Introduction

The cement industry has been implementing strategies to reduce its environmental impact, particularly the significant release of greenhouse gases during its production. Approximately 8% of global CO₂ emissions are attributed to the production of Portland cement, due to the combustion and decarbonation of its raw materials at high temperatures (Aïtcin, 2000; Boden; Andres; Marland, 2017; Gartner, 2004). One of the strategies is the replacement of cement with other materials that can act as fillers and/or react with the binder constituents, known as supplementary cementitious materials (SCM). However, the availability of SCMs may be insufficient to meet the required cement consumption, necessitating the discovery of new sources to enhance their strategic potential (Scrivener; John; Gartner, 2018). Residues from different industrial processes become an interesting alternative, combining the strategy of safely reusing these by-products with the potential for large-scale application.

Some industrial by-products are widely used as SCM (Juenger; Snellings; Bernal, 2019; Li et al., 2021), such as granulated blast furnace slag (Chen; Liu; Yan, 2024; Cheng et al., 2024; Joseph et al., 2024), silica fume (Jeong et al., 2020; Shelote; Bala; Gupta, 2023), and fly ash (Chen; Jia, 2024). The bauxite residue from the Bayer process (BR) has been studied as an alternative to replace cement in cementitious compositions. According to data from the U.S. Geological Survey (2024), global alumina production in 2023 reached 140 million tons, generating approximately 190 million tons of BR (IAI, 2015), which is equivalent to 3,5% of cement production (CEMBUREAU, 2023).

Manfroi, Cheriaf, and Rocha (2014) and Romano et al. (2016) demonstrated the potential of BR in products with improved microstructural, mechanical, and hygroscopic properties. Ribeiro, Labrincha, and Morelli (2012) indicated that the high alkalinity of BR can increase the corrosion resistance of reinforcement concrete. Viyasun et al. (2021) found that partial replacement of cement with BR increases compressive, tensile, and flexural strengths, as well as improves electrical resistivity. Ghalehnovi et al. (2019), Liu and Poon (2016), Tang et al. (2018, 2019), and Yao et al. (2013) observed that combinations of BR with fly ash improve the mechanical properties of self-compacting concrete, despite some rheological losses. Hou et al. (2021) noted that the addition of BR in high-performance concrete reduces workability and initial strength, but this increases with curing time. Kang and Kwon (2017) studied the efflorescence potential in alkali-activated slag cement mortars, finding that BR intensifies efflorescence due to increased capillary pores.

BR is generated during the process of obtaining alumina, when a hot NaOH solution, at temperatures ranging from 140 °C (high-gibbsite bauxite) to 240 °C+ (high boehmite), and up to 280 °C (high diaspore bauxite), is mixed with bauxite ore to convert it to aluminum hydroxide, and the other constituents that are not capable of dissolving are eliminated during the filtration process, generating the BR (Ruys, 2019).

The association of BR with Portland cement is complex, posing significant challenges in fixing soluble elements and potentially hazardous leaching ions. One approach to mitigate the deleterious effects of BR is through an additional process of calcination and/or vitrification, aiming to inert leachable products and improve some properties of cementitious products (Manfroi; Cheriaf; Rocha, 2014; Snellings; Suraneni; Skibsted, 2023; Venkatesh; Nerella; Sri Rama Chand, 2020). However, this additional BR treatment process is energy-intensive, which can lead to an increase in CO₂ emissions and increase the costs of using the residue in combination with cement (IAI, 2020). A low-energy intensity alternative is the combination of the natural, untreated BR with materials that can chemically interact with it, reducing the medium’s alkalinity (Tapas et al., 2021) and encapsulating heavy metals (Samantasinghar; Singh, 2023). That’s the case of this article: use an untreated BR in combination with another material that could inert leachable ions.

Considering the scalability of products that combine diverse materials is fundamental for their successful implementation by the industry. Thus, it is crucial to confront the local reality of BR-generating plants, identifying cements and other SCMs used in the area that could chemically interact with BR to achieve ion fixation. Studying the chemical reactions among all materials used is vital for understanding these combinations and ensuring the safe, large-scale application of these materials in civil construction.

