Aqueous Carbonation of Waste Incineration Residues: Comparing BA, FA, and APCr Across Production Scenarios

1. Introduction

In recent decades, mineral carbonation—also referred to as CO₂ mineralization—of industrial alkaline waste (IAW) has evolved from a promising waste management strategy into a commercially viable solution [1,2]. Accelerated Carbonation Technology (ACT) not only stabilizes these wastes but also significantly reduces CO₂ emissions, all while producing valuable byproducts for reuse [3,4]. Leading companies include Carbon8 and O.C.O Technology (UK) in the Municipal Solid Waste Incineration (MSWI) and cement sectors [5], Orbix (Belgium), Mineral Carbonation International (Australia), and RHI Magnesita in the steel and refractory products sectors [6], as well as Neustark (Switzerland), Heidelberg Materials (Germany), and Carbonaide (Finland) in the concrete sector [7]. Aqueous carbonation is widely favored for its scalability, as it requires milder conditions than gas-solid carbonation and enables further intensification through additives, catalysts, process integration, alternative energy, and innovative reactor designs [8]. While the scale-up of ACT is ongoing, relevant efforts are still required to fully realize its potential in addressing key environmental challenges related to carbon capture and storage (CCS), waste management, and the circular economy at large.

In particular, the chemical and mineral heterogeneity, along with the toxicity of MSWI residues (MSWIr), presents a significant challenge to their widespread integration into this emerging industry [9,10]. MSWI generates three main types of alkaline solid residues: bottom ashes (BA), fly ashes (FA), and air pollution control residues (APCr) [11]. Their production scenarios—ranging from the initial MSW feedstocks to their collection and storage at the outlet of the incineration process—along with factors such as the waste inflow, combustion conditions, process layout and solid additives, significantly influence their physicochemical properties, which ultimately determine their suitability for ACT. Furthermore, increasing quantities of sorted MSW have led to higher production of refuse-derived fuels (RDF), which are then incinerated. This results in BA with distinct characteristics compared to non-pretreated MSWI [12,13,14,15,16,17]. Key factors include grain sizes, mineral composition, and leachability, which all play a critical role in any carbonation process [18,19].

Figure 1 illustrates the MSWI process diagram for a typical electricity and heat cogeneration plant, highlighting the collection points for BA, FA and APCr. BA are collected beneath the incineration chamber, while FA are typically retrieved from the boiler and electrofilters [20,21,22]. The finer fractions of BA are both toxic and highly reactive during carbonation, thus presenting a challenge for their management but also an opportunity for potential use [23,24]. FA are defined as “the particulate matter carried over from the combustion chamber and removed from the flue gas stream prior to the addition of any type of sorbent material” [25]. If sorbent additives are used upstream, electrofilter residues should be classified as APCr. Additionally, residues gathered in the bag filters located downstream of the gas cleaning system are likewise categorized as APCr [26]. Residues from the boiler, electrostatic filters, and bag filters are either blended or maintained apart in big bags or silos by the facilities. Either lime or sodium bicarbonate are typically employed as sorbent for acid gas removal, alongside lignite powder to control dioxins and mercury emissions [27,28]. Lime plays a crucial role in forming calcium (chloro)(hydr)oxide phases (CCHO), which are highly reactive in FA, and APCr [29]. Ca(OH)₂ is widely recognized as the primary driver of carbonation in BA, FA, and APCr, although in most cases, Ca-aluminates serve as the main source of calcium in BA [30,31,32,33]. The products of the acid gas removal reactions involving lime, HCl and Cl_2, are identified as a solid solution of CaClOH and CaCl₂·2H₂O [34,35]. These compounds also play a key role as Ca-providers in the carbonation of FA and APCr [36,37,38,39].

While ACT has been commercially implemented for FA and APCr by UK companies such as Carbon8 and OCO Technology, there are currently no successful industrial case studies involving BA. The most significant full-scale advancements were reported by THM University of Applied Sciences in Gießen and the Department of Geosciences at the University of Cologne [40,41,42]. Additional pilot-scale trials were reported by the Korea Institute of Geosciences and Mineral Resources (KIGAM) [43,44,45]. Besides, despite existing valorization solutions through ACT such as lightweight aggregate (LWA), cement and geopolymer manufacturing [46,47,48,49,50,51], as well as enhanced metal recovery (EMR) processes like FLUWA and FLUREC [52], FA and APCr are still largely disposed of in landfills, including underground repositories such as salt mines [53,54], as these methods remain less widespread due to logistical challenges and technological barriers. To support the sustainable development of aggregate production from BA, FA and APCr as a viable valorization pathway, advancements are required in removing potentially toxic elements prior to manufacturing and assessing the suitability of various MSWI streams. To achieve this, combination of EMR and ACT for advanced processing of diverse feedstocks—including both waste and naturally occurring materials—is gaining attention [55,56,57,58,59,60].

