The increased world population has generated significant energy consumption in all areas of industry, which has contributed to the global climate changes linked to gas emissions. The production of Portland cement (PC) results in the emission of approximately 700 kg of carbon dioxide (CO₂) into the atmosphere for every ton of PC produced (1). The cement industry accounts for approximately 7% of all global emissions, with a heat demand of 3500 to 5000 MJ per ton of clinker (2). This environmental impact makes the use of alkali-activated binders very interesting, and these are produced through the reactions of aluminosilicates such as slag with an alkaline activator. This represents a good option for reducing CO₂ emissions. Commercialization of the alkali-activated binders would require a detailed understanding of the chemical properties of these binders, such as the setting times, workabilities, hydration kinetics and durabilities (3). The reactivity and mechanical strength development of the slag depend on its chemical composition, fineness, curing temperature, and type and concentration of the activator (4-8). Moreover, slag, which is a byproduct of the iron industry, has variable chemical compositions, and these depend mainly on the raw material and the industrial process used (9). The hydration of slag in water is slow, and a glassy shell that forms around the slag grain prevents advanced hydration (10); therefore, it is necessary to use an alkaline activating agent such as sodium hydroxide or sodium silicate to promote the reaction. The use of an alkaline activator accelerates dissolution of the silica and alumina ions by breaking the Si-O and Al-O bonds and leads to precipitation of the hydrates (11-14). Bai et al. (15) show that the initial pH plays an important role in the dissolution of slag and the development of the initial hydration products, but advanced hydration depends primarily on the reaction of the Ca²⁺ dissolved from the slag with the activator anion.

Traditional alkaline activators such as NaOH and Na₂SiO₃ are normally used, but they have several limitations, including their corrosive natures, energy-intensive production methods, high costs and CO₂ emission levels, and complex production systems because they are not available in nature (16, 17). Na₂CO₃ has a negligible carbon footprint, is abundant in nature, is easy to handle, is potentially beneficial in terms of health and safety, has proven effective in the activation process and can be utilized as an alkaline activator (18). However, its use with AASs leads to low mechanical strengths at early stages and fast setting times. In general, the behavior of the AAS depends mainly on the nature and dosage of the activator, the curing conditions and the chemical composition of the slag. These parameters have prevented commercialization and universal formulation of these binders, which need to be optimized.

Alkali-activated slag materials prepared from sodium hydroxide and/or sodium silicate are also characterized by higher mechanical strengths, faster setting times and reduced susceptibilities to chemical attack than PC (19-21). However, shrinkage is the main problem limiting their use instead of PC; this higher shrinkage results from the nature of the gel formed and the higher mesopore content in the AAS, which tend to result in more significant shrinkage compared to PC (22, 23). In addition, there are few studies in the literature that evaluated the durabilities of sodium carbonate-activated slags exposed to acids and sulfates, and more research is needed.

Acid resistance is important because concrete structures can be exposed to aggressive environments during mineral processing, mining, and other industrial processes and experience deterioration. The action of acids on cement paste involves attacks on the hardened paste components; this leads to conversions of all calcium compounds in the unreacted residue and hydration products to form calcium salts of the attacking acid, which destroys the binding capacity of the hardened cement (19). Siad et al. (24) reported that portlandite (Ca(OH)₂) dissolved completely in a 5% sulfuric acid solution, while calcium silicate hydrate gel underwent progressive decalcification and the formation of other hydrates. A binder’s resistance to acid attack can be evaluated by measuring the weight loss, compressive strength, and corroded depth. Aliques-Granero et al. (25) reported increases in the masses and expansion degrees of AAS samples activated by Na₂SiO₃ with increasing dosage (3% and 5% wt.) of the sulfuric acid solution, and the degradation appeared as cracks on the cube edges and extended toward the centers of the specimens, but no erosion-like phenomena were observed. In another study (26), NaOH was used to activate slag, and more strength reduction was observed for treatment with sulfuric acid (5% H₂SO₄) than with hydrochloric acid (5% HCl). The high loss of strength in sulfuric acid was attributed to deterioration of the concrete due to the formation of expansive gypsum from the calcium ions in the slags and the sulfate ions in the acid solutions. In contrast, when the AAS mortars were exposed to hydrochloric acid, calcium chloride (CaCl₂) was formed in smaller amounts, the salt hardly leached out of the concrete through the pores, and minimal deterioration resulted. Pereira et al. (27) demonstrated that activation of the slag by Na₂SiO₃ and NaOH provided lower mass losses in 0.5 M hydrochloric acid than those seen for PC, and this lower mass loss of the alkali-activated binder was related to the absence of portlandite and the lower Ca/Si ratio relative to that in PC.