In this study, BR generated by Alumar, in São Luís do Maranhão, was chosen. Based on its location, the most commonly used cement in the region and an SCM were selected, ensuring they were sourced within a radius of no more than 200 km from the alumina plant. This distance, although arbitrary, was designed to make the transportation of materials for the production of cement components both economically and environmentally viable. In the same city as Alumar, CP IV cement is commonly used, and the alumina plant generates electrical energy for self-production with a coal-fired thermoelectric plant, which produces fly ash. Therefore, these materials were chosen to be combined and studied for their chemical reactions. The use of fly ash and BR from the same industry, and the combination of these two by-products, in addition to being logistically viable, seeks to properly dispose of two wastes generated by a company in different industrial processes.

Thus, this study aimed to analyze the chemical reaction of different cementitious paste systems with partial replacement of Portland cement by BR, fly ash and BR with fly ash, using the methods of isothermal conduction calorimetry, X-ray diffractometry (XRD) and thermogravimetric analysis (TGA).

Materials

A Brazilian Portland cement classified as CP IV (addition of pozzolan, 42% of amorphous phase) (ABNT NBR 16697, 2018), commonly used in the region of São Luís do Maranhão, along with BR and fly ash (FA) generated at Alumar’s alumina production plant (referred to as supplementary cementitious materials – SCM), located in São Luís/MA, were the materials used. The SCMs were dried in an oven at 105 ºC and sieved in a sieve with an opening of 106 μm.

Table 1 specifies the physical characteristics of raw materials. The real density was determined in a gas He pycnometer, Micromeritics – AccuPyc II 1340. The specific surface area (SSA) was determined in a Belsorp-Max (Bel Japan) equipment according to the BET method, after sample treatment in a Belprep-vac II equipment at 40 °C and 10^-2 kPa vacuum for 24 h. The particle size distribution (PSD) was evaluated in a Helos (Sympatec) laser granulometry, with a detection range of 0.1–350 μm. The shape factor (SF) was calculated as the ratio between the volumetric surface area (VSA), obtained from the product between real density and SSA, and theoretical VSA_esf, obtained from PSD, assuming a spherical particle model.

The chemical composition of the materials is presented in Table 2, and was obtained by XRF, in a PANalytical spectrometer, model Zetium, with sample preparation using pressed tablet, according to the general guidelines of ISO 29581-2/10 (ECS, 2010). The loss on ignition (LOI) was determined at 1020 °C.

CP IV has a typical composition of this type of cement, fitting the specifications of NBR 16697 (ABNT, 2018). For BR, it is worth highlighting the high levels of Fe and Al from the bauxite ore, and Na close to 6%, resulting from the treatment of the Bayer process. The percentage of sodium indicated in the chemical analysis represents the total amount of the metal in the form of oxide. However, only a small portion of this value corresponds to soluble sodium, which will actually participate in the cement reaction. Fly ash is composed mostly of silica, with significant amounts of calcium and alumina. Given its origin in the northeast of the country, it is important to note that it has different characteristics from fly ashes typically found in Brazil, which commonly come from the southern region where most of the national coal deposits are located (Filho, 2008; Gobbo, 2009).

The TGA, shown in Figure 1, was performed using a Netzsch equipment, T209 F1 model apparatus. The heating rate was maintained at 10 °C.min^-1 up to the temperature of 1000 °C, in an analytical N₂ atmosphere, flow of 20 ml.min^-1 and the volatized gases purged at a rate of 10 ml min^-1. The calculation of cement mass losses, excluding the volatile material mass from the mass loss calculations (normalized by residual mass), is presented on its TGA graph of Figure 1a. The amount of portlandite in the CP IV is less than 2%, which is an acceptable pre-hydration value for the use of the material without compromising quality, likely originating from the cement production process, which involves cooling with water, and/or from its storage after packing, which can absorb some ambient humidity, even if properly stored.

