Optimizing the Leaching Parameters of Asbestos Tailings for Maximizing the Recovery of Critical Metals

Abstract

Asbestos tailings represent a historical liability in many countries. Canada aims at transforming this industrial legacy into an opportunity to both mitigate the environmental footprint and recover critical (such as magnesium, nickel, chromium, and cobalt) and strategic metals, which represent significant economic development potential. This study aimed to investigate the recovery of critical and strategic metals (CSMs) from asbestos tailings using hydrochloric (HCl) acid leaching, with acid concentration (2–12 mol/L), leaching temperature (20–90 °C), and solid–liquid ratio (10–40%) as key process parameters. The tailing samples studied is composed mostly of chrysotile and lizardite. It contains about 40% magnesium (as its oxide MgO) and nickel and chromium showing contents 52 and 60 times higher than their respective average crustal abundances (Clarke values). Iron content is 8.7% (expressed as its ferric oxide Fe₂O₃). To optimize key factors influencing the leaching process, a statistical experimental design was employed. The designed leaching experiments were subsequently performed, and results were used to define leaching conditions aiming at maximizing Mg and Ni recoveries while minimizing iron contamination using response surface methodology (RSM) based on the central composite design (CCD). A quadratic polynomial model was developed to describe the relationship between the process parameters and metal recoveries. Among the tested effects of acid concentration, temperature, and pulp density on magnesium recovery, the modeling indicated that both hydrochloric acid concentration and leaching temperature significantly enhanced metal recovery, whereas increasing pulp density had a negative effect at low temperature. The empirical mathematical model derived from the experimental data, accounting for the uncertainties on chemical data, indicated that high magnesium recovery was achieved at 90 °C, with 10–12 N hydrochloric acid and a solid-to-liquid ratio of 33.6–40%. These findings reveal the potential for the recovery of critical and strategic metals, both in terms of efficiency and economic viability.

1. Introduction

Asbestos has been widely used in various industries, including thermal and electrical insulation, roofing, ceramics, cement, textiles, flooring, and coatings [1,2]. It is still produced in China, Kazakhstan, and Brazil [3]. Its widespread use is due to its desirable properties, such as heat resistance, incombustibility, insulation capabilities, durability, and high tensile strength [1,4]. From 1881 to 1950, Québec was the only producer of asbestos in Canada [5]. The province of Québec dominated the asbestos global market until the end of the eighties [2]. In 1989, the US banned the use of asbestos for health and safety reasons. Progressively, asbestos use decreased all around the world. The last mining sites in Québec were finally closed in 2012. The health hazards associated with asbestos include serious respiratory diseases and cancers and have led to stringent regulations and a substantial decline in its use [4,6,7]. “The sale, import and use of asbestos and asbestos-containing products has been prohibited—in Canada—since 30 December 2018. The federal regulations do not apply to tailings that contain asbestos, with some exceptions” [8]. Between 800 million tons and two billion tons of asbestos tailings over about 2000 ha according to different public sources [9,10] remain as huge stockpiles in the Thetford Mine-Val-des-Sources (former Asbestos City) geological corridor, located in south-central Québec, Canada. Huge amounts of asbestos tailings also remain in many other places in Canada [11,12] and in the world [13].

The host rock of asbestos ores is serpentinite, a rock resulting from the alteration of ultramafic rocks by saline water under low-temperature metamorphism. In the Thetford region (Southern Québec, Canada), asbestos host-rocks had a variety of compositions as illustrated by the geology synthesis of Riordon [14] and more recently by that of Schroetter et al. [15]. The serpentine group of minerals is a wide group which includes the chrysotile sub-group according to the Dana Classification [16]. Main minerals of the serpentine group are antigorite which is not asbestos and chrysotile which is asbestiform [17]. The serpentine group also includes other phyllosilicates like the lizardite sub-group of minerals [18]. These phyllosilicates are composed of iron (Fe) and magnesium (Mg), with various contents of nickel (Ni) and other potential strategic metals (chromium, manganese, cobalt) in their crystal structure. Such critical and strategic metals (CSM) are essential for the development of green energy and the electrification of transportation in many jurisdictions [19]. Chrysotile is the main fibrous asbestos mineral found in Québec asbestos tails [2]. Chrysotile, also commonly called asbestos or serpentine (which leads to some confusion), is represented by several polymorphs which are clinochrysotile (the most common) or orthochrysotile or parachrysotile [20] from the phyllosilicate family (in the serpentine group and chrysotile subgroup). Other regulated asbestos minerals are amphiboles and include crocidolite, also called riebeckite [21], amosite, also called grunerite [22], actinolite, also called hexagonite, nephrite jade, smaragdite or emerald green [23], anthophyllite [24] and tremolite [25] from the amphibole family. The basic structural units of amphiboles consist of tetrahedral (T) layers of silica oxide ions [Si₄O₁₁]⁶⁻ arranged in double chains [26]. Chrysotile also contains these tetrahedra organized into six-membered rings that extend into virtually infinite sheets, interlayered with an octahedral (O) brucite-like layer of magnesium oxyhydroxide ions [Mg₆O₄(OH)₈]⁻⁴ [26]. Through the serpentinization of peridotite rocks, a natural metamorphism geological process, chrysotile is expected to be accompanied by residual olivine (possibly depending on the level of geological metamorphism and alteration), pyroxenes and other serpentine minerals, as well as by magnetite and brucite [27].

The asbestos tailings, which are residues from the mining and mineral processing of asbestos ore [14,28], still contain high content of chrysotile, the main asbestos mineral in this region, and can still contribute to public health concerns and environmental contamination, if not handled properly [29]. The persistence of these tailings in soil and water systems [30] presents challenges for remediation and poses potential health risks to nearby communities. Aside from the environmental management of stockpiles, many projects have been launched in the last decades, to transform this industrial legacy into valuable opportunities [31]. The degradation of asbestos by various processes may lead to final tailings being recycled or reused [32,33]. For example, this is the case of the commercial transformation of asbestos by “Les Sables Olimag” [34]. Such processes enhance metal extraction while reducing both operational costs and environmental impacts [35]. Other processes under development aim at the destruction of fibers, or at the carbonation of Mg primary minerals, or both. Magnesium-rich minerals, which are abundant in ultramafic mining waste, if not leached during the process, offer a safe and cost-effective solution for carbon dioxide (CO₂) sequestration [36,37,38,39,40]. Such technological solutions are encouraged in Québec and Canada as exemplified by the development of compagnies like Exterra Carbon Solutions [41].

On the other hand, the recovery of metals through hydrometallurgical processes is a promising area of research due to both their effectiveness in processing low-grade ores and their potential to render leached chrysotile fibers harmless. The treatment of asbestos tailings has been conducted by leaching methods with strong mineral acids, including hydrochloric acid, sulfuric acid, nitric acid, oxalic acid, and phosphoric acid [1,42,43,44,45,46,47,48,49,50,51]. Quite recent reviews are provided in [4,33,51].

Magnesium metal can be commercially recovered from pure MgCl₂ solutions via molten salt electrolysis, a high-temperature process [52,53]. In contrast, precipitation methods are typically used to recover other metals from leach solutions, where metals like nickel, cobalt, and chromium are extracted through chemical reactions that allow the formation of solid metal compounds for further recovery [47]. In addition, hydrochloric acid (HCl) can be recovered and reused in the process [54]. For these reasons, extensive research has focused on using HCl as a leaching agent to extract magnesium from asbestos tailings. Several studies and patents have reported on and summarized this approach over the years [43,47,49,54,55]. In the academic literature, Shayakhmetova et al. optimized the leaching process of chrysotile asbestos waste using hydrochloric acid, the results showed that, under the optimal conditions hydrochloric acid concentration of 18%, temperature of ~85–90 °C, pressure of 1 atmosphere, and duration of 2 h, the leached percentages of Mg and Ni, reached, respectively, 96.0% and 81.3% [47].

Baigenzhenov et al. improved the production of magnesium chloride from wastes generated in chrysotile asbestos production. The leaching process with 25% concentrated hydrochloric acid at a temperature of 80–85 °C resulted in 92% of the magnesium passing into solution [56]. Abdel-Aal et al. [57] determined the optimum conditions for the decomposition of serpentine ore with hydrochloric acid. The results showed that under the optimal conditions—hydrochloric acid concentration of 12%, temperature of 95 °C, and duration of 4 h—the maximum MgO recovery achieved was 96.5%. Cheng and Lamy-Morissette [49] who also leached asbestos from the Thetford region emphasized the need to use high strength HCl (10 M) to obtain an extraction yield higher than 90% Mg. Their sample was composed mostly of lizardite, although other minor minerals were present. Unfortunately, the content of chrysotile is not clearly given for that sample but described to be broadly between 1% and 38% by volume. They also show that very little silica is released during acid leaching, as expected [49]. Aside the optimization of leaching, an interesting study was led by Morgan [58], describing the effect of various chrysotile mineralogical features on the dissolution rates. Currently, different methods are available for magnesium extraction, including pyrometallurgical, hydrometallurgical, and electrometallurgical processes as reviewed recently by Taheri and Larachi [51].

In the present study, the focus is on the complete recovery of valuable metals, including magnesium, by using HCl, towards a zero-waste strategy of asbestos tailings. There are many factors that influence the dissolution of the minerals present and the subsequent retention in solution of the metals to be recovered:

• The nature of the minerals to be dissolved;
• The mineralogical site to which the value (or non-desired) metals are associated with, and their readiness to be leached or not;
• Their size, and the grain size distribution of the powder;
• The proportion of solid to acid (solid percentage in % by mass, or L/S ratio in g/mL);
• Acid concentration;
• Leachate washing intensity;
• Physico-chemical conditions such as temperature, pressure, oxidation–reduction state, acidity, saturation index regarding secondary phases;
• Contact conditions between solid and acid: agitation mode, agitation duration.

Thus, the outcome of the leaching process depends on several factors and on significant interactions between them.

From previous research work, optimal leaching conditions appear to differ widely from one site to another. It is thus important to optimize leaching conditions for a complete recovery of values (CSM and silica) for the sample under study, knowing that interactions between leaching factors are significant in this kind of material under HCl acid leaching. Many of the already published data do not report all of their parameters and operating conditions to the specific composition of their feed and simply refer to it generically as asbestos. However, it is deemed critical to have mineralogical knowledge and a scientific perspective of the feed (mineralogical, geochemical, and crystal structure) when elaborating and discussing the mechanisms involved during leaching of complex ore matrices. The present study included a mineralogical approach on the feed material.

