Lab-grade calcium carbonate with >99% purity was procured from Thermo-Fisher Scientific and magnesium carbonate and amorphous silica gel were procured from Fisher-Scientific. Seven biomineralized calcium carbonate powder samples were provided by Prometheus Materials and prepared by their proprietary process using various algae strains. One mineralized calcium carbonate powder sample prepared by proprietary accelerated carbonation process was provided by the Gadikota Research Group. The steel slag and Cu-Ni mine tailing were received from Nucor Steel and Eagle Mine, Michigan respectively.

Model system experiments were performed with synthetic lab-grade CaCO₃, MgCO₃∙nH₂O and amorphous SiO₂ samples. Simplistic mix designs were used to assess the effects of varying quantities of MgCO₃ and amorphous silica on CaCO₃ decomposition as shown in Table 2. The synthetic samples used in this study had a top particle size of 75 µm or less. The mineralized CaCO₃ received from research partners were also passed through 200 mesh (75 µm) to ensure similar particle size across samples. In some cases, samples were hand-ground to achieve the particle size threshold. The industrial waste samples, mine tailing and steel slag, were ground and sieved with 200 mesh to achieve similar top particle size.

The quantification of CaCO₃ was analyzed in a heterogeneous matrix by dosing steel slag and mine tailing with known amounts of lab-grade CaCO₃. The mix designs for CaCO₃ dosed steel slag and mine tailing are listed in Table 3. The appropriate amounts of samples were weighed and mixed by gentle grinding in an agate mortar and pestle to ensure homogeneous mixing. For each matrix, i.e., steel slag and mine tailing, three samples were prepared to have approximately 2, 5, and 10 wt% CaCO₃ respectively.

Material characterization was performed using Thermogravimetric Analysis (TGA), X-ray Powder Diffraction (XRPD), and scanning electron microscope (SEM). The TGA (TA instrument TGA 5500) was used for carbonate quantification via thermal decomposition of CaCO₃. 8–12 mg of sample was loaded in a platinum pan at room temperature. The TGA experiments were performed at heating rates of 10 °C/min and 20 °C/min using N₂ at a 25 mL/min flowrate. The sample was initially heated to 105 °C, followed by an isothermal hold for 15 min to determine the moisture content. Post isothermal hold, the sample was heated to 1000 °C at the specified ramp rates above to ensure complete thermal decomposition of MgCO₃ and CaCO₃. Triplicate TGA measurements were performed to ensure repeatability and reproducibility of this work.

CaCO₃ content was calculated using the sample mass loss in specified temperature range as shown in Equations 1, 2. Δ m = w o n s e t − w e n d s e t w i × 100 (1) % C a C O 3 = Δ m 43.97 × 100 (2) where, Δm = % mass loss between the onset and end temperature.

wonset = sample mass at the onset decomposition temperature point.

Wendset = sample mass at the endset decomposition temperature point

wi = initial weight of the sample.

Onset and endset temperature of carbonate decompositions were calculated using tangent intercept method in TRIOS software, and was further used to calculate the range of decomposition as shown in Equation 3: R a n g e = T e n d s e t − T o n s e t (3)

In this expression, T_endset = endset decomposition temperature level and T_onset = onset decomposition temperature level. For steel slag and mine tailing samples dosed with known amounts of lab-grade CaCO₃, % error of measurement was calculated according to Equation 4: %   e r r o r = % C a C O 3 − Q C a C O 3 Q C a C O 3 × 100 (4)

In this expression, Q_CaCO3 = known amount of CaCO₃ dosed into the samples.

XRPD was performed on Rigaku Ultima IV diffractometer with Cu- Kα source (40 kV, 40 mA) to collect the diffraction pattern. Profex, open access software, and PDF4+ mineral database were used to perform Rietveld refinement on the XRPD data. Hitachi S4800 SEM was used to investigate morphological properties. The powders were evenly dispersed using a cotton swab on adhesive copper tape followed by sputter coating to decrease the charging effects.