In this context, the present work aims to compare the reactivity of BA, FA, and APCr toward aqueous carbonation under typical MSWI production scenarios by evaluating:

• Total CO₂ uptake, assessed through measurements of carbonation capacity, carbonation degree achieved, and resulting efficiency.
• Kinetic behavior, evaluated by measuring the CO₂ absorption rate.
• Leachability of heavy metals (HMs), determined by analyzing HMs concentrations in the carbonated wastewater and assessing the stability of the carbonated product.

We selected five representative samples from three facilities—A, B, C, D, E. Each sample underwent thorough physicochemical characterization to correlate its properties with performance. Relying on previous studies, regular water washing and aqueous carbonation tests were conducted, with CO₂ bubbling as the sole variable [61]. The aim was to evaluate and compare the effectiveness of aqueous carbonation versus regular washing in the removal and stabilization of heavy metals (HMs). Carbonation tests were carried out under dynamic conditions, using high-precision gas flow sensors for continuous CO₂ flow monitoring, completing previous methods adapted to pressurized vessels [62]. A novel method, based on previous studies [63,64], was implemented to measure in-situ the degree of carbonation. A novel mechanism for Cr leaching during and after BA carbonation was proposed, and its implications for designing a more sustainable management process layout were discussed.

2. Materials and Methods

2.1. Chemical Reagents

The A, C, E samples were collected concurrently from the same French plant (plant 1), while B and D were provided by an Italian plant (plant 2) and another French plant (plant 3), respectively. Figure S1 illustrates the samples collected from Plant A, including all discarded coarse BA fractions, while Figure S2 shows the grain size distribution curve of these BA. Figure 2 illustrates the methodology used to select the samples, as well as the five chosen samples, which represent the most common MSWIr production scenarios. Ca(OH)₂, used as a standard to calculate the carbonation degree achieved during the experiments, was provided by ERBApharm with a purity of 98% and a specific surface area (SSA) of 16.035 m²/g ± 0.09 (R² = 0.999). CO₂ was provided by Sapio at a purity of 99.9 %. Ultrapure and degassed water was used throughout the experiments. All samples were collected a few days after generation, dried in an oven at 378 K and eventually analyzed.

The BA fraction under 0.25 mm, sample A, represents 11.15 wt.% of the total BA mass, as shown by Figures S1 and S2. This fraction was chosen for its superior specific surface area, calcium availability, and toxicity compared to coarser fractions (Caviglia, et al., 2019). The sample B originates from Italian plant 2 utilizing the NEUTREC® technology, which uses sodium bicarbonate (NaHCO₃) downstream as additive for acid gas removal. These residues are known for their high leachability of chlorides, sulfates, fluorides, and oxyanions (e.g. from As, Cr, Sb, Se) [27,66]. The leachate typically has a pH around 8.5, similar to NaHCO₃ in equilibrium with water. The APCr from plant 2 are normally processed to produce sodium hydroxide (brine) for the salt industry, by SOLVAL S.p.A. The sample C consists of boiler and electrofilter residues, while the sample E is composed of APCr from the bag filter of French plant 1. In plant 1, Ca(OH)₂ is injected by cyclones both before the electrofilter and the bag filter. The sample D derives from French plant 3 applying a semi-dry process injecting CaO and steam for acid gas removal [67]. Sample D results from combining boiler and electrofilter ashes with the APCr from the bag filter in shared storage silos.

2.2. Experimental Procedures

Each experiment was conducted in a double-wall Pyrex reactor with a capacity of 1.5 L, a height of 16.8 cm, and a diameter of 10 cm, equipped with a thermostatic bath, as reported in previous studies [63]. Each of the five samples undergone two tests: regular water washing and aqueous carbonation, with CO₂ bubbling as the sole difference. Ten experiments were therefore conducted, as listed in Table 2 with the following identifiers : A1, A2, B1, B2, C1, C2, D1, D2, E1 and E2. The detailed experimental conditions are provided in Table 1. Table 2 provides the experiment identifiers with their corresponding duration and liquid sample names. The methods for calculating the average CO₂ absorption rate during the aqueous carbonation experiments, the theoretical carbonation capacity, and the carbonation degree of the solid samples were developed and implemented in previous studies [63,64]. The relevant formulas are detailed in the Supplementary Information as Eqs. 1-6. Pure CO₂ was used at a constant partial pressure of 1.2 bar and a flow rate of 0.4 L/min. The stirrer speed was 300 rpm, with a MSWIr concentration of 2.5 wt.% (L/S ratio of 40). The temperature was maintained at 333K, within field-scale carbonation ranges. Reactants were combined with ultra-pure water under Ar atmosphere. Subsequently, the suspension was poured into the reactor. The solution was stirred thoroughly for at least 10 min to ensure proper mixing of the suspension. During this time, Ar was bubbled through the solution to prevent premature carbonation caused by exposure to ambient air. Before conducting the carbonation experiments —A1, B1, C1, D1, and E1—, CO₂ was flushed through the reactor to remove residual air. At t=0, CO₂ bubbling into the reactor began. These experiments were considered completed when the gas flow sensors indicated cessation of CO₂ absorption by the solution (i.e. inlet flow = outlet flow). Subsequently, regular water washing experiments —A2, B2, C2, D2, and E2— were carried out for a duration comparable to that of their respective aqueous carbonation counterparts. Upon completion of each experiment, 10 mL of the resulting wastewater was sampled, filtered and collected for further analysis. A portion of the solid samples was also filtered, dried, and subsequently collected for further analysis.

