Heat-induced phase transitions in mining tailings to create alternative supplementary cementitious materials

The present study investigated the mineralogical changes in five different mining tailings (i.e., bauxite, gold, copper, and lead) with varying heating conditions (i.e., non-heating, 600 ◦ C, and 900 ◦ C) to explore the feasibility of using thermally treated tailings as supplementary cementitious materials. In particular, among the used heating conditions, bauxite tailings heated to 600 ◦ C showed the best reactivity as supplementary cementitious material and thus rigorously studied the fundamentals of the increased reactivity. Well-balanced Al and Si dis-solutions from the thermal decompositions of gibbsite, boehmite, and kaolinite seem to be the result of the best reactivity at the bauxite tailings heated at 600 ◦ C among used heating conditions. It is also noted that, although tailings originated from the same types of ore or contained high Al 2 O 3 and SiO 2 contents, their supplementary cementitious reactivity differed depending on the contents of highly (i) soluble, (ii) thermally decomposable, and (iii) Al or Si-bearing minerals such as boehmite, gibbsite, kaolinite, and chamosite.


Introduction
The increasing urgency to address environmental concerns and promote sustainable practices has prompted a renewed emphasis on managing industrial waste.In particular, tailings, the residual material left after extracting valuable minerals from ore through mining or mineral processing, present a notable challenge in industrial waste management but are also a potential resource.More than 300 million tons of mine waste are generated in the European mining industry alone (Shengo, 2021).Moreover, global red mud stocks exceeded 2.5 billion tons in 2007 and continue to grow rapidly (Power et al., 2011).In particular, among ~150 million tons of bauxite residue generated yearly, only ~4 million tons were utilised (Evans, 2016).Additionally, when producing 1 ton of copper metal, approximately 129 tons of copper tailings were generated (Gordon, 2002).Furthermore, industrial waste management of tailing is significant since hazardous substances contained in tailings can affect the surrounding ecosystem.Until now, tailings have been stored in tailings ponds, which can adversely affect the environment (Vogt, 2012); thus, various techniques utilizing tailings as construction material have been developed to reduce the amount of tailings disposal (Martins et al., 2021).
One promising way to reduce solid wastes destined for landfills is to use tailings in sustainable construction materials.OPC concrete currently accounts for over 5-8% of global carbon dioxide emissions (IEA, 2019), and most of this emission results from producing Portland cement clinkers.Developing Low-carbon concrete is urgently needed to achieve net zero by 2050 (Bouckaert et al., 2021).For instance, certain tailings can be utilized as supplementary cementitious material (SCM) that can replace a portion of Portland cement in concrete (Ramanathan et al., 2021;Vargas and Lopez, 2018).In addition, they were used as aggregates or inert fillers in concrete (Kuranchie et al., 2015;Peys et al., 2022).However, utilization of tailings is relatively limited in the construction material due to lack of specification, hazardous heavy metal inclusions (e.g., Pb, Cd, or As (Kyle et al., 2012)), and their complex mineralogical and chemical composition (Simonsen et al., 2020), which often requires chemical or mechanical treatments to enhance their reactivity for SCMs.
The milling process, one of the mechanical pretreatments, has been used to improve the reactivity of tailings by changing their physical and chemical properties (Yao et al., 2020).For example, it can increase the specific surface area and reduce particle size in tailings.The prolonged milling process damages the lattice of the mineral, reduces its crystallinity, and increases the amorphous phases, which can help improve the reactivity in alkaline activation (Ramanathan et al., 2021).Yao et al. (2019) conducted a study investigating the impact of prolonged grinding on the particle size and pozzolanic reactivity of gold tailings.The particle size and specific surface area of gold tailing were sharply reduced after 20 min of milling, but they gradually decreased after further grinding after 60 min of milling.Moreover, there was a decrease in crystallinity across all phases, coupled with an increase in amorphous content.The alteration in micro-structural characteristics manifested as an enhancement in pozzolanic reactivity.Specifically, an extension of the milling period from 40 to 80 min resulted in a notable escalation in the pozzolanic activity index, elevating it from 62 % to 79 % (Kyle et al., 2012).Concomitantly, Chen et al. (2022) explored the effects of mechano-chemical treatment on the pozzolanic reactivity of iron tailings.Their observations mirrored those of Yao et al. (2019), elucidating that prolonged grinding precipitated an augmentation in reactivity.The pozzolanic activity index demonstrated a substantial increase, progressing from 60.1 % to 88.9 %.
Meanwhile, calcination involves subjecting tailings to heat treatment, which enhances their reactivity by modifying their chemical properties.This process typically involves exposing the tailings to high temperatures over a period.The heat induces chemical reactions like decomposition, dehydration, and oxidation, leading to changes in the mineral composition of the tailings.For example, heat treatment can affect the reactivity of tailings through changing chemical and mineralogical properties, such as thermal decomposition of minerals, release of chemically bound water, and formation of new phases (Chen et al., 2022).Ye et al. (2017) found that the calcination of bauxite tailings can decompose the hydroxide-bearing mineral phases, including kaolinite and diaspore phases, which enhance the reactivity of the bauxite tailings.Through alkali activation, the compressive strength of the blended mixture composed of 70 % calcined bauxite and 30 % slag was over 45 MPa after 28 days (Ye et al., 2017).Thus, mechanical and heat treatment seems to improve the reactivity of tailings, making them more suitable for alternative SCMs.However, in the previous study by Ye et al. (2017), there was still a lack of considering the calcination temperature selection and the effects of the different temperatures.In addition, despite several previous studies (Yao et al., 2020;Yao et al., 2019;Chen et al., 2022;Ye et al., 2017;Cheng et al., 2016;Dabbebi et al., 2020;Perumal et al., 2019), the fundamental understandings of improved reactivity have not been fully understood.This paper will complement and advance the understanding of the increased reactivity of tailings by mainly focusing on phase changes in the components containing Al and Si and their dissolution in the alkaline solution.
A wide range of waste-derived materials (including tailings) have been tested to develop sustainable low-carbon concrete.However, two barriers to entering various waste-derived materials into the established concrete industry are their heterogeneity and the lack of reactivity testing methods.Recently, the rapid, robust, and relevant (i.e., R3) reactivity test was introduced to evaluate the reactivity of SCMs by simulating similar conditions of hydrated mixed cement (Snellings and Scrivener, 2016).This method was initially designed to assess the reactivity of calcined clay but has now been extended to assess the reactivity of a wide range of SCMs (Snellings et al., 2019).For instance, according to the R3 reactivity test on fly ash, slag, metakaolin, and calcined clay by Li et al. (2018), the 7-day R3 isothermal calorimetry test results were highly correlated with the 28-day compressive strength testing result.Ramanathan et al. (2021) also used the R3 test to evaluate the reactivity of various types of tailings.They reported that mechanical pretreatment contributes to the generated accumulated heat of the R3 tailings mixture increasing significantly, and the amount highly depends on the milling duration.
The present study determined the effects of two pretreatment methods (mechanical grinding and thermal treatment) on the reactivity of bauxite, gold, copper, and lead tailings.In particular, this paper focuses on advancing the fundamental understanding of the increased reactivity by thermal treatment.The grinding and heating pretreatments, particle size distribution, X-ray fluorescence (XRF), X-ray diffraction (XRD), and thermogravimetry (TG) analysis were conducted to characterize the change in tailings by the treatments.In particular, the cementitious reactivity of tailings was independently evaluated by two tests using i) the dissolution test in an alkaline solution and ii) the R3 isothermal calorimetry test.Moreover, promising reactivity was observed in thermally treated bauxite tailings.Thus, the fundamental mechanism of enhanced reactivity of treated bauxite tailings was further investigated using in-situ high-temperature XRD (HT-XRD) and 27 Al solid-state nuclear magnetic resonance (NMR).