The peak of DTG at 911 °C in the CP IV (Figure 1a) refers to the presence of the mineral wollastonite, found in the fly ash of the region, detected in subsequent XDR results, and, according to Scrivener, Snellings and Lothenbach (2018), decomposes near 800 °C. In BR (Figure 1b), it is possible to verify the presence of gibbsite and goethite minerals, due to the significant presence of Fe and Al from the bauxite ore, along with calcite, minerals commonly found on TGA of various BR (Romano et al., 2021). In FA (Figure 1c), the minerals gypsum, calcite, muscovite, enstatite and wollastonite were identified. It is known that at the beginning of the burning process at the FA thermoelectric plant, limestone and sand are added to the boiler, which explains the presence of these minerals.

The mineralogical composition is shown in Figure 2 and was determined by XRD in a PANalytical diffractometer, model Empyrean III. The samples were prepared on a display with manual pressing and leveled with the aid of a metal blade, with the set fixed to a metal base. The tests were performed with copper radiation, automatic gap of 0.5°, nickel filter and rotation frequency of 2 seconds per revolution, step of 0.02°, remaining in each step for 300 s, in the range of 5° < 2θ < 70°.

In CP IV (Figure 2a), characteristics phases of clinker were found, such as alite (C₃S), belite (C₂S), calcium aluminate (C₃A) and brownmillerite (C₄AF). The intensity of calcite was not high enough to be detected in XRD, due to its low presence (Figure 1a), but it is known that this mineral is found in all types of Brazilian cements. In BR (Figure 2b), hematite and goethite were identified, due to the significant presence of iron (more than 30 %), sodalite from the ore treatment with NaOH, and gibbsite, indicating the aluminum present in the bauxite ore. Anatase, a titanium mineral, and calcite were also found (Romano et al., 2021). In FA (Figure 2c), the mineralogical phases quartz, hematite, enstatite, muscovite, wollastonite, gypsum and calcite were identified. The presence of an amorphous halo indicates a material without a defined crystalline structure, characteristic of pozzolans.

Methods

All pastes were produced with fixed water-to-solid ratio of 0.5 by weight: the reference paste (cement only) and the others with 70% cement and 30% SCM (by volume), named as 100CP, 70CP_30BR, 70CP_30FA, 70CP_15BR_15FA, as detailed in Table 3.

To mix the pastes, the powders were homogenized in a plastic container, and then water was added. A manual premix was carried out for 1 minute. Then, the container was placed in a Flacktek SpeedMixer^® equipment, where it was mixed for 2 minutes at a rotational speed of 1500/minute.

To quantify the heat flow released during the chemical reactions of cement, an isothermal conduction calorimeter, Calmetrix I-CAL 8000 HPC, was used, with a controlled temperature of 23 ºC for a period of 72 hours. The duration of the induction period was determined according to C1679 (ASTM, 2017). This involves drawing lines tangent to the deceleration phase after the wetting peak and to the acceleration period, then drawing a line parallel to the x-axis and tangent to the induction period (low dissolution rate). The intersection of these lines determines the duration of the induction period.

The evolution of hydrated crystalline phases was monitored using the XRD and TGA methods, with the same equipment described previously. The goal was to identify which crystalline phases were being formed or consumed during the analyzed hydration period at the events of greater heat release indicated by calorimetry. Therefore, the hydration was stopped shortly after these events to perform XRD and TGA analyses.

The hydration was stopped by using the solvent method. Portions of the previously mixed paste were placed into Erlenmeyer flasks. At the desired times, isopropyl alcohol was added to the flasks, changed after 1 hour, and then replaced again with a more diluted form of isopropyl alcohol after another hour. After 24 hours of immersion in the more dilute solvent, it was replaced by ether, where it remained for another 24 hours. After this time, the material was dried in an oven at 40 °C, crushed and passed through a sieve with a mesh opening of 106 µm to perform the tests (Maciel et al., 2019).