Although previous studies have demonstrated the feasibility of extracting magnesium and nickel from asbestos tailings using hydrochloric acid, most lacked systematic optimization and did not account for the influence of mineralogical composition on leaching performance. The interaction effects between key parameters such as acid concentration, temperature, and solid-to-liquid ratio also remain underexplored. This study addresses these gaps by integrating detailed mineralogical characterization with a statistical optimization approach RSM to maximize CSM recovery while minimizing iron contamination.

The optimization was performed through a minimal set of experiments thanks to using a statistical design of experiments (DOE) method. The effects of three factors—acid concentration, temperature, and solid-to-liquid ratio-on the leaching process of asbestos tailings were investigated. In this first attempt, the duration and the particle size distribution of the feed were kept constant. Based on this, the recovery response for magnesium and other elements from the acid leaching process was optimized using a standard response surface model (RSM) design involving a central composite design (CCD). It is known that when the expected response is affected by the combination of several independent factors and their interactions, the response surface method (RSM) is an effective tool proposed for the optimization process [59]. The central composite design (CCD) method is a commonly applied approach within the response surface model (RSM) to optimize the leaching of valuable metals from solid wastes [60,61,62]. RSM is an optimization method combining experiments with statistics [62,63]. It is well suited to fit a quadratic surface. In parallel, the mineralogical and geochemical characterizations of the feed sample was approached by coupling X-ray diffraction, Mössbauer spectroscopy and microprobe analyses on hand-picked typical grains of the samples, in addition to the whole sample geochemistry by major and Ni, Co, Cr trace elements quantifications.

2. Materials and Methods

2.1. Collection and Preparation of Homogeneous Head-Samples

About 60 kg of asbestos tailings from the Thetford region were provided for the project by the industrial partner and his collaborators. The material was wet. The totality of this material was sieved wet, for safety reasons, to keep only the fraction finer than 2.36 mm as the head-sample for this study. The head-sample moisture content was estimated at 21% (dry sample of about 47 kg).

2.1.1. Preparation of Homogeneous Head-Samples

• Wet method preparation for the first series (W-series);

After chunks larger than 2.36 mm were removed from the collected material, the undersized head material was quartered manually into a final set of about 1 kilo samples, using a shovel according to a rigorous quartering plan. On about 0.5 kg of sub-sample, also collected by quartering, the characterization of the particle size distribution was performed by wet sieving. Then 3 aliquots of 10 to 20 g, collected by core-drilling along the whole depth of 2 different samples bags using a soil micro-coring tool. These were used for the estimation of moisture, for the asbestos content estimation or safety risk analysis and for the geochemical analysis (selected metals) of head-samples used for the leaching study. Sub-samples were stored in sealed plastic bags before use, as about 1 kg samples.

• Dry method preparation for the second series (D-series);

Approximately 7.7 kg of the initial head-sample (−2.36 mm) was dried. Batches of 320 g of this residue were prepared to obtain a representative particle size distribution for leaching. For sample homogenization, the asbestos residue was divided into 12 batches by a carousel-type divider. Using a rifle, the 12 batches were mixed so that 2 opposite batches from the carousel were intertwined to form 24 batches of 320 g each. Twenty batches out of 24 were then divided into 8 smaller 40 g batches using a rotary divider. Granulometric and chemical analyses of the samples were carried out to verify the homogeneity and representativeness of each batch during all leaching operations.

2.1.2. Preparation of Fibers, Magnetic, Non-Magnetic Fractions and Hand-Picked Grains for the Mineralogical Study

Finally, to get more information about the iron-containing phases, magnetic or not, a sample was prepared in the following way: the asbestos fibers fraction (30–35%) was removed from two sub-samples by using its natural floatability. On the sink fraction, the magnetic fraction (40–50%) was separated from the non-magnetic one (50–60%) by using a handheld magnet. The magnetic and the non-magnetic fractions were further classified by handpicking under a microscope [64]. An aliquot of each mineral class was characterized by X-ray diffraction (XRD) and Mössbauer spectroscopy. Electron probe microanalysis (EPMA) was used to quantify Ni, Cr (chromium), Fe, Mg, Al (aluminum), Si (silica) of some grains as a complement to this study [64].

2.2. Methods of Characterization of Samples

The head-sample was characterized in more details (particle size distribution—PSD, moisture, mineralogy, geochemistry) than leached products for several practical reasons, while only chemistry was determined for leaching head-samples and products (solid tails, leachates and wash waters alike). The characterization of the head-samples allowed to evaluate occupational safety and health risks and the level of preventive measures to be taken to work with that dangerous material in the laboratory, to evaluate hydrometallurgical risks and opportunities, and evaluate how risks could be minimized or avoided in a sustainable way, to assess how mineral concentration processes prior to leaching could benefit the whole sustainability of the process, to determine the content of the various CSMs, and of value minerals, to evaluate the level of recovery effort to be put on each of them. For each sample, the geochemical analysis aims at balancing the mass of chemical elements (or their oxides) between the leaching feed and products.

2.2.1. Solid Samples

• Particle sizes distribution

The PSD was determined by wet sieving on the 8, 12, 20, 35, 60, 140, 400 U.S. mesh screens on the W-series. This analysis was duplicated. From the resulting curves, particle size distribution criteria of interest for the process (passing sizes: D80, D50, D38 and top size) were estimated. On the D-series, to verify the homogeneity of the sample, dry sieving of the 5 random batches of the (head sample) asbestos residue was also carried out.

For both W and D series, PSD results were used to get the specifications of studied samples but also to validate the homogeneity of samples quartered from the head material.

• Moisture

Moisture was determined by weighing a known wet mass of sub-sample before and after full drying at 105 °C. The stable weights were taken after the sample was brought back to room temperature and humidity. This latter value was first used to estimate the dry mass but led to major uncertainties. Therefore, the following samples were dried immediately prior to use in the leaching tests. The first leaching series was conducted on the W-series. It was performed under the following conditions: leaching temperature was maintained between 25–90 °C, hydrochloric acid concentration ranged from 2–12 mol/L, and the solid–liquid ratio varied between 10–40% (Table 1, Section 2.4).

• Mineralogy

The minerals inventory was approached by various techniques. First, the whole sample was analyzed to evaluate the safety risk while working with asbestos tailings. This is commonly performed by polarized microscopy, although near infrared/shortwave infrared (NIR-SWIR) hyperspectral scan imaging [65] or Raman spectroscopy could be more accurate [66,67,68,69]. The semi-quantitative volume content of fibers in the head-sample was achieved by a commercially certified laboratory (Laboratoire Silica, Montréal) to evaluate the risk and how samples will be handled during the project [70]. Then the crystalline fraction of the pulverized sample was determined by XRD followed by Rietveld data treatment at a commercial laboratory (Actlabs, Activation Laboratories Ltd., Ancaster, ON, Canada) [71]. “The X-ray diffraction analysis was performed on a Bruker D8 Endeavour diffractometer equipped with Cu X-ray source and operating at the following conditions: 40 kV and 40 mA; range 4–70 deg 2-theta; step size 0.02 deg 2-theta; time per step 0.5 s; fixed divergence slit, angle 0.30; sample rotation 15 rotation per minute (rpm). The PDF4/Minerals ICDD database was used for mineral identification. The quantities of the crystalline mineral phases were determined using Rietveld method. The Rietveld method is based on the calculation of the full diffraction pattern from crystal structure data.” Corundum was added to the sample as an internal standard [71].

To complete the mineralogical composition of the head-sample, a Mössbauer spectroscopy study, coupled with XRD and EPMA on hand-picked grains, was performed. Grains specimens were hand-picked under a microscope. They were hand-ground and mounted on a zero-background single-crystal silicon plate using a small amount of vacuum grease [64].

The ⁵⁷Fe Mössbauer measurements were made on a conventional spectrometer operated in constant-acceleration mode using a 50mCi ⁵⁷Co(Rh) source. All spectra were acquired at ambient temperature. The instrument was calibrated using a thin alpha-Fe foil and isomer shifts are quoted relative to the centre of the calibration spectrum. Samples were hand ground and mixed with boron nitride powder to form a uniform absorber and then mounted in acetal holders with thin windows. A technical challenge for natural fibers and micaceous minerals, side the fact that they may be far from stoichiometric composition, is that they are uneasy to break down in a homogeneous powder: such textured samples result in less precise Mössbauer spectra than synthetic minerals. A summary of the principle of Mössbauer Spectroscopy (MS) with a reference to measured features (δ, Δ, data referred to in the MS detailed table results) is given in Appendix A (Appendix A.2.1).

The EPMA used was a Cameca SX 100 FIVE FE having 5 wavelengths dispersive X-rays spectrometers. This field emission microprobe has large dispersive crystals which make it easier to measure trace elements [72]. Analyses were obtained on separated grains mounted in polished thick sections. Data acquisition conditions were 15 kV accelerating voltage, 20 nA beam current (10 µm beam size), 10 s counting time for each element (Mg, Fe, Si, Ni, Co, Cr). Standards used for the calibration of the wavelength dispersive spectroscopy (WDS) detector were olivine (with the TAP crystal, hematite (with the LLiF crystal), olivine (with LTAP crystal), NiO (with the LLiF crystal), co-metal (with the LLiF crystal), chromite (with the LPET crystal) for Mg, Fe, Si, Ni, Co, Cr respectively. Detection limits were defined according to Ancey et al. method [73].

• Geochemistry

  ○

  Wet chemistry quantification of selected elements and CSM

The geochemistry of the head sample and the leached solid residues of the W-series were analyzed using a conventional digestion method involving dissolution in a mixture of three acids (HNO₃, HCl, and HF) at 95 °C and atmospheric pressure overnight. To enhance this approach and reduce uncertainty in the elemental mass balance, an improved protocol was developed. Subsequent post-leaching solid residue leftover from the W- and D-series were digested using a more comprehensive acid blend (HNO₃, HCl, HF, and H₃BO₃) under the same conditions. After completely evaporating the digestate, the residue was redissolved in HNO₃, filtered through a 0.45 µm membrane (acquired from Avantor/VWR International, Mississauga, ON, Canada), and analyzed in the same manner as the aqueous leachates (see Section 2.2.2). This method (Method #9 HP 0.5 g H₃BO₃) was performed in triplicate and validated for Mg, Ni, Co, Cr, Fe, and Mn. The standard deviation among replicates was below 5% for Mg, Ni, Co, Fe, and Mn, and 8% for Cr.