Tailings
Five different mine tailings in Australia [i.e., bauxite tailings (BT), gold tailings (GT), two different copper tailings (CT1 and CT2), and lead tailings (LT)] were investigated.BT and GT were supplied by the AILLIQ ARC Linkage project, while other tailings (i.e., CT1, CT2, and LT) were provided by Glencore Australia Holding PTY Limited.BT and LT were received as chunks of rock caused by the consolidation inside a storage facility, while others were coarse grain.

Preparation of materials
Although it was challenging to mill all the tailings to precisely the same size, they were carefully milled for further characterization.Then, five mine tailings (i.e., BT, GT, CT1, CT2, and LT) were dried at 105 • C for three days to remove any moisture and then milled using a ring mill (RM2000, Scott, New Zealand) for five minutes.These milled tailings were then characterized for their physical and chemical properties, as summarized in Table 1.On the other hand, the effects of thermal treatment on the chemical properties of tailings were investigated.The dried tailings were placed in a muffle furnace (CF1600 muffle furnace, Across International, United States), and the temperature in the furnace increased to either 600 • C or 900 • C at 3 • C/min with two hours of dwelling time at each specific temperature.The thermally-treated tailings were then cooled inside the furnace to reach the ambient temperature.As shown in Table 1, Group U included five different tailing samples after milling (i.e., BT-U, GT-U, CT1-U, CT2-U, and LT-U), while Group H1 and Group H2 involved the milled samples subjected to the thermal treatment at either 600 ºC (i.e., BT-H1, GT-H1, CT1-H1, CT2-H1) or 900 ºC (i.e., BT-H2, GT-H2, CT1-H2, CT2-H2, and LT-H2), respectively.It should be noted that more heating conditions can be helpful in selecting the optimal heating condition, but the present study carefully selected these temperatures based on the XRD and TG/DTG results [see Sections 3.1 and 3.2, respectively] for the many experiments using 5 different tailings."

Experimental methods
This paper characterized five different tailings and three different treatments (Group U, Group H1, and Group H2 in Table 1).Therefore, various characterization techniques were implemented to improve the fundamental understanding of the changes in reactivity of tailings.To facilitate the understanding of various tests conducted in this paper, Table 1 summarizes the experimental program conducted in this study.

Particle size distribution
Particle size distributions of Group U samples (i.e., BT-U, GT-U, CT1-U, CT2-U, and LT-U) were evaluated by the laser diffraction particle size analyzer (Mastesizer 2000, Malvern Panalytical, United Kingdom) under wet conditions.Due to the slow or very low reactivity of tailings in water, distilled water was selected as a dispersant medium for the measuring procedure.Also, the agglomeration of powder particles was prevented by using ultrasonic fragmentation and stirring.Given that refractive index (RI) was needed to predict the diffraction from the particle surface, it was estimated based on the chemical composition obtained from XRF results and the RI value for each oxide (Malvern, 2000).The absorption factor was 1.0 because all milled tailings were colored powders (Malvern, 2000).

X-ray fluorescence (XRF) and X-ray diffraction (XRD)
The chemical oxide compositions of Group U samples were analyzed using a wavelength-dispersive XRF spectrometer (Axios, Malvern Panalytical, United Kingdom).
In addition, the XRD patterns of all Group U, H1, and H2 samples were detected by an X-ray diffractometer (PANalytical Empyrean 2, Malvern Panalytical, United Kingdom) equipped with a Co-Kα (wavelength = 1.79 Å) radiation from 5 • to 100 • Due to iron-bearing minerals in some tailings, the present study used cobalt (Co) as an X-ray source to minimize the noise effect between copper (Cu) source and iron-bearing minerals [e.g., hematite (Fe 2 O 3 ), biotite [K(Mg,Fe) 3 AlSi 3 O 10 (F,OH) 2 ], chamosite [(Mg,Fe) 5 Al(Si 3 Al)O 10 (OH) 8 ], ankerite [Ca(Fe,Mg,Mn) (CO 3 ) 2 ], and pyrite (FeS 2 )].The step time and size were set at 197 s and 0.026 • , respectively.Then, the PANalytical X'pert HighScore Plus Program (Seven, 2011) was used to analyze the acquired X-ray scan by comparing it with the reference patterns from the International Center for Diffraction Data (ICDD) PDF-4 + 2023 database (Kabekkodu et al., 2002).The mineralogical contents of Group U samples were determined through Rietveld refinement.Some minerals of Group H1 and H2 samples could not be fully quantified because the structure information of these reference patterns was not provided in the used database.Therefore, Group H1 and H2 samples were only compared with Group U samples to identify mineralogical changes with different heating conditions.