Results and discussions

Calorimetry

Figure 3 shows the chemical reaction by calorimetry of the four compositions analyzed, normalized by the cement mass, to evaluate the effect of the SCM. Figure 3a represents the heat flow and Figure 3b the heat released. The symbols on the graphic of Figure 3a indicate the maximum formation of some hydration products: * stands for calcium silicates hydrates (C-S-H) and portlandite (CH); # is for ettringite (AFt); and + stands for monosulfoaluminate (AFm). In the case of heat released (Figure 3b), the first hour of hydration was disregarded in the graphic representation, and the curves were normalized by the initial amount of heat released of the reference composition. In this way, all results are integrated, being represented with a start at the same time and with the same initial heat released. Additionally, the value of cumulative heat at the end of the experiment is written above the curve of each sample. The blended compositions release more heat than the reference, indicating physicochemical interactions between the materials, which will be discussed below.

Figure 4 shows some parameters calculated from the calorimetry experiment. Figure 4a is the duration of induction period of each paste, Figure 4b is the reaction rate at the acceleration period and Figure 4c is the time at C-S-H/CH peak.

For the reference composition, the induction period lasted 1 hour and 36 minutes, ending at 2 hours and 24 minutes, with setting time, according to the C1679 method (ASTM, 2017), at 4 hours and 13 minutes. The peak of C-S-H/CH formation occurred at 9 hours 34 minutes, releasing a heat flow of 1.85x10^-3 W/g. El Fami et al. (2022) monitored the hydration of cement pastes with replacement of Portland cement by fly ash up to 40%, and, even though the composition of the materials is different, for comparison purposes, they obtained similar heat flow values. Furthermore, it was possible to observe a peak at 32 hours, releasing 0.70x10^-3 W/g of heat, which corresponds to the slower formation of AFt, due the addition of fly ash in CP IV, which influences the formation of aluminates.

For the blended pastes, it is evident that the SCMs caused an acceleration of reactions. Figure 4a shows a shorter induction period for all blends, with 70CP_30FA having the shortest time. During the induction period, C-S-H nucleates rapidly, increasing the calcium concentration in the solution, making the C₃S dissolution rate very small (Scrivener; Juilland; Monteiro, 2015). With a reduced induction period, it can be inferred that the effects of SCMs were to increase this dissolution rate and promote faster C-S-H nucleation. Martin (2017) investigated the effect of fly ash on the hydration of calcium sulfoaluminate cement, finding that increasing fly ash contents accelerated the hydration of the cement due to the filler effect, caused by the presence of additional surfaces that accelerate nucleation and formation of hydrates. The shape factor evaluates how far the particles are from the perfect spherical shape. When cement grains are replaced by grains with greater surface area and greater irregularities, which is the case of the SCMs, it promotes greater shear on its surface, favoring the formation and growth of hydrates (C-S-H nucleation) (Berodier; Scrivener, 2014; Deschner et al., 2012; Scrivener; Juilland; Monteiro, 2015). Both SCMs have bigger SSA than cement, providing more surface for hydrates to grow, but the fly ash has a much higher shape factor than cement and BR, likely causing the most significant influence on the acceleration of reaction at this point. The ternary system (70CP_15BR_15FA) results in a balance of both SCM’s effects.

Figure 4b presents the values of reaction rate during the acceleration period, equivalent to the slope of the graphic of Figure 3a at this moment. When the heat curves are essentially the same, with the same shape, it is understood that there are no changes in hydration related to chemical interactions, with the changes observed in heat flow essentially due to physical factors, such as the filler effect (Scrivener; Juilland; Monteiro, 2015). In compositions with BR, a change in the curve of Figure 3a is observed at this moment, as shown by significantly higher reaction rate values presented in Figure 4b. This indicates physicochemical interactions with cement that will be explored in the sequence, related to when BR begins to react, with possible formation of new hydration products in cementitious pastes due to the addition of aluminum and sodium in the solution (Romano et al., 2019, 2018). The reaction rates of pastes with BR are the same and higher than the binary system with fly ash (70CP_30FA), demonstrating that the BR effect predominates at this time.

Figure 4c shows the time at which C-S-H and CH reach their maximum formation (peak), and again, it is seen that the SCMs accelerate this process. In the next session, the results of the XRD and TG analyses will be presented, comparing them with the calorimetry results, so that the formation of hydrated products and consumption of the cementitious phases can be confirmed.

