E}}_{\mathit{e}\mathit{x}\mathit{p}}$	${\mathbf{E}}_{\mathit{c}\mathit{a}\mathit{l}}$
	s		$\mathbf{m}\mathbf{o}\mathbf{l}·{\mathbf{L}}^{-1}$				$\times {10}^{-7}\mathbf{m}\mathbf{o}\mathbf{l}·{\mathbf{m}}^{-2}·\mathbf{P}{\mathbf{a}}^{-1}·{\mathbf{s}}^{-1}$
E	406	9.146	0.007	0.002	0.003	0.103	0.51	0.52	1.38	1.50
	2206	8.771	0.017	0.003	0.008	0.102	0.44	0.45	1.19	1.50
	4006	8.761	0.027	0.004	0.015	0.102	0.44	0.45	1.19	1.53
	5806	8.786	0.041	0.007	0.022	0.101	0.58	0.60	1.57	1.60
	7606	8.755	0.058	0.009	0.030	0.101	0.58	0.60	1.57	1.68
	9406	8.715	0.059	0.009	0.041	0.105	0.58	0.60	1.57	1.64
	11,206	8.763	0.075	0.012	0.046	0.102	0.58	0.60	1.57	1.72
	12,406	8.760	0.092	0.015	0.051	0.102	0.58	0.60	1.57	1.81
	13,606	8.776	0.110	0.018	0.055	0.099	0.65	0.67	1.77	1.90
	14,806	8.784	0.109	0.018	0.060	0.099	0.65	0.67	1.77	1.87
	16,006	8.753	0.130	0.020	0.065	0.098	0.65	0.67	1.77	1.95
	17,206	8.731	0.151	0.023	0.062	0.089	0.65	0.67	1.77	2.01
	18,406	8.747	0.149	0.023	0.057	0.078	0.65	0.67	1.77	1.98
	19,606	8.746	0.212	0.033	0.054	0.068	0.65	0.67	1.77	2.25
	20,806	8.766	0.184	0.030	0.053	0.062	0.65	0.67	1.77	2.13
	22,006	8.765	0.217	0.035	0.051	0.056	0.65	0.67	1.77	2.25
C	316	9.456	0.015	0.007	0.003	0.102	0.54	0.56	1.47	1.71
	1516	9.038	0.020	0.005	0.009	0.103	0.54	0.56	1.47	1.62
	2716	8.996	0.033	0.008	0.013	0.102	0.54	0.56	1.47	1.70
	3916	9.004	0.039	0.010	0.023	0.107	0.54	0.56	1.47	1.72
	5116	8.999	0.059	0.015	0.026	0.105	0.54	0.56	1.47	1.89
	6316	8.999	0.068	0.017	0.032	0.105	0.61	0.63	1.66	1.91
	7516	9.000	0.098	0.024	0.038	0.105	0.68	0.71	1.86	2.14
	9316	8.999	0.107	0.026	0.049	0.105	0.75	0.78	2.05	2.12
	10,516	8.999	0.149	0.036	0.052	0.101	0.75	0.78	2.05	2.38
	11,716	9.000	0.156	0.038	0.049	0.092	0.82	0.86	2.25	2.36
	12,916	8.999	0.169	0.042	0.040	0.075	0.90	0.93	2.46	2.39
	14,116	9.000	0.186	0.047	0.035	0.062	0.90	0.93	2.46	2.46
	15,316	9.000	0.254	0.063	0.036	0.054	0.90	0.93	2.46	2.79
	16,516	8.998	0.255	0.064	0.037	0.048	0.82	0.86	2.25	2.77
	17,716	9.000	0.273	0.069	0.039	0.042	0.90	0.93	2.46	2.84
	18,916	9.000	0.297	0.074	0.043	0.038	0.90	0.93	2.46	2.92
F	263	9.485	0.012	0.006	0.003	0.102	0.58	0.60	1.58	1.60
	1463	9.172	0.024	0.008	0.008	0.102	0.51	0.53	1.39	1.70
	2663	9.161	0.034	0.011	0.014	0.102	0.58	0.60	1.58	1.72
	3863	9.161	0.048	0.016	0.019	0.102	0.65	0.67	1.77	1.84
	5063	9.155	0.069	0.022	0.025	0.102	0.65	0.67	1.77	2.02
	6263	9.164	0.096	0.031	0.031	0.101	0.72	0.75	1.97	2.23
	7463	9.155	0.090	0.029	0.041	0.104	0.87	0.90	2.37	2.11
	8663	9.164	0.130	0.042	0.047	0.102	0.87	0.90	2.37	2.41
	9863	9.157	0.110	0.035	0.056	0.102	0.94	0.98	2.58	2.19
	11,063	9.162	0.151	0.048	0.058	0.096	1.01	1.06	2.78	2.47
	12,263	9.154	0.145	0.046	0.056	0.084	1.08	1.14	3.00	2.38
	13,463	9.155	0.183	0.059	0.052	0.069	1.16	1.22	3.21	2.61
	14,663	9.160	0.250	0.080	0.052	0.060	1.23	1.30	3.43	2.98
	15,863	9.165	0.347	0.111	0.054	0.050	1.30	1.39	3.65	3.43
	17,063	9.158	0.341	0.108	0.059	0.044	1.38	1.47	3.87	3.37
	18,263	9.158	0.356	0.112	0.067	0.039	1.38	1.47	3.87	3.41
G	179	9.460	0.009	0.005	0.002	0.091	0.52	0.53	1.40	1.57
	1979	9.703	0.052	0.033	0.010	0.089	0.88	0.91	2.40	2.16
	3779	9.690	0.130	0.081	0.026	0.091	1.17	1.24	3.26	3.01
	4979	9.696	0.152	0.095	0.035	0.089	1.40	1.49	3.93	3.15
	6179	9.690	0.260	0.161	0.043	0.084	1.63	1.76	4.63	3.97
	7379	9.700	0.315	0.197	0.042	0.068	1.87	2.04	5.37	4.31
	8579	9.689	0.370	0.229	0.045	0.054	2.11	2.34	6.14	4.60
	9779	9.693	0.417	0.257	0.055	0.045	2.28	2.54	6.68	4.83
	10,979	9.693	0.557	0.338	0.067	0.037	2.45	2.75	7.24	5.50
	13,979	9.699	0.815	0.477	0.110	0.024	2.79	3.19	8.41	6.47