2.3. Analytical Methods

The SSA of the MSWIr samples was determined using a Micromeritics ASAP 2020 instrument, following the Brunauer-Emmett-Teller (BET) method with nitrogen adsorption at -196°C. Prior to analysis, samples were degassed at 150°C for 240 minutes. Elemental composition was measured via X-ray fluorescence (XRF) using a Rigaku Supermini200. To identify and quantify the solid-state phases, X-ray powder diffraction (XRPD) was performed with a Rigaku MiniFlex 600 benchtop diffractometer, scanning between 3° and 90° at a speed of 0.2°/min. Three separate XRPD data collections were performed on the raw APCr samples, using the Rietveld method with corundum as internal standard for phase quantification. Surface structure and morphology were investigated using a TESCAN VEGA 3 scanning electron microscope (SEM) paired with Oxford Xplore 15 energy dispersive X-ray spectroscopy (EDS) for elemental mapping. TCLP leaching tests were conducted following the LS EN-12457 standards. For ion quantification in solution, a Metrohm 883 Basic IC Plus with an 863 Compact autosampler was used in conjunction with Metrosep C/A Supp 4 – 250/4.0 columns, detecting cations and anions including sodium, potassium, calcium, magnesium, chloride, sulfate, and bromide. For HMs detection, inductively coupled plasma mass spectrometry (ICP-MS) was employed, utilizing a Thermo Fisher iCAP-TQe system with an ASX 560 autosampler. Al, As, Cd, Cu, Co, Cr, Fe, Mn, Ni, Pb, Sb, Zn were detected. After filtering and diluting the samples in ultrapure water, they were acidified with 0.1% nitric acid (HNO₃), and calibration was performed using five standards ranging from 1 to 50 ppm. pH levels were measured with a Hanna HI H-ORP meter, while electrical conductivity (EC) was assessed using a Mettler Toledo FiveEasy EC meter. To monitor the CO₂ flow and total uptake, a Bronkhorst modular system was used, comprising a Mass Stream controller (MFC D-6321) for flow regulation, which had a maximum capacity of 2 Ln/min and an accuracy of ± 1.0% RD plus ± 0.5% FS, and a flow meter (F101E) with a maximum flow rate of 3 Ln/min and an accuracy of ± 1.0% FS. The system also included a digital interface based on the FLOW-BUS protocol, enabling diagnostics and communication via RS232. CO₂ absorption data was recorded every second throughout the experiments.

3. Results and Discussion

3.1. MSWIr Physicochemical Characterization

3.1.1. Mineralogy and Chemistry

The XRPD patterns of the raw MSWIr samples are presented in Figure 3. The SSA, the loss on ignition (LOI), the Rietveld quantification of mineral phases, and the XRF chemical composition are provided in Table 3. XRPD allowed the recognition of twenty-seven mineral phases, thus showing the relevant mineralogic heterogeneity of MSWIr. The most abundant minerals were CaClOH, portlandite, larnite, calcite, halite, sylvite, and anhydrite. XRF identified and quantified twenty-one oxides, with CaO leading the chemical composition in samples A, C, D, and E, at 50.2, 54.8, 76.7, and 61.4 wt.%, respectively. Using the modified Steinour equation (Eq. S4), XRF quantification allows the calculation of the theoretical carbonation capacity $\phi$ (g CO₂/kg MSWIr) of the samples. The $\phi$ values for samples A, B, C, D, and E were 389.5, 64.4, 392.9, 524.6 and 443.1g CO₂/kg, respectively. However, these values do not account for the CO₂ amount already present as carbonate, which can be derived from the Rietveld analysis and should be subtracted for accuracy. Thus, the adjusted $\phi$ values are 365.3, 56, 392.9, 458.2 and 425.3 g CO₂/kg, respectively.

Sample A contains a low proportion of CCHO, with only 0.8 wt.% Ca(OH)₂. Most of the calcium is bound to Ca-Al hydroxides, specifically hydrocalumite (Hcl) (14.2 wt.%), katoite (Kto) (12.2 wt%) and Ca-Al nitrate (2.8 wt.%), or silicates like larnite (3.7 wt.%) and akermanite (1.5 wt.%). Hcl is known to precipitate from Ca(OH)₂ and Al(OH)₃ in water, in competition with the formation of Kto [68]. Hcl is known to slowly transform into Kto in alkaline solutions, where Kto is thermodynamically more stable and less soluble [69]. Jiménez[70] proposed that Cl-bearing Hcl may decompose thermally in mayenite (Ca₁₂Al₁₄O₃₃), CaClOH and CaO. Note that ettringite (Ett) was not detected in sample A, even though it is one of the most common Ca-Al hydroxides found in BA and specifically bears sulfates [71,72,73]. Calcite is already present at 5.5 wt.%, indicating rapid early natural carbonation after quenching and exposure to ambient CO₂ at the outlet of the incineration chamber. Chemically, sample A has a significant content of Al₂O₃ (18.67 wt.%) and HMs, including Fe₂O₃ (3.17 wt.%), TiO₂ (1.78 wt.%), ZnO (0.852 wt.%), CuO (0.25 wt.%), Cr₂O₃ (0.088 wt.%) and Ag₂O (0.16 wt.%). Finally, phosphorus may potentially be worth recovering, with P₂O₅ at 2.81 wt.% present as hydroxyapatite (4 wt.%) and archerite (0.1 wt.%).