The TGA analyses of mineralized CaCO₃ powders revealed slight differences in decomposition onset temperature compared to the lab grade CaCO₃. The TGA mass loss curves are shown in Figure 4 and the onset and endset temperature of the thermal decomposition of mineralized and lab-grade CaCO₃ samples are listed in Table 5. Note that the TGA measurements were performed at heating rate of 10 °C/min under N₂ atmosphere. The onset and endset temperature levels have been considered parameters for comparison between different CaCO₃ sample types. It was observed that the onset temperatures of CaCO₃ samples decomposition varied between 515 °C and 577 °C whereas the endset temperatures varied between 690 °C and 763 °C. The highest onset and endset temperature levels were observed in the case of lab grade CaCO₃ corresponding to widest temperature range of decomposition. The wide decomposition temperature range for lab-grade CaCO₃ could be due to kinetic limitations at the highest CaCO₃ concentration as well as the higher crystallinity of calcite as confirmed with SEM analyses.

In addition to physical parameters such as particle size, sample size, and run conditions, the temperature range of CaCO₃ decomposition not only depends on the crystallinity of calcite but also on the presence of other polymorphs such as vaterite and aragonite (Kemp et al., 2022). The powder XRD patterns of mineralized CaCO₃ samples are shown in Figure 5. Calcite was found to be the major CaCO₃ polymorph in most of the samples except MC1 and MC2 where vaterite is the major CaCO₃ polymorph. Since the mineralized samples were crystalline, the XRD results showed that calcite and vaterite could easily be identified with the XRD as characteristic diffraction peaks of calcite and vaterite are obtained at 2θ angle of 29.4° and 27° respectively.

Table 6 presents detailed QXRD results of mineralized CaCO₃ showing minor components present in the mineralized CaCO₃ samples and include MgCO₃, MgCO₃·nH₂O, NaNO₃, Na₂SO₄, CaMg₃(CO₃)₄, SiO₂, and NaHCO₃. In general, the presence of minor mineral phases in mineralized CaCO₃ depends a lot on process and chemical reagents used. The QXRD results indicated that MC1, MC2 and MC3 samples contain 59%, 54.5% and 21.3% vaterite respectively. The presence of vaterite in samples MC1, MC2 and MC3 did not lead to decrease in the onset decomposition temperature. The vaterite content slightly decreased from 59% in MC1 to 54.5% in MC2. However, the onset decomposition temperature increased from 515 °C in MC1 to 563 °C in MC2. The vaterite content did not appear to decrease the onset decomposition temperature of mineralized CaCO₃ samples. The maximum calcite as well as CaCO₃ content in mineralized samples was observed in MC5, i.e., 97%, which corresponds to the highest decomposition onset temperature (571 °C) compared to other mineralized CaCO₃ samples analyzed in this study.

Overall, CaCO₃ contents obtained from QXRD and TGA were in good agreement except for MC7 sample (Figure 6). The CaCO₃ content in MC7 obtained from QXRD and TGA were about 86.1% and 80.5%, respectively. Particularly for MC7 sample, QXRD and TGA measurements were not in good agreement. A minor amount of NaNO₃ (4%–5%) was detected in MC6 and MC7 samples (Table 6) that also have a similar temperature range of decomposition as CaCO₃. The thermal decomposition of sodium nitrate has an onset in the range of 580 °C–600 °C (Bartos et al., 1944; Gimenez and Fereres, 2015). Gimenez et al. used a similar TGA decomposition temperature profile to that in this study (10 °C/min; N₂) and found the onset decomposition temperature of NaNO₃ being close to ∼613 °C (Gimenez and Fereres, 2015). However, in case of MC7, the deviation in CaCO₃ measurement should not be entirely attributed to presence of these minor components. The presence of organic components may also result in overestimation in case of QXRD. In this case, QXRD measurement clearly appears to overestimate the CaCO₃ content where TGA measurement can also be deemed an underestimation due to possible overlap with decomposition of NaNO₃.

The overlap between decomposition temperature ranges of CaCO₃ and NaNO₃ or other minor components appears to be another important factor affecting carbonate quantification when using only TGA. Although the magnitude of error in this study is relatively smaller, higher concentrations of NaNO₃ or other components may affect the TGA measurement of CaCO₃ content significantly.