• ○

  Whole-rock geochemistry by wavelength-dispersive X-ray fluorescence (XRF)

Furthermore, and to get a better knowledge of the composition of major elements of the head-sample, the whole-rock geochemistry was assessed by Wavelength Dispersive X-ray Fluorescence (WD-XRF, PANanytical Axios^mAX), on glass disks prepared by borated fusion (Method 50/50 de Li₂B₄O₇ et LiBO₂) using a Katanax fusion fluxer (model K1 Prime). The certified reference material (CRM) Andesite JA-1 [74] was used to countercheck the analytical response for major and trace elements. It was supplied by Brammer Standard Company, Inc., Houston, TX, USA. In complement to that whole-rock analysis approach, the loss of ignition (LOI) was determined at 1050 °C.

• ○

  Validation of head-subsamples homogeneity

The validation of head-subsamples homogeneity was also performed by geochemical analysis of important elements for the process development. For the W-series, the content in Mg, Ni, Cr, Co, Fe, and Mn in a total of six sub-samples drilled into two different 1 kg bags were determined. The standard deviation on the five sub-samples is equal or better than 5% for Mg, Ni, Co, Fe, Mn and equal to 8% for Cr. For the D-series, XRF and wet-chemical analysis were run on 5 sub-samples (certificates 30801_A1 to 30810_A1) and validated by the analysis of the CRM.

2.2.2. Aqueous Samples

Leachates and wash waters (after acid full digestion and having been brought back to a dilute HNO₃ aqueous matrix) were filtered (0.45 µm porosity) and analyzed by microwave induced plasma interfaced to an atomic emission spectrophotometer (Agilent MP-AES 4200 acquired from Agilent Technologies Canada Inc., Toronto, ON, Canada with its software MP Expert) for major and minor elements, and by inductively coupled plasma mass spectrometry (Agilent ICP-MS 7800 acquired from Agilent Technologies Canada Inc., Toronto, ON, Canada with its software MS Hunter) for trace elements. The analytical results obtained were used to calculate metal recovery percentages.

2.3. Leaching Methods

There were two series of experiments, the first to optimize the leaching process and study the parameters (W-series), and second (D-series) to confirm first results and conduct a sensitivity study around the designed optimal set of conditions.

2.3.1. Experimental Protocols

Using the W-series samples, twenty-three experiments were conducted under various conditions as designed by the Stat-Ease Design of Experiment software. Using the W-series samples, twenty-three leaching tests were performed under variable conditions of acid concentration, temperature and solid to liquid ratios. Ranges for these parameters and conditions of each test are indicated in Table 1. The leaching time was set at 90 min.

Using the D-series samples, four leaching tests were run in triplicate with highly concentrated hydrochloric acid (HCl) at different concentrations (10 N and 12 N) were carried out in batch mode. The experimental protocol was about the same for both series: Acid temperature was maintained with a round-bottomed flask heater or with a water bath system, using a suitably sized metal bowl, and a hot plate with an electromagnetic stirrer. To avoid acid loss through evaporation, a condensation system was used to condense the vapor. Agitation rate was set between 300 and 360 rpm for the D-series and W-series respectively, high enough to keep the solid dispersed while avoiding a mixing vortex, using a magnetic bar. After the leaching process, hot vacuum filtration was carried out to separate the solid from the liquid, using a vacuum system and a silicone heating blanket to maintain the leaching temperature during filtration. The liquid fraction was analyzed by ICP to quantify the metals dissolved in the leachate, while the solid fraction was rinsed twice with a solution of 1 N hot HCl. The amount of rinse water was recovered separately (D-series) or not (W-series) and analyzed by ICP as well. The resulting residual solid was then dried at 105 °C in a conventional oven for 24 h. After drying, the mass of the resulting leached solid was determined and analyzed by ICP and XRF to assess the amount of metal dissolution in the residue versus leaching.

For each leaching run, a specific acid solution made from analytical grade commercial concentrated hydrochloric acid (37%) was used. The sequence of preparation of the leaching operation started by preparing the solution to the desired concentration, heating the solution to the targeted temperature, and then adding the prepared feed-sample to the reactor in respect to the solid–liquid ratio. At the end of the leaching process, solid and liquid were filtrated under vacuum, while maintaining the temperature of the filtration system at about 65 °C (D-series), using 0.45 µm pore size filter papers made from quartz fibers. The leached tails were washed using hot DI water and dried at approximately 105 °C.

2.3.2. Set-Ups

The experimental set-up to perform the leaching experiments is shown in Figure 1. It includes a 250 mL laboratory reactor connected to a reflux condenser (with water circulation) and equipped with a thermometer, positioned on top of a hotplate with magnetic stirrer. Temperature is controlled by placing the reactor on a heater in a water bath.

2.3.3. Reagents

The leaching agent used in all tests was hydrochloric acid (HCl, +37%). To determine the metal composition of the solid fraction, a three-acid digestion technique using hydrochloric acid (HCl 34–37% for trace metal analysis), nitric acid (HNO₃ 60–70% and hydrofluoric acid (HF 47–51%). These 3 acids are high purity grade used for trace metals analysis, followed by complexation with ACS grade boric acid (H₃BO₃ +99.5%). Reagents are from VWR International (VWR International, 2360 Argentia Road, Mississauga, ON, Canada).

2.3.4. Operating Variables and Conditions

The effects of three factors—acid concentration, temperature, and solid-to-liquid ratio-on the leaching process of asbestos tailings were investigated.

2.4. Design of Experiments (DOE): Modeling of Some Asbestos Tails Leaching Experiments to Optimise Leaching Conditions

The choice of parameters, their ranges (Table 1), and other operating conditions (Section 2.3) are based on preliminary experiments using the same kind of sample and based on the findings of other researchers about similar materials [60,75,76,77,78,79,80,81].

Table 1. Independent factors and their levels.

Parameters	Unit	Low Level	High Level
HCl concentration	mol/L (%)	2 (6.1)	12 (37)
Solid to liquid ratio	%	10	40
Temperature	°C	20	90

Table 1. Independent factors and their levels.

Parameters	Unit	Low Level	High Level
HCl concentration	mol/L (%)	2 (6.1)	12 (37)
Solid to liquid ratio	%	10	40
Temperature	°C	20	90

The experimental tests for the leaching process were designed using the response surface methodology (RSM) with Design Expert software (version 23.0.0; Stat-Ease Inc., Minneapolis, MN, USA). Central composite design (CCD) was selected as the experimental design domain for leaching process optimization of the experimental parameters. In this design, several runs were suggested to cover low, high, and central data points in addition to the extra replicates so a model could be suggested to relate the parameters to the leaching efficiencies of Mg, Al, Cr, Mn, Fe, Co, and Ni (as responses of the model). A regression analysis of the data based on using a second-order polynomial equation was used by the model [80]:

$$Y={\beta }_0{\sum }_{i=1}^k{\beta }_i{X}_i+{\sum }_{i=1}^k{\beta }_{ii}{X}_i^2+{\sum }_{i=1}^{k-1}{\sum }_{j=2}^k{\beta }_{ij}{X}_i{X}_j+\epsilon$$

(1)

where Y denotes the value of the response variable, the regression coefficients (β); X are the independent parameters; k is the number of optimized variables; ε represents the random error.

Here, Fischer’s test value (F-value) and the probability value (p-value) in the analysis of variance (ANOVA) are the basis to evaluate the effects of coefficient (or parameter) in the above equation [76].

3. Results and Interpretations

3.1. Characterization of Samples

3.1.1. Particle Sizes Distribution (PSD)

The particle size distribution shows that particles of samples to be leached are coarse, having 52% of the head sample over 425 µm and about 25% over 1 mm mesh sieves (Figure 2). The particle size distributions are similar with a D₈₀ of 1000 µm for the 5 random batches of residues.

Using tailings as received as a feed of the hydrometallurgical process, without any re-grinding prior to leaching, is significantly eco-energetic for a mine. If value metals can be leached readily enough, the milling of such coarse particles will not be necessary. The economics and the greenhouse gas balance of the process might require a compromise between grinding and strong leaching conditions.

3.1.2. Mineralogy

• Inventory of main minerals by X-ray diffraction (XRD) in the head-sample

According to the mineralogy of the crystalline material provided by XRD [71], the main group of minerals is identified as being serpentine. The rest of the crystalline matter is represented by minor minerals such as chlorite, forsterite, brucite, spinel (including magnetite) and accessory quartz (

Table 2. Crystalline minerals identified in the head-sample.

Identified Minerals [64,71]	Estimated Amounts
serpentine (lizardite, chrysotile mostly)	Most of the sample
chlorite	Few percents
forsterite (Mg-olivine)	Few percents
brucite (Mg (OH)2)	Few percents
magnetite (Fe3O4, and other spinels)	Minor
quartz (SiO2)	Accessory

; Figure A1 and

Table A1. Minerals determined by quantitative analysis of the bulk head sample XRD [71].

Identified Minerals by XRD	Recovery of Elements (% by Weight)
Serpentine	44.7
Chlorite	2.0
Forsterite (Mg-olivine)	2.0
Brucite (Mg (OH)2)	1.6
Magnetite (Fe3O4, and other spinels)	1.1
Quartz (SiO2)	0.2
Amorphous	48.4

; Appendix A.1). Forsterite is the magnesian pole of the olivine (Fe-Mg)-series. Chlorite is an iron-containing phyllosilicate. Half of the sample was identified as amorphous [71]. The nature of the amorphous fraction remains unknown.

Estimated amounts of crystalline material are thus only qualified according to all observations collected on samples (Table 2). The main serpentine minerals were identified by coupling several techniques as indicated) in next sections [64].

Asbestos crystals, as characterized by polarised microscopy, indicated the presence of mainly chrysotile fibers for 45 to 50% (by volume) [70]. This is consistent with the large volume of chrysotile fibers separated from the bulk by flotation in the present study.