Discussion

To highlight the changes caused by the SCMs in cement hydration, Figure 5 was formulated, as a different way of visualizing these data. It represents the difference between the heat flow of the reference paste and each blended system, where is possible to see the most important events shown in calorimetry, and therefore, when the SCMs most influence the hydration and heat released. The greater the difference (the higher values shown in the graphic), the greater the influence of SCMs on the corresponding event, and positive values indicate the intensification on chemical reaction caused by the SCMs.

In the first hours, which embraces setting time and acceleration period, FA has the most impact on hydration. Both SCMs have larger SSA than cement, which accelerates the reaction itself by filer effect (BERODIER, 2015), but the shape factor of FA is considerably higher than the other materials. This implies more irregularities on its surface, which facilitates the precipitation of hydrates to form clusters (Maciel; Romano; Pileggi, 2023), and, therefore, the reaction intensifies. As the reaction continues (from 6 hours onwards), Figure 5 shows that BR becomes the material that has the most influence on cement hydration, corresponding to a chemical effect added to the physical one. A sharp peak at 14 hours, corresponding to maximum AFt formation, and then, the positive values around 30 hours, corresponding to AFm phases formation, show how BR affects the aluminate’s reactions.

The formation of these hydrated products can be monitored by TG and XRD analyses, shown, respectively, in Figure 6 (derivative of TGA curve – DTG) and Figure 7. Figure 6a and Figure 7a show the results of the reference paste (100CP), Figure 6b and Figure 7b show the results of paste 70CP_30BR, Figure 6c and Figure 7c show the results of paste 70CP_30FA, and Figure 6d and Figure 7d show the results of paste 70CP_15BR_15FA.

From TGA data, it is possible to calculate the chemically combined water (Cw) (Figure 8a) and portlandite content (CH) (Figure 8b), according to de Weerdt et al. (2011) (Equations 1 and 2):

Where:

M is the molar mass; and

W_X is the percentage of mass loss at temperature x °C.

As seen in Figure 5, the profile curves of reference and 70CP_30FA are essentially the same, but with anticipation of reactions. As mentioned before, this suggest that the FA intensifies the reactions due to filer effect (Berodier, 2015), and, in fact, that are no chemical changes observed in the hydration period, as Figure 6c and Figure 7c demonstrate: the mass losses occur at the same temperatures and the diffractograms identified the same crystalline phases when compared to the reference (Figures 6a and 7a, respectively).

Thus, the discussion will be focused on the chemical influence of BR.

Figure 8b shows that the binary system with BR has more Portlandite content that the other pastes in the first ours of hydration, which encompasses the acceleration period. The increased alkalinity of the medium promoted by BR may explain the acceleration of hydrates formation (Lothenbach; Scrivener; Hooton, 2011). Besides that, Figure 6b shows mass losses at temperatures equivalent to the dehydroxylation of OH^- of some aluminates. Since these mass losses are increasing (indicated by the grey arrows in the Figure 6b), it is suggested that aluminate hydrates are being formed, such as calcium aluminosilicate hydrate (CASH), because of the absorption of the extra Al from BR in the C-S-H, AH (aluminate hydrate), CAH (calcium aluminate hydrate), and sodium aluminosilicate hydrate (NASH), due to the high amounts of Na in the residue (in this case, only the soluble sodium participate in the reaction) (Lothenbach; Scrivener; Hooton, 2011; Romano et al., 2019, 2016). Besides that, for the BR pastes, there is still gibbsite remaining from the anhydrous material in this temperature range. Figure 8b shows more Cw in compositions with BR during all hydration period, which might be because of these new hydration products. All of this implies a change in the reaction rate during the acceleration period quantified in Figure 4b. Such hydration products were not found in the diffractograms of Figure 7b and Figure 7d, probably due to their crystalline structure and overlap with other crystalline phases. Additional tests were not performed to confirm the existence of such aluminates.

As the reaction continues, it can be seen in Figure 7b that the intensity of sodalite peak begins to decrease at 10 hours, which may indicate that it is soluble and reacting to form NASH, whose intensity in TGA (Figure 6b) becomes more evident from 24 hours onwards.