lists the experimental results of the enhancement factor (${\mathrm{E}}_{exp}$) at different sampling times. To link reaction kinetics to E, two additional parameters, the Hatta modulus (${\mathrm{M}}_{\mathrm{H}}^2$) and an enhancement factor for an infinitely fast reaction (${\mathrm{E}}_i$), are introduced below:

$${\mathrm{M}}_{\mathrm{H}}^2 = \frac{k_{total}{D}_{{\mathrm{CO}}_2,\mathrm{L}}}{{\left(k_l^0\right)}^2}$$

(22)

$$k_{total} = k_{{\mathrm{H}}_2\mathrm{O}}{C}_{{\mathrm{H}}_2\mathrm{O}} + k_{{\mathrm{OH}}^{-}}{C}_{{\mathrm{OH}}^{-}} + k_{app}{C}_{{\mathrm{NH}}_3}$$

(23)

$${\mathrm{E}}_i = 1 + \frac{D_{{\mathrm{NH}}_3,\mathrm{L}}{C}_{{\mathrm{NH}}_3}^{}{H}_{{\mathrm{CO}}_2}}{D_{{\mathrm{CO}}_2,\mathrm{L}}{P}_{{\mathrm{CO}}_2}^{int}},$$

(24)

where $D_{{\mathrm{CO}}_2,\mathrm{L}}$ is the diffusion coefficient of CO₂ in the liquid phase, $D_{{\mathrm{NH}}_3,\mathrm{L}}$ and $C_{{\mathrm{NH}}_3}$ are the diffusion coefficient and the bulk concentration of ammonia in the liquid phase, respectively, and $P_{{\mathrm{CO}}_2}^{int}$ is the partial pressure of CO₂ in the liquid phase at the gas-liquid interface. Although, $D_{{\mathrm{CO}}_2,\mathrm{L}}$ and $D_{{\mathrm{NH}}_3,\mathrm{L}}$ are inversely proportional to solution viscosity (i.e., the Stokes–Einstein equation), which is a function of solution composition and temperature. The diffusion coefficient of CO₂ and NH₃ in water at 25 °C is used as approximation of diffusion in NH₃–MgCl₂ solutions due to the limited concentration effect on viscosity [70]. To calculate the theoretical enhancement factor (${\mathrm{E}}_{cal}$), we refer to an explicit approximate expression [71]:

$${\mathrm{E}}_{cal} = -\frac{{\mathrm{M}}_{\mathrm{H}}^2}{2\left({\mathrm{E}}_i-1\right)} + \sqrt{\frac{{\mathrm{M}}_{\mathrm{H}}^4}{4{\left({\mathrm{E}}_i-1\right)}^2} + \frac{{\mathrm{E}}_i{\mathrm{M}}_{\mathrm{H}}^2}{{\mathrm{E}}_i-1} + 1}.$$