The particle morphological examination of mineralized CaCO₃ samples provided more insight into the presence of different polymorphs, cubic as well as rhombohedral calcite crystals. SEM images of MC1, MC4, MC5, and MC8 are shown in Figures 7–9 respectively. MC1 was mostly observed to contain cubic calcite crystals where calcites mostly appear as clusters, however, isolated vaterite crystals were also found. The vaterite crystals present were not perfectly spherical and were seemingly fused together as shown in Figure 7c. Due to the presence of vaterite clusters, the grain size calculated from XRD data is different from that observed under the SEM. MC4 and MC5 exhibit very different particle morphologies as compared to other samples. MC4 contained two different types of calcite crystals, i.e., poorly crystallized cubic calcite and hexagonal calcite rose-like morphology (Figure 8a). On the other hand, MC5 was mostly observed to have layered rhombohedral calcite crystals. Figure 9 shows the cubic crystallites of calcite observed in MC8. Most of the calcite crystallites are smaller than a micron and appear in clusters. Some isolated calcite crystals bigger than a micron were also observed but were found to contain cubic ultrafine crystallites clustered together. As also calculated from the XRD data, calcites with the smallest grain size were present in MC8, i.e., 36.3 nm MC2, MC6, MC7, and MC8 were found to exhibit crystallite size of less than 100 nm (Supplementary Table A2).

The Cu-Ni mine tailing exhibited a complex mineralogy, having a mix of various ultramafic silicate minerals, sulfides, sulfates, and other basaltic minerals such as chlorite. The qualitative XRD analysis confirmed the presence of lizardite, augite, forsterite, bytownite, chlorite, pyrrhotite as major crystalline phases (Figure 10). A complete list of mineral phases identified in Mine Tailing sample is mentioned in Supplementary Table A3. Note that there was no characteristic CaCO₃ peak observed in this case. A similar class of mineral phases were observed in a study by Mends et al. on flotation tailing from Eagle mine (Mends et al., 2025).

The TGA thermal decomposition results of raw mine tailing showed a continuous mass loss especially in the range of 450 °C–1000 °C. Two major derivative peaks at 514 °C and 679 °C were observed as indicated in Figure 11a. The first mass loss was observed in the range of ∼410 °C–625 °C whereas the second mass loss in the range of 615 °C–751 °C, followed by two small derivative peaks ranging up to ∼850 °C. Majority of the mass loss in the range of 450 °C–800 °C can be attributed to the decomposition of lizardite and chlorite. However, a small part of mass loss can also be attributed to dolomite decomposition that occurs in the similar temperature range. Lizardite, the major mineral phase in the mine tailing sample, decomposes in a similar temperature range to that of CaCO₃. Zhou et al. studied the dehydroxylation of lizardite and observed a continuous mass loss in the range of 580 °C–700 °C and a peak at ∼624 °C with 15 °C/min heating rate in N₂ atmosphere (Zhou et al., 2017). Lempart et al. studied the dehydroxylation of three different chlorite samples and observed mass loss in two steps with major mass loss being in the range of ∼400 °C–600 °C with a heating rate of 10 °C/min in N₂ atmosphere (Lempart et al., 2018). Note that the dehydroxylation of both chlorite and lizardite releases water vapor that is known to have a catalytic effect on the decomposition of CaCO₃ (Silakhori et al., 2021), albeit it may not have significant effect in this case given the small amount of sample.

Figures 11b,c shows the derivative curves of samples with lab-grade CaCO₃ dosed into mine tailing and pure lab-grade CaCO₃. The decomposition of lab-grade CaCO₃, irrespective of concentration, overlaps with that of the lizardite dehydroxylation step. This indicates that baseline correction is needed in order to quantify the fresh carbonates. As also mentioned in previous sections, in addition to operating conditions such as heating rate, particle size, CO₂ partial pressure, crystallinity etc., the decomposition temperature range of calcium carbonate varies significantly given the presence of polymorphs and impurities. The CaCO₃ dosing was helpful in order to recognize its decomposition temperature range in a specific matrix, especially in complex matrices such as tailings and slags. The dosing technique could be helpful for carbonation studies to deconvolute temperature ranges of decomposition of calcium carbonates and other components in a given matrix. The thermal decomposition data of mine tailing samples dosed with calcium carbonate is also mentioned in Table 7.