Thermogravimetric analysis (TGA)
The weight losses of Group U samples were measured by a thermogravimetric analyzer (Q600, TA instrument, United States) to evaluate the thermal decompositions of their minerals.All samples were maintained at 105 ºC for 30 min to remove any evaporation water.Then, they were placed on the alumina pan and heated from room temperature to 1050 • C at 10 • C/min under N 2 gas.Data was analyzed using TA Universal Analysis software.

Dissolution tests of tailings
To investigate elements (Al, Si, Ca, Fe, and Mg) dissolved in the alkaline environment (similar to the cementitious system), dissolution tests for all Groups (M, H1, and H2) were conducted.The dissolution tests can also be used to assess the reactivity of tailings along with R3 tests.4.0 gs of powder tailing samples were mixed with 400.0 ml of 1.0 M sodium hydroxide (NaOH) solution in the glass bottle.Next, the mixed solution was sealed with plastic wrap and stored in a 40 • C oven.At specific times (i.e., 4, 20, 24, 44, 48, 72, and 168 h), 2.0 ml of the solution was collected and filtered by a 0.22 μm syringe filter.Afterward, the collected filtered solution was diluted 50 times with 4 % nitric acid (HNO 3 ).Then, the elements in the solution were analyzed using the ICP-OES analyzer (Agilent 700 series, Agilent Technologies, United States).

R3 isothermal calorimetry test for reactivity of tailings
The cementitious reactivities of Group U, H1, and H2 samples were evaluated using the R3 test (Snellings et al., 2019).Among different types of R3 tests (e.g., isothermal calorimetry, Ca(OH) 2 consumption, bound water, and chemical shrinkage), isothermal calorimetry analysis was selected in this study since heat output across a wide range of materials can correlate best with compressive strength development (Snellings et al., 2019).For the R3 test, the potassium solution was prepared by dissolving 4.0 g of potassium hydroxide (KOH) and 20.0 g of potassium sulfate (K 2 SO 4 ) in 1.0 L of deionized water.Then, 1.0 g of powder tailing sample was mixed with 3.0 g of calcium hydroxide [Ca (OH) 2 ] and 0.5 g of calcium carbonate (CaCO 3 ) in a potassium solution using a constant speed mixer at 1,600 rpm over two minutes.The normalized heat flow and total accumulated heat of the R3 mixture were recorded for 7 days at 40 • C using an isothermal calorimeter (TAM Air, TA instrument, United States) according to RILEM TC 267 (Li et al., 2018).

In-situ high-temperature XRD for the bauxite tailings
Due to the improved reactivity of the thermally treated bauxite tailings, the phase transition of the BT-U sample was further analyzed using an X-ray diffractometer (PANalytical MPD, Malvern Panalytical, United Kingdom) equipped with a high-temperature chamber.The insitu HT-XRD test was conducted using a Cu-Kα (wavelength = 1.54 Å) radiation source, which is different from the Co-Kα in the previous XRD test, and this scan result might be slightly affected by the iron content in the tailings (Mos et al., 2018).During the in-situ HT-XRD test, measurement range, step time, and step size were set at 10-100º, 147 s, and 0.026º, respectively.After mounting the powder BT-U sample on the platinum holder, it was placed in the XRD instrument and heated from 26 • C to 1200 • C with a constant heating rate of 30 ºC/min.The heating process proceeded in 150 • C increments and held for 15 mins before scanning at the specific temperatures (e.g., 26, 300, 450, 600, 750, 900, 1050, and 1200 • C).This holding time (i.e., 15 min) was set to be sufficient for the reaction under continuous heating conditions.Then, the same software and database used in the previous XRD test were utilized

Group
Label Treatments Testing to analyze the acquired XRD scans at different temperature conditions.The details of the HT-XRD scans for bauxite tailing can be found in Appendix C.

27 Al solid-state NMR for the bauxite tailings
27 Al solid-state NMR was used to trace the transition of aluminumbearing phases (boehmite, gibbsite, and kaolinite) into various alumina phases.The solid-state NMR measurements were acquired on a Bruker Neo Spectrometer fitted with a 9.4 Tesla superconducting magnet operation at 400 MHz and 104 MHz for the 1 H and 27 Al nuclei, respectively.The samples were packed into 4 mm zirconia rotors fitted with a Kel-F cap and spun to 12 kHz at the magic angle.The spectra were acquired with a hard 1 µs pulse, up to 4 k signal transients were co-added for sufficient signal-to-noise, and 0.2 s recycle delays were sufficient for complete relaxation.The 27 Al spectra were referenced to a 1 M aqueous solution of AlNO 3 at 0 ppm.The DMFit software (Massiot et al., 2002) and Sola software within the Bruker Topspin suite were used to simulate and fit the spectra.The details of the simulation fit can be found in Appendix D.

Basic analyses of milled tailings (Group U) and thermally treated tailings (Group H1 and H2)
All detailed results of basic analyses of raw tailings and thermally treated tailings can be found in Appendices A and B (due to the size limitation of the Journal), which include the particle size distribution, chemical oxides compositions, and mineralogical composition of milled tailings (Group U) and the changes in the mineralogical compositions caused by thermal treatment (Group H1 and H2).In this section, several significant findings are summarized below.
• All tailings had similar particle size distributions (D 50 = 7.37 ± 1.73) in the range of 0.1 to 100.0 μm (see Fig. A1), although LT-U was slightly finer than other tailings.Thus, the particle size would have a negligible effect on the following characterization and reactivity.

Thermogravimetric analysis
In summary, it should be noted that in the present study, the heating conditions of milled tailings were determined to be 600 ºC to complete the decompositions of gibbsite, boehmite, and chamosite and 900 ºC to finish the decomposition of ankerite.Moreover, the weight changes at 600-900 • C of BT-U, GT-U, and CT1-U were minor (i.e., 0.97-1.29 wt%), while that of CT2-U and LT-U were considerable (i.e., 11.42-12.78wt %).Therefore, considering the following R3 test results, the temperature range between 600 and 900 • C not necessarily selected.