(25)

Since Equations (22)–(25) only contain one variable $k_{app}$, one can fit ${\mathrm{E}}_{cal}$ to ${\mathrm{E}}_{exp}$ by tuning this parameter. Fitting results show that ${\mathrm{E}}_{exp}$ can be reproduced satisfactorily using the models built above (Figure 8), but notable deviations exist when ${\mathrm{E}}_{exp}$ is larger than 4.

Table 4. Comparison of kinetic constants in the literature at 25 °C.

${\mathit{C}}_{\mathbf{N}{\mathbf{H}}_3\left(\mathbf{a}\mathbf{q}\right)}$	T	${\mathit{k}}_{\mathit{a}\mathit{p}\mathit{p}}$1	${\mathit{k}}_2$2	Sources
$\mathbf{m}\mathbf{o}\mathbf{l}·{\mathbf{L}}^{-1}$	°C	${\mathbf{m}}^3·\mathbf{m}\mathbf{o}{\mathbf{l}}^{-1}·{\mathbf{s}}^{-1}$
0.027–0.19	0–40	0.44	-	[72]
0.002–0.016	15–45	0.45	-	[62]
0.1–7	5–25	0.37–1.30	$12.8$	[39]
0.9–5.4	25–49	0.94–0.95	$1.23\times {10}^7$	[73]
0.6–5.6	6–31	0.95–1.14	$59.0$	[40]
0.002–0.62	25	0.32–0.95	-	This work

¹ Apparent kinetic rate constant between CO₂ and NH₃(aq) within concentration range in this work ($C_{{\mathrm{NH}}_3\left(\mathrm{aq}\right)}=0.002–0.62\mathrm{mol}·{\mathrm{L}}^{-1}$); ² Rate constant of zwitterion formation reaction (Equation (7)).

shows the fitted $k_{app}$ value together with the values compiled from other studies. The $k_{app}$ value obtained in this work is similar to but slightly higher than in earlier studies conducted at relatively low NH₃ concentrations [62,72]. Concerning the deviations in Figure 8a, it should be noted that we assume a constant $k_{app}$ in the fitting; in reality, however, $k_{app}$ may vary when conditions change. Specifically, due to accelerated deprotonating (Equation (8)), the overall rate constant between CO₂ and NH₃ (Equation (11)) is expected to correlate positively with the NH₃ concentration. Thus, $k_{app}$ should increase with NH₃ concentration. Such dependency is also shown by the observation of greater $k_{app}$ under higher NH₃ concentrations in previous work [40,73] (Table 4).

Knowing $k_{app}$’s dependence on NH₃ concentration, we now refine the calculation of the theoretical enhancement factor (${\mathrm{E}}_{cal}$) by assuming a linear relationship between $C_{{\mathrm{NH}}_3}$ and $k_{app} \left({\mathrm{m}}^3·{\mathrm{mol}}^{-1}·{\mathrm{s}}^{-1}\right)$:

$$k_{app} = 1.022 C_{{\mathrm{NH}}_3}+0.414.$$

(26)

The intercept and slope of Equation (26) are obtained through fitting ${\mathrm{E}}_{cal}$ to ${\mathrm{E}}_{exp}$ With this correction, the fit between ${\mathrm{E}}_{exp}$ and ${\mathrm{E}}_{cal}$ is significantly improved (with a smaller sum of squared errors, or SSE), especially when ${\mathrm{E}}_{exp}>4$ (Figure 8). Notably, the linear form of Equation (26) is similar to the rate constant expression of Equation (3) developed within a “termolecular mechanism” frame, where the initial product is not zwitterions but a loosely bounded termolecular complex involving CO₂, ammonia, and base molecules [73,74,75].