Important mineralogical features for mineral processing and hydrometallurgical process developments were still missing at this stage as several of the identified minerals could contain Mg, Ni, Cr but also Fe and Mn which are considered undesirable companion elements in the leaching stage of the process.

Furthermore, hand-picking and classification of grains presenting similar visual features was also performed [64]. The mineralogy of the head-sample was thus further studied by combining optical microscopy observations, Mössbauer spectroscopy, XRD and EPMA on hand-picked minerals [64].

Thanks to this study, the inventory of serpentine minerals has been brought to a more advanced level of characterization. A compilation of data from the literature [82,83,84,85,86,87] for the main serpentine minerals observed is given in Figure A2 and

Table A2. Room-temperature hyperfine Mössbauer parameters. Crystallographic sites hosting iron ions in the crystal lattice are indicated by O and T, which stand for octahedral and tetrahedral coordination respectively.

Mineral	Crystal Site	Fe Valency	δ (mm/s)	Δ (mm/s)	Bhf (T)	Reference
Chrysotile	O	Fe2+	1.13	2.75		[82]
	O	Fe3+	0.31	0.86
	T	Fe3+	0.18	0.33
Lizardite	O	Fe2+	1.14	2.70		[83]
	O	Fe3+	0.40	0.70
	T	Fe3+	0.24	0.39
Antigorite	O	Fe2+	1.13	2.69		[81,82]
	O	Fe3+	0.41	0.70
	T	Fe3+	0.21	0.36
Magnetite			0.28		49.0	[83]
			0.66		45.9

(Appendix A.2). Experimental data on chrysotile-free, magnetic and non-magnetic, fractions of the head-sample are partly presented in Appendix A.2. A synthesis of processed results is given below.

This study allowed us to separate a first class of grains without any particular geometric shape, with a wide diversity of colors (dark green, light green, yellow-brown and even pale greyish green) and transparency. They showed conchoidal to smooth fractures and greasy surfaces. They presented a variable chemical composition by EPMA but were closest to lizardite. The combined spectral features (EPMA, XRD, Mössbauer) led us to conclude it was mostly lizardite (

Table 3. Room-temperature hyperfine parameters determined by Mossbauer spectroscopy on chrysotile free head-sample (significant digits are reported followed by the incertitude value into parentheses).

Mineral	δ (mm/s)	Δ (mm/s)	Bhf (T)	Area
Lizardite 1				This study
Component 1 (Fe2+)	1.15(1)	2.75(1)	0	60.0
Component 2 (Fe2+)	1.16(1)	2.20(1)	0	19.3
Component 3 (Fe3+)	0.32(2)	0.59(4)	0	20.7
Antigorite				This study
Component 1 (Fe2+)	1.14(1)	2.71(1)	0	90
Component 2 (Fe3+)	0.25(2)	0.42(3)	0	10
Magnetic fraction of the head sample				This study
Component 1	0.28(1)	−0.02(1)	48.7(1)	24	Assigned to magnetite
Component 2	0.64(1)	+0.02(2)	45.7(1)	37	Assigned to magnetite
Component 3 (Fe2+)	1.15(1)	2.84(1)	0	25	Assigned to antigorite
Component 4 (Fe3+)	0.38(1)	0.46(1)	0	8	-
Component 4 (Fe3+)	0.45(1)	1.05(1)	0	6	-

and Figure A3, Appendix A.2.2). Aside chrysotile fibers, the dark green, opaque grains of lizardite are abundant in the head sample. Light green (LG) crystals showed a similar composition in Mg and Si (EPMA) but variable contents of Fe, Ni, Cr, and a lower crystallinity (XRD). Some grains, showing the same EPMA features were more yellow brown (YB+LG). They were classified as altered lizardite. Indeed, they contain iron (and some traces of nickel and chromium) in low content (

Table A3. EPMA specifications and results on selected area of some hand-picked grains [64].

Instrument	Cameca Sx100
Accelerating Voltage	15 kV
Beam Currents	20 nA
Beam Size	10 um
	Mg		Cr	Fe	Ni	Si	Co
Counting Time (seconds)	20		20	20	20	20	20
Analytical Crystals	TAP		LPET	LLiF	LLiF	LTAP	LLiF
Standards	Olivine		Chromite	Hematite	NiO	Olivine	Metal
Standard Sources	CM Taylor		CM Taylor	CM Taylor	Cameca	CM Taylor	Cameca
		Oxide %
Name	Point#	MgO	Cr2O3	FeO	NiO	SiO2	CoO	Total
Pale Green-G1-1	13	40.50	0.003	1.601	0.035	42.39	0.018	84.55
Pale Green-G2-2	14	40.39	0.055	2.010	0.034	41.63	0.000	84.12
Pale Green-G3-3	15	38.39	0.000	3.072	0.051	40.19	0.000	81.70
Antigorite-G1-1	16	37.99	0.000	3.579	0.025	43.56	0.009	85.17
Antigorite-G2-2	17	38.37	0.000	3.499	0.000	43.26	0.030	85.16
Antigorite-G3-3	18	37.79	0.002	4.045	0.402	42.83	0.000	85.06
(Fe-OX) Antigorite-Mag-1	19	0.077	0.000	91.54	0.012	0.073	0.022	91.72
(Fe-OX) Antigorite-Mag-2	20	0.442	0.000	90.48	0.010	0.429	0.063	91.42
Lizardite-G1-1	21	39.97	0.003	2.054	0.137	42.16	0.032	84.36
Lizardite-G2-2	22	39.85	0.049	2.140	0.549	42.74	0.050	85.38
Lizardite-G3-3	23	38.43	0.002	3.448	0.154	43.98	0.000	86.01
YB+LG-G1-1	24	38.69	0.519	2.106	0.088	40.50	0.000	81.90
YB+LG-G2-2	25	38.65	0.491	1.940	0.011	40.11	0.026	81.22
YB+LG-G3-3	26	39.43	0.566	2.192	0.004	40.70	0.001	82.90
YB+LG-G4-4	27	39.06	0.068	2.994	0.067	42.12	0.026	84.34
YB+LG-G5-5	28	39.82	0.000	2.246	0.000	42.31	0.017	84.40
YB+LG-G6-6	29	40.34	0.000	1.628	0.029	42.03	0.004	84.04

Legend: YB stands for Yellow-Brown, LG stands for Light Green, Gx stands for Grain number x; Each selected grain, or association of grains (e.g., like YB+LG or Antigorite-Mag-1) was numbered. Point# means Point number.

in Appendix A.2.2).

Changes in colors and shapes are commonly interpreted as resulting from some alteration of the serpentine minerals [81,82,83,84]. The Mössbauer spectra on lizardite grains, altered or not, were fitted using three components, while antigorite and magnetite were fitted using two components, the hyperfine parameters of which are given in Table 3.

However, the identification of lizardite by Mössbauer analysis could not be fully satisfying when compared to literature data [81,82,83] as reported in Appendix A.2.2 (Table A2). This may be explained by the complexity of these natural minerals which, in addition to that, clearly contain some minerals inclusions and are not fully liberated (e.g., black dots on pale green grains seen on the lower picture of Figure A3; Appendix A.2.2). Lizardite was assigned after having confronted and combined all available data from this study.

A second class of serpentine mineral was represented by acicular green crystals of antigorite which were readily recognized (Figure A4, Appendix A.2.2). A third one was represented by platy, soft and fragile grains identified as being tremolite/actinolite with copperish tints (Figure A5, Appendix A.2.2). These crystals were also readily found in quite good abundance. A few conglomerates of mixed whitish silicates (optical observations and XRD) were also objectified. This study showed the presence of iron in each main serpentine minerals of the tailings sample studied.

Aside serpentine minerals, two spinel phases were identified with good confidence using Mössbauer spectroscopy in the sample: magnetite and chromite [84,85]. Magnetite is from far, the most abundant one and its Mössbauer response is characterized by typical sextets (Table 3 and Figure A6 in Appendix A.2.2). Room temperature hyperfine parameters obtained were consistent with data from the literature [85]. Trevorite (Ni, Fe-oxide) was also detected by EPMA on some spinel grains. Finally, it was noted that these oxides were often not liberated from silicates. This observation indicates that re-grinding of tailings prior to leaching could allow us to remove magnetite from the feed of the leaching stage as an efficient way to remove some of the iron. Chromite is non-magnetic and would remain in the feed of the hydrometallurgical process. As a conclusion, the variability of the spectral features of serpentine minerals reflects the complexity of the mineralogy of such environments, and this is consistent with what can be found in the literature [86,87]. Mössbauer and EPMA measurements indicate that some iron, chromium and nickel are distributed in many minerals, including magnesium bearing serpentine minerals.

• Variable Fe²⁺/Fe³⁺ ratios in minerals:

Mössbauer spectroscopy allows us to quantify the distribution of ferrous and ferric iron in iron-bearing minerals from the various sets of sub-samples (magnetic, non-magnetic and fibers; cf. separation method explained in Section 2.2.1). In the magnetic sub-sample, oxide grains were hand-picked and analyzed. This study illustrates that up to 61% of iron reports to magnetite. The rest of the iron reports to non-magnetics iron-containing minerals which were not liberated from magnetite, or which were physically entrained with it. The feed subsample (159.4 mg) analyzed showed contents of 16% Fe³⁺ and 30% of Fe²⁺. A proportion of 53% of the iron is contained in magnetite and the rest is in serpentine minerals. A proportion of 61% of the iron is in magnetite in the magnetic fraction, with 14% of Fe³⁺ and 25% of Fe²⁺. In the sub-sample of the non-magnetic fraction, there was 30% Fe³⁺ and 70% Fe²⁺ and magnetite is absent. As a conclusion, although the sample shows some alteration in tailings stockpiles, the feed of leaching still shows reducing features with Fe²⁺ being systematically higher than Fe³⁺. This is consistent with what has been described in literature as well [82,83,86].

3.1.3. Geochemistry of the Head Sample

• Major elements geochemistry by energy-dispersive X-ray fluorescence (XRF)

The major elements, expressed as oxides, are SiO₂, MgO and Fe₂O₃, amounting to 36.1% ± 0.3%. and 40.4% ± 0.3% and 8.7% ± 0.3% respectively (

Table 4. Chemical analysis results for 5 random batches (certificates 30806 to 30810) obtained by XRF (%).