Dissolution of tailings in 1 M NaOH solution
The dissolution tests were conducted to compare the relative reactivities of tailings.Fig. 2 presented Al and Si concentrations obtained from the dissolution test for Group U, H1, and H2 tailings at 4 to 168 h.The Ca, Mg, and Fe concentrations were negligible and lower than the detection limit; thus, they were not reported in this paper.
Fig. 2(a) shows that BT-U had a significantly higher Al concentration than other milled tailings.The Al concentration of BT-U was 31.6 mmol/ l after 168 h, while the Al concentrations of other tailings were lower than 0.62 mmol/l even after 168 h.The XRF result showed that BT-U had the highest Al 2 O 3 content due to Al-bearing minerals compared to other tailings (see Table A1).However, Al 2 O 3 content in XRF could not always be correlated with the Al ion dissolved in the solution.The Al 2 O 3 content in GT-U (15.45 wt.%) was higher than that in CT1-U (9.75 wt.%), but the concentration of dissolved Al for GT-U (no detection) was less than that of CT1-U (0.62 mmol/l) shown in Fig. 2(a).The mineralogical properties of tailings should also be considered together to assess their reactivity.GT-U contained 49.5 % albite (NaAlSi 3 O 8 ), an Alcontaining feldspar mineral insoluble in water or alkaline solution, which resulted in the low reactivity of GT-U (Holm and Kleppa, 1968).
Fig. 2(b) illustrates that BT-U showed the highest Si concentration, approaching 7.4 mmol/l after 168 h.It should be noted that BT-U had the lowest SiO 2 content based on the XRF result (i.e., 12.69 % in Table A1).The Si concentrations could be affected by the solubility of Sibearing mineral contents; for example, the Si-bearing minerals in Group U tailings, except for BT-U, consisted of insoluble minerals such as quartz and albite, whereas those in BT-U were mainly of kaolinite, a soluble mineral, with minor quartz (see Table B1).Meanwhile, Si concentrations were similar between GT-U and CT1-U and between CT2-U and LT-U, along with consistency with previous XRD and TG results (see Fig. B1 and Fig. 1), indicating only thermal decompositions of chamosite.
Fig. 2 (c-f) shows Al and Si-dissolved concentrations of Group H1 and H2 tailings, which were milled tailings after heating at 600 • C and 900 • C, respectively.Despite heating conditions, BT always showed higher Al and Si concentrations than other tailings.However, as displayed in Fig. 2(c) and (e), the Al concentrations of BT decreased while increasing the heating temperature; for example, that of BT-U, BT-H1, and BT-H2 after 168 h were 31.6,17.4, and 6.2 mmol/l, respectively.On the other hand, Fig. 2(d) and (f) show that the Si concentration in BT was remarkably increased with increasing heating temperature, given that 168-hour Si concentrations of BT-U, BT-H1, and BT-H2 were 7.4, 15.6, and 16.5 mmol/l, respectively.Thus, it should be noted that high heating conditions significantly affected the Al and Si-dissolved concentration of bauxite tailing (BT).
On the contrary, GT, CT2, and LT showed negligible changes in Al and Si concentration regardless of heating conditions.Thus, although chamosite or ankerite were thermally decomposed below 600 • C or 900 • C, respectively (see Fig. 1), their decompositions were ineffective in Al and Si dissolutions due to the low amount of decomposable phases.However, CT1 had a notable increase in Si concentration after 900 • C heating since the 168-hour Si concentration of CT1-U was 4.5 mmol/l while those of CT1-H1 and CT1-H2 were 3.8 and 13.3 mmol/l, respectively.Thus, proper heating temperature selections could improve the dissolution of Si species depending on tailing types.

R3 reactivity test
Fig. 3(a)-(d) shows the first 24 h of normalized heat flow of the R3 mixture for all tailings in Group U, H1, and H2.The results were collected 45 mins after inserting samples into the calorimetry channels to stabilize them from the initial heat disturbance (Hamdan et al., 2023).Except for BT-U and BT-H1 tailings-contained R3 mixtures, other tailing mixtures showed only one initial peak within two hours.However, BT-U and BT-H1 mixtures showed an initial peak within two hours and second peaks at ~3 and ~10 h, respectively.In the R3 test, the initial peak indicates the released heat due to the wetting process from the rapid dissolution of the reactive components in the tailing mixture.The initial peak of the mixture varied from 3 to 8.5 mW/g depending on the tailings type and heating conditions.On the other hand, the second peak Z. Yao et al. indicates the formation of hydration products (Avet et al., 2016).In other words, except for BT-U and BT-H1, other tailings with different treatments produced no or only minor amounts of hydration or pozzolanic reactions in R3 tests.
In Fig. 3(a), among the BT mixtures with different heating conditions, the BT-H1 mixture showed the highest initial heat flow value (8.5 mW/g), indicating the most rapid reaction rate at an early.Meanwhile, BT-U, BT-H1, and BT-H2 had heat flows approaching 1 mW/g after 24 h.On the other hand, unlike the BT mixture, GT, CT1, CT2, and LT mixtures in Fig. 3(b), (c), and (d) showed similar and low reactivity regardless of heating conditions; thus, heat treatments had minor effects on the changes in their reactivity.Only the CT2-H1 mixture showed slightly higher reactivity.Although the initial heat flow of CT2 slightly increased when it was heated at 600 • C, the heat flow diminished fast, similarly for CT2-U and CT2-H2.Furthermore, thermal treatment at 900 • C negatively affected the CT2 and LT mixtures.Fig. 3(e) presents the accumulated heat of R3 mixtures for tailings with different heating conditions, and 4 data points at 5, 24, 72, and 168 h are plotted to facilitate the comparison.In this study, accumulated heat was used as an indicator of relative reactivity for used tailings because this heat at 7 days (i.e., 168 h) was proved to have a reasonable correlation with the strength development in the previous study (Avet et al., 2016).
It should be noted that only BT mixture showed remarkably more considerable accumulated heat than others and exhibited a similar cumulative heat to fly ash.In other words, bauxite tailing was promising to be used as an SCM, while other tailings used in this paper were less reactive.Among the tailings used in this paper, bauxite tailing was promising to be used as an SCM.Moreover, heat treatment at 600 • C was most appropriate to improve the SCM performance of BT because the BT-H1 mixture showed the highest accumulated heat at around 241.5 J/ g at 168 h.Therefore, more rigorous studies on BT tailings were Z. Yao et al. conducted in this study to understand their changes in reactivity by thermal treatment.