4.2.2. Factors Controlling CO₂ Absorption

In the two-film model, the mass transfer coefficients on the gas side ($k_g$) and the solution side ($k_l$) are important factors controlling the CO₂ absorption rate under a fixed pressure gradient. Since the gas phase fluid dynamic parameters directly determine $k_g$, the gas phase fluid conditions are kept the same ($k_g = 2.23\times {10}^{-6} \mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{Pa}}^{-1}·{\mathrm{s}}^{-1}$) between all absorption experiments. The much higher $k_g$ values, as compared to experimental $k_l$ values ($0.45-3.19\times {10}^{-7} \mathrm{mol}·{\mathrm{m}}^{-2}·{\mathrm{Pa}}^{-1}·{\mathrm{s}}^{-1}$, Table 3), indicate that the mass transfer in solution side is the slow step for CO₂ absorption in the wetted-column device. Thus, an absorption device designed with high $k_l$ values (e.g., bubble tank) should be able to promote CO₂ mass transfer through MgCl₂–NH₃–NH₄Cl solutions.

As discussed in Section 4.2.1, three reactions (Equations (1)–(3)) are contributing to CO₂ mass transfer on the solution side. Figure 9a shows the individual rate constant of these reactions in a pseudo first-order form (i.e., $k_x{C}_x=k_{{\mathrm{H}}_2\mathrm{O}}{C}_{{\mathrm{H}}_2\mathrm{O}}=k_{{\mathrm{H}}_2\mathrm{O}}^{\text{'}}$ for ${\mathrm{H}}_2\mathrm{O}$; $k_x{C}_x=k_{{\mathrm{OH}}^{-}}{C}_{{\mathrm{OH}}^{-}}$ for ${\mathrm{OH}}^{-}$; $k_x{C}_x=k_{\mathrm{app}}{C}_{{\mathrm{NH}}_3}$ for NH₃). The rate constant of Equation (1) is fixed at $0.026{\mathrm{s}}^{-1}$ for experiments C, E, F, and G as we set the temperature and water concentration to be constant. Similarly, the rate constant of Equation (2) for any specific experiment is constant, but shows a positive correlation with solution pH among the different experiments (from 0.064 ${\mathrm{s}}^{-1}$ of experiment E, to 0.615 ${\mathrm{s}}^{-1}$ of G). The rate constant of OH⁻ is ~3–30 times larger than H₂O. The rate constant of NH₃(aq) is highest, up to 431 ${\mathrm{s}}^{-1}$(Figure 9a), thus the CO₂–NH₃ reaction should be the dominant reaction among reactions (1)–(3). As discussed in Section 4.2.1, the reaction rate between CO₂(aq) and NH₃(aq) increases with NH₃(aq) concentration through enhanced reaction rate and greater apparent rate constant $k_{app}$ (Equations (19) and (26), Figure 9b). Thus, higher NH₃(aq) concentration would facilitate CO₂ mass transfer on the solution side, either through increasing the total ammonium nitrogen concentration (e.g., adding ammonium salts) or raising the pH.

Although the effect of temperature is not investigated in this work, its effect on CO₂ absorption is expected to be positive (i.e., higher temperature, faster CO₂ absorption), as shown by previous studies (e.g., [40,73]).

4.3. Mg-carbonate Precipitation and Phase Transformation

4.3.1. Nesquehonite Precipitation Kinetics

Our results show that Mg-carbonate only precipitates as nesquehonite (Table 1) in the absence of NH₄Cl. We further discuss the effect of nesquehonite precipitation on CO₂ absorption.

The initial Mg concentration does not affect the CO₂ absorption process (Figure 3) but impacts on carbonate precipitation (e.g., timing of nesquehonite’s first appearance). Figure 10 shows the amount of total absorbed carbon (as CO₂) and precipitated carbon (as nesquehonite). Although CO₂ transfers from gas phase to solution continuously, precipitation does not start until a certain time point (Figure 6d). At this stage, the solution quickly becomes supersaturated (relative to nesquehonite), overcomes the nucleation energy barrier, and starts precipitating carbonate. The initial precipitation rate of nesquehonite (${\phi }_p$) is smaller compared to the CO₂ absorption rate (${\phi }_{{\mathrm{CO}}_2}$), making the absorbed CO₂ accumulate in the solution as dissolved inorganic carbon (DIC) (Figure 10a,b). After initial precipitation, ${\phi }_p$ increases rapidly and overwhelms ${\phi }_{{\mathrm{CO}}_2}$, efficiently converting DIC to nesquehonite. By default, no further Mg is added into the reactor. As a result, the solution gradually gets less saturated, and ${\phi }_p$ becomes smaller than ${\phi }_{C{O}_2}$ again, leading to increasing DIC. The accumulation of DIC in the solution also reduces the CO₂-carbonate conversion efficiency.

