Direct Aqueous Carbonation of Heat-Activated Lizardite; Effect of Particle Size and Solids Loading on Magnesite Yield

Abstract

In this study, we investigated the effect of particle size and solids loading on the magnesite yield in the direct aqueous mineral carbonation of heat-activated lizardite. Experimentation was conducted under single-step reaction conditions (130 bar partial pressure of carbon dioxide (CO₂) and 150 °C, with 0.64 M sodium bicarbonate (NaHCO₃) and 15 wt% solids) as developed by the Albany Research Center (ARC). The objective of the study was to enhance the understanding of the direct aqueous mineral carbonation process in heat-activated lizardite. Furthermore, we aimed to shed light on how variations in particle size could affect the reaction rate, yield, and the development of protective silica layers. Our experimental data suggest that the extraction of magnesium from finer particles (sub 20 µm) is marginally more effective than from the larger size fractions. This difference likely stems from the larger surface area of fine particles (sub 20 µm) in both low and high solids loading experiments. The highest magnesite yield was 50% after 60 min, and this was achieved for both solids loadings (5 and 15 wt%), demonstrating that the solids loading had no impact on the yield. Our findings indicate rapid heat-activated lizardite reaction within 20 min, which achieved 34% and 40% conversion for 5 wt% and 15 wt% solids loading, respectively. This is followed by declining rates with increasing solids loading.

1. Introduction

Ex situ mineral carbonation involves capturing carbon dioxide (CO₂) and reacting it with minerals like serpentine or olivine in controlled industrial settings to form stable carbonates, effectively sequestering the CO₂. While research has advanced our understanding of reaction kinetics and optimising conditions, challenges, such as high energy requirements and costs, hinder large-scale implementation. Pilot projects have demonstrated feasibility, but further technological and economic developments are necessary for widespread adoption [1,2].

The projected high cost of ex situ mineral carbonation is one of the most significant barriers to the commercialisation of this technology. Costs ranging from USD 50 to USD 300 per ton of carbon dioxide (CO₂) have been estimated [3,4,5,6,7], and significant contributors to these costs are the energy requirements associated with thermal and mechanical activation and high capital and operating costs. In the case of using serpentine as feedstock, thermal activation requires an energy input of at least 541 MJ t⁻¹ serpentinite with 20% OH_res (degree of dehydroxylation) and with recovery of approximately 80% of sensible heat [8]. Slow reaction rates and inadequate transformation of feed material are two factors impeding the process. In one-hour carbonation experiments, the conversion rates were recorded at 62% and 40% when utilising antigorite and lizardite particles smaller than 75 µm, respectively [7].

The overall aqueous carbonation reactions of olivine and serpentine (in molecular form) are as follows [9]:

Mg₂SiO_{4 (s)} + 2CO_{2 (aq)} → 2MgCO_{3 (s)} + SiO_{2 (s)}

(1)

Mg₃Si₂O₅(OH)_{4 (s)} + 3CO_{2 (aq)} → 3MgCO_{3 (s)} + 2SiO_{2 (s)} + 2H₂O _(l)

(2)

In the direct aqueous carbonation process, the carrier solution contains sodium bicarbonate (NaHCO₃) and sodium chloride (NaCl) mixed with the fine particles of mineral silicate (serpentine or olivine) in an aqueous solution of (typically) 15 wt% solids. CO₂ is fed into the batch reactor at elevated pressures (up to 180 bar) [10]. The carbonation reaction is composed of two overall steps: the dissolution of the feed mineral and the precipitation of the Mg-carbonate product. Carbonic acid is formed due to the dissolution of CO₂ in water, with the acid dissociating to liberate protons (H⁺), which react with the mineral to release magnesium (Mg) cations. The Mg²⁺ ions can then react with carbonate ions (CO₃²⁻) to produce Mg-carbonate (MgCO₃). Both these steps occur concurrently in the ARC batch process [9].

The rate of direct aqueous carbonation may be expected to be surface area dependent, and subsequently, prior to the reaction, a grinding of the mineral is performed to increase mineral surface area. The fresh ore is normally ground to a size below 70 µm and sometimes as fine as ~4 µm [1,11,12].