Unheated tailings as a supplementary cementitious material
Supplementary cementitious materials (SCM) are generally classified into two groups based on their chemical composition: (1) hydraulic materials and (2) pozzolanic materials.The former contains high calcium content, while the latter has low Ca content and noticeably high reactive Si content (Simonsen et al., 2020;Mehta and Monteiro, 2006).Thus, regarding chemical oxide composition (see Table A1), the five tailings used in the present study may have potential candidates for pozzolanic materials.
Among the widely used pozzolanic materials, fly ash was reported to generate normalized accumulated heat of at least 150 J/g at 7 days in the R3 isothermal calorimetry test (Londono-Zuluaga et al., 2022;Ahmaruzzaman, 2010).However, as illustrated in Fig. 3(e), the 7-day normalized accumulated heat of unheated bauxite tailing (BT-U) was 208.4 J/g, while those of other tailings were only 31.9-75.6J/g, significantly lower than that of fly ash.Among the unheated tailings, dissolution testing results (see Fig. 2) showed that BT-U had the highest dissolved Al and Si ions despite having the lowest SiO 2 content (see Table A1).Thus, BT-U showed the most promising reactivity as an SCM.
As shown in Fig. 4, the present study conducted TG/DTG analysis on the BT-U after the R3 test and collected the XRD patterns of BT-U after the dissolution and R3 tests.After the dissolution test, the solid residue of BT-U was prepared by drying it in a vacuum oven at 105 • C for 24 h.In addition, after the 7-day R3 isothermal calorimetric test, the solid residue of BT-U was prepared by a hydration stop process using isopropanol and diethyl ether and dried in a vacuum oven at 40 • C for 1 h.Then, the same equipment and experimental conditions were used for TG and XRD analyses.Fig. 4(a) compares the XRD patterns of BT-U under different heating conditions and after the dissolution and R3 tests.In BT-U dissolved residue after the dissolution test, there was no detection of the gibbsite phase and reduction of the kaolinite phase, indicating their dissolutions in the alkaline environment (i.e., 1 M NaOH solution).Moreover, these dissolved phases may participate in the reaction of the R3 test because monocarboaluminate (3CaO⋅Al 2 O 3 ⋅CaCO 3 ⋅11H 2 O) was noticeably identified in the R3 mixture of the BT-U.Unfortunately, calcium aluminosilicate hydrate (CASH) was hardly identified because it is an amorphous phase with minor Si content.For example, the R3 mixture for BT-U only accounts for 2.8 % of SiO 2 based on XRF data and the R3 mixture proportion. of boehmite and kaolinite, although their peak intensities in DTG were much smaller than the corresponding peaks from the BT-U.Meanwhile, since the R3 mixture was prepared by mixing BT-U, Ca(OH) 2 , and CaCO 3 , it contained two more DTG peaks at ~450 • C due to dehydration of Ca(OH) 2 (Irabien et al., 1990) and at ~700 • C due to decarbonization of CaCO 3 (Karunadasa et al., 2019).Although Ca(OH) 2 consumption in the R3 test can be used as an indicator of the reactivity of tested material, TGA was challenging to evaluate the reactivity of BT-U because DTG peaks of boehmite and kaolinite in BT-U overlapped with that of Ca (OH) 2 .
Next, it should be noted that the R3 mixture of BT-U showed one DTG peak at ~140 • C, which was not detected in the BT-U.This peak indicates the formation of the hydration products, given that the intensity of the DTG peak of the gibbsite was relatively reduced after the R3 test.F. Avet et al. (2016) reported the formation of monocarboaluminate from the R3 test for calcined clay.Therefore, it can be inferred that the gibbsite was the main mineral phase that would react with calcium hydroxide and form monocarboaluminate.Moreover, since monocarboaluminate was thermally decomposed at 140-170 • C (Afroz et al., 2022), TG result was consistent with the XRD result.In addition, the gradual mass decrease from ~140 • C to ~350 • C also indicated the formation of hydration products, mainly calcium aluminosilicate hydrates (CASH) (Hamdan et al., 2023a(Hamdan et al., , 2023b;;Haha et al., 2011), supporting the existence of CASH which was not identified in the XRD pattern [see Fig. 4(a)].