Oxide	Unit	1	2	2	4	5	Mean	Std. Dev.
SiO2	%	36.37	36.08	35.79	35.83	36.43	36.1	0.3
Al2O3	%	0.86	0.81	0.78	0.70	0.67	0.76	0.07
CaO	%	0.21	0.34	0.20	0.16	0.21	0.22	0.06
MgO	%	40.27	40.00	40.32	40.99	40.42	40.4	0.3
Na2O	%	-	-	-	-	-	-	-
K2O	%	0.02	0.05	0.05	0.02	0.04	0.04	0.01
Fe2O3	%	8.47	8.91	9.08	8.50	8.28	8.7	0.3
TiO2	%	-	-	-	-	-	-	-
MnO	%	0.12	0.11	0.11	0.10	0.12	0.11	0.01
Cr2O3	%	0.28	0.30	0.32	0.28	0.40	0.31	0.04
NiO	%	0.28	0.29	0.31	0.33	0.32	0.31	0.02
LOI	%	12.96	12.96	12.96	12.96	12.96	-	-

). The LOI was found to be 12.96%. The whole-rock analysis ends up with total major oxide values between 98.53% and 98.87% (Appendix A.3,

Table A5. XRF analytical certificate (CTRI Certificate 24900, 21 August 2024).

Sample Code		Standard JA-1	Standard JA-1	29400A-E1	29400A-E2	29400A-E3	29400A-E4
Date of Preparation				8 March 2024	8 March 2024	8 March 2024	8 March 2024
Identification	Unit	Obtained	Expected	Head-Sample	Head-Sample	Head-Sample	Head-Sample
Na2O	%	5.055	3.84	0.51	0.49	0.30	0.34
MgO	%	1.508	1.57	42.46	42.35	42.93	43.31
Al2O3	%	15.058	15.22	1.03	1.00	0.69	0.63
SiO2	%	61.844	63.97	41.06	42.35	42.20	41.83
P2O5	%	0.088	0.165
SO3	%	0.019		0.13	0.17	0.08	0.07
K2O	%	0.692	0.77	0.13	0.13	0.06	0.05
CaO	%	5.367	5.7	0.47	0.40	0.24	0.30
TiO2	%	0.791	0.85
Cr2O3	%			0.48	0.40	0.30	0.28
MnO	%	0.194		0.14	0.15	0.15	0.16
Fe2O3	%	7.541	7.07	11.73	10.88	10.90	10.88
NiO	%	5.055	3.84	0.40	0.36	0.38	0.43
CuO	%
ZnO	%			0.33	0.19	0.32	0.24
Total	%			98.87	98.87	98.54	98.53

).

• Geochemistry of metals followed during leaching by wet preparation and inductively coupled plasma mass spectrometry (ICP-MS)

Contents of the selected elements Mg, Al, Cr, Mn, Fe, Co, Ni in the head sample are reported in

Table 5. Contents of metals of interest for the project in the head sample, as given for the W-series and the D-series (expressed in ppm, plus or minus the standard deviation and, under brackets, followed by the relative error).

Run	Mg (ppm)	Al (ppm)	Cr (ppm)	Mn (ppm)	Fe (ppm)	Co (ppm)	Ni (ppm)
W-series	210,000	2345	670	885	73,242	98	2227
An. Series (n = 3)	209,036 ± 1364 (0.7%)	2763 ± 604 (22%)	593 ± 31 (5%)	854 ± 34 (4%)	62,932 ± 2335 (4%)	97 ± 3 (3%)	2144 ± 67 (3%)
D-series (n = 5)	236,038 ± 4991 (2%)	2748 ± 231 (8%)	706 ± 41 (6%)	814 ± 20 (3%)	60,286 ± 1750 (3%)	95 ± 3 (3%)	1868 ± 97 (5%)

and their contents will be tracked in the leaching products. Metals of interest were designated as magnesium (Mg), chromium (Cr) and nickel (Ni) with contents 9, 52 and 60 times higher than the Clarke levels [88], respectively. Iron (Fe), aluminum (Al), manganese (Mn) and cobalt (Co) were also recorded because Fe, Mn and Al secondary minerals (or precipitates) may have significant impacts on filterability and accompanying CSM hydrometallurgical extractions.

The analytical series (An. Series) refers to the certificates 29,400 A, B, C (Table A5 and

Table A6. MP-AES and ICP-MS analytical certificate after wet digestion (CTRI Certificate 24900, 21 August 2024; Method#9HP-0.5 gH₃BO₃; Bold: quantitative; Normal: semi-quantitative).

Sample #		29400A	29400B	29400C
Date of Preparation		8 March 2024	8 March 2024	8 March 2024
Identification	Unit	Head-Sample Test #9HP-0.5 gH3BO3	Head-Sample Test #9HP-0.5 gH3BO3	Head-Sample Test #9HP-0.5 gH3BO3
Na	mg/Kg	12	12	12
Mg	mg/Kg	211,082	207,132	208,894
Al	mg/Kg	2400	2221	3669
K	mg/Kg	450	420	1349
Ca	mg/Kg	7391	8979	6222
Ti	mg/Kg	39	47	50
V	mg/Kg	13	13	12
Cr	mg/Kg	557	582	640
Mn	mg/Kg	828	829	904
Fe	mg/Kg	63,913	59,430	65,454
Co	mg/Kg	100	99	93
Ni	mg/Kg	227	2162	2043
Cu	mg/Kg	6	6	37
Zn	mg/Kg	106	112	90
As	mg/Kg	4	6	1
Se	mg/Kg	<1	<1	<1
Mo	mg/Kg	<0.2	0.4	<0.2
Cd	mg/Kg	<0.1	<0.1	<0.1
Sn	mg/Kg	<0.1	<0.1	<0.1
Sb	mg/Kg	<0.1	<0.1	<0.1
Te	mg/Kg	<0.1	<0.1	<0.1
Ba	mg/Kg	10	11	27
Pb	mg/Kg	<0.1	0.3	2.7
Bi	mg/Kg	<0.1	<0.1	<0.1
U	mg/Kg	<0.1	0.1	<0.1

) in Appendix A.3. These were first set of results. Aluminum analysis was particularly difficult to analyze quantitatively. After the analytical method has been improved, the relative mean deviation results on triplicates for the W-series, as compared to the analytical series (An. Series; Table 5), are 0.7% for Mg, 4% for Fe and Mn, 22% for Al, 5% for Cr, and 3% for Ni. For the D-series (Table 5) standard deviation results are 6.5% for Mg and Ni, 9.5% for Cr, 2% for Fe and Mn, and 0.3% for Al. These standard deviation values are indicative of an uncertainty level better than 10–20% on experimental results for most targeted elements.

3.1.4. Check of the Homogeneous Division Quality

For both W- and D-series of samples to be leached, the similarity of particle size distribution and the geochemistry of several sub-samples after the division of sub-samples indicated that divided batches were homogeneous and well mixed. The risk of having different batches during leaching tests is therefore low (see also PSD, Figure 2).

3.2. Leaching Experiments and DOE Treatment of Data

3.2.1. Metals Leached from the W-Series Feeds

The experimental conditions based on central composite design (CCD), as well as their corresponding metal recoveries, are presented in

Table 6. Experimental recoveries values of metal species in leaching solutions. Selected experimental conditions for acid concentration, solid to liquid ratio and temperature levels were designed by using Stat-Ease prior to experiments (next paragraph).

Run	Acid Concentration (mol/L)	Solid to Liquid Ratio (%)	Temperature (°C)	Mg (%)	Al (%)	Cr (%)	Mn (%)	Fe (%)	Co (%)	Ni (%)
1	2.0	10	60	69	26	25	59	27	65	79
2	5.5	40	90	64	26	25	58	46	71	76
3	12.0	13	30	31	13	21	51	68	65	78
4	5.5	10	20	18	7	9	36	14	42	53
5	7.0	25	55	68	31	31	65	71	78	85
6	2.0	40	50	17	3	7	27	5	30	37
7	12.0	36	83	123	65	58	104	115	118	124
8	7.0	25	55	66	27	32	62	67	79	86
9	5.5	10	20	21	7	12	42	17	50	62
10	8.6	25	20	20	7	12	41	41	56	64
11	7.0	25	55	55	27	33	66	68	79	78
12	10.2	10	90	85	44	48	84	86	100	95
13	8.7	13	30	19	8	16	43	49	60	58
14	8.7	40	55	63	27	31.	66	74	79	77
15	8.5	25	90	90	45	53	78	82	81	82
16	2.0	23	90	28	0	1	34	7	37	37
17	12.0	36	85	80	39	38	71	74	80	74
18	5.1	40	20	19	6	10	40	17	49	54
19	7.0	25	55	70	28	35	71	83	79	83
20	10.4	40	20	21	8	15	46	55	59	59
21	7.0	25	55	67	27	32	70	78	78	82
22	5.0	10	90	110	49	51	96	96	100	106
23	2.0	27	20	18	7	9	41	13	47	54

Note: There are some values greater than 100% recoveries. This is explained by unavoidable uncertainties in sampling and characterization of such samples (refer to Section 3.1.3). The order of magnitude of the uncertainty on recoveries (a quotient) is as expected.

. Results show mass percentages of the leached metals and include Mg, Al, Cr, Mn, Fe, Co, and Ni. The range of recoveries is broad for each individual metal and covers from almost zero to a full recovery. This is a good indication that the test design and parameters ranges were well selected, hence the effect of each parameter could be observed.