Re-serpentinisation of the heat-activated material can occur as a competing reaction, depending on the reaction conditions, and this reaction is noted below [13]:

2Mg₂SiO_{4 (s)} + CO_{2 (aq)} + 2H₂O _(l) → Mg₃Si₂O₅(OH)_{4 (s)} + MgCO_{3 (s)}

(3)

Several researchers have observed the formation of a silica (SiO₂) rich layer around the surface of the reacting particles [14,15]. This layer inhibits the diffusion of Mg cations from within the mineral into the bulk solution, thus preventing the reaction with the carbonate ions [14,15]. It has been speculated that the formation of a Si-rich passivating layer is a critical impasse to the further development of the process [15]. Possibly, it is partly as a consequence of this that direct aqueous carbonation, although a promising technology, remains commercially unrealised [16].

Concurrent grinding has been shown to enhance magnesite yields from natural feedstocks, enabling olivine carbonation without pre-treatment and achieving some reactivity in untreated lizardite [17]. Zirconia and stainless-steel media were most effective, with minimal additional energy consumption required, compared to alumina [17,18]. Similarly, Yang et al. and He et al. presented an innovative wet grinding carbonation technique for the treatment of alkaline solid wastes. The proposed method achieves dual objectives: rapid carbonation of alkaline metal ions and enhanced CO₂ capture. The high efficiency in CO₂ uptake is primarily attributed to the continuous removal of the Si-rich passivating layer through wet grinding, which consistently exposes new surface areas to the reaction [19,20]. A recent study examined the effect of NaHCO₃ on the yield and product phases in the direct aqueous mineral carbonation of heat-activated lizardite under single-step conditions (130 bar partial pressure of CO₂, 150 °C). Using advanced analytical techniques, the research found that NaHCO₃ increases CO₃²⁻ concentration, thus enhancing supersaturation and promoting greater precipitation of MgCO₃. Additionally, NaHCO₃ influenced the formation of specific magnesium carbonate phases, such as hydromagnesite or magnesite, demonstrating its role in optimising reaction efficiency and product selectivity [21]. Qian et al. showed that natural minerals like kaolinite, montmorillonite, and wollastonite offer significant potential for CO₂ capture and mineralisation due to their availability, low cost, and stability. Various modification methods (e.g., acid and alkali treatments) enhance their surface area and active sites, improving CO₂ uptake. Future research aims to optimise these methods and explore composite materials, indirect mineralisation techniques, and economic gains from additive recovery and by-products [22]. Balucan et al. investigated the optimisation of thermal activation of serpentine ores for reducing energy costs for CO₂ mineralisation [23]. They identified conditions that enhance Mg availability while minimising crystallinity loss. The findings support the feasibility of using Great Serpentinite Belt resources for cost-effective CO₂ sequestration [23]. Another study examined mineral carbonation in New South Wales, using magnesium silicates for CO₂ capture [24]. Formic acid proved to be the most effective leaching agent, but acetic acid recycling was impractical. While dissolution can be optimised, carbonation remains challenging due to carbonate solubility [24]. A recent study demonstrated a method for mineral carbonation, where CO₂ removal induces Mg-carbonate precipitation without the need for alkaline agents. By utilising MgSO₄, NaHCO₃, and Scenedesmus microalgae, CO₂ degassing was accelerated, with nitrogen sparging enhancing the process [25]. The method, which involves heat-activated serpentinite, functions under mild conditions and can effectively sequester CO₂ in different streams [25].

This investigation aims to understand the mechanism of direct aqueous mineral carbonation of heat-activated lizardite by tracking the evolution of aqueous and solid species during experiments. It also examines the impact of particle size on reaction rate, yield, and the formation of passivating silica layers.

2. Material, Characterisation, and Methodology

There are several serpentinite deposits in New South Wales (NSW), Australia, with a notable concentration of serpentinite rocks located less than 200 km from multiple power stations within the state [26]. In this study, lizardite was sourced from the Great Serpentinite Belt in NSW, which stretches from the townships of Barraba to Bingara. A bulk sample of serpentinite (predominantly lizardite, designated as 1A), composed mainly of olivine and lizardite, was collected from an abandoned dunite quarry situated at Doonba Station, northeast of Barraba.