Like mine tailings, steel slags also possessed a complex mineralogy, however, the measurement of carbonates in steel slag must take into account a slightly different silicate matrix. A qualitative XRD analysis revealed that calcite, merwinite, belite and akermanite are the major crystalline phases present whereas minor phases such as wustite, mayenite, magnetite, magnesite, quartz, were also observed (Figure 12). A complete list of mineral phases identified in a mine tailing sample is mentioned in Supplementary Table A3. In addition to all the possible minor crystalline phases, a part of the sample was amorphous or poorly crystalline as possible amorphous humps were observed at 2θ close to 11° and 32°. The accuracy of QXRD methods is typically high, however, the presence of amorphous mineral phases could result in uncertainty in carbonate quantification.

Figure 13a shows the TGA decomposition curve of steel slag with step wise mass losses in different temperature range. Although qualitative mineralogical analysis is conducive in understanding mass losses in respective temperature ranges, it is difficult to dedicate mass loss in the range of 105 °C–850 °C to a particular thermal event due to their potential overlaps. As shown in Figure 13a, based on the derivative curve, TGA mass loss of the steel slag can be segmented into five temperature ranges. The mass loss in the isothermal segment at 105 °C can be attributed to loosely adsorbed moisture in the sample. The mass loss ranging from 105 °C to 304 °C can be dedicated to the beginning of dehydration or dehydroxylation of schwertmannite, nacrite and some organic components that may be present (Lázaro et al., 2022; Ben, 1997). In the temperature range of 304 °C–405 °C, several possible overlaps between the dehydration of brucite and continued dehydration of schwertmannite should be expected (Harrison et al., 2013; Lázaro et al., 2022; Siauciunas et al., 2023). Amara et al. observed two major dehydration derivative peaks of nacrite at 108 °C and 245 °C (Ben, 1997). The final dehydroxylation peak of nacrite was observed at 627 °C. The brucite decomposition window can range from 315 °C to 450 °C (Turvey et al., 2022). The decomposition of Magnesite and brucite may or may not overlap with each other but can severely overlap with that of portlandite in the temperature range of ∼400 °C–500 °C (Villagrán-Zaccardi et al., 2017). For temperature ranges pertaining to CaCO₃ decomposition in the steel slag matrix (∼550 °C-∼750 °C in this case), simultaneous decomposition of schwertmannite, nacrite, magnesite and CaCO₃ should be expected. The loss of structural sulfate of schwertmannite occurs between 400 °C and 700 °C, centered at ∼615 °C (Lázaro et al., 2022). The above discussion clarifies the role of presence of other components on CaCO₃ and MgCO₃ quantification in steel slag samples. Therefore, baseline correction is important for accurate determination of fresh carbonates. However, for carbonation studies, it is also important to ensure that the components potentially interfering with carbonate measurements are not undergoing carbonation.

Figures 13b,c show the TGA decomposition curves of steel slag sample dosed with known amounts (∼10%, ∼5% and ∼2%) of lab-grade CaCO₃ and comparison of their derivatives with lab-grade CaCO₃. It can be clearly observed that the decomposition temperature range of the same lab-grade CaCO₃ in the steel slag matrix shrinks as compared to the pure lab-grade CaCO₃. The calculated onset, endset, and peak temperatures for the derivative curve of CaCO₃ dosed steel slag samples are also mentioned in Table 7. The span of temperature range of CaCO₃ decomposition in steel slag matrix varies from 130 °C to 138 °C whereas that of pure lab-grade CaCO₃ is ∼243 °C (Supplementary Table A1). Additionally, the decomposition of CaCO₃ in the steel slag matrix was found to have onset temperature in the range of 611 °C–626 °C whereas the pure lab-grade CaCO₃ had an onset temperature at around 593 °C. This is similar to the results obtained in the case of mine tailings, suggesting the silicate matrices tend to slightly delay the onset of CaCO₃ decomposition.