Knowledge of the mineralogy of the feed and the applicable mechanisms in HCl acidic medium can help better understand the leaching recoveries, while statistical models offer a quantitative expression of the measured effects. Asbestos tailings consist of a structure formed by the alternate stacking of tetrahedral silica layers and octahedral brucite [Mg(OH)₂] layers, which are linked into sheets. Under acidic conditions, the brucite layers of Mg(OH)₂ dissolve from the asbestos, leaving behind the silica as a residual product. Leaching involves the dissolution of minerals using HCl. The dissolution reaction of Mg²⁺ ions leached from asbestos tailings in HCl solution can be expressed as follows (Equation (2)):

$${\mathrm{M}\mathrm{g}}_3{\mathrm{S}\mathrm{i}}_2{\mathrm{O}}_5\left({\mathrm{O}\mathrm{H}}_4\right)+6\mathrm{H}\mathrm{C}\mathrm{l}\to 3\mathrm{M}\mathrm{g}{\mathrm{C}\mathrm{l}}_2+2\mathrm{S}\mathrm{i}{\mathrm{O}}_2+5{\mathrm{H}}_2\mathrm{O}$$

(2)

In this reaction, the magnesium silicate minerals react with hydrochloric acid to produce magnesium chloride, silica, and water. Hydrochloric acid also reacts with various impurities to produce different products, according to the following equations (Equations (3)–(5)):

$${\mathrm{F}\mathrm{e}}_2{\mathrm{O}}_3+6\mathrm{H}\mathrm{C}\mathrm{l}\to 2\mathrm{F}\mathrm{e}{\mathrm{C}\mathrm{l}}_3+3{\mathrm{H}}_2\mathrm{O}$$

(3)

$$\mathrm{C}\mathrm{a}\mathrm{O}+2\mathrm{H}\mathrm{C}\mathrm{l}\to \mathrm{C}\mathrm{a}{\mathrm{C}\mathrm{l}}_2+{\mathrm{H}}_2\mathrm{O}$$

(4)

$${\mathrm{A}\mathrm{l}}_2{\mathrm{O}}_3+6\mathrm{H}\mathrm{C}\mathrm{l}\to 2\mathrm{A}\mathrm{l}{\mathrm{C}\mathrm{l}}_3+3{\mathrm{H}}_2\mathrm{O}$$

(5)

3.2.2. Statistical Analysis of Leaching

An analysis of variance (ANOVA) was conducted to determine the effects of individual parameters and their interactions on metals recoveries.

Table 7. ANOVA for Mg leaching behavior from the asbestos tailings using HCl using Stat-Ease.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	18,705.92	5	3741.18	14.71	<0.0001	Significant
A: HCl concentration	1573.45	1	1573.45	6.19	0.0235
B: Solid to liquid ratio	283.08	1	283.08	1.11	0.3062
C: Temperature	12,073.9	1	12,073.9	47.48	<0.0001
AB	1199.19	1	1199.19	4.72	0.0443
AC	1081.88	1	1081.88	4.25	0.0548
Residual	4323.02	17	254.3
Lack of Fit	3177.61	10	317.76	1.94	0.1954	Not significant
Pure Error	1145.41	7	163.63
Cor Total	23,028.93	22

and

Table 8. ANOVA for Fe leaching efficiency from the asbestos tailings using HCl using Stat-Ease.

Source	Sum of Squares	df	Mean Square	F-Value	p-Value
Model	18,885.06	9	2098.34	9.23	0.0002	Significant
A: HCl concentration	10,558.25	1	10,558.25	46.45	<0.0001
B: Solid to liquid ratio	251.43	1	251.43	1.11	0.3121
C: Temperature	4187.34	1	4187.34	18.42	0.0009
AB	262.69	1	262.69	1.16	0.3019
AC	234.32	1	234.32	1.03	0.3285
BC	843.73	1	843.73	3.71	0.0762
A2	904.44	1	904.44	3.98	0.0675
B2	109.57	1	109.57	0.482	0.4997
C2	1009.37	1	1009.37	4.44	0.0551
Residual	2955.09	13	227.31
Lack of Fit	1753.6	6	292.27	1.7	0.2507	Not significant
Pure Error	1201.49	7	171.64
Cor Total	21,840.15	22

present the results for magnesium (Mg), the most valuable component of the feedstock, and iron (Fe), the primary impurity (or co-product eventually, depending on the easiness of removal as a useful product downstream). The F-values, along with their associated p-values, indicate the strength and significance of each factor: p-values < 0.005, 0.01, and 0.05 correspond to confidence levels of at least 99.5%, 99%, and 95%, respectively [89]. These values suggest that the model constitutive terms are meaningful, and the residual errors are negligible. In this analysis, HCl concentration (A) and temperature (C) were identified as significant contributors. A strong interactive effect was observed between acid concentration (A) and the solid-to-liquid ratio (B), indicating that the influence of one depends on the level of the other. The overall model fit was sound, with no significant lack of fit. Despite varying levels of statistical confidence, acid concentration and temperature consistently played critical roles in Mg recovery. The smallest p-values for Mg and Fe highlight these two factors as the most influential. For Mg, the interactions AB (acid concentration × solid-to-liquid ratio) and AC (acid concentration × temperature) were important. For Fe, the same interactions were relevant, along with BC (solid-to-liquid ratio × temperature). A more detailed discussion of these interactive effects follows in the next subsections.

The following equation (Equation (6);

Table A7. Key factors and statistical values for the RSM.

Element	Significant Factors	Model p-Value	Factor p-Value	Actual Equation	R2
Mg	[HCl]	<0.0001	0.0031	59.5716 − 6.582 × [HCl]	0.9225
	[S/L]		0.0525	−1.6775 × [S/L]
	T		<0.0001	+0.3260 × T
	[HCl] × [S/L]		0.0228	+0.0193 × [HCl] × [S/L]
	[HCl] × T		0.0112	+0.0781 × [HCl] × T
Ni	[HCl]	0.0076	0.0037	41.7971 + 2.7266 × [HCl]	0.7560
	[S/L]		0.0366	−0.3259 × [S/L]
	T		0.0071	+0.3751 × T
Cr	[HCl]	<0.0001	<0.0001	−17.3530 + 2.9614 × [HCl]	0.8657
	[S/L]		0.0495	−0.2017 × [S/L]
	T		<0.0001	+0.8469 × T
	[HCl] × T		0.0036	+0.0559 × [HCl] × T
	[S/L] × T		0.0749	−0.0121 × [S/L] × T
	[HCl]2		0.0665	−0.0089 × [HCl]2
	T2		0.0380	−0.0060 × T2
Co	[HCl]	0.0002	<0.0001	25.2560 + 3.3626 × [HCl]	0.8688
	[S/L]		0.0390	−0.1962 × [S/L]
	T		0.0002	+0.4561 × T
Fe	[HCl]	0.0001	<0.0001	−74.1267 + 10.7060 × [HCl]	0.8791
	[S/L]		0.2121	+1.2700 × [S/L]
	T		0.0005	+2.1591 × T
	[HCl] × [S/L]		0.3679	+0.0909 × [HCl] × [S/L]
	[HCl] × T		0.4028	+0.0365 × [HCl] × T
	[S/L] × T		0.0422	−0.0197 × [S/L] × T
	[HCl]2		0.0286	−0.5914 × [HCl]2
	[S/L]2		0.4755	−0.0226 × [S/L]2
	T2		0.0368	−0.0127 × T2
Mn	[HCl]	0.0001	0.0018	46.1771 − 0.8993 × [HCl]	0.7821
	[S/L]		0.1196	−0.2866 × [S/L]
	T		0.0001	+0.0464 × T
	[HCl] × T		0.0426	+0.0612[HCl] × T
Al	[HCl]	<0.0001	0.0017	29.6489 − 3.6024 × [HCl]	0.8211
	[S/L]		0.2583	+0.7849 × [S/L]
	T		<0.0001	−0.02145 × T
	[HCl] × [S/L]		0.0862	+0.0852 × [HCl] × [S/L]
	[HCl] × T		0.0055	+0.06314 × [HCl] × T

Legend: [HCl]: HCl concentration (mole/L); [S/L]: Solid to liquid ratio (%); T: Temperature (°C).

in Appendix B) was proposed to predict magnesium (Mg) recovery based on acid concentration, temperature, and the solid-to-liquid ratio. Since the coefficients in the equation reflect the importance of each parameter, temperature has the greatest impact.

$$\begin{array}{cc}\hfill \mathrm{M}\mathrm{g} \mathrm{r}\mathrm{e}\mathrm{c}\mathrm{o}\mathrm{v}\mathrm{e}\mathrm{r}\mathrm{y}\left(\%\right)& \\ & =-6.58\left(\mathrm{H}\mathrm{C}\mathrm{l} \mathrm{C}\mathrm{o}\mathrm{n}\mathrm{c}.\right)-1.68\left(\mathrm{S}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{d} \mathrm{t}\mathrm{o} \mathrm{l}\mathrm{i}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{d} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\right)\hfill \\ & +0.33\left(\mathrm{T}\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}\right)+0.19\left(\mathrm{H}\mathrm{C}\mathrm{l} \mathrm{C}\mathrm{o}\mathrm{n}\mathrm{c}.\right)\left(\mathrm{S}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{d} \mathrm{t}\mathrm{o} \mathrm{l}\mathrm{i}\mathrm{q}\mathrm{u}\mathrm{i}\mathrm{d} \mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\right)\hfill \\ & +0.08\left(\mathrm{H}\mathrm{C}\mathrm{l} \mathrm{C}\mathrm{o}\mathrm{n}\mathrm{c}.\right)\left(\mathrm{T}\mathrm{e}\mathrm{m}\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{u}\mathrm{r}\mathrm{e}\right)+59.57\hfill \end{array}$$

(6)

3.2.3. Effect of Individual Factors on Metal Recovery

Figure 3 shows how the concentration of acid, the solid-to-liquid ratio, and temperature affect the recovery of magnesium (Mg), the primary metal of interest, and iron (Fe), the main impurity. Increasing the HCl concentration positively affects the leaching of both metals, especially Fe recovery, up to approximately 8 M. Beyond this concentration, however, the effect plateaus. Increasing the HCl concentration from 2 M to 12 M improved Mg recovery from 40% to 65% and Fe recovery from 20% to 75%. This indicates that higher acid concentrations significantly enhance metal extraction. A similar pattern is observed with temperature: higher temperatures increase the leaching efficiency of both Mg and Fe, though the effect on Fe diminishes beyond 55 °C. In contrast, the influence of the solid-to-liquid ratio appears relatively minor and slightly negative. The observed trends in Mg recovery as a function of acid concentration and temperature align with the findings of Baigenzhenov et al. [56], who reported up to 92% Mg extraction from serpentine using 25% HCl (6.8 M), a solid-to-liquid mass ratio of 4:1, and three hours of leaching at 80 °C. Similarly, Shayakhmetova et al. [47] achieved maximum Mg and Fe recoveries of 96.0% and 87.8%, respectively, in a pilot-scale HCl leaching trial using 18% HCl (5.4 M) on chrysotile asbestos waste at 85 °C for two hours.