The degradation processes of cementitious materials in the presence of sulfates can result in expansion, cracking and spalling. The degradation of cement by sulfate occurs through the formation of expansive products, including gypsum (CaSO₄·2H₂O) and ettringite (AFt, 3CaO·Al₂O₃·3CaSO₄·32H₂O). Portlandite reacts with sulfates to form gypsum, and ettringite forms from a monosulfate (AFm, 3CaO·Al₂O₃·CaSO₄·12H₂O) and sulfate ions. These products occupy larger volumes than the compounds they replace (28-30). Moreover, crystallization of these products in highly saturated pores or in mesopores generates damaging pressures, causing damage to the material (31). In the literature, the evaluation of sulfate resistance for AAS is based on the ordinary cement standard ASTM C1012 (32). AAS prepared with NaOH has higher stability in sodium sulfate (5% Na₂SO₄) than in magnesium sulfate (5% MgSO₄), and the latter leads to expansion due to the formation of gypsum (33). Several studies (27, 34) have shown that slag specimens activated with Na₂SiO₃ and NaOH showed lower reductions in mechanical strength and expansion than PC specimens in sulfate environments. The low compressive strengths of the PC specimens are due to the formation of expansive products such as gypsum or ettringite. In contrast to the alkali-activated binders, which did not form portlandite, reduced amounts of the expansive products resulted in limited material deterioration. Another study (35) indicated that an AAS paste showed low levels of ettringite formation in Na₂SO₄ solutions, especially when Na₂CO₃ was employed as the alkaline activator; this prevented ettringite formation and led to stronger resistance against sulfate attack. On the other hand, the attack of Na₂CO₃-activated slag paste by a MgSO₄ solution was more aggressive than that of a Na₂SO₄ solution, which led to greater degradation and loss of strength due to the formation of gypsum via the reaction between SO₄ ²⁻ and Ca²⁺ ions and transformation of the C-A-S-H gel into noncementitious and fibrous magnesium aluminosilicate-hydrate (M-A-S-H) gel (36).

The durabilities of AAS materials are mainly related to the hydration products formed, their sizes, and the degrees of porosity. These factors also depend on the curing conditions, nature, and dosage of the activator. In addition, few studies of slag activation by Na₂CO₃ have appeared in the literature, resulting in limited information on the behaviors of Na₂CO₃-activated slags exposed to chemical attack, such as their corrosion depths. This lack of information confirms the need for this work. Therefore, this study was designed aid in the global commercialization of AAS by determining the behavior of Na₂CO₃-activated Algerian slag mortar cured at 95% relative humidity (RH) and 23 °C; these were subjected to aggressive environments, such as sulfates, hydrochloric and sulfuric acids, and the results obtained were compared under the same exposure conditions for slags activated with Na₂SiO₃ and NaOH. For comparative purposes, a similar study was carried out with PC as the reference material. In addition, the effects of activator dosage and the natures and silicate moduli on the properties of fresh and aged samples were evaluated with the setting times, workabilities, mechanical properties and drying shrinkage of the AAS.

This section presents a scientific discussion of the properties of the AAS in its fresh and hardened states as a function of the nature and dosage of the activator and the silica modulus. These properties were mainly related to reactions of the slag with the components of the alkaline activator.

The high fluidity of the AAS was due to the alkaline activator, which had a plasticizing effect. For example, when sodium silicate was used as an activator, silicate ions were adsorbed onto the surfaces of the slag grains. They adsorbed on the surfaces of the particles, which increased the magnitude of the electric double layer repulsive forces and led to deflocculation of the particles and consequently to increases in the workability (40). The presence of silicate ions in the sodium silicate activator justified the higher fluidity of the AAS compared to those treated with the other activators (NaOH and Na₂CO₃). As for the setting time, the rapid setting of the AAS was attributed to faster dissolution of the slag and formation of more initial hydrates compared to the PC compounds. Particularly for sodium silicate, Palacios et al. (41) detected the formation of a highly polymerized product after only 40 min (sodium aluminosilicate gel and C-A-S-H). It was also mentioned that activation of the slag by NaOH required more time to form C-A-S-H than activation of the slag by Na₂SiO₃. There were initially no silicate species in solution, and the concentrations of silicate and aluminum species, as well as that of calcium ions, slowly increased as the slag dissolved, which explained the faster setting time of the latter.