Effects of thermal treatment on the reactivity of tailings
The R3 testing results (see Fig. 3) showed that thermal treatments significantly affected the reactivity depending on the types of tailings.Although tailings except for BT were not worth calcining nor to be considered as SCM, their reactivity changes with different heating conditions were given in this section.
First, CT1 and GT showed relatively little change in reactivity before 28 h compared to other tailings [see Fig. 3(b)] but continuously increased reactivity until 168 h [see Fig. 3(e)].For example, the accumulated heat of CT1-H2 was increased by about 165.1 % compared to CT1-U.Moreover, CT1-H2 showed about 2.95 times higher amounts of dissolved Si species than CT1-U, though it contained insufficient Si to be used as SCM (see Fig. 2).Meanwhile, CT1-U contained approximately 31.0 % chamosite [(Fe 2+ ,Mg) 5 Al(AlSi 3 O 10 )(OH) 8 ] and 69.0 % quartz (SiO 2 ) (see Table B1).Chamosite is a less reactive mineral, given that it is one of the most common minerals in phyllosilicate chlorite groups, and phyllosilicate-type minerals were reported to have weak reactivity (Xu and Van Deventer, 2000).Since quartz is a low-reactive mineral, CT1-U showed less reactivity in the R3 test.However, heat treatment induced iron phase transitions in chamosite such as iron oxidation, two-step dehydroxylation process, decomposition, and recrystallization.Notably, the recrystallization stage would produce amorphous silica, providing sufficient reactive silica (Steudel et al., 2016); thus, 1st copper tailing heated to 900 • C (i.e., CT1-H2) showed enhanced reactivity and dissolved Si species compared to the unheated copper tailing (i.e., CT1-U).
Then, the reactivities of CT2 and LT increased when they were heated at 600 • C but degraded when they were heated at 900 • C (see Fig. 3).Meanwhile, the dissolution of Al and Si of both tailings showed fewer changes regardless of heating conditions because they were mainly composed of a non-reactive mineral (e.g., quartz) and mineral excluding Al or Si oxides [i.e., ankerite; Ca(Fe,Mg,Mn)(CO 3 ) 2 ] (see Fig. 2 and Table B1).In the XRD test [see Fig. B1(d) and (e)], when both tailings were heated at 600 and 900 • C, ankerite was mainly decomposed, while magnesioferrite [Mg(Fe 3+ ) 2 O 4 ] and anhydrite (CaSO 4 ) were newly formed.Since magnesioferrite is a stable mineral in the spinel group (Sokhatskaya et al., 1972), the phase transition to magnesioferrite from ankerite might slightly worsen the reactivity of CT2 and LT.Anhydrite (CaSO 4 ) could be used as a retarder for alkaline activation of ground granulated blast furnace slag (i.e., GGBFS) by reducing pH (Hamdan et al., 2023) or as a setting retarder for Portland cement hydration (Huang et al., 2022).Thus, it could react with water, affecting the reaction like gypsum (CaSO 4 ⋅2H 2 O).Meanwhile, chamosite was decomposed when CT2 and LT were heated at 600 • C [see Fig. B1(d) and (e)].The thermal decomposition of chamosite may enhance the reactivity of CT1 as the chamosite decomposition may produce amorphous.Then, it might improve the reactivity of 2nd copper and lead tailings heated to 600 • C (i.e., CT2-H2 and LT-H2, respectively).However, this hypothesis should be further studied.
BT showed remarkably higher reactivities than both tailings.Thus, the present study focused more on the reactivity of BT in the next section using further analysis, including in-situ high temperature (HT)-XRD and 27 Al solid-state nuclear magnetic resonance (NMR).

Effect of phase transition of bauxite tailing on the reactivity
It should be noted that heat treatment at 600 • C showed the highest reactivity.However, heat treatment at 900 • C lowered the reactivity of BT and even worsened it than the initial condition (see Fig. 3).In particular, the dissolution test (see Fig. 2) showed that dissolved Al species decreased while dissolved Si species increased when BT was heated at higher temperatures.This section rigorously discusses reactivity changes of bauxite tailing due to thermal treatments.

Decomposition of gibbsite and boehmite due to the thermal treatment
As illustrated in Fig. C1, in-situ HT-XRD analysis was conducted on the BT to identify its in-situ phase transition at different temperatures from 26 • C to 1200 • C. It should be noted that, as shown in Fig. C2, the XRD pattern measured with other X-ray sources could be easily converted to another source using Highscore Plus software, according to Bragg's law.Thus, the X-ray scan measured by HT-XRD with a copper Xray source could be compared with that measured by normal XRD with a cobalt X-ray source.
In the HT-XRD result (see Fig. C1), gibbsite was decomposed below 300 • C, while boehmite and kaolinite disappeared between 450 and 600 • C, showing consistency with normal XRD (after cooling of thermally treated BT) and TGA results [see Fig. 4(a) and Fig. 1(a), respectively].Unfortunately, in-situ HT-XRD could not identify alumina formation because the phases in the BT-U sample exposed to the high temperature were yet to be crystallized; thus, all phases present poorly crystalline multiple broadened peaks overlapping each other.Furthermore, the phase identification of BT-U was challenging because most XRD peaks detected in the normal XRD were broadened and shifted leftward with the elevated temperature.Despite the limited information from the result of in-situ HT-XRD, the HT-XRD results provided direct evidence of the decomposition of gibbsite and boehmite, consistent with the results obtained from the TGA in Section 3.2.