Sample B does not contain any CCHO. The silicate content includes larnite (0.7 wt.%), melilite (1.3 wt.%), and merwinite (0.8 wt.%). Chlorides are present as halite (25.1 wt.%) and sylvite (12.1 wt.%), while sulfates comprise anhydrite (11.3 wt.%) and bassanite (4.5 wt.%). Sample B contains only 15.1 wt.% CaO; however, Na₂O, K₂O, and Cl are concentrated at 14.7 wt.%, 24.2 wt.%, and 14.1 wt.%, respectively, consistent with the use of a NaHCO₃ additive instead of lime. This sample shows potential for EMR, containing ZnO, Sb₂O₃, and PbO at 6.382 wt.%, 0.314 wt.%, and 0.831 wt.%, respectively.

Sample C contains a high proportion of CCHO, with 14.3 wt.% CaClOH, 2.1 wt.% Ca(OH)₂, and 1.7 wt.% CaO. It also has significant silicate content, including larnite (6.7 wt.%), gehlenite (3 wt.%), and merwinite (2.4 wt.%). Although sample C has lower Al₂O₃ (1.83 wt.%) and HMs content compared to sample A, it still contains Fe₂O₃ (1.23 wt.%), TiO₂ (1.25 wt.%), and ZnO (1.32 wt.%), with antimony at 0.13 wt.%. The sample has typical alkali-chloride levels for boiler/electrofilter residues, including halite (6.4 wt.%) and sylvite (4.9 wt.%), contributing to a total Cl content of 15.14 wt.%. Sulfates are present as anhydrite (4.9 wt.%) and bassanite (1.8 wt.%), resulting in a total SO₃ content of 7.4 wt.%.

Sample D has the highest CCHO proportion, containing 44.3 wt.% CaClOH, 7.2 wt% Ca(OH)₂ and 1.5 wt.% CaCl₂∙xH₂O. Sulfur is present as the sulfite mineral hannebachite at 19.1 wt.%. This sample also has the highest proportion of calcite at 15.1 wt.%, consistent with the use of a semi-dry process involving CaO and steam, which favors early carbonation from flue gas. Chemically, this sample has the highest proportion of CaO (76.7 wt.%) and does not indicate any potential for EMR. Notably, alternative alkali oxides were almost absent compared to the other samples, with Na₂O and K₂O present at 0.098 wt.% and 0.455 wt.% respectively. Additionally, it has the highest crystalline content at 88 wt.%.

Sample E features a high proportion of CCHO, with 20.8 wt.% CaClOH and 9.8 wt.% Ca(OH)₂. Silicate content includes larnite (1.5 wt.%) and hatrurite (1.8 wt.%). The total Cl content is elevated at 19.5 wt.%, consistent with the presence of CaClOH, halite (4.9 wt.%), and sylvite (1.3 wt.%). This sample also has the highest amorphous content at 55.5 wt.% and the lowest HMs proportions. Interestingly, the comparison of hazardous trace metal content between samples A, C, and E produced results consistent with Zhao[74], who classified the volatility of seven metals in the following order: Cd, As > Ni, Zn > Pb > Cr, Cu. Consequently, BA contains higher proportions of Cr and Cu, while boiler and bag filter residues have comparatively higher concentrations of Pb and Zn.

3.1.2. Microstucture

Figure 4 illustrates the morphology of the particles in sample A. Layered Ca-Al particles, densely dispersed across the sample, were the predominant morphology detected. Similar particles were observed by Moon[75] and identified as Kto whereas Jiménez[70] ascribe them to Hcl. As Hcl and Kto exhibit similar morphology, such phases are distinguishable from one another by their chemical composition. Notably, Hcl contains a significant amount of Cl, approximately 12.6 wt.%, while Kto does not. EDS scans of the layered Ca-Al particles reveal Cl concentrations ranging from 0 to 7.2 wt.%, suggesting that Hcl and Kto particles were packed together. Additional Ca-Al morphologies were identified, as illustrated by the botryoidal particle in Figure 4 C. Apatite was also detected, consistently with the XRPD results. Highly heterogeneous particles, mainly composed of alkali chlorides and molten glasses, were found. Finally, the XRPD and SEM-EDS patterns of the sample A magnetic fraction from dry extraction, as shown in Figure S3, indicate the presence of iron oxides, including visible crystals of hematite and magnetite.

Figure 5 depicts the typical SEM view of the sample B. SEM-EDS patterns confirmed the prevalence of alkali chlorides and Zn-containing phases, along with the lower Ca concentration quantified by XRPD and XRF. A large proportion of nanoparticles were mixed or adsorbed onto glass spheres and large silicate single crystals.