The quantification of carbonates requires more attention and needs to be revisited given the growth in number of accelerated carbonation studies in the last 2 decades. Most of the carbonation studies use TGA for carbonate quantification with assumptions that all or most of the mass losses in the approximate temperature range of ∼500 °C–850 °C are due to CaCO₃ decomposition which may not always be true. The temperature range of CaCO₃ decomposition is also dependent on the impurities present in the carbonated product which can have catalytic effects (e.g., MgO). TGA alone is insufficient for carbonate quantification, especially in the case of industrial waste that exhibits complex mineralogy. The focus of this study was on the characterization and evaluation of factors affecting carbonate quantification in relatively higher purity samples as well as heterogeneous matrices of industrial waste samples consisting of silicates. A variety of samples including mineralized CaCO₃, steel slag, mine tailing, synthetic samples such as lab-grade CaCO₃, MgCO₃, amorphous silica, CaCO₃ dosed mine tailing and steel slag were analyzed. The CaCO₃ was dosed in different amounts (2, 5 and 10 wt%) in steel slag and mine tailing samples. The variation in CaCO₃ content did not have significant effect on the decomposition temperature levels. In general, the presence of impurities significantly lowered the endset temperature of CaCO₃ decomposition as shown in Figure 14 whereas the onset of decomposition was variably affected depending on the type of impurity. The TGA results revealed significant variation in the decomposition temperature range of CaCO₃ in the presence of different impurity types. It is worth noting that the decomposition temperature range of pure lab-grade CaCO₃ was widest, suggesting a kinetic component related to the sheer quantity of carbonates (∼99%). The presence of MgCO₃ resulted in a catalytic effect on the decomposition of lab-grade CaCO₃ whereas the industrial wastes (steel slag and mine tailing) delayed the onset decomposition temperature. The addition of amorphous silica gel along with MgCO₃ still had a slight catalytic effect on the onset of decomposition. These findings are crucial in terms of carbonate quantification and improve understanding on the influence of impurities on CaCO₃ decomposition temperature ranges.

The increasing focus on accelerated or natural carbonation studies has necessitated the development of a standardized protocol for quantification of carbonates, especially in complex and heterogeneous matrices. Guo et al. assessed the progress on carbonation of tailings and observed a four-fold increase in number of publications on carbonation of mine tailings (Guo et al., 2024). In addition to mine tailings, different types of slags, municipal solid waste incinerator residues, natural rocks, and other heterogenous materials have been utilized in various carbonation studies (Fang et al., 2024; Raza et al., 2022; Qin et al., 2022; Li et al., 2007; Sexsmith et al., 2003). However, there is a lack of publications and studies on precise quantification of carbonates in samples that often possess complex minerology coupled with wide ranges of impurities. Studies by Ferrara et al. and Kemp et al. are relevant to this study as they explored the quantification of carbonation using various methodologies (Ferrara et al., 2023; Kemp et al., 2022). The development of a standardized protocol will benefit the quantification of carbonates in carbonated products and ensure safe and durable concrete production that conform to required standards when they are used as SCM.

Based on our results, a generalized workflow was drafted for improved quantification of carbonates in the CO₂ mineralized samples (Figure 15). The workflow highlights the importance of pre-characterization and utilization of analytical methods to identify and quantify the mineral phases that interfere with TGA measurement of CaCO₃. Quantitative methods such as QXRD should be utilized when possible. However, for samples with amorphous mineral phases, the use of FTIR or organic elemental analyzer is recommended.

Based on the characterization and carbonate quantification results obtained in this study, the following are the proposed best practices identified for calcium carbonate quantification in complex and heterogeneous matrices:

Use of a fixed temperature range is not accurate for carbonate quantification. Temperature range of decomposition of lab-grade CaCO₃ varies with types of impurities and matrices.