The dissolution of Al due to the thermal treatment
The dissolved Al species contents in BT-U, BT-H1, and BT-H2 in 1 M NaOH solution kept decreasing with the increase in the temperature for the thermal treatment.This trend resulted from the change in the phases of aluminum-bearing minerals (i.e., boehmite, gibbsite, and kaolinite).In particular, BT-U was predominantly composed of boehmite [AlO (OH)] [see Fig. 4(a)].According to the previous study (Mercury et al., 2020;Wilson, 1979;Lamouri et al., 2017), Al 2 O 3 from the thermal decomposition of boehmite was transformed to γ- therefore, bauxite heated to 600 • C (BT-H1) and 900 • C (BT-H2) was expected to mainly contain γ-Al 2 O 3 and δ-Al 2 O 3 , respectively.
In addition, the phase change of Al 2 O 3 was also observed in the XRD result, indicating the peak splitting at 54 • with elevated temperatures [see the green arrows in Fig. 4(a)].In particular, an alumina phase for bauxite tailing heated to 600 • C (i.e., BT-H1) was mostly matched with γ-Al 2 O 3 (cubic unit cell structure with a = b = c = 3.95 Å).On the other Z. Yao et al. hand, that phase heated to 900 • C (i.e., BT-H2) was γ-Al 2 O 3 (tetragonal unit cell structure with a = b = 5.65 Å and c = 7.87 Å) along with δ-Al 2 O 3 (tetragonal unit cell structure with a = b = 5.59 Å and c = 23.67Å).In other words, the crystalline structure of alumina was changed from cubic to tetragonal (Prasetya et al., 2020).
Furthermore, the phase transition of alumina with elevated temperatures might be the result of diminished Al concentration in the dissolution test (see Fig. 2).First, since gibbsite [Al(OH) 3 ] and kaolinite [Al 2 Si 2 O 5 (OH) 4 ] were fully dissolved in the 1 M sodium hydroxide solutions [see Fig. 4(a)], they became the primary sources contributing to the dissolved Al ion concentration in BT-U.Then, when bauxite was heated to 600 • C, gibbsite, kaolinite, and boehmite were thermally decomposed and became γ-Al 2 O 3 and metakaolin (i.e., aluminosilicate mineral).However, since γ-Al 2 O 3 in BT-H1 had less solubility than gibbsite in BT-U (Carrier et al., 2007;Bénézeth et al., 2008), BT-H1 showed less Al ion concentration than BT-U.Likewise, BT-H2 contained δ-Al 2 O 3 and metakaolin, offering much lower Al ion concentration than BT-U and BT-H1.Moreover, the Al dissolution changes from the phase transition of Al 2 O 3 could affect the reactivity of bauxite tailings with different heating conditions, according to the previous paper reporting reactivities of the various phases of alumina (Amrute et al., 2020).
In this study, XRD analysis was challenging in identifying alumina phase change with different heating conditions because they overlapped.Thus, the present study conducted 27 Al NMR analysis to identify the alumina polymorph in detail.Fig. 5 presents the 27 Al solid-state NMR spectra for bauxite tailings with different heating conditions.Detailed deconvolution of NMR spectra can be found in Appendix D. First, as shown in Fig. 5(a), the BT-U contained predominant octahedral Al VI aluminate species and relatively low tetrahedral Al IV species.Then, boehmite, gibbsite, and kaolinite spectra were simulated since the XRD analysis of BT-U identified the presence of those aluminum-bearing minerals [see Fig. 4(a)].Interestingly, all the aluminate sites were sixcoordinate octahedral sites, with boehmite and kaolinite consisting of one Al VI site and gibbsite consisting of two Al VI sites.The simulation yielded a near complete spectral assignment of the 27 Al NMR spectrum for the BT-U sample, based on the specific NMR parameters known from literature values (Chandran et al., 2019) or measurement of reference compounds (i.e., kaolinite in the presence case).This fit not only well described the central transition (in the ~100 to − 100 ppm range), which was particularly sensitive to the quadrupolar coupling (C Q ) and the quadrupolar anisotropy (η Q ).However, the weak 27 Al signal corresponding to the Al IV site was not accounted for by the known aluminate phase; thus, it was assigned to an amorphous silica-alumina phase.
However, considering the expected transformation sequence of boehmite and gibbsite to the final α-Al 2 O 3 (Lamouri et al., 2017)  ] is a crystalline aluminosilicate phase, and metakaolin is an amorphous aluminosilicate phase (Abbas et al., 2020).Kaolinite is the primary soluble Si source, but it could only be partially dissolved in the dissolution test [see Fig. 4(a)].On the other hand, after heat treatments, kaolinite was transformed into metakaolin, widely used in SCM due to its superior pozzolanic reactivity.Then, metakaolin formation could significantly contribute to increasing Si ion concentration (see Fig. 2).The simulation fit of 27 Al NMR (see Fig. 5) shows the disordered aluminum-silicate phases [labeled as amorphous silica-alumina phase (Si-Al)].However, it should be noted that the Si-Al phase was generated to fit the remaining part of the spectra for bauxite tailing with different heating temperatures because there are significant Z. Yao et al. challenges in using the metakaolin signal to fit the spectrum (Rawal et al., 2016;Souayfan et al., 2022).This unidentified Si-Al phase was negligible in the BT-U [see Fig. 5(a)] but distinctive in both BT-H1 and BT-H2.This amorphous Si-Al phase (including metakaolin) seems to participate in the reaction and the dissolution, increasing the Si concentration in 1 M NaOH.Moreover, it is worth noting that BT-H1 and BT-H2 had similar Si ion concentrations supporting the dissolution of metakaolin (the primary source of soluble Si).

Summary of reactivity of the bauxite tailings due to the thermal treatment
The dissolution test (see Fig. 2) showed that thermal treatment increased the Al concentration of BT-U and decreased the Si concentration of it.In particular, the R3 test results (see Fig. 3) showed that bauxite heated to 600 • C (i.e., BT-H1) had the highest accumulated heat, indicating the best reactivity as SCM among used heating conditions due to well-balanced amounts of dissolved Al and Si ions from the thermal decompositions of gibbsite, boehmite, and kaolinite [see Fig. 1(a)].Meanwhile, unheated bauxite (i.e., BT-U) did not show sufficient reactivity due to the insufficient amount of Si ions dissolved from kaolinite, though it had the highest amount of Al ions dissolved from gibbsite and kaolinite [see Fig. 4(a)].Moreover, bauxite heated to 900 • C (i.e., BT-H2) showed the highest amount of dissolved Si from kaolinite but decreased Al concentration due to the phase transition approaching a higher crystalline Al 2 O 3 phase [see Fig. 4(a)].

Conclusion
The present study investigated the effects of thermal treatment on the properties of five different tailings, including bauxite, gold, copper, and lead tailings (i.e., BT, GT, CT1, CT2, and LT, respectively), to explore the potentials of using the tailings as SCMs.This study mainly focused on the rigorous phase changes caused by the thermal treatment and its effects on the reactivity of tailings.Heat treatments were performed at 600 • C and 900 • C depending on the thermal decomposition of significant minerals (e.g., gibbsite, boehmite, kaolinite, chamosite, and ankerite) contained in tailings.Then, PSD, XRF, XRD, and TG analyses were conducted to characterize their chemical and mineralogical properties.Moreover, their reactivity was evaluated by dissolution tests and R3 isothermal calorimetric tests.Detailed conclusions are given as follows.
• Although tailings contained high Al 2 O 3 and SiO 2 contents, their Al and Si dissolutions and supplementary cementitious reactivity could significantly depend on the types of mineral phases containing those species.Most of the Al and Si-bearing mineral phases in bauxite tailing were gibbsite, boehmite, kaolinite, and highly soluble and thermally decomposable minerals, while other tailings contained albite and quartz, which are none-or low-reactive minerals.• It should be noted that BT had remarkably high Al and Si dissolutions and supplementary cementitious reactivity.In particular, heating at 600 • C most effectively enhanced the reactivity of BT compared to non-heating and heating at 900 • C.This seems to result from the well-balanced amounts of dissolved Al and Si in the treated tailings at 600 • C by the thermal decompositions of gibbsite, boehmite, and kaolinite.
• On the other hand, all tailings except for bauxite tailings used in this study is not worth calcining nor considered SCM because they were ineffective based on the heat release from the R3 test and dissolution results.Although thermal treatment can lead to a minor improvement in their reactivity due to thermally decomposed minerals, the overall reactivity of these tailings is too low to be sufficient for acting as SCMs.• In addition to confirming previous studies on the reactivity of different tailings and the effects of thermal treatment for using the tailings as alternative SCMs, this study further advanced the fundamental understanding of changes in the reactivity of Al-rich tailings.
Understanding the mineralogical compositions of tailings and their phase changes over temperature can be crucial for screening the potential tailings for alternative SCMs.In this study, the possible incorporation of Al-rich BT tailings/calcined BT tailings as SCMs seems to be one of the promising pathways to develop an alternative SCM to replace conventional SCMs such as blast furnace slag and fly ash.
Many mining tailings have been investigated for the construction application.However, most approaches are based on trial-error methods without an in-depth understanding of reactivity.This approach also requires extensive time and costs.The current study suggested that their potential application may be understood by relatively easy tests, including mineralogy and R3 test, and thus applicable tailings can be selected before the extensive trial-error experiments.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.As shown in Table A1, different types and tailings sources resulted in different chemical oxide compositions.This is expected because the chemical composition of tailings is highly dependent on the mineralogy of the ore body and the method of processing to extract metals.The BT-U contained the highest amount of Al 2 O 3 (55.68wt.%) as the main component, followed by SiO 2 (12.69 wt.%) and Fe 2 O 3 (9.26wt.%).On the other hand, other milled tailings (i.e., GT-U, CT1-U, CT2-U, and LT-U) contained SiO 2 at 38.36 to 64.33 wt.%.GT-U, CT1-U, CT2-U, and LT-U were mainly composed of SiO 2 , Al 2 O 3 , Fe 2 O 3 , CaO, and MgO with minor amounts of other oxides.CT1-U and LT-U showed relatively high amounts of MgO and SO 3 .