Figure 6 presents a typical SEM view of the sample C, highlighting key morphological features. Notably, distinct CCHO particles with a sheared morphology were concentrated in the sample. Glass spheres and iron oxides, common phases in boiler and electrostatic filter residues, were also observed. Additionally, cubic crystals of sodium chloride and unreacted lime particles were identified.

Figure 7 illustrates the typical features observed in the sample D, revealing a relatively homogeneous bulk matrix. Trigonal prismatic portlandite and trigonal calcite crystals are clearly visible in Figures 7 C, E, F, I. Additionally, sheared structures of CaClOH, similar to those found in samples C, are densely concentrated, consistent with the 44.3 wt.% of CaClOH in this sample.

3.2. MSWIr Reactivity to Aqueous Carbonation

3.2.1. Theoretical Background

In MSWIr, most reactive minerals are calcium (chloro)(hydr)oxides, including lime, portlandite and sinjarite (CaCl₂·2H₂O) [76]. Other reactive minerals include calcium silicates such as larnite and hatrurite, as well as magnesium (hydr)oxides like periclase or brucite. In BA, calcium aluminates such as Hcl, Kto, mayenite and ettringite can be the main Ca-providing minerals for carbonation [77,78]. The stoichiometric carbonation capacity (gCO₂/kg) reported for 1 kg of MSWIr, based on free CaO, Ca(OH)₂, CaClOH, CaCl₂·2H₂O, Ca₂SiO₄, Ca₃SiO₅, MgO and Mg(OH)₂, Ca₂Al(OH)₆Cl.H₂O, Ca₃Al₂(OH)₁₂ and Ca₁₂Al₁₄O₃₃ content in MSWIr, is shown in figure 8. Thus, the carbonation capacity of MSWIr and other alkaline waste can typically range from 0 to 400 gCO₂/kg, depending on the Ca and Mg content that is not already bound to carbonates, sulfates, or phosphates [79]. The overall carbonation reactions involving these major Ca/Mg-bearing reactive minerals present in MSWIr systems were listed in Table 4.

3.2.2. CO₂ Absorption Rate and Total Uptake

Table 5 presents the weight loss during carbonation and water washing experiments, reflecting the removal of soluble salts (weight loss) and the uptake of CO₂ (weight gain). As anticipated, the samples A1-A2 exhibited minimal weight loss (2.6–5.4 wt.%), attributed to the absence of significant soluble sulfate and chloride phases. In contrast, samples C1-C2 showed higher weight losses (21.8–25.8 wt.%), which correlate with its substantial chloride content. Samples E1-E2 also demonstrated considerable weight loss (13.8–27.8 wt.%); however, the loss was lower during carbonation, likely due to the CO₂ uptake. Samples B1-B2 experienced the highest weight loss (51.4–52.2 wt.%), consistent with its elevated Na-/K-chloride and sulfate content. Finally, samples D1-D2 also exhibited significant weight loss (34.2–28.4 wt.%), aligning with its high Ca-chloride content.

Figure 9 illustrates the CO₂ absorption rate patterns $\eta \left(t\right)$ of the five samples. Table 6 reports the carbonation performances in terms of absorption rate $\overline{\eta }\left(t\right)$, degree of carbonation achieved $\omega$ (g CO₂/kg), and carbonation efficiency $\epsilon$ (%), calculated using the carbonation capacity $\phi$. The detailed formulas for $\left(\eta |t\right),\overline{\eta }\left(t\right),\omega ,\phi ,\epsilon$ are provided in Eqs. S1-6. Note that it is recognized in the literature that the Steinour equation tend to overestimate the carbonation capacities of studied materials, resulting in low apparent carbonation efficiency [80].

Samples C, D, and E exhibit similar kinetic behaviour, as indicated by the common trends in their $\eta \left(t\right)$ patterns and $\overline{\eta }\left(t\right)$ values of 29.9%, 24.7% and 23.3% respectively. This aligns with the high CCHO content and the Ca(OH)₂ aqueous carbonation $\eta \left(t\right)$ patterns observed in previous studies [63]. Note that $\eta \left(t\right)$ can be easily increased by using a sparger or other static gas-liquid mixers, which produce smaller CO₂ bubbles entering the suspension [81,82]. Additionally, carbonation degrees of 93, 167 and 244.5 gCO₂/kg, with corresponding efficiencies of 23.7%, 36.4% and 57.5% were achieved for these samples, respectively. The sample E achieved the highest CO₂ carbonation degree of 244.5 gCO₂/kg, consistent with its origin from a bag filter where lime sorbent was used upstream. By comparison, the sample D achieved $\eta \left(t\right)$ values smaller than expected, primarily due to calcium being retained in water due to two factors: the high solubility of calcium chloride phases and the low concentrations of alternative alkali oxides, which limited their ability to bind with chloride ions instead of calcium. Indeed, Na₂O and K₂O concentrations in the sample D were significantly lower than in any other sample, at 0.098 wt.% and 0.455 wt.% respectively. This suggests that the wastewater from carbonation experiment D1 could be re-carbonated with an alkali-hydroxide (NaOH) to precipitate nearly pure CaCO₃ and co-precipitate HMs. These findings indicate that the reactivity of samples C, D and E makes them prime candidates for ACT. In contrast, the sample B exhibited minimal reactivity to carbonation, and chemical data suggest its better suitability for FLUWA-FLUREC processes [83].