Based on the above highlighted observations, we suggest that the concentration of Mg in solution needs to be controlled for enhanced CO₂-carbonate transformation efficiency. In industrial practices, timely input of concentrated-Mg solutions should be exercised to guarantee the condition of ${\phi }_p> {\phi }_{{\mathrm{CO}}_2}$. Nucleation barrier is another concern during industrialization, as the relatively long induction time of nesquehonite (1.2–3.6 h, Section 3.1 and Section 3.2) indicates quite a high nucleation energy barrier. The addition of crystal seeds might help to overcome such an energy threshold.

Besides nucleation, nesquehonite precipitation experiments conducted in the literature under similar CO₂ partial pressure by spraying gas into brucite slurry [34] exhibited a smaller precipitation rate (~0.01$\mathrm{mol}·{\mathrm{L}}^{-1}·{\mathrm{h}}^{-1}$, P_CO2 = 0.1 atm) compared to this work (~0.04 $\mathrm{mol}·{\mathrm{L}}^{-1}·{\mathrm{h}}^{-1}$, P_CO2 = 0.15 atm), indicating the kinetic advancement of our process. On the other hand, brucite carbonation under elevated pressure (15 atm) resulted in a much higher precipitation rate (~0.66 $\mathrm{mol}·{\mathrm{L}}^{-1}·{\mathrm{h}}^{-1}$) [76]. Overall, due to quite different technical settings, we could only do a first-order comparison, but further studies are needed to normalize the effects of scales and other factors (e.g., CO₂ pressure and gas-solution contact area) for a more meaningful comparison.

4.3.2. Phase Transformation of Less Stable to More Stable Carbonation Products

The chief Mg-carbonate precipitated under high TN conditions (>2 $\mathrm{mol}·{\mathrm{L}}^{-1}$) is roguinite, which was formed by milling epsomite (${\mathrm{MgSO}}_4·7{\mathrm{H}}_2\mathrm{O}$) with ${\mathrm{NH}}_4{\mathrm{HCO}}_3$ [77]. Roguinite is very unstable under ambient conditions and can lose all of the ammonium carbonate and most of the water, forming amorphous magnesium carbonate with a very high surface area (~500${\mathrm{m}}^2·{\mathrm{g}}^{-1}$) via thermal decomposition [78]. Our results show that roguinite can be transformed into nesquehonite at 50 °C and further converted to dypingite at 80 °C by washing with water. During phase transformation, the ammonium component is also removed from the solid, allowing for recycling for CO₂ absorption.

Under low TN conditions (<2 $\mathrm{mol}·{\mathrm{L}}^{-1}$), nesquehonite is the dominant Mg-carbonate precipitated besides roguinite. Nesquehonite is also unstable at ambient conditions and can gradually lose CO₂, transforming to its basic form (e.g., dypingite and hydromagnesite) at temperatures above 50 °C [36,55] or even at room temperature [34,79]. Thus, once washed in hot water (>50 °C), both nesquehonite and roguinite can be transformed into basic magnesium carbonates.

4.4. Implications for Industrial Applications

Experimental data show that liquid-side mass transfer process is the rate-limiting step of CO₂ absorption. Thus, to achieve more efficient CO₂ scrubbing, reactors with a higher liquid-side mass transfer coefficient, $k_l^0$, are needed. Various kinds of reactors are available for the CO₂ absorption depending on how the gas is in contact with the solution/slurry [80]. For aqueous droplets falling in the gas phase, i.e., spray tower, the gas phase side has a lower mass transfer barrier due to stronger gas phase convection. Thus, a spray tower is suitable for fast reaction systems, e.g., SO₂ absorption. On the other hand, for the liquid-side mass transfer limited CO₂–MgCl₂–NH₃–NH₄Cl system, letting the gas phase rise within the solution (i.e., bubbling) would guarantee a more effective liquid-side mass transfer process. Thus, a bubble tank is an ideal reactor and works better with agitation. Besides the economic advantages, a bubbled tank with agitation can handle solid precipitation without being clogged up, compared with staged tower reactors. In addition, similar to this study, parameters including NH₃ concentration, Mg concentration, and pH can also be tuned to fulfill a CO₂ emissions reduction target.