The crushed lizardite (under 10 mm) was further processed through wet grinding using a ball mill (MTI Corporation). This involved operating the ball mill for 4 h at a rotational speed of 200 rpm, with 3 kg of stainless-steel balls of various sizes (from 8 to 25 mm in diameter). The procedure typically maintained a 50% solids concentration in the ball mill, achieved by mixing 600 g of crushed lizardite with 600 mL of distilled water in a milling jar, which was filled to two-thirds of its capacity. The lizardite-rich serpentinite obtained from the Great Serpentinite Belt in New South Wales (NSW) underwent wet sieving to separate it into size fractions of 45–75 µm, 20–45 µm, and less than 20 µm. Subsequently, each sieved fraction was subjected to overnight oven-drying at 100 °C. A particle size analyser (Malvern Mastersizer 2000) was used to measure the particle size distribution of each fraction by dispersing the particles in water. To evaluate the effect of varying particle size on the surface reaction rate, small particles (less than 20 μm) were removed from the distinct fractions using a cleaning protocol. This process involved wet sieving 25 g of the ground material in a single batch through a 20 μm sieve. After sieving, the sieve was placed in an ultrasonic water bath for 5 min, and the resulting solution was analysed using the particle size analyser. If fine particles were still detected in the solution, the procedure was repeated until particles smaller than 20 μm were completely eliminated from the targeted fraction.

Different size fractions of lizardite were analysed by XRD analysis to identify and quantify the main crystalline phases present in each sample. X-ray diffraction (XRD) was used to qualitatively and quantitatively determine the crystalline phases of the different size fractions of lizardite. XRD analysis of the samples were accomplished on a Philips X’Pert Pro diffractometer (Philips Analytical, Eindhoven, The Netherlands) with Cu Kα radiation operated at a current of 40 mA and a voltage of 40 kV. Samples were continuously scanned from 5° to 90° 2θ with a step size of 0.002° or 0.008° by a goniometer (Philips Analytical, Almelo, Netherlands). XRD patterns were analysed using X’pert Highscore software, matching them with the International Center for Diffraction Data (ICDD) to identify the crystalline phases in the material based on standard intensity patterns.

Carbonation experiments were carried out using a commercially available high-pressure stainless-steel reactor (Parr 4560, 600 mL). The reactor head was equipped with a thermocouple, mechanical pressure gauge, vent needle valve, rupture relief disk, and a dip tube for sampling. A cooling tube supplied water to stabilise the reactor’s temperature, while an electric jacket, managed by a temperature controller, heated the reactor to the set temperature [24,25,26,27,28,29].

Carbonation experiments were then undertaken under Albany Research Center (ARC) conditions (130 bar CO₂ and 150 °C with 0.64 M NaHCO₃) and conducted with distinctive particle size fractions (45–75, 20–45 and sub 20 µm) with two different solids loading of 5 and 15 wt%. These experiments were conducted over a period of 1 h. The aliquots of the sample from the reacting vessel were taken after heating (AH) at t₀, t₁₀, t₂₀, t₄₀ and t₆₀, where subscript sample denotes the time (in min) from when the reaction mixture reached the desired temperature (150 °C) without any addition of CO₂. Time zero (t₀) was taken as the time when the reaction mixture was pressurised to the desired pressure (130 bar).

Inductively coupled plasma optical emission spectrometry (ICP-OES) with argon plasma was used to determine the total concentrations of Mg and silicon (Si) in the liquid samples obtained from the carbonation experiments. This technique uses high-temperature plasma to ionise and excite aqueous species, causing them to emit photons at characteristic wavelengths, which enables the identification and quantification of the elements.

A thermogravimetric analyser (TGA) (Mettler Toledo Star e System—TGA, Switzerland), combined with a mass spectrometer (Thermostar™ PFEIFFER, GSD 301 T3), was used to analyse the solid samples and quantify carbonate content in the product. This TGA-MS combination enabled the identification of exhaust gases generated from the carbonated sample. The quadrupole mass spectrometer (MS), composed of four parallel rod electrodes, was connected to the TGA through a heated quartz capillary line (up to 200 °C) to prevent gas condensation, allowing the detection of condensable gases at low concentrations. The MS detects gases within a molecular weight range of 1–300 g/mol. It measures the mass-to-charge (m/z) ratio and identifies ions in the gas phase by ionising the exhaust gas. A quadrupole field creates a static electrical field that accelerates ion transfer through the quadrupole, while a magnetic field facilitates ion separation based on their m/z ratio. Finally, an electron multiplier detects the ion intensity (ion current), which is displayed as a spectrum. Each of the representative solid samples was filtered, washed with distilled water to remove any NaHCO₃, and then oven-dried overnight at 100 °C. The dried solid samples were milled using a mortar and pestle to achieve a homogenous sample prior to TGA-DTG-MS analysis to quantify the Mg-carbonate phase and characterise the evolved gases.