On the other hand, sample A displays a slower absorption rate, with an average $\overline{\eta }\left(t\right)$value of 12.85%. The slower kinetic behaviour of BA compared to APCr was previously noted by Baciocchi[31]. Indeed, Ca-aluminates dissolve more slowly than CCHO, which results in slower CO₂ absorption rates. However, the calculated degree of carbonation achieved is still significant, at 101.1 gCO₂/kg, corresponding to an efficiency of 27.7%. Notably, the ω value closely matches the theoretical carbonation capacity based on the Rietveld quantification of calcium aluminates (90.4 gCO₂/kg), portlandite and larnite (23.6 gCO₂/kg), totalling 114.1 gCO₂/kg. These results strongly support that Ca-aluminates, primarily Hcl and Kto, are the main drivers of carbonation in BA. This value is about three times higher than commonly reported for BA in the literature, which normally range in 25-45 gCO₂/kg [84,85]. However, using the fine BA fraction (< 150 µm), Baciocchi[31] also reported a significant 14% CO₂ uptake driven by the carbonation of Hcl and Ca(OH)₂.

In BA, the leachability intensely increases with decreasing particle size [86]. Additionally, smaller particle sizes are associated with higher calcium content [87,88]. This contributes to increasing both the ecotoxicity and reactivity of fine BA fractions during carbonation. Thus, the finer BA fractions are both a challenge and an opportunity, given that separating them from the bulk BA for carbonation could significantly reduce the leachability of the bulk fraction and help mitigate CO₂ emissions. According to our data and the literature, the fraction of BA particles under 250 µm accounts for approximately 12 wt.% [89], which represents about 5 kt of material annually for a plant processing 200,000 tons of MSW per year. On a larger scale, this could amount to approximately 350 kt in a country like France. Considering an effective carbonation capacity comparable to that achieved in the present study, this amounts to an emission reduction potential of 36,000 tCO₂. Besides, the amount suitable for ACT could be even larger, as coarser fractions can adsorb fine particles and have been shown to contain Ca-aluminates, which may result in significant reactivity. The ideal granulometry for BA selection in ACT might exceed 0.5 mm, depending on the composition of specific BA streams, the overall BA processing layout, and how the carbonated material will be repurposed.

Finally, the amount of CO₂ absorbed by the sample B, if any, is not quantifiable. The brief decreasing pattern corresponds only to CO₂ dissolving in water up to saturation, as demonstrated by previous studies [63]. The values of $\omega$ and $\epsilon$ were therefore approximated to be 0.

3.2.3. Mineralogy Changes in Carbonated And Washed Samples

Figure 10 shows the XRPD patterns of the carbonated and washed samples. Calcite was the only crystalline phase identified in the carbonated sample A, while Hcl and Kto remained present in the washed sample. Um[90] also compared the efficiency of carbonation with water washing, focusing on Cl removal in BA. They noted that Hcl, ettringite, and sodalite are the main insoluble Cl-containing phases, which decompose only in the presence of CO₂. Aluminum reprecipitates as amorphous Al-oxides [45]. The carbonated sample C produced both calcite and anhydrite, whereas calcite, portlandite, hydroxyapatite, and merwinite remained in the washed sample. The low solubility of portlandite at 333 K, combined with the dissolution of the CaClOH phase and reprecipitation as Ca(OH)₂, explains its consistent presence in the washed samples C2, D2, E2. In the carbonated sample E, only calcite was detected, whereas water washing resulted in portlandite, calcite, larnite, and mullite. As expected, the carbonated (B1) and washed (B2) samples exhibited nearly identical patterns. The main difference for sample B was a reduction in the intensity of the Hcl peak after carbonation, resulting in a slight increase in the calcite intensity peaks. Anhydrite, quartz, periclase, and even halite persisted in both cases. Finally, the carbonated sample D1 produced both calcite and hannebachite, while the washed sample also contained portlandite.

3.2.4. HMs Leaching During and After Carbonation

Table 7 and Figure 11 present the EC/pH and HM concentrations from the TCLP analysis of raw samples, carbonated samples, and wastewater from the aqueous carbonation and water washing experiments. Table S1 provides the physicochemical properties and the respective leaching limit values pf HMs for the use of recycled waste in France, Italy, and Germany. Detailed concentrations are reported in Table S2. The raw sample B exhibited the highest leachability among all the tested samples, with HM concentrations several orders of magnitude higher measured for Cr, Mn, Co, Ni, Cu, Zn, As, Cd, and Sb. EC values were highest, ranging from 20.8 to 21.5 mS/cm in wastewaters, consistent with the weight loss, as well as the chloride and sulfate concentrations measured by XRF. Additionally, it produced less alkaline leachates, with washing wastewaters at a pH of 10.1. Leaching concentrations were consistent among the raw C, D and E samples. However, sample A tended to produce notably higher concentrations of Al, Cu, and Sb, along with lower concentrations of Mn, Fe, Co, Ni, Zn, and Pb.