Table A1
Chemical oxide compositions of Group U tailings (%, in weight).LOI is the loss of ignition.Appendix B. Mineralogical analysis of the tailings (Group U, H1, and H2) Fig. B1 shows the XRD patterns of Group U, H1, and H2 tailings for the crystalline changes of tailings with different heating temperatures (i.e., 600 • C for H1 and 900 • C for H2).Table B1 displays the mineral compositions (without considering an amorphous phase) of only Group U tailings, quantified by Rietveld refinement using Highscore software.
Two copper tailings (i.e., CT1-U and CT2-U) showed different mineral compositions depending on their sources.Moreover, CT2-U showed mineral compositions similar to LT-U.For example, while CT1-U was composed of quartz and chamosite, both CT2-U and LT-U were composed of high contents of quartz and ankerite with minor amounts of chamosite and pyrite.
As shown in Fig. B1(a), while BT-H1 had the appearance of alumina (at 54 and 79.5 • 2θ), aluminum-bearing minerals (kaolinite, boehmite, and gibbsite) disappeared.The decompositions of aluminum-bearing minerals in BT-U produced alumina (Al 2 O 3 ) in BT-H1.BT-H2 showed an XRD pattern similar to BT-H1, but the alumina pattern split at 54 • 2θ (see green arrows in Fig. B1(a), which seems to result from changes in the lattice parameters of alumina at 900 • C (Prasetya et al., 2020). In

Table B1
Mineral compositions of five unheated tailings calculated from the XRD result (see Fig. B1).

Appendix C. HT-XRD for bauxite tailing
In the present study, the phase transition of bauxite tailing was evaluated by high-temperature-XRD techniques using the X-ray diffractometer with a copper X-ray source equipped with a high-temperature chamber.Fig. C1 presents collected high-temperature XRD patterns of bauxite tailing at 26, 300, 450, 600, 750, 900, 1050, and 1200 • C.Moreover, the sample holder was also measured by the same X-ray diffractometer at 26 • C to compare it with the HT-XRD patterns of bauxite tailing.
Since normal XRD patterns of other tailings were collected by another X-ray diffractometer with a cobalt X-ray source, Fig. C2 compares the XRD pattern of BT-U between cobalt and copper sources.
4.3.3.The dissolution of Si due to the thermal treatmentCompared to the dissolution of Al, the dissolution of Si of BT-U, BT-H1, and BT-H2 in 1 M NaOH solution increased after the thermal treatment.Given that the main phases of bauxite tailing were gibbisite [Al(OH) 3 ], boehmite [AlO(OH)], and kaolinite [Al 2 Si 2 O 5 (OH) 4 ], only kaolinite (or even metakaolin) can contribute to Si dissolution.In particular, kaolinite [Al 2 Si 2 O 5 (OH) 4

Fig. 5 .
Fig. 5. Experimental 1D 27 Al MAS NMR (plotted in blue) and the simulated fit (plotted in red) for (a) BT-U, (b) BT-H1, and (c) BT-H2.Individual components of the fit are plotted in various colors beneath the spectra.
Fig.C2(a) displays the XRD pattern of BT-U measured by a cobalt X-ray source, while Fig.C2(b) illustrates the XRD pattern of BT-U converted into copper X-ray source using Highscore Plus software according to Bragg's law.

Fig. C2 .
Fig. C2.XRD result of BT-U using a normal X-ray diffractometer with a cobalt source: (a) raw data and (b) converted data from cobalt source to copper source using Highscore Plus software.
, BT-H1 contained γ-Al 2 O 3 , χ-Al 2 O 3 , and η-Al 2 O 3 and, while BT-H2 was included δ-Al 2 O 3 , θ-Al 2 O 3 , and κ-Al 2 O 3 due to decompositions of boehmite and gibbsite.Moreover, given that BT-U contained 48.4% boehmite (see TableB1), BT-H1 and BT-H2 were predominantly composed of γ-Al 2 O 3 and δ-Al 2 O 3 , respectively.However, since a satisfactory fit was not obtained, additional spectral components representing disordered sites at Al IV , Al V , and Al VI positions were required.This can be attributed to the increase in amorphous content associated with metakaolin due to kaolinite decomposition.Therefore, the change in the atomic structure of Al obtained from 27 Al NMR spectra well agrees with the phase changes of aluminum-bearing minerals in BT-U (increased γ-Al 2 O 3 and δ-Al 2 O 3 by the decomposition of gibbsite and boehmite, and kaolinite), which also support the observed dissolution trend of Al in this study.