**1. Introduction**

The increasing greenhouse gas emissions and particularly anthropogenic carbon dioxide (CO2) in the atmosphere are known to play a major role in climate change [1]. Mitigation solutions are needed more than ever. Among the methodologies proposed for mitigation, mineral carbonation appears to be one of the most sustainable [2,3]. This natural and spontaneous phenomenon involves the reaction between CO2 (aqueous or gas) and divalent cations bearing minerals in order to form the associate carbonates [3]:

Equation (1) Carbonation reaction [4]

$$\rm{x(Mg, Ca)\_xSi\_yO\_{x+2y} + xCO\_2 = >x(Mg, Ca)CO\_3 + yCO\_2} \tag{1}$$

The reaction products are stable and inert solids where CO2 is sequestered. The composition of the resulting carbonates depends on the major cations present in the reactant mineral [5]. Carbonation reaction can be divided in three main steps: (i) the CO2 dissolution in water (ii) the material dissolution and (iii) the precipitation of carbonates as final products. The process is essentially controlled by the first two steps [6]. Serpentine minerals, due to their high amount of Mg2<sup>+</sup> [7] are considered for carbonation [8]. Thermal treatment acts on serpentine dissolution by enhancing Mg2<sup>+</sup> availability, making it a key step for the process [9]. Serpentine dissolution first results in a rapid exchange of surfacing Mg2<sup>+</sup> with protons (H+) before being extracted from the structure into the solution, during a much slower phase [10,11]. The dissociation of CO2 added to the solution will generate protons and HCO3 - ions, therefore enhancing Mg2<sup>+</sup> availability (Pasquier et al., 2014b).

Lizardite, antigorite, and chrysotile are the main minerals of the serpentine group (Mg3Si2O5(OH)4), belonging to the phyllosilicate class [7,12–14]. Serpentine structure is made of stacked layers composed of two sheets: the tetrahedral layer composed of silicon tetrahedral (SiO4), linked to the lateral Mg of the octahedral layer by its apical oxygen atoms, forming a covalent bond [14,15]. Outer hydroxyl groups contribute to Van der Waals interactions between the two layers, whereas inner hydroxyl groups contribute to intrafoliar Van der Waals interactions [15–17].

Under high temperatures, hydroxyl groups, linked to Mg atoms, escape the structure. During this dehydroxylation process, serpentine transformed into amorphous phases (between 550 and 750 ◦C—Equation (2)), and then recrystallized into forsterite (Mg2SiO4 > 750 ◦C), associated with enstatite (MgSiO3 > 800 ◦C) as the temperature increased (Equation (3)) [18–20]. Two types of amorphous phases have been described [21]: pseudo-amorphous phases, named α-meta-serpentine, appearing at 50% of the total dehydroxylation reaction, and amorphous meta-serpentine, appearing at 90% of the total dehydroxylation. The formation of αmeta-serpentine component can be observed at a temperature close to 580 ◦C visualized on a diffractogram by a feature in the lower angle domain (2θ = ± 6◦) [21].

Equations (2) and (3): Serpentine dihydroxylation

$$\mathrm{Mg\_3Si\_2O\_5(OH)\_4(s)} \to \mathrm{Mg\_3Si\_2O\_7(s)} + 2\mathrm{H\_2O}\_{(g)}\tag{2}$$

$$2\text{Mg}\_3\text{Si}\_2\text{O}\_{7\text{(s)}} + \text{SiO}\_{2\text{(s)}} \rightarrow 3\text{Mg}\_2\text{SiO}\_{4\text{(s)}} + \text{MgSiO}\_{3\text{(s)}} + \text{SiO}\_{2\text{(s)}}.\tag{3}$$

It has been observed that amorphous meta serpentine tends to promote Mg2<sup>+</sup> leaching and thus carbonation [21–23]. Therefore, optimized conditions for carbonations have been prescribed to be between 630 ◦C and 650 ◦C for 30 to 120 min [22,24,25]. However, in the previous studies, carbonation reactions have essentially been performed using pure CO2 gas at high temperature and high pressure [21,25–27], strong acids or salts to promote dissolution [22,28]. To date no studies have been conducted on optimizing thermal activation from the mineralogical point of view, especially for direct aqueous mineral carbonation using diluted gas. In these conditions, serpentine dissolution is only promoted by carbonic acid at room temperature and low/mild CO2 partial pressure and a good activation is more than ever critical for reaction.