Table 8 presents the ratio of HM leaching concentrations between aqueous carbonation and water washing experiments, expressed as a percentage. Overall, the results indicate that carbonation generally increases the leaching of HMs into the washing water and leads to a significant reduction in leaching during the TCLP tests. The sample A exhibited contrasting behaviour in comparison with the samples C, D and E, where TCLP leachate concentrations significantly decreased after carbonation compared to water washing. On the other hand, sample B displayed comparable HM concentration values between carbonation and water washing, thus highlighting its unreactivity and the similar phase compositions.

However, Al and Pb were strongly stabilized in both wastewater and TCLP leachates, while Mn and Sb were mobilized, with Sb showing the most critical behaviour under carbonation. Co, Ni and Cd were highly mobilized in the wastewaters but were significantly stabilized in the TCLP leachates. Conversely, Cu and Zn showed no consistent trends in the wastewater but exhibited enhanced stabilization in the TCLP leachates. As displayed slight mobilization in the wastewater, which resulted in a modest reduction in TCLP concentrations. Fe and Cr exhibited the most contrasting leaching behaviours.

After carbonation, Al predominantly exists as Al(OH)₃ [90,91], either as gibbsite or in amorphous form, owing to the low solubility of Al-hydroxides at pH levels of 7-8 [92]. Khan[93] reported the incorporation of As oxyanions into calcite crystals after carbonation treatment. Cd can be stabilized through octavite precipitation (CdCO₃) or by coprecipitation with calcite, forming (Cd,Ca)CO₃ [94]. Mn leaching was found to be highly pH-dependent, significantly increasing as carbonation lowered the pH [95]. Pb is the most stabilized HM during aqueous carbonation, as cerussite (PbCO₃) serves as a solubility-controlling mineral in the pH range 6–12 [96,97,98,99,100]. Notably, Pb is significantly more stabilized in APCr by wet carbonation compared to gas-solid carbonation [101]. Sb is the most critically mobilized HM, primarily attributed to romeite (CaSb₂O₆) dissolution [102,103]. Sb solubility primarily depends on pH and Ca^2⁺ availability [104]. Sb may pose a potential toxic risk in secondary building materials containing MSWIr, necessitating long-term monitoring of its release [105]. Overall, our results are consistent with previous studies. Specificaly, using APCr, Baciocchi[106] reported that the main effects of carbonation on trace metal leaching included a significant decrease in the mobility of Pb, Zn, and Cu at high pH, a slight change or mobilization for Cr and Sb, and no major effects on the release of As and soluble salts.

3.2.5. Mechanism of Chromium Leaching in BA Carbonation: A Novel Perspective

Stabilizing the finer BA fractions should be further enhanced to support systematic reuse and ensure the long-term safety of BA-amended constructions. Cr and Sb are among the most critical elements in BA for HM leaching, often exceeding inert waste quality standards [107]. Other HMs may include Ba, Cu, Ni, Mo, Pb, and and Zn [108,109,110].

Cr exists primarily in two oxidation states: Cr(III)) and Cr(VI). While Cr(VI) is widely recognized for its acute toxicity and carcinogenic potential, Cr(III) is less harmful, although it also poses health risks under certain conditions [111,112]. Using µ-XANES, De Matteis[113] detected Cr(III) only, in the finer BA grain sizes, occuring as chromite. Notably, Cr leaching behaviour during and following carbonation was more critical in comparison with other samples. Using BA, Um[114] found that Cr leaching initially increases due to the decomposition of layered double hydroxides (LDHs), ettringite and Hcl, but may decrease as Cr ions are adsorbed onto the resulting amorphous Al-material. To verify the Cr leaching control mechanism, Figure 12 and Figure 13 show backscattered electron SEM images of Hcl and Cr-Mn-Fe oxide, respectively, in a polished section of the raw sample A.

Figure 12 illustrates that no Cr was detected in the Hcl particles analyzed via EDS. Instead, Cr was concentrated in the Cr-Mn-Fe oxides particles, which were coated by Ca-aluminates (Figure 13). We attribute this coating to the dissolution-precipitation of CaO and Al₂O₃ during quenching, promoting to the formation of Ca-aluminate precipitates on the surfaces of HM oxide particles (Figure 14). The presence of Cr-Mn-Fe oxide particles aligns with the known affinity of Cr for Mn-Fe oxides [115,116,117,118]. These findings indicate that in our sample A, the increased leaching of Cr, Mn, and Fe during carbonation is primarily controlled by the decomposition of these Cr-Mn-Fe oxide particles, a process that is itself driven by the breakdown of the Ca-aluminate coatings. This highlights the complex interplay between mineral phases and the leaching behaviours of HMs in the carbonation process. This newly proposed mechanism is illustrated in Figure 14.