On the one hand, an increase in the modulus of the silicate led to an increase in the number of silicate ions adsorbed on the particle surface and an increase in the magnitude of the repulsive force, which explains the improvements in workability seen with increasing silicate modulus. In addition, the increased setting time seen with a higher silicate modulus may have been related to the decreased pH of the alkaline solution, which slowed the formation of hydrates and consequently the setting phenomenon. The silicate species were more highly polymerized and reacted slowly with the calcium released from the slag, and the time needed to induce gel formation was elongated (42).

On the other hand, increases in the dosages of the activators containing sodium hydroxide and sodium silicate accelerated dissolution of the slag by breaking the Si-O and Al-O bonds and forming hydration products (C-(A)-S-H) that enhanced the binding properties of the material, which led to reductions of the workabilities and rapid hardening of the specimens with short setting times (43, 44). When Na₂CO₃ was used to activate the slag, the same effect was seen for increased Na₂O dosages, leading to better workability and a reduction in the setting time. Bernal et al. (45) observed that at an early age, the reactions between carbonate (CO₃ ^2-) ions from the Na₂CO₃ with calcium (Ca²⁺) ions in the slag occurred to form calcite and gaylussite. Moreover, zeolite was also formed from the aluminate and silicate species in the slag due to the sodium from the Na₂CO₃. Subsequently, once the CO₃ ²⁻ was exhausted, a C-A-S-H gel was formed via a reaction involving calcium, aluminum and silicate dissolved from the slag. Increasing this activator dosage accelerated dissolution of the slag, consumption of the carbonate ions and formation of these hydration products. The diffractograms and the thermogravimetric analyses corroborated the precipitation of these products, gaylussite and carbonates. The presence of gaylussite in the hydration products from Na₂CO₃-activated slags has also been reported in the literature (38, 46, 47).

The main factors affecting the mechanical strength are the hydration products formed during hydration of the binder and the porosity of the material. The different mechanical strengths of the AASs were related to the nature of the hydrates formed with each activator. Taking into account the XRD and TG data for the samples, the main reaction product for all of the samples was the C-A-S-H gel, but the secondary reaction products differed according to the type of activator used. Strantligite and hydrotalcite were detected when sodium hydroxide was employed as the activator, while gaylussite and another carbonate were formed when the slag was activated with sodium carbonate or a mixture of sodium silicate and sodium hydroxide. The low mechanical strength seen at an early age for the slag activated by sodium carbonate was due to the low alkalinity of the activator solution, which delayed the dissolution of slag and hydrate precipitation (48), which enabled the generation of gaylussite and calcite. These phases did not give the high degree of cohesion necessary for the development of high early mechanical strength. At later ages, the increased strengths were due to formation of the C-A-S-H gel, which led to an increased strength (45).

The increased compressive strengths seen with higher silicate moduli were related to the incorporation of silicate anions into the C-A-S-H gel structure, resulting in a decrease in the Ca/Si ratio and thus an improvement in the binding capacity of C-A-S-H (21); the M6 sample exhibited the highest mechanical strength. Additionally, the structures of the samples were densified with higher Ms, leading to lower porosities (21). In addition, over time, more silicate anions were dissolved in the medium, and they were incorporated into the gel structure, refining the matrix and improving the compressive strengths of all samples.

Greater mechanical strength development was observed for increasing activator dosages. This may have been due to the increased alkaline concentrations in the activation solutions. This enabled the dissolution of more calcium and silicate ions in the slag grains and generated more gel, which was responsible for the binding properties; consequently, a higher strength of the system was achieved.

The unremarkable variations in flexural strength with increases in the Na₂O dosage and silicate modulus of the sodium silicate activator may have been related to greater sensitivity of the flexural strength to cracking compared to the compressive strength. In a previous study (21), SEM images showed significant formation of microcracks in the matrix and matrix-aggregate interfaces in samples with high Ms values.

The drying shrinkage resulted from the decreased volume of the material caused by chemical reactions and the exchange of moisture with the environment (23). This occurred through two mechanisms: disjunction pressure, mainly when the material had a very fine porosity, and capillary suction (49). The higher shrinkage of the alkali-activated slag compared to the PC was related to the higher amount of mesopores (with pore diameters between 2.5 and 50 nm) of the AAS, which resulted in higher capillary stresses leading to higher shrinkage (50). In addition, the formation of a calcium silicate gel rich in aluminum during hydration of the sodium silicate-activated slag led to a drying shrinkage higher than that of the PC. This gel had a high water content, which may dry out with water loss (51).