Moreover, when the deceleration period takes place for 70CP_30BR, Figure 5a shows the maximum formation of monosulfoaluminate (AFm), with peak at 31 hours, which is 7 hours after the AFt peak (maximum formation of AFt at 14 hours). AFm phase formation occurs when, after sulfate depletion by the reaction with C₃A to maximum formation of AFt, C₃A continues its reaction with AFt to form this hydrate (Scrivener; Juilland; Monteiro, 2015; Zunino; Scrivener, 2020). At 24 hours, it is already possible to identify in the diffractogram of Figure 7b an AFm phase, the hemicarboaluminate (Hc), and at 48 hours, besides the presence of Hc, also another AFm phase is identified, the monocarboaluminate (Mc).

Analyzing the chemical changes in the ternary system to see the impacts of the combination of the SCMs on hydration, Figure 6d shows mass losses in the range of temperature of dehydroxylation of OH^- of aluminates, but with much less intensity than the binary system with BR (Figure 6b). In addition, the diffractogram of Fkigure 7d does not indicate accentuation of sodalite (2θ of 14º) during the period analyzed, as Figure 7b suggested. These facts may be due to the lower content of both SCMs in ternary systems compared to binary ones, but the possibility of FA interacting with BR, promoting lower solubility of the phases with sodium, is not ruled out. More studies must be conducted to determine whether there is chemical interaction between the SCMs, and, consequently, if the ternary blend promotes a low-energy intensity alternative for the safe use of BR, by reducing its potential of leaching dangerous elements.

Conclusions

This study investigated the influence of Supplementary Cementitious Materials (SCMs), specifically fly ash (FA) and bauxite residue (BR), on the hydration of Portland cement. Three systems were formulated: reference paste (only cement), two binary pastes with 70% cement and 30% SCM, and a ternary paste with 70% cement and 15% of each SCM. The primary focus was on understanding how these SCMs affect the hydration process, the formation of hydrated products, and the overall heat released during the reaction.

The use of SCMs accelerated the hydration reactions, evidenced by the shortened induction periods, and increased reaction rates during the acceleration period. Additionally, blended compositions with FA and BR released more heat compared to the reference, indicating that these materials actively participate in the hydration process and enhance reaction kinetics. FA and BR possess larger specific surface areas (SSA) than cement, which accelerates the reaction due to the filler effect. In the initial hours, encompassing the setting time and acceleration period, FA demonstrated the most substantial impact on hydration, because of its considerably higher shape factor, leading to more surface irregularities, facilitating the precipitation of hydrates into clusters and intensifying the reaction. As hydration progresses beyond six hours, BR exhibits the most pronounced influence on cement hydration, adding a chemical effect to the physical one.

During the acceleration period, the binary composition with BR formed more Portlandite (CH) than all pastes, resulting in a higher reaction rate compared to the reference. Additionally, the compositions with BR had more chemically combined water than all other pastes suggesting the formation of additional aluminate hydrates, such as calcium aluminosilicate hydrate (CASH) and sodium aluminosilicate hydrate (NASH), due to the extra aluminum and sodium provided by the residue. However, further tests are needed to confirm the existence of these aluminates.

While the study provided insights into the physicochemical interactions of SCMs with cement, the exact mechanisms, especially in ternary blends, remain partially understood. More detailed investigations are needed to fully elucidate these interactions. The potential for leaching of hazardous elements from BR-containing blends was not fully addressed, necessitating further studies to ensure the long-term environmental safety of using BR in cementitious systems.

In conclusion, the incorporation of FA and BR as SCMs in Portland cement presents promising advantages in enhancing the hydration process, promoting the formation of beneficial hydrates, and contributing to environmental sustainability. The use of these materials also allows the scalability of the cementitious product, as it offers logistical benefits by using by-products from the same region. These results highlight the importance of understanding the chemical interactions between materials for the safe application of BR on a large scale in civil construction, aiming at both environmental sustainability and technical performance.

Acknowledgments

Many thanks to Alcoa and Alcoa Foundation, as well as the late PhD Maria Alba Cincotto. We hope it did she proud.

References

Edited by

Publication Dates

History