A significant factor in the leaching process is the concentration of reagents, as it affects both the rate and efficiency of metal extraction from the feed. The positive effect of acid concentration is explained by referring to Equations (2) and (3) and knowing the fact that the more the reagent is available, the better chance of pushing the leaching reaction forward. The impact of reagent concentration on the leaching efficiency in HCl system can be attributed to both H⁺ and Cl⁻ ions. The increase in hydrogen ions significantly influences the rate of metal extraction by making the system more acidic (improving the acidolysis leaching mechanism). However, the chloride anions are strong complexing reagents and can increase the solubility and mobility of the metals in the system. Therefore, Cl⁻ concentration is important when evaluating the behavior of dissolved elements. In addition to the thermodynamics of the system, it was suggested that kinetic limitations and the need for the reagents to diffuse through the unreacted core of the chrysotile particles are the other major role players [47]. Having more reagent volume available in the bulk of the system can potentially help with their diffusion into the particles and increase the metal recovery.

Although the impact of pulp density is limited, at higher ratios (e.g., 40%), the recovery of magnesium (Mg) and iron (Fe) decreases to approximately 50%. This is primarily due to reduced acid availability per unit mass of solids. The limited liquid phase restricts contact between the reagents and solid particles, hindering dissolution. Conversely, lower solid-to-liquid ratios increase the amount of available acid relative to the solids, promoting more efficient leaching. While this parameter’s influence seems modest compared to acid concentration or temperature, its importance becomes clearer in the context of parameter interactions, to be examined in the next subsection.

Increased temperatures from 20 °C to 90 °C prompted metal extraction and increased Mg recovery from 20% to 80%. This improvement is largely because of the increased reactivity of acids with the brucite serpentine moiety at higher temperatures. The greater abundance of magnesium relative to iron in the feedstock may also explain why temperature has a more pronounced effect on magnesium recovery. Baigenzhenov et al. [60] reported that the chemical interaction between hydrochloric acid and the silicate matrix is significantly promoted during chrysotile HCl leaching at elevated temperatures (40 °C to 100 °C). Their thermodynamic analysis revealed that the Gibbs free energy (ΔG) of the reactions governing SiO₂ dissolution and chrysotile decomposition becomes increasingly negative with rising temperatures, thereby enhancing metal liberation, especially magnesium. In contrast, iron (Fe) recovery showed only modest improvement beyond 70 °C. This plateau may be due to the accumulation of FeCl₃ in the solution approaching saturation, as described by Equation (3), thereby limiting further dissolution.

3.2.4. Analysis of Interactive Factors on Metal Recovery

The efficiency of metal recovery can be affected by more than a single factor effects. Interactions between factors like acid concentration and solid to liquid ratio (AB), acid concentration and temperature (AC), and solid-to-liquid ratio and temperature (BC) were studied here. Figure 4 illustrates the interactive influence of acid concentration and temperature on Mg and Fe recovery in both 2D and 3D view when solid-to-liquid ratio is held constant. The results clearly indicate that the interaction effect of these factors is significant and positive. This means that the effect of temperature increases with rising acid concentration, suggesting that both higher acid concentration and elevated temperature enhance metal recovery. On the other hand, at lower temperatures, HCl concentration did not affect Mg recovery significantly, where this effect of acid concentration on Mg recovery was almost 50% greater when temperature was higher, which agrees with the findings of other researchers in literature [75,76]. A similar behavior, but less intense, was observed for Fe leaching. Both parameters of acid concentration and temperature, up to 70 °C, increase Fe leaching efficiency and temperature seems to have no influence on Fe leaching when acid concentration is lower than 4 M. To see the effect of parameters on the recovery of other metals, refer to Appendix B.

Higher acid concentrations and elevated temperatures enhance metal dissolution by increasing proton activity, accelerating metal dissolution kinetics, as well as making thermodynamics more favorable [47,56]. While these conditions favor the release of Mg²⁺, they also lead to greater acid consumption and higher neutralization demands [75].

The dual effect of acid concentration and solid-to-liquid ratio at a constant temperature on Mg and Fe recovery is presented in Figure 5. Changing the solid-to-liquid ratio from 10% to 40% has little negative effect on magnesium recovery when acid concentration is at the lowest level. This influence was slightly different when a high solid to liquid ratio of 40% was used. In this condition, an increase in the molarity of acid could meaningfully influence the Mg recovery: the increase was from about 50% at 2 M to almost 100% at 10 M. At higher solid to liquid ratio, more solid is available to be leached therefore there is higher concentration of dissolved metals in the system compared to when there is less feed available to be leached. The behavior for iron in response to the interactions between acid concentration and solid to liquid ratio is as expected. In the graph related to Fe leaching, both parameters of solid-to-liquid ratio and acid concentration have an influence on metal recovery. The graph reveals that as acid concentration increases and solid-to–liquid ratio goes down, iron recovery increases significantly.

The effect of the interaction between temperature and solid to liquid ratio on leaching behavior is shown in Figure 6. The graphs reveal that elevated temperatures and higher solid to liquid ratios lead to greater Mg recovery. However, the parameters do not seem to have any interactions (as BC was not selected as a significant interaction in the ANOVA Table 6). The effect of temperature suggests that Mg release in leachate is controlled by the activation energy of the reaction. As temperature increases, the thermal excitation of many more species drives the reaction to overcome the activation barrier, and the reagent becomes more mobile (diffusion and transport of the lixiviant and ions become faster). The positive effect of a higher solid-to-liquid ratio on Mg recovery may indicate that the particle-particle interactions stimulate further the interparticle mass transfer of secondary Mg-products out of the particle surface and result in a new fresh surface ready to react with the leaching acid. In the case of Fe, solid-to-liquid ratio has no meaningful effect on metal recovery if temperature is less than about 50 °C; however, as temperature increases, a negative effect of solid-to-liquid ratio becomes evident. This may be due to reduced reagent availability at higher solid-to-liquid ratios, limited reagent access to reactive sites, and/or the formation of stable, non-friable, iron-rich precipitates on leached particle surfaces. This indicates that a significant fraction of the iron present in the feed sample is occurring in minerals that do not contain Mg. The mineralogy shows a significant amount of magnetite which could explain that.

The graphs also show that an increase in the solid to liquid ratio leads to lower recovery of Fe while the recovery of Mg is not significantly affected by this factor. Noteworthy, applying both a high temperature and a high solid-to-liquid ratio (Figure 6) is even more favorable to high recovery of Mg and a lower recovery of Fe. This suggests the possibility to optimize the leaching of Mg in a selective way regarding Fe.

3.2.5. Behavior of CSM (Cr, Ni, Co, Mn) and Al

Although chromium behaves similarly to iron under leaching conditions, it is more challenging to recover (Figure A8, Figure A9, Figure A11, Figure A12 and Figure A13 in Appendix B). Higher chromium recovery is observed at elevated temperatures and lower solid-to-liquid ratios (Figure A8, Figure A9, Figure A11 and Figure A13). This may be due to the refractory nature of chromite, the primary chromium-bearing mineral in the feedstock, indicating that the phase controlling chromium leaching is different from the phase governing iron dissolution.

The behavior of Al, Ni, Co, Mn under leaching conditions is close to Mg one regarding HCl concentration and temperature (Figure A10, Figure A11, Figure A12 and Figure A13 in Appendix B). However, a lower solid to liquid ratio is favoring Al, Ni, Co, Mn release.

3.2.6. Parameter Optimization

The desirability function approach was used to determine the maximum values of input variables to achieve optimal performance for one or more responses (metal leaching efficiencies). During the optimization phase, each response is translated into a unique desirability function. The highest ideal response is represented by a value of 1, with the function ranging from 0 to 1, intermediate value indicates results that are roughly acceptable [63,90,91].

Figure 7 illustrates the independent variable in response surface methodology. By adjusting the parameters for maximum magnesium (Mg) leaching, we determined the operating conditions and the desired outcome. The model predicts full Mg recovery under operating conditions involving HCl concentrations ranging from 10 to 12 M, a solid-to-liquid ratio ranging from 33.6% to 40%, and a temperature of 90 °C. The desirability value of 0.80 demonstrates the applicability of the developed model.

3.3. Experimental Validation and Sensitivity Analysis Around Optimal Experimental Conditions

The optimal conditions for a high leaching recovery as defined during the first leaching series and their RSM modeling (W-series) were selected to carry out the second series of leaching tests (D-series). The aim here was to check experimentally the optimal conditions and to evaluate the robustness of the process, or its sensitivity to changes in temperature or concentration of acid, around the optimal values. Four triplicate leaching tests with highly concentrated hydrochloric acid (HCl) at different concentrations, i.e., 10 N and 12 N with two different temperatures i.e., 80 °C and 90 °C, were carried out in batch mode. The results obtained are shown in

Table 9. Average results of triplicate trials with 10 N at 80 °C (Tests 24 A, B, C).

		% Recovery
		Mg	Al	Cr	Mn	Fe	Co	Ni
Analysed Feed		90	62	66	81	92	88	92
		(σ = 2)	(σ = 3)	(σ = 3)	(σ = 10)	(σ = 2)	(σ = 2)	(σ = 2)
Calculated Feed		88	59	64	83	91	87	91
		(σ = 3)	(σ = 2)	(σ = 3)	(σ = 8)	(σ = 3)	(σ = 2)	(σ = 2)
Mean		89	61	65	82	92	88	92
Sigma tot		4	4	4	13	4	3	3

,

Table 10. Average results of triplicate trials with 10 N at 90 °C (Tests 25 A, B, C).

		% Recovery
		Mg	Al	Cr	Mn	Fe	Co	Ni
Analysed Feed		92	65	71	90	94	90	93
		(σ = 1)	(σ = 2)	(σ = 1)	(σ = 1)	(σ = 1)	(σ = 1)	(σ = 2)
Calculated Feed		91	61	69	90	93	92	92
		(σ = 2)	(σ = 1)	(σ = 1)	(σ = 1)	(σ = 1)	(σ = 3)	(σ = 2)
Mean		92	63	70	90	94	91	93
Sigma tot		2	2	1	1	1	3	3

,

Table 11. Average results of triplicate trials with 12 N at 80 °C (Tests 26 A, B, C).