3. Results and Discussion

3.1. Particle Size Analysis

In the present study, it was necessary to remove all fine particles from each sieve fraction, prior to any carbonation reaction experiments, to determine the effect of varying particle size on the surface reaction rate. This was achieved through the development and application of a fine particle (sub 20 μm) removal protocol, as mentioned in the methods section.

3.2. X-Ray Diffraction Analysis

The results of the XRD analysis are presented in Figure 1. This figure shows the main crystalline phases identified in the lizardite-rich serpentinite sample as lizardite, clinochlore, and magnetite for each of the size fractions.

The Reference Intensity Ratio (RIR) method is a practical and widely used approach for quantitative phase analysis in X-ray diffraction (XRD). It simplifies analysis by comparing phase intensities to reference standards, reducing matrix effects and requiring minimal sample preparation [30].

Quantitative analysis of the crystalline phases was undertaken by applying the Reference Intensity Ratio (RIR) method [30] and confirmed lizardite as the dominant phase present in all fractions, within an error estimated to be approximately 10%. Sources of error are related to particle size, especially as the particle size of the Si standard used is in the nanometer size range, while the material size of the lizardite is in the micrometer scale, which can lead to different RIR values to those published in the literature. The analytical results, summarised in

Table 1. Composition of crystalline phases for raw lizardite with different fractions.

	Crystalline Phases %				Amorphous Phase %
Particle Size (µm)	Clinochlore	Lizardite	Magnetite	Total Crystalline % *
45–75	2.0	98.8	3.0	103.8	0.0
20–45	1.7	107.0	3.0	111.7	0.0
<20	0.80	84.0	3.8	88.6	11.0

*: The error is influenced by particle size, particularly because the Si standard used is in the nanometer range, whereas the lizardite material is on the micrometer scale.

, also showed clinochlore and magnetite as secondary phases, which represented up to 3.75% of the mineral composition.

Each fraction of raw lizardite was heat activated under a vacuum in a rotary kiln at 630 °C for 4 h. The heat-activated samples were again analysed by XRD to identify and quantify the main crystalline phases. Figure 2 shows the XRD patterns for heat-activated lizardite for the different particle size fractions, with the main crystalline phases identified as lizardite, forsterite, clinochlore, hematite and magnetite, as listed in

Table 2. Crystalline and amorphous phases for heat-activated lizardite at different particle size fractions.

	Crystalline Phases %						Amorphous Phase %
Particle Size (µm)	Clinochlore	Lizardite	Forsterite	Magnetite	Hematite	Total Crystalline %
45–75	3.7	7.2	14.1	0.3	0.6	25.9	74.1
20–45	8.1	5.8	12.8	0.3	0.6	27.6	72.4
<20	6.8	0.9	8.7	0.2	0.6	17.2	82.8

.

It is evident from the quantitative XRD analysis results of the heat-activated lizardite listed in Table 2, the heat-activation step transforms the material into a product that is approximately 77 ± 6% amorphous for all fractions. Based on the results of XRD analysis for the heat-activated lizardite with different particle size fractions, it is reasonable to conclude that any change in reactivity observed during experimentation for the 45–75 and 20–45 µm fractions is unlikely to be the result of differences in the amount of the amorphous phase in these fractions. It can be observed that the formation of forsterite occurs as a result of this dehydroxylation process and is a subsequent step in the heat activation of lizardite.