These findings suggest that to sustainably enhance the stabilization of fine BA fractions, metal separation techniques should be implemented prior to carbonation. Additionally, Allegrini[119] found a considerable non-recovered resource potential in fine fraction (< 2 mm), where approximately 12% of the total non-ferrous potential in the BA were left. Using a similar BA fine fraction (< 2 mm), Pienkoß[120] and Sierra[121] demonstrated that HMs can be recovered by enriching the denser fraction via wet density separation, yielding a metal concentrate containing Fe, Zn, Cu, Pb, Zr, Cr, Ni, Sn, W, and V. The most critical process parameters were the stroke frequency and tilt angle of the wet shaking table. Further improvement in HM enrichment could be leveraged by using dry-discharger or non-aged wet-discharged BA. Thus, particles like those shown in Figure 13 appear to be suitable candidates for density or magnetic separation, effectively reducing the leaching of Cr, Mn, and Fe. Bruno[122] showed that EMR of Cu and Fe from BA would result in significant energy savings, reductions in CO₂ emissions, and economic profitability driven by recycling and avoided landfill costs. Indeed, Cu is known to undergo fractionation both in finer (< 0.45 mm) and coarse (> 4 mm) BA fractions [123]. Beikmohammadi[124] showed that average concentrations of HMs in BA, including Zn, Cu, Ba, Pb, Cr, Ni, Sn, V, As, and Sb, were higher in the < 4 mm fraction, while Au and Ag were significantly higher in the < 0.5 mm fraction, making their extraction economically viable. This underscores the importance of appropriate EMR-based pre-treatment before utilizing these materials in ACT, as well as in construction or landfilling [125]. However, while carbonation results in density and strength improvement of fine BA [126], Weiksnar[127] demonstrated that EMR has minimal to no impact on the compressive strength of BA-amended concrete.

4. Conclusions

This study aims to assess the reactivity of Municipal Solid Waste Incineration residues (MSWIr) in aqueous carbonation under typical production conditions. Mixtures of boiler, electrofilter, and bag filter residues were selected to compare their reactivity in terms of CO₂ absorption rates, CO₂ uptake, and resulting heavy metal (HM) leaching behaviour. The effectiveness of aqueous carbonation was compared with conventional water washing for HMs removal and stabilization, further highlighting the stabilization and mobilization effects induced by carbonation.

Results show that bag filter residues from plant 1 using lime as a sorbent are the most reactive type of MSWIr, achieving a carbonation degree of 244.5 gCO₂/kg (sample E). In comparison, a carbonation degree of 101 gCO₂/kg was achieved for the BA fine fraction (sample A), 93 gCO₂/kg for the boiler/electrofilter APCr (sample C), 0 gCO₂/kg for the electrofilter/bag filter FA from plant 2 using sodium bicarbonate, and 167 gCO₂/kg for the mixed boiler/electrofilter/bag filter APCr from plant 3 using lime (sample D). Thus, samples from plant using lime are prime candidates for Accelerated Carbonation Technology (ACT), while MSWIr from plants using sodium bicarbonate are unsuitable and better suited for the FLUWA/FLUREC processes focused solely on EMR.

All MSWIr with high calcium (chloro)(hydr)oxide content exhibited similar kinetic behaviour during carbonation, as shown by the comparable trends and intensities in their CO₂ absorption rate patterns. However, contrary to common findings in the literature, BA carbonation was primarily driven by Ca-aluminates—specifically by hydrocalumite and katoite—rather than by lime. The carbonation kinetics of BA were slower due to the slower dissolution rate of Ca-aluminates compared to lime, as reflected in the BA sample's CO₂ absorption rate pattern. Despite this, the dissolution of Ca-aluminates led to a carbonation degree (101 gCO₂/kg) about three times higher than that commonly reported for BA in the literature, which normally ranges in 25-45 gCO₂/kg.

Overall, our results demonstrate that HM leaching significantly increases during carbonation compared to regular washing but subsequent TCLP tests indicate a relevant decrease in toxicity. The samples C, D and E showed the most substantial improvements, with marked reductions in leaching of Cr, Fe, Co, Ni, Cu, Zn, Cd, and Pb post-carbonation. However, Sb exhibits a critical leaching increase in all samples, both in wastewater and TCLP leachates. Additionally, the leaching behaviour of Fe, and especially Cr, pose significant concerns about BA carbonation. We attribute the elevated Cr, Mn, and Fe leaching during BA carbonation primarily to the decomposition of Cr-Mn-Fe oxide particles, a process driven by pH lowering and breakdown of Ca-aluminate coatings on slag particles. To our knowledge, this mechanism is being reported for the first time in this work.

In conclusion, BA and APCr from plants using lime as a sorbent are successful candidates for ACT. However, additional treatments must be integrated into their waste management sequences, in order to develop a fully sustainable process, in both environmental and economical terms. Notably, integrating EMR from the fine fraction of BA through density or magnetic separation prior to carbonation has the potential to sustainably address excessive HM leaching, particularly of Fe and Cr and even potentially Sb, while recycling raw materials.