		% Recovery
		Mg	Al	Cr	Mn	Fe	Co	Ni
Analysed Feed		88	64	69	82.5	91.5	87.5	91
		(σ = 1)	(σ = 3)	(σ = 1)	(σ = 0.5)	(σ = 0.5)	(σ = 0.5)	(σ = 1)
Calculated Feed		86	56	66	83.5	90.5	86.5	91
		(σ = 1)	(σ = 1)	(σ = 2)	(σ = 0.5)	(σ = 0.5)	(σ = 0.5)	(σ = 1)
Mean		87	60	68	83	91	87	91
Sigma tot		1	3	2	1	1	1	1

and

Table 12. Average results of triplicate trials with 12 N at 90 °C (Tests 27 A, B, C).

		% Recovery
		Mg	Al	Cr	Mn	Fe	Co	Ni
Analysed Feed		92	77	71	86	94	91	93
		(σ = 1)	(σ = 15)	(σ = 3)	(σ = 2)	(σ = 1)	(σ = 0.5)	(σ = 1)
Calculated Feed		90	72	68	86	93	90	93
		(σ = 1)	(σ = 18)	(σ = 3)	(σ = 2)	(σ = 1)	(σ = 1)	(σ = 1)
Mean		92	75	70	86	94	91	93
Sigma tot		1	23	4	3	1	1	1

. Recoveries are calculated both according to the geochemistry of the analyzed feed and of the back-calculated feed (Table 9, Table 10, Table 11 and Table 12).

By setting the percentage of solids at 36% and varying the temperature and hydrochloric acid (HCl) concentration, the results showed that the percentage of metal dissolution in the leachate changes significantly as a function of varying temperature and acid concentration. An increase in temperature can accelerate the metal leaching process, with increments of 10 °C. Especially at the level of Mg recovery. This temperature effect is also related to HCl concentration; the higher the HCl concentration, the better the metal recovery. Acid concentration therefore has a positive effect on metal leaching, including leaching of Al, Cr, Mn, Fe, Co and Ni.

The extent of leaching for the two best conditions (10 N and 12 N at the same temperature of 90 °C) was calculated based on the ratio of extracted metals to initial metal contents. For Mg, Al, Cr, Mn, Fe, Co and Ni, the leaching degree for the condition of 10 N at 90 °C is in the order of 92% ± 2%, 63% ± 2%, 70% ± 1%, 90% ± 1%, 94% ± 1%, 91% ± 1% and 93% ± 3%, respectively. For the condition of 12 N at 90 °C, the leaching degree is, correspondingly, in the order of 92% ± 1%, 75% ± 23%, 70% ± 4%, 86% ± 3%, 94% ± 1%, 91% ± 1% and 93% ± 1%. As shown in Figure 8, there are no statistically significant differences in the leaching results for the two conditions.

According to Figure 8, the values of these elements are closer to those obtained by DOE. However, there is still some un-leached Mg, probably due to the presence of some incompletely leached Mg-minerals. For the other metals leached, such as Mn, Al, Fe, Co and Ni, the percentage recovery value appears to be significant, except for chromium, with a leaching rate of around 71%. The low recovery of the latter can be explained by the host mineral of Cr which is chromite, a refractory mineral that resists acid attack and high temperatures. As the material used contains mostly 40–50% chrysotile, acid leaching can transform chrysotile into amorphous hydrated silica at such high HCl concentrations.

As shown in Figure 9, a rapid decrease in the acid concentration of the leachate was observed between 0 and 20 min, dropping from 12 N to 3.5 N. Thereafter, the residual acid concentration gradually stabilized, reaching approximately 2.8 N after 80 min. In parallel, the recovery percentage of certain metals such as Mg, Fe, Co, and Ni increased significantly during the first 20–40 min, reaching about 90%, and then stabilized between 87 and 92% beyond 80 min, except for Cr, which behaved as a refractory element. We observed that the strong acid consumption during the first 20 min corresponds to the rapid recovery phase of the metals, while the subsequent stabilization of both parameters reflects on the one hand, an equilibrium between the residual acid and the remaining possible reactions, and on the other hand, a probable passivation of certain reactive surfaces.

Figure 10 illustrates the comparison between the results predicted by the DOE model and the experimental results obtained under fixed conditions (HCl = 12 N, temperature = 90 °C, and solid content = 36%). Overall, the two curves exhibit a satisfactory agreement, thereby confirming the relevance of the model. However, slight discrepancies are observed for certain elements, particularly Cr and Mn, which can be attributed to experimental variability as well as to the inherent limitations of the model predictions.

4. Discussion

The DOE study, based on a central composite design response surface methodology (CCD RSM), identified that, among the 3 factors studied, the most influential factors for Mg recovery are the temperature and HCl concentration. A concentration of HCl of at least 10 M is necessary to reach a total Mg recovery when the temperature is above 80 °C and the solid-to-liquid ratio is above 30%. The mathematical representation predicts that increasing the temperature even above 90 °C could be beneficial for a selective recovery of Mg. Moreover, while high temperature leads to higher Mg recovery, it limits the leaching of iron. Avoiding iron leaching is desirable because it complicates the subsequent purification stage of the leaching process. If the limitation of iron release is wished, the determination of the main mineral that releases iron in the leachate becomes crucial. Indeed, if this mineral is magnetite, as revealed by the mineral characterization, it could be beneficial to the whole process to remove it by low intensity magnetic separation before the leaching stage. As observed in this study, magnetite should however be liberated from Mg-bearing minerals which are mainly serpentine minerals. This can be done by a targeted grinding to the liberation mesh of magnetite. This study also predicts that, when the solid-to-liquid ratio is high, the process may be more sustainable due to improved selective recovery of magnesium (at the expense of iron) and the possibility of regenerating the acid. Furthermore, this study revealed that nickel, the primary CSM constituent, is more prevalent in the feed than in the Earth’s crust, and exhibits leaching behavior like magnesium’s. Chromium, the second most enriched CSM, exhibits a leaching response in concentrated HCl similar to that of iron. However, this study suggests that this response is not primarily associated with the iron-rich mineral phase. Elevated temperatures enhance its dissolution. Since the primary chromium-bearing phase in asbestos tailings is non-magnetic chromite, using a magnet to separate iron minerals before leaching is likely effectless in terms of chromium recovery. Nevertheless, chromium recovery is expected to be lower than nickel recovery because chromium is found within the refractory spinel structure of chromite [84].

The combination of high HCl concentrations and high temperatures was necessary for this ore to maximize the recovery of its Mg, Ni, Cr contents. This result is consistent with the ones obtained by Cheng and Lamy-Morissette [49] on some asbestos tailings of the same region. Although the geochemistry of the major elements and the CSM they provided are similar to those in our current study, the description of the mineralogy of the feed sample is insufficient to determine whether it could have originated from the same area or tailings pond.

The use of high-strength acid for magnesium recovery likely stems from multiple factors. Notably, the asbestos tailings examined contain a relatively high proportion of chrysotile compared to other deposits in the Thetford region, suggesting a probable origin from the Jefferson mine. This reflects the compositional heterogeneity of asbestos tailings across mining sites [14,28], a variability demonstrated at the scale of individual stockpiles of other mine sites [91]. Since such details are not well documented in academic literature, a more comprehensive characterization of asbestos tailings piles is recommended. When treating tailings as a resource for hydrometallurgical extraction, they should be approached as complex geological formations, with distinct mineralogical and geochemical variations that can influence processing performance. For instance, variations in chrysotile content affect pulp rheology [92,93,94], and the resulting increase in viscosity in concentrated HCl may hinder the efficient recovery of dissolved ions.

Second, the tailings underwent leaching without grinding beforehand to minimize processing costs. While this approach is relevant, the economics must be revised once the leaching and purification of pregnant leach solutions behaviors of the ore are better understood. Adding grinding to the process could improve magnetite liberation and subsequent removal by magnetic separation prior to leaching. This would result in lower iron recovery, possibly benefiting the entire process. However, milling further such tailings may also present some drawbacks, according to our experience in mineral processing. First, the presence of chrysotile fibers makes milling of such rocks less efficient than other hard-rock ores [95]. This is due to the rheological behavior of rocks containing such minerals: their presence induces high-yield stress slurries [94,95,96]. Solutions could be to disintegrating fibers [94] or removing fibers from the process feed, by industrial flotation [94,97] prior to milling but both solutions add major complexities to the flowsheet, and thus cost, and potential loss of recovery downstream. The Mössbauer study also shows that floated fibers still bear magnetite, thus showing that iron impurities will still be present in the leaching feed even after flotation of fibers and subsequent magnetic separation of magnetite. On the other hand, milling is the process stage that consumes the most of energy [94,95] cost. This is not wished for the reprocessing of low-grade tailings. Adding regrinding to the flowsheet will increase carbon tax and make the process somewhat less sustainable. Finally, while the project progressed, iron was shown to be a potential valuable by-product. Considering all these points, the physical separation effort of magnetic minerals was not pursued for this study. It finally was used only for qualitative mineralogy assessment. Instead, the research effort was put on the chemical separation of iron in the downstream hydrometallurgical flowsheet.

Third, this study also suggests that the acid strength could be decreased if the temperature of HCl is further increased. Aside from economics, the boiling temperature of aqueous HCl solutions, according to their concentration, will be the limiting factor.

5. Conclusions

This study investigated the recovery of magnesium and other critical (such as nickel, chromium and cobalt) and strategic metals (CSMs) from an asbestos tailings sample through hydrochloric acid leaching. response surface methodology (RSM), combined with central composite design (CCD), was employed to optimize the recovery conditions. The model provided a satisfactory fit to the experimental data. Additionally, multiple response optimization using the desirability function was applied to determine the optimal leaching conditions for magnesium, iron, and nickel. The optimal parameters—hydrochloric acid concentration of 12 N, leaching temperature of 90 °C, and a solid-to-liquid ratio of 33.6%—resulted in a magnesium recovery slightly more than 90%. Under these conditions, approximately about 90% of both iron and nickel were also leached, together with 70% Cr. Results were interpreted and discussed according to liquid-solid interactions, including the particle size of tailings grains and the mineralogy of asbestos tails.