Phase transformation experiments show that an unstable form of nesquehonite or roguinite can be readily converted to stable basic-form Mg-carbonates (dypingite in this study), which are important industrial raw materials with diverse applications. Dypingite-like Mg-carbonates can serve as a flame-retardant or fire-retardant additive for polymers [81,82,83] or simply as backfill [84]. Hydromagnesite is also a promising material to replace CaO in calcium-based cement, which usually has a net release of CO₂, as opposed to Mg-based cement, which could potentially absorb nearly as much CO₂ during its service life as was emitted during its manufacture (carbon-neutral) [85].

Since the working medium used for CO₂ absorption is Mg solution in this study, a broad range of Mg-rich minerals could act as a raw material. Thus, the theoretical sequestration ability is almost unlimited. However, considering the tremendous energy that goes into harvesting Mg from silicates [11], Mg-containing brines resulting from silicate weathering might be more promising. For example, the Qinghai Lake (water capacity 119.6 km³) in western China has an extremely high Mg concentration of 16$\mathrm{g}·{\mathrm{L}}^{-1}$ [86], which could sequestrate 3.5 Gt CO₂ in total while producing 9.3 Gt hydromagnesite. In addition, a solid byproduct of potassium salt production named bischofite (MgCl₂·6H₂O ) is readily available for use. Notably, 320–480 Mt of bischofite are produced each year around Chaidam basin in China [87], equivalent to a CO₂ sequestration potential of 70–104 Mt producing 186–276 Mt hydromagnesite.

5. Conclusions

This study systematically investigated the reaction kinetics of CO₂ absorption and Mg-carbonate precipitation in MgCl₂–NH₃–NH₄Cl solutions using a novel pH-controlled wetted-wall column device. The obtained results from and conclusions about the studied system are summarized as follows:

(1)

Careful design and theoretical modeling of solution chemistry led to the precipitation of high-purity nesquehonite in NH₄Cl-free experiments.

(2)

The gas-side and liquid-side mass transfer coefficients were estimated through combining experimental data and a two-film mass transfer model. Notably, the CO₂ mass transfer resistance on the liquid side was found to be greater than the gas side, suggesting that the liquid-side reaction is the major limiting step in the overall reaction and that a higher liquid-side mass transfer coefficient should promote CO₂ absorption.

(3)

Interaction of CO₂(aq) with NH₃(aq) is a critical process controlling CO₂ mass transfer from the gas to the solid phase. Based on the zwitterion mechanism, NH₃(aq) concentration correlates closely with the apparent rate constant of the CO₂–NH₃(aq) reaction ($k_{app}$), pointing to the importance of NH₃ in the overall reaction. Together with the literature data, our results suggest that increasing temperature, pH, and NH₃(aq) concentration would facilitate CO₂ absorption.

(4)

The solid products are sensitive to ambient conditions and treatments. If the initial ammonium nitrogen concentration is low ($C_{{\mathrm{NH}}_4\mathrm{Cl}}^0$ < 2$\mathrm{mol}·{\mathrm{L}}^{-1}$), only nesquehonite would precipitate; if $C_{{\mathrm{NH}}_4\mathrm{Cl}}^0$ > 2$\mathrm{mol}·{\mathrm{L}}^{-1}$, roguinite would appear. High pH value and high Mg concentration could help nesquehonite overcome the nucleation energy barrier and facilitate carbonation. Thus, besides adjusting pH, adding dissolved Mg is likely an effective way to enhance the precipitation rate of Mg-carbonate.

(5)

Industrially useful and more stable materials could be obtained with designed treatments of the solid products. Specifically, washing nesquehonite and roguinite in hot water (>50 °C) could convert those minerals to more basic forms of Mg-carbonate (e.g., dypingite).

Overall, our work provides a more comprehensive understanding of the kinetics of a carbonation reaction in the MgCl₂–NH₃–NH₄Cl system, which would allow for the development of an applicable CO₂-minimization routine. The studied factors and their associated behavior can directly guide the design and optimization of reaction systems aiming at more efficient CO₂–Mg-carbonate conversion, and help with building facilities for capturing and sequestering atmospheric CO₂ at the industrial scale.