This study is part of the follow-up work on direct flue gas carbonation process initiated by Mercier et al. at INRS, Québec [29]. Using mining residues available in the Province of Québec, the process uses a simulated cement plant flue gas to perform direct flue gas aqueous carbonation [30]. Carbonation reaction parameters have been optimized by Pasquier [31], optimized conditions for the precipitation of carbonates have been determined by Moreno [32] whereas a technical and economical evaluation of the process have shown its feasibility and sustainability in the Province of Québec [33]. However, a pilot scale test revealed that thermal treatment conditions needed to be optimized for the INRS process as well [34,35].

In the present paper, only the proportion of magnesium prior to precipitation will be studied and considered as an intermediate product of the carbonation, as thermal activation can only acts on enhancing serpentine dissolution. Therefore, post-carbonation solids were not considered in the present study for the given reasons. Furthermore, it serves to give a novel approach of evaluating the influence of amorphous phases on serpentine dissolution and thus Mg2<sup>+</sup> leaching during direct flue gas aqueous mineral carbonation by introducing a new quantifying method of those phases. Those new mineralogical data will provide a further understanding of the relation between thermal activation and serpentine dissolution and therefore, improve this step in the INRS carbonation process.

#### **2. Materials and Methods**

#### *2.1. Sample Preparation, Characterization and Analytical Methods*

Serpentinite residues were sampled on stockpiles from Jeffrey Mine, near the town of Asbestos, southern of the Province of Québec. Lizardite is identified to be the major serpentine polymorph [36]. Fibres, representing 20 wt % of the residue, were removed by gravimetric separation based on their buoyancy. Iron oxides were also removed by wet gravimetric separation using the Wifley table, due to their potential commercial value for the process.

The sample was ground using a ring mill (Retsch RS200, Dusseldorf, Germany). The grain size distribution is given in Table 1 Values were obtained and measured using a particle size distribution analyzer (LA-950V2 Horiba, Kyoto, Japan).


**Table 1.** Size distribution of the sample.

The chemical composition of the starting material is given in Table 2. The chemical composition of liquid and solid samples was obtained using inductively coupled plasma-atomic emission spectrometry (ICP-AES) analysis (Varian, Palo Alto, CA, USA). Solid samples were first fused using the Claisse Method [37]. Loss on ignition (LOI) was obtained from mass difference after placing the sample into a ceramic crucible inside a muffle furnace for 6 h at 1025 ◦C.


**Table 2.** Composition of the raw solid feedstock 1.

<sup>1</sup> Major compounds only.

Phases were identified using XRPD analysis (Bruker AXS, 2004, Karlsruhe, Germany), performed at the University of British Columbia. To prepare the sample, 1.6 g were mixed with 0.4 g of pure corundum (Al2O3) [38], used as an internal standard, representing a 20.0 wt % spike. Samples were ground in ethanol using agate grinding pellets for seven minutes, in a McCrone micronizing mill to ensure homogenization. Scans were acquired for 30 min with 2θ ranging from 3◦ to 80◦ with scanning step size of 2θ = 0.3◦ with a counting time of 7 s per step, on a Siemens D5000 Bragg-Brentano θ-2θ diffractometer (Bruker AXS, 2004, Karlsruhe, Germany) with radiation CuKα (40kV, 40mA). Matches were obtained using Bruker identification software DIFFRACplus EVA and the ICDD PDF-2 database.

Quantification of phases was performed using the Rietveld method ([39–42]) based on a calibration factor obtained from the mass and volume of each phase's unit cell. However, this method requires that all of the phases show high degrees of crystallinity with well-defined crystal structures [42]. Serpentine minerals are known to show discrepancies from their ideal crystal structures [38,43]. Therefore, when the crystalline structure of a phase is unknown or partially known, it can be quantified through the use of the Partial Or No Known Crystalline Structure method (PONKCS), combined with the Rietveld method [44]. A standard sample of pure chrysotile (90.0 wt %) and fluorite (10.0 wt %), provided by The University of British Columbia (Vancouver, British Columbia, Canada)., whose composition is well known [38] was used in order to calibrate the PONKCS model. A calibrated mass value for the unit cell of both phases was acquired by Rietveld refinements and the chrysotile peaks were fitted using the Le Bail method [45]. The unit cell parameters and the space group were extracted from Falini [46]. The generated PONKCS model was then used in the Rietveld refinements as a crystallographic information files in the software TOPAS (Bruker AXS) [44,47].