3.3. Elemental Analysis of the Supernatant Solutions with ICP-OES

Slurry samples were taken at regular intervals to understand the reaction mechanism. Each slurry sample was filtered using a 0.22 µm syringe filter, and the filtrate acidified using 2 wt% nitric acid and analysed by ICP-OES. As seen from the results of ICP-OES analysis in Figure 3 and Figure 4 (See Supplementary Materials, Tables S1–S6 for details), a rapid rate of extraction of Mg from the heat-activated mineral was observed initially. The concentration of Mg in solution then reached a maximum value, corresponding to high levels of supersaturation for various Mg carbonate phases (hydromagnesite and magnesite). This is followed by a rapid decline in Mg concentration levels to a somewhat uniform and significantly lower concentration, suggestive of Mg-carbonate precipitation occurring. The precipitation of magnesite was confirmed by XRD, as shown in Figure 5.

Low rates of Mg extraction were observed during the heating of the solution (to 150 °C), that is, prior to time zero (t₀) and prior to the introduction of CO₂ to the reactor. It appears that Mg extraction from fine particles (sub 20 µm) is slightly higher compared with the larger size fractions (20–45 and 45–75 µm) under the ARC conditions of the carbonation reaction (130 bar CO₂, 150 °C, and 0.64 M NaHCO₃) and this is possibly due to the relatively high surface area of the fine particles (sub 20 µm) for both the low and high solids loading experiments (5 and 15 wt% solids) as shown in Figure 3. Figure 4 shows the percentage of dissolved Mg in the supernatant relative to the total Mg in the material.

It is evident from the results of ICP-OES analysis in Figure 6a,b that a rapid rate of extraction of Si from the heat-activated lizardite was observed initially, as was observed for Mg extraction (see Figure 3). The concentration of Si in solution then reached a maximum value and then, after 3 min, remained relatively constant, demonstrative of the precipitation (polymerisation) of Si from the solution to form Si-rich layer on the particles. It is hypothesised that the reaction leading to Mg-silicate formation results in the development of a single protective layer around the reacting particle. This Mg-silicate layer could impede the reaction rate by hindering the diffusion of protons to the surface of the dissolving particle [14,15].

3.4. Thermogravimetric and Mass Spectroscopic Analysis (TGA-MS)

Typical TGA-DTG-MS traces for the carbonated solid samples prepared from experiments conducted for 60 min under ARC conditions (130 bar CO₂, 150 °C, 0.64 M NaHCO₃ and 15 wt% solids) are shown in Figure 7 and Figure 8. This analysis method was used in determining the carbonate content (magnesite) in the reaction product for each particle size fraction. It can be seen from Figure 9 and

Table 3. Decomposition temperatures of Mg-carbonate product of carbonation experiments at standard conditions (130 bar CO₂ and 150 °C with 0.64 M NaHCO₃) conducted with distinctive particle size fractions.

Carbonation Experiment Undertaken with Distinctive Particle Size (µm)	Temperature at Which the Maximum Rate of Decomposition of Mg-Carbonate (°C) 5 wt%	Temperature at Which the Maximum Rate of Decomposition of Mg-Carbonate (°C) 15 wt%
45–75	489	490
20–45	493	493
Sub 20	495	497

that the thermal decomposition characteristics of the carbonated samples are very similar. According to TGA, DTG and MS traces for coarse particles (45–75 µm), the Mg-carbonate decomposed at 490 °C (peak temperature), while for the moderately sized fraction (20–45 µm), the decomposition occurred at 493 °C. With fine particles (sub 20 µm), the peak decomposition temperature of the Mg-carbonate occurred at 497 °C. This (slight) variation in the decomposition temperatures between the different fractions might be due to a different mass of the material in the TGA sample pan and varying proportions of reacted and unreacted phases in each sample. This variation may be an artefact caused by the effect of heat transfer (due to thermal conductivity differences between samples) on the mass loss curve.

From a comparison of the TGA analyses, the mass loss increases dramatically from the coarse particles (45–75 µm) compared to the fine particles (sub 20 µm Figure 7a). This could translate to a greater conversion to magnesite for the fine particles compared to the coarse particles. It is evident that grinding of the mineral can disrupt the crystalline phases and produce a large fraction of fine particles (sub 20 µm) with a relatively high surface area (typically BET area of 11 m²/g). This, in turn, leads to an increase in the rate of the carbonation reaction [7].

In order to study the influence of the mass fraction of solids of the reacting slurry on the carbonation reaction for different size fractions, carbonation experiments were performed using a reduced solids loading (5 wt% solids) under otherwise standard reaction conditions. Figure 8 shows the TGA-DTG-MS output for the carbonated solid samples, after 60 min of reaction time for each particle size fraction. A similar thermal behaviour for the recovered solids from these low solids loading experiments was observed when compared to carbonation experiments conducted at relatively high solids loading (15 wt% solids). On the other hand, TGA analysis shows that the mass loss increased dramatically with the fine particles demonstrating greater conversion of the Mg-carbonate phase (magnesite) in these samples. MS analysis of the products for the low solids loading experiment showed that the intensity of the CO₂ peaks increased for fine particles (sub 20 µm), which confirms that more magnesite was formed during the carbonation reaction for this experiment.

3.5. Magnesite Yield Calculations

Magnesite (product) is the phase observed to precipitate under the ARC conditions used in the carbonation reaction for heat-activated lizardite. The fraction of Mg precipitated (after 1 h) was calculated based on TGA-MS analysis as shown in Equation (4), which determines the amount of product (i.e., magnesite) by measuring the amount of CO₂, because of the molar equivalence of Mg and Carbon in magnesite.

$$Magnesite\mathrm{ }Yield={\displaystyle \frac{Moles\mathrm{ }of\mathrm{ }\mathrm{ }magneiste\mathrm{ }produced}{Theoretical\mathrm{ }\max imum\mathrm{ }moles\mathrm{ }of\mathrm{ }magnesite\mathrm{ }produced}}$$

(4)

The amount of magnesite produced was obtained based on the TGA mass loss between 400 and 550 °C where CO₂ was released from the sample.

Figure 9 shows the magnesite yield calculated based on the TGA-MS data using Equation 1 (See Supplementary Materials, Tables S7–S12 for details). As can be seen in this figure, the highest magnesite yield achieved was 50% after 60 min for both solids loading (5 and 15 wt%). Figure 9 shows that the reaction of heat-activated lizardite is rapid for the first 20 min, achieving 34% conversion (for sub 20 µm particle size fraction), after which point the reaction rate declines dramatically. It has been proposed by various researchers [14,15] that the decline in reaction rate is caused by the formation of a Si rich shell around a core of unreacted particle. Figure 9 summarises the effect of the solids loading on the yield percentage. This figure shows that there is little effect on the yield of precipitated magnesite for each particle size fraction. It can be noted that the magnesite yield at t₀ represents the yield prior to the introduction of CO₂ into the reactor. It can be observed in this study that the reaction kinetics exhibited a two-step pattern, which was characterised by a swift reaction rate in the initial 20-min period, yielding magnesite ranging from 20 to 50% yield followed by a drop in the reaction rate due to the formation of a Si-rich layer. It can be observed that, with a decrease in particle size, a noticeable rise in the percentage of Mg extraction occurred, resulting in a higher magnesite yield [31,32]. Following this, the reaction rate experienced a significant decrease, likely due to the formation of a Si-rich layer.

4. Conclusions

The carbonation experiments with heat-activated lizardite under ARC conditions (130 bar CO₂, 150 °C, 0.64 M NaHCO₃) examined the impact of particle size (45–75, 20–45, and sub 20 µm) on reaction rate, yield, and Si passivation layer formation. Reaction kinetics displayed a two-step pattern, an initial rapid rate for 20 min, which yielded 20–50% magnesite, followed by a significant slowdown. Yield and rate increased as particle size decreased, with a peak yield of 50% for particles sub 20 µm after 60 min. Mg concentration spiked at 4 min and then stabilised, indicating that magnesium dissolution is the rate-limiting step in the carbonation process. Solids loading (5 wt% and 15 wt%) was found to have no significant impact on the magnesite yield as both achieved similar yield values.

Results suggest magnesite precipitation is relatively quick, with Mg dissolution as the rate-limiting step. As observed by several researchers [33,34], an Mg-silicate layer forms around particles, inhibiting reaction progress and potentially affecting proton diffusion. This layer significantly slows down reaction rates over time, which emphasises the need for strategies to overcome its inhibitory effects. Implementing a wet grinding carbonisation process could offer an effective solution as it has been shown to break the passivation layer and enhance reaction efficiency [17,18].