As for the influence of the silicate modulus and the Na₂O dosages for the samples activated with sodium hydroxide or a mixture of sodium hydroxide and sodium silicate, a high Ms and dosage precipitated more C-A-S-H gel, which may explain the increased shrinkage of the specimens. Li et al. (52) found that sodium silicate-activated slag mortar exhibited more drying shrinkage than sodium hydroxide activation. The high shrinkage of the Na₂SiO₃-activated slag mortars was related to the percentage of mesopores. These mortars had higher percentages of mesopores than the slag mortars activated with NaOH, which implied an increase in the water evaporation rate (23). However, when sodium carbonate was employed as the activator, the increased Na₂O dosage had a negligible effect on the shrinkage, which may have been related to the increased reaction rate, as has been explained previously; this resulted in an increase in the amount of chemically bound water reduced the rate of water evaporation (53).

The durabilities of the AAS mortars exposed to sulfates and acid depended on the nature of the alkaline activator employed and the reaction products formed. For sulfate attack, the expansion of the PC mortar was due to the formation of expansive products (gypsum and ettringite), which exerted great pressure on the pore walls. These expansive products were formed by reactions of the portlandite and AFm phases with sulfate ions (30, 54). The opposite behavior was observed for the AAS mortars, and the low expansion rates of these specimens may have been due to the reaction products formed. The absence of portlandite during hydration of the AAS mortar, independent of the activator (sodium hydroxide, a mixture of sodium hydroxide and sodium silicate, and sodium carbonate) used, resulted in the formation of few expansive products in the sulfate environment and consequently low expansion, as shown by the XRD and TG/DTG analyses. In addition, the AAS mortars exhibited low permeabilities, which enabled less penetration of the aggressive ions compared to the PC mortar (55, 56). The alkaline activator employed did not seem to exert any effect on the durabilities of the AAS mortars during sulfate attack; the damage levels were similar for the three samples.

The AAS mortars exhibited higher resistance than the PC mortar when they were exposed to HCl and H₂SO₄, and the AAS mortars exhibited lower mass losses and larger nondamaged cores than the PC mortar. This high resistance was due to the lower calcium content in the slag, the absence of portlandite indicated by the diffractograms and the TG analysis, and the formation of a C-A-S-H gel with a low Ca/Si ratio. This precluded the precipitation of salts, resulted in low initial permeability and high alkalinity of the pore solution, and provided high stability to an acidic medium (20). The anion of the acid exerted a great influence on the durabilities of the samples because it reacted with the calcium ions of the slag to form a salt of the attacking acid. Greater mass losses occurred for the mortars exposed to H₂SO₄ relative to those exposed to HCl. In the sulfuric acid solution, the reaction between sulfuric acid and the calcium ions formed ample amounts of expansive products such as gypsum, which caused great deterioration of the specimens. In addition, the cracks formed on the edges of these specimens facilitated penetration of the acid ions, which further deteriorated the material due to leaching of the calcium ions (26). The formation of expansive products also resulted in increased cross-sections, as with the M3 mortar. This effect of sulfuric acid on the Na₂SiO₃-activated slag was observed by Aliques-Granero et al. (25). For hydrochloric acid, the chloride ions reacted with the calcium ions and formed small quantities of calcium chloride. This salt was highly soluble and leached out of the concrete through its pores or by creating voids that deteriorated the specimen (26). In these cases, the small amounts of calcium chloride precipitated resulted in less leaching and less deterioration of the samples compared to sulfuric acid exposure.

The specimens activated with sodium hydroxide or a mixture of sodium hydroxide and sodium silicate behaved similarly in the hydrochloric acid, while activation with sodium carbonate provided different results. These samples showed better resistance to sulfuric acid and worse resistance to hydrochloric acid than the other two samples. Ye et al. (35) observed that the carbonate anions suppressed ettringite precipitation from the sodium carbonate-activated slag, and therefore, this sample developed a high resistance toward the aggressive environment. However, the use of the sodium carbonate activator resulted in higher total pore volumes and increased permeabilities of the ions (57), e.g., chloride ions in this case, and therefore enabled increased formation of the chloride salts and further deterioration of the material.

In light of these results, the use of sodium carbonate as the alkaline activator to provide a cementitious material from blast furnace slag is very attractive from the perspectives of environmental and economic factors and material durability. It has the advantages of low drying shrinkage and high resistance to sulfate attack and sulfuric acid, as indicated by the mass losses and deterioration levels of these specimens compared to those made with other activators (sodium silicate and sodium hydroxide), despite the low mechanical strengths at early ages.

This paper presents a study of Algerian slag activated by Na₂CO₃ and its behavior toward aggressive conditions involving sulfates, hydrochloric acid and sulfuric acid, and a comparison is provided with slags activated by Na₂SiO₃ and NaOH. In addition, the effects of dosage and activator type on the workabilities, setting times, mechanical strengths, drying shrinkage levels, and hydration products formed were also evaluated. The main conclusions derived from this investigation were: