AC Ambiente Construído Ambiente Construído 1415-8876 1678-8621 Associação Nacional de Tecnologia do Ambiente Construído - ANTAC Resumo A substituição do cimento Portland comum (CPC) por ligantes alternativos é uma solução técnica e economicamente viável para reduzir o impacto ambiental causado pela construção civil. Dentre esses ligantes, encontram-se aqueles pertencentes à família das Cerâmicas de Fosfato Quimicamente Ligadas (CFQL), na qual está incluído o cimento de fosfato de magnésio (CFM), originado por reações ácido-base. Porém, devido à formação de fases diferentes em relação ao CPC, um estudo aprofundado da durabilidade deste material é essencial para compreender a nova matriz e suas possibilidades de aplicação. Este artigo propõe adaptar o método usual de avaliação da reação álcali-sílica (RAS) válido para matrizes de cimento Portland para avaliar matrizes CFM. Para tanto, foram moldadas argamassas de referência com CPC e duas argamassas de CFM com diferentes relações molares M/F (magnésio/fosfato) com agregado reativo para avaliar a ocorrência de RAS. Observou-se que após 56 dias de ensaios, sob baixas temperaturas, as argamassas de CFM com maior relação molar M/F apresentaram baixa expansão, enquanto nas argamassas com CPC e CFM com menor razão molar M/F, os agregados apresentaram reações deletérias. A proposta de adaptação do método de ensaio RAS, considerando as particularidades das argamassas de CFM, foi satisfatória. Introduction Portland cement, the product most used in civil construction worldwide, generates several environmental impacts in its production. Those impacts include high energy and natural resource consumption, besides excessive release of gases that contribute to the greenhouse effect. This concern has sparked interest in using alternative binders as substitutes for Portland cement, as the acid-base reaction cements, including magnesium phosphate cement (MPC), which belongs to the Chemically Bonded Phosphate Ceramics (CBPC) family. This family is composed by ceramics formed at room temperature by the reaction between metallic cations and water-soluble phosphate sources (Ribeiro; De Paula; Morelli, 2019). MPC matrixes are formed from an acid-base reaction between magnesium oxide (MgO) and a soluble phosphate acid (usually ammonium or potassium phosphate), resulting in a magnesium phosphate salt with cementing characteristics. In the presence of water, the main reaction between magnesium oxide and ammonium phosphate results in the struvite phase, summarized in Equation 1 (Qian, 2020). MgO + NH 4 H 2 PO 4 + 5 H 2 O → NH 4 MgPO 4 ⋅ 6 H 2 O Eq. 1 Among the MPC components, MgO consists of the alkaline component of the acid-base reaction. This oxide requires calcination to reduce its high specific surface area, which makes it very reactive when used without heat treatment (Ribeiro; De Paula; Morelli, 2019), with a very fast setting, which makes its use in civil construction unfeasible. Thus, calcination reduces the material’s specific surface area by increasing the particles’ size and generating agglomeration, reducing the reactivity and solubility of the material (Wagh; Jeong, 2003). The setting time of this cement is directly linked to the texture of the MgO, which is influenced not only by calcination but also by the grinding process of this material (Qian, 2020). Besides the MgO calcination, retarders such as sodium tripolyphosphate (Na5P3O10 or STPP), boric acid (H3BO3) and borates (such as Na2B4O7·10H2O and Na2B4O7·5H2O), also slow down the reaction, due to the reduction of dissolution rate of MgO, contributing for a more controlled hydration reaction. Another factor that increases the reactivity of MgO is the presence of impurities, such as calcium oxide and magnesium carbonate, which dissociation is quicker than that of MgO (Qian, 2020). In relation to the acid component of MPC, the most often used is ADP (NH3H2PO4), which corresponds to the source of phosphate anions that react with magnesium oxide in the acid-base reaction (Ribeiro; De Paula; Morelli, 2019). Matrices containing ADP have high initial mechanical strength and better strength development (Tang et al., 2020; Fan; Chen, 2014), in addition to a relatively low reaction rate (Fan; Chen, 2014), with a relatively low cost (compared to other phosphate sources used in MPC matrices, such as potassium dihydrogen phosphate (KDP, KH2PO4)). However, when reacting with water, ADP releases ammonia gas (NH3), which smells unpleasant (Tang et al., 2020; Jianming; Luming; Jie, 2019; Fan; Chen, 2014; Wagh; Jeong, 2003). Due to the release of gaseous ammonia, MPC made from MgO and NH4H2PO4 faces environmental and hygienic limitations for indoor use. Consequently, it is mainly used in outdoor applications like rapid repair of roads and runways (Qian, 2020). The MPC hydration reaction generates different products. The struvite (NH4MgPO4·6H2O) is the main of them (Jun et al., 2018), which plays an essential role in the mechanical properties (Ribeiro; De Paula; Morelli, 2020) and in the adhesion when used in coatings (Jun et al., 2018). In addition to struvite, intermediate hydrated products such as dittmarite (NH4MgPO4·H2O), sterkorite [Na(NH4)HPO4·4H2O] and schertelite [(NH4)2MgH2(PO4)2·4H2O] can also be formed. Less frequently, hanaite [(NH4)2Mg3(HPO4)4·8H2O] and newberyite (MgHPO4·3H2O), among other phases, can also be found (Soudée; Péra, 2000). Struvite and dittmarite are chemically and structurally similar, tending to transform into each other without any microstructural damage to the cementitious matrix (Ribeiro; De Paula; Morelli, 2020). Dittmarite is formed as the main product if the setting is fast (without an efficient setting retarder), while struvite is the main product when the setting is slow (Walling; Provis, 2016). When thermally treated in water at ~60 ºC, struvite is decomposed, releasing only water molecules, forming a new phase called dittmarite, more thermally stable, decomposing only at 221 ºC. Differently, when the heating occurs in dry air, the struvite is decomposed, releasing ammonia (NH3) and water, forming Mg-phosphate hydrates, as Equation 2 (Sarkar, 1991). NH 4 MgPO 4 ⋅ 6 H 2 O → MgHPO 4 + NH 3 ↑ + 6 H 2 O ↑ Eq. 2 In addition to magnesium oxide and the phosphate source, MPC blends can have other materials that influence the composites’ characteristics, such as boric acid and sodium tripolyphosphate (STPP). Boric acid is a retarder added to the mixture to control the setting of the MPC and limit the temperature increase, given the high reactivity of this cement (Lahalle et al., 2016). Consequently, this retarder improves the conditions of workability of the mixture, allowing the reactions to occur effectively with the formation of phosphates (Qian, 2020). STPP is a dispersant that also acts as a retarder in MPC matrices, and its mechanism of retarding action is associated with the chelation of Mg+2 ions (it increases the pH of the mixture and absorbs heat during STPP dissolution, as well). However, the primary function of STPP is as a deflocculant in the matrices, improving the compaction of the mixes in the fresh state and, consequently, reducing their porosity in the hardened state (Ribeiro, 2006). The possibilities of application of MPC composites in civil construction are not consensus, and their use is currently limited to repair structures, mainly rapid repair of roads and runways, due to their properties (Qian, 2020). In this scenario, therefore, studies need to focus on durability parameters that allow the evaluation of the behavior of those matrices in adverse environmental conditions, predicting the service life of MPC composites. It is worth noting that different cementitious matrices may have distinct durability parameters; therefore, these parameters should be studied closely. Due to the common applications of MPC composites in the repair of roads and runways, contact with soil can be worrisome, mainly if this soil has high moisture and alkalis in its composition because this scenario is favorable for alkali-aggregate reaction (AAR). In addition, studies that evaluate the occurrence of AAR in concrete produced with MPC are necessary since the elements that make this matrix have significant amounts of alkalis, mainly magnesium (MgO) and sodium (from STPP), which may cause damage to MPC composites when exposed to a deleterious environment, with high humidity and high alkalis content. In ordinary cement Portland (OPC) matrices, the AAR occurs between the cement alkalis (generally sodium and potassium) present in the pore solution and the reactive minerals of the aggregate (Marinoni et al., 2012). There are two possible types of AAR: alkali-silica reaction (ASR) and alkali-carbonate reaction (ACR). Because of the higher incidence of ASR, it is often considered a synonym of AAR. As OPC and MPC matrices present different hydrated reaction products, it is expected that the ASR mechanism is not the same in those materials, so a deeper study is needed. According to Thomas (2011), ASR in OPC matrices starts with the reaction between the hydroxyl ions from the pore solutions and the reactive silica present in the aggregate, attacking and breaking the siloxane bonds (Si – O – Si) and causing one of the four bonds that silicon makes with oxygen to be occupied by the OH- ion (Equation 3). Next, silanol groups (Si – OH) are broken by OH- ions, originating SiO- ions on the surface of the aggregate (Equation 4). When alkali is present in excess, the amount of OH– in the alkaline solution increases, which in turn dissolves more silica, enhancing the ASR (Venkatachalam et al., 2022). Si 2 O + H 2 O → 2 Si ( OH ) Eq. 3 Si ( OH ) + OH − → SiO − + H 2 O Eq. 4 The final product of ASR is a hydraulic and amorphous alkaline silica gel, hygroscopic, which expands after reacting with water (Marinoni et al., 2012). This gel expansion generates internal stresses in the matrix that can exceed the tensile strength of concrete, causing progressive cracking and associated deterioration (Fanijo; Kolawole; Almakrab, 2021). The cracks have a mapped appearance, spreading across the entire surface of the structural element (Helene; Carvalho; Pacheco, 2017), facilitating the entry of other deleterious substances and compromising the performance of the structure (Marinoni et al., 2012). This paper aims to establish appropriate test parameters for magnesium phosphate cement matrices regarding the occurrence of the alkali-aggregate reaction. Additionally, it evaluates the behavior of the matrixes with different M/P molar ratios in the presence of reactive aggregates. Materials and methods Materials The cement used to produce the reference mortar was the Brazilian Portland cement containing filler (CP II-F, equivalent to the cement ASTM type IL) since this cement has no mitigating effect on AAR. This cement meets the specifications of C 595 (ASTM, 2008). A reactive aggregate with a maximum aggregate size of 4.8 mm and water was used in the reference mortar. To produce MPC mortars, the raw materials were dead-burnt MgO, ADP, H3BO3, STPP, reactive aggregate (Dmax = 4.8 mm), and water. Tables 1 and 2 present the physical characteristics and chemical composition, respectively, while Figures 1 and 2 present the particle size distribution and the X-ray diffractograms of the raw materials. The particle size distribution of the raw materials was obtained through an Anton Paar Laser Diffraction Particle Size Analyzer, model PSA 1190, using ethanol and acetone as MgO and ADP dispersants, respectively, and dry mode for boric acid and STPP. Chemical analyses were performed using a Bruker S1 Titan 800 XRF Analyzer. For mineralogical analyses, a Bruker model D2 Phaser diffractometer was used, with a copper target tube (wavelength, λ, equal to 0.154060 nm), with a current of 10 mA and voltage of 30 KV. To identify the crystalline phases of the raw materials sieved with a #200 mesh size, the DIFFRAC plus-EVA software was applied, which has the Crystallography Open Database (COD) as a database. Table 1 and Figure 1 show the high fineness of MgO, close to the values obtained for OPC, reflecting in MgO high reactivity and low setting time. The ADP, the acid element of the acid-base reaction of MPC, also presents particles of high fineness, contributing to the high dispersion of this reagent in the mixture. Besides, a broader particle size distribution of MgO among the other raw materials can be noticed, benefiting the packing of the mixture. According to Chellappah and Aston (2012), this happens because when the size distribution is broader, the spaces between the larger particles are better filled by the smaller particles of various sizes. Table 1 Physical characteristics of the raw materials Property OPC MgO ADP H3BO3 STPP Sand Specific Gravity (g/cm3) 3.12 3.61 1.81 1.52 2.51 2.70 BET Specific SurfaceArea (m2/g) 1.64 0.99 0.12 0.06 0.32 - D50 (mm) 15.58 18.63 38.37 207.37 50.67 - Table 2 Chemical composition of the raw materials Material Oxide (%) B2O3 MgO SiO2 Fe2O3 CaO Al2O3 MnO TiO2 Na2O P2O5 SO3 K2O OPC - - 14.18 5.87 68.43 3.62 0.13 0.41 - 0.39 6.42 0.54 MgO - 91.28 5.98 1.37 0.62 0.62 0.09 0.03 - - - - ADP - - 15.62 0.16 0.12 3.95 0.02 0.01 - 78.58 1.38 0.15 H3BO3 56.90 - - - - - - - - - 0.03 - STPP - - 0.03 - - - - - 42.00 57.60 0.23 - Figure 1 Particle size distribution of the materials used for the mortar production Figure 2 presents the diffractograms of the raw materials applied in the production of MPC matrices. Regarding the diffractograms of ADP, MgO, and boric acid (Figure 2), the predominance of mineralogical phases biphosphammite (NH4H2PO4), periclase (MgO), and sassolite (H3BO3), respectively, was detected, indicating a high purity of these materials. STPP, as observed by Checchinato et al. (2002), is essentially composed of an anhydrous phase in form II, in addition to traces of NaOH·H2O and Na3PO4 as impurities before the hydration. Considering the potentially reactive aggregate, this sample presented four crystalline phases: albite (NaAlSi3O8), quartz (Si2O), annite [KFe2+3AlSi3O10(OH)2], and actinolite [Ca2(Mg4.5–2.5Fe2+0.5–2.5)Si8O22(OH)2]. Among those minerals found, the silica present in albite can be reactive, according to Zheng et al. (2023). Figure 2 XRD diffractogram of the materials used in the MPC matrix: (a) MgO, (b) ADP, (c) boric acid, (d) STPP (e) reactive aggregate Methods Preparation of aggregate and mortar proportions Potentially reactive coarse aggregate was crushed in a ball mill and sieved to separate the fractions according to C1260 (ASTM, 2022), thus increasing its reactivity due to the increase of its specific surface area. The OPC (M0) and MPC (M1 and M2) mortars were proportioned by mass as 1:2.25:0.47 (cement: sand: water). Concerning the MPC mortars, the consumption of boric acid, STPP, sand, and water was calculated in relation to MgO mass, since the other binder component (ADP) is soluble in water. The MPC matrices produced had two different molar ratios, M/P = 7 (M1) and M/P = 3 (M2), aiming to evaluate the influence of those proportions in ASR. The consumption of mortar materials, in kg/m3, is presented in Table 3. Table 3 Consumption of materials (kg/m3) of mortars Mix OPC MgO ADP Boric Acid STPP Sand Water M0 440.0 - - - - 990.0 206.8 M1 - 523.0 224.0 52.3 52.3 1177.0 247.0 M2 - 470.0 385.0 47.0 47.0 1058.0 221.0 Mortar-bars casting Mortar bars with dimensions of 25mm x 25mm x 285mm, traditionally used in the accelerated test method for potential alkali reactivity of aggregates, were cast, three for each mixture, without any procedure of compactness due to the fluidity of the mixture. However, due to the high adhesion between the MPC matrix and the metallic surfaces, it is recommended to use plastic or wooden molds for casting the specimens (Ma et al., 2014; Ribeiro, 2006), avoiding problems during demolding (Figure 3). In this research, wooden molds with a thin waterproofing layer were used for mortar-bars casting. Mixtures M1 and M2 presented similar visual consistency, being fluid and easy to cast. Figure 3 Wooden mold for MPC-mortars casting Test for potential alkali reactivity of aggregates Immediately after casting the mortar bars, the molds containing OPC mortars were placed in the humid chamber, with a controlled temperature (23 ± 2)ºC and humidity higher than 95%. In contrast, the molds containing MPC mortars were covered and placed in a room with a controlled temperature (23º ± 2)ºC and humidity (40 ± 5)% for (24 ± 2) hours. Then, OPC and MPC mortar bars were demolded and placed inside a container for water curing at (80 ± 2)ºC, and kept in this condition for additional 24 hours. After this period, following the guidelines of C1260 (ASTM, 2022), the bars were removed from the curing container, and their initial lengths were measured using an appropriate device with a dial indicator (Figure 4a) in an acclimatized room at (23 ± 2)ºC. After the initial reading, the bars were submerged in a container with aqueous 1N NaOH solution at (80 ± 2)°C, and their lengths were measured daily for 30 days after casting (Figure 4b). Figure 4 Alkali-silica reaction test, composed of (a) a measuring device with a dial indicator and (b) a metallic container with NaOH 1N solution at (40 ± 2)°C for submerging the mortar bars The expansion of the bars was determined from the percentage increase in their length throughout the test. At the end of the procedure, the results were analyzed based on the expansion ranges according to C1260 (ASTM, 2022). Kuperman et al. (2005) state that although this worldwide known method has a reduced cost, it has some drawbacks, such as low reliable results. The lack of reliability in the results is due to: the failure to detect aggregates that show slow reactivity; excess leaching of alkalis present in the mortar of the bars; and the granulometric composition of crushed aggregates not reproducing reality. Mineralogical composition analysis After the test, fragments of the MPC mortar bars were extracted, pulverized, and sieved with a #200 mesh size for mineralogical analysis. The same equipment and parameters presented in the characterization of the materials were used. The crystalline phases were quantified using the Rietveld method, with the aid of the TOPAS software and the corresponding Crystallographic Information Files (CIF) of each phase found. TOPAS also provided statistical indicators that allowed evaluation of the quality of the refinement results: RWP (Weighted Profile R-factor) and REXP (Expected R-factor). Thus, the RWP/REXP ratio provides the GOF (Goodness of Fit), which is ideal when closer to the unit value. Results and discussions Adaptation of the ASR Test Method The test method originally developed to evaluate AAR in Portland cement matrices, proposed by C1260 (ASTM, 2022), when applied in MPC matrices, it promoted premature disintegration of the mortar bars, making it impossible to perform the test. According to Sarkar (1991), when thermally treated in water at ~60 ºC, struvite decomposes, releasing only water molecules and forming a new, more thermally stable phase called dittmarite, which decomposes only at 221 ºC. However, the decomposition of struvite to dittmarite substantially degrades the performance of MPC matrices (Qian, 2020). The decomposition of struvite could be verified by analyzing the diffractograms of the MPC mortar bars submerged in a 1N NaOH solution at (80 ± 2)°C after 28 days, as the peaks of struvite were not detected (Figure 5). Figure 5 Diffractogram of MPC mortar after immersion in water and 1N NaOH aqueous solution at (80±2) °C for 28 days Therefore, the AAR test method presented in C1260 (ASTM, 2022) was unsuitable for MPC matrices, as its main crystalline phases (struvite and dittmarite) dissolved. The dissolution was proved by the absence of struvite- and dittmarite-related peaks in the diffractogram. It is known that if the reaction between MgO and ADP does not occur effectively, it will result in a residual phosphate that dissolves in water, making the pore solution of the MPC matrix slightly acidic. This acidic environment promotes the slow dissolution of the main hydrated products, increasing porosity within the MPC matrix and, consequently, decreasing its mechanical resistance (Qian, 2020). Therefore, based on the preliminary results obtained, and due to the physical and chemical characteristics of MPC matrices, it was necessary to make adaptations to the method proposed by C1260 (ASTM, 2022), such as: use wooden molds to facilitate the demolding of MPC mortar bars, which must occur 24 hours after casting; apply air cure to the mortar bars at a controlled temperature of (23 ± 2)°C and humidity (40 ± 5)% for 28 days, to avoid the dissolution of unreacted raw materials and the decomposition of the struvite. After this period, the initial length of the bars (zero) must be read. Then, the bars can be immersed in a 1N NaOH solution; and reduce the test temperature from (80 ± 2)°C to (40 ± 2)°C, increasing the period of bar length readings to at least 56 days (counted after the initial 28 days of curing) and carrying it out every two days. Potential alkali reactivity of aggregates Figure 6 shows the expansion curves of OPC (M0) and MPC (M1 and M2) regarding the alkali-silica reaction. The adapted parameters proposed previously were applied only to the new bars cast with MPC mortars. Results were evaluated according to the limits determined by C1260 (ASTM, 2022). Figure 6 Expansion curves of OPC (M0) and MPC (M1 and M2) mortars to ASR assessment As can be seen in Figure 6, the aggregate used confirmed its high reactivity when introduced into the Portland cement matrix, reaching almost 0.30% expansion after 28 days. This is also confirmed by the presence of the mineral albite in the aggregate’s diffractogram (Figure 2e). When added to the MPC matrix, the aggregate showed two distinct behaviors: in mortar M1, the aggregate was harmless in terms of alkali-silica reactivity, with total expansion less than 0.10%; while in M2 mortar bars, the maximum expansion was similar to that observed in OPC mortar-bars (0.30%), however, with slower development (56 days), since the temperature was reduced for MPC matrices (40 ± 2)°C. This expansive behavior in M2 mortars may be associated with their greater porosity, which has a higher ADP content and whose surface porosity is shown in Figure 7. The higher the ADP content, the greater the release of NH3 from the mixture when reacting with water, increasing the porosity of the matrix (TANG et al., 2020; Jianming; Luming; Jie, 2019; Ribeiro De Paula; Morelli, 2019; Fan; Chen, 2014). According to Fanijo, Kolawole and Almakrab (2021), typically, highly porous aggregate and concrete allow easy entry of moisture from the environment to activate AAR, increasing the susceptibility and rate of this reaction. In this present research, the aggregate’s porosity was not evaluated, but the observed matrix’s porosity validated the citation of Fanijo, Kolawole, and Almakrab (2021). Figure 7 MPC-based mortar bars (a) M1, and (b) M2 On the other hand, mortar M1, with a higher MgO content, showed excellent performance in terms of ASR. This behavior is also linked to the greater compactness of this mixture, benefiting from the better particle size distribution of MgO (Figure 1) when compared to that of ADP, resulting in a less porous matrix, thus making it difficult for moisture to enter in the matrix, taking the alkalis inside, which is essential for the AAR occurrence. Figure 8 shows the diffractograms with the mineralogical composition of mortars M1 and M2 after the ASR test. Although the M/P molar ratios of M1 and M2 are different, the phases formed were identical, with the struvite, periclase, albite, quartz, and brucite phases being identified. Figure 8 XRD diffractograms of MPC-based mortars after 56 test days: (a) M1 and (b) M2 The adaptations in the test method of C1260 (ASTM, 2022) proposed for evaluating the MPC-based matrix were efficient since the struvite and brucite crystalline phases were present until the end of the test without undergoing decomposition. Through the quantitative analysis of the crystalline phases in the diffractograms of samples M1 and M2, a higher content of periclase was found in sample M1 (71.99%) compared to sample M2 (67.15%) due to the higher M/P molar ratio of sample M1. Regarding the struvite phase, comparing specimens of M1 and M2, a reduction in struvite content was observed (19.47% versus 3.86%); it is related to the higher M/P molar ratio of M1 that contributes to the formation of this crystalline phase. The mineral albite, composed of reactive silica, comes from the aggregate and presented a higher content in M2 than in M1 (5.14% versus 22.16%). In addition to these phases, 0.02% brucite and 3.39% quartz were identified in matrix M1, and 0.02% and 6.83%, respectively, in matrix M2. The amorphous content measured in the XRD quantification was equal to 51.6% (sample M1) and 60.5% (sample M2), which may have significantly influenced the reactivity (AAR) of the mortars. MPC specimens with a higher amorphous content (M2) showed greater reactivity. This increase in the amorphous content can also be caused by an expansive gel formation. When evaluating the quality of the refinement results in the quantitative analysis of the diffractograms, an RWP of 7.76 and GOF of 3.47 for matrix M1 and an RWP of 8.44 and GOF of 3.79 for matrix M2 were observed, indicating good reliability of this analysis. The presence of albite in M2, three times higher than in M1, must be highlighted, since in this sample was registered the highest expansion. According to Zheng et al. (2023), the SiO2 of albite in contact with NaOH is subject to an ASR, generating internal micro-expansions for gel formation. As the albite content was higher in M2, the internal tension was also higher, thus generating cracks in these mortar bars. More studies must be performed at a microscopic scale to better explain the gel formation, including the mechanism of formation and chemical composition. Figure 9 shows one of the M2 mortar bars at the end of the test, along with a magnification-contrasted image of its central region, showing the formation of cracks similar to those typical of alkali-silica reactions. Such damages were not identified in the M1 mortar bars. Figure 9 Cracking is evidenced in the M2 mortar bar Conclusions From the results presented regarding the alkali-silica reaction in magnesium phosphate cement matrices, it is concluded that: the method parameters proposed to evaluate ASR in MPC matrices proved to be suitable. MPC mortar specimens must be air-cured at a controlled temperature of (23 ± 2)°C for 28 days before immersion in a 1N NaOH solution for 56 days; the M/P molar ratio influences the alkali-aggregate reactivity of the MPC matrix in the presence of reactive aggregates. A higher M/P molar ratio presented a better performance in relation to ASR. However, further studies testing other M/P molar ratios could help to have more conclusive results; MPC mortars with a lower M/P molar ratio and higher amorphous content (M2) showed greater alkali-aggregate reactivity. This behavior can be attributed to the more significant formation of expansive products that generated cracks in the mortar bars with a lower M/P molar ratio; the mortar with the highest M/P molar ratio performed better in the presence of reactive aggregate, avoiding the harmful consequences of ASR; and more studies are needed to explain better the mechanism of gel formation and its chemical composition. References AMERICAN SOCIETY FOR TESTING AND MATERIALS. C1260: standard test method for potential alkali reactivity of aggregates (Mortar-Bar Method). Pensilvânia, 2022. AMERICAN SOCIETY FOR TESTING AND MATERIALS C1260: standard test method for potential alkali reactivity of aggregates (Mortar-Bar Method) Pensilvânia 2022 AMERICAN SOCIETY FOR TESTING AND MATERIALS. C595: standard specification for blended hydraulic cements. Pensilvânia, 2008. AMERICAN SOCIETY FOR TESTING AND MATERIALS C595: standard specification for blended hydraulic cements Pensilvânia 2008 CHECCHINATO, F. et al. A hidratação do tripolifosfato de sódio vista como uma alternativa para otimizar as ações de estabilização de suspensões cerâmicas. 2002. In: CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 15., Natal, 2002. Anais [...] Natal, 2002. CHECCHINATO F A hidratação do tripolifosfato de sódio vista como uma alternativa para otimizar as ações de estabilização de suspensões cerâmicas 2002 CONGRESSO BRASILEIRO DE ENGENHARIA E CIÊNCIA DOS MATERIAIS, 15 Natal Anais Natal 2002 CHELLAPPAH, K.; ASTON, M. A new outlook on the ideal packing theory for bridging solids. Materials Science, v. 12, p. SPE-151636-MS, 2012. CHELLAPPAH K. ASTON M. A new outlook on the ideal packing theory for bridging solids Materials Science 12 SPE-151636-MS SPE-151636-MS 2012 FAN, S.; CHEN, B. Experimental study of phosphate salts influencing properties of magnesium phosphate cement. Construction and Building Materials, v. 65, p. 480–486, Aug. 2014. FAN S CHEN B Experimental study of phosphate salts influencing properties of magnesium phosphate cement Construction and Building Materials 65 480 486 08 2014 FANIJO, E. O.; KOLAWOLE, J. T.; ALMAKRAB, A. Alkali-silica reaction (ASR) in concrete structures: mechanisms, effects and evaluation test methods adopted in the United States. Case Studies in Construction Materials, v. 15, p. e00563, 2021. FANIJO E. O KOLAWOLE J. T ALMAKRAB A Alkali-silica reaction (ASR) in concrete structures: mechanisms, effects and evaluation test methods adopted in the United States Case Studies in Construction Materials 15 e00563 e00563 2021 HELENE, P.; CARVALHO, M.; PACHECO, J. Engineering field tests for alkali-aggregate reaction. Structural Concrete, v. 18, n. 2, p. 349–355, 2017. HELENE P CARVALHO M PACHECO J Engineering field tests for alkali-aggregate reaction Structural Concrete 18 2 349 355 2017 JIANMING, Y.; LUMING, W.; JIE, Z. Experimental study on the deformation characteristics of magnesium potassium phosphate cement paste at early hydration ages. Cement and Concrete Composites, v. 103, p. 175-182, 2019. JIANMING Y LUMING W JIE Z Experimental study on the deformation characteristics of magnesium potassium phosphate cement paste at early hydration ages Cement and Concrete Composites 103 175 182 2019 JUN, L. et al. Microstructure Evolution of a Magnesium Phosphate Protective Layer on Concrete Structures in a Sulfate Environment. Coatings, v. 8, p. 1–10, 2018. JUN L Microstructure Evolution of a Magnesium Phosphate Protective Layer on Concrete Structures in a Sulfate Environment Coatings 8 1 10 2018 KUPERMAN, S. C. et al. A CESP e a detecção da reação álcali-agregado. In: SEMINÁRIO NACIONAL DE GRANDES BARRAGENS, 26, Goiânia, 2005. Anais [...] Rio de Janeiro: CBDB, 2005. KUPERMAN S. C A CESP e a detecção da reação álcali-agregado SEMINÁRIO NACIONAL DE GRANDES BARRAGENS, 26 Goiânia 2005 Anais Rio de Janeiro CBDB 2005 LAHALLE, H. et al. Investigation of magnesium phosphate cement hydration in diluted suspension and its retardation by boric acid. Cement and Concrete Research, v. 87, p. 77–86, Sep. 2016. LAHALLE H Investigation of magnesium phosphate cement hydration in diluted suspension and its retardation by boric acid Cement and Concrete Research 87 77 86 09 2016 MA, H. et al. Effects of water content, magnesia-to-phosphate molar ratio and ageon pore structure, strength and permeability of magnesium potassium phosphate cement paste. Materials and Design, v. 64, p. 497-502, 2014. MA H. Effects of water content, magnesia-to-phosphate molar ratio and ageon pore structure, strength and permeability of magnesium potassium phosphate cement paste Materials and Design 64 497 502 2014 MARINONI, N. et al. Influence of aggregate mineralogy on alkali–silica reaction studied by X-ray powder diffraction and imaging techniques. Journal of Materials Science, v. 47, n. 6, p. 2845–2855, Mar. 2012. MARINONI N Influence of aggregate mineralogy on alkali–silica reaction studied by X-ray powder diffraction and imaging techniques Journal of Materials Science 47 6 2845 2855 03 2012 QIAN, J. Magnesium phosphate cement. In: SHAND, M. A. et al. (ed.). Magnesia cements: from formulation to application. New York: Elsevier, 2020. QIAN J. Magnesium phosphate cement SHAND M. A. Magnesia cements: from formulation to application New York Elsevier 2020 RIBEIRO, D. V. Influência da adição de pó de retífica em uma matriz de cimento de fosfato de magnésio. São Carlos, 2006. 128 f. Dissertação (Mestrado em Engenharia de Materiais) – Universidade Federal de São Carlos, São Carlos, 2006. RIBEIRO D. V Influência da adição de pó de retífica em uma matriz de cimento de fosfato de magnésio São Carlos 2006 128 f Dissertação (Mestrado em Engenharia de Materiais) Universidade Federal de São Carlos São Carlos RIBEIRO, D. V.; DE PAULA, G. R.; MORELLI, M. R. Effect of MgO/NH4H2PO4 ratio on the properties of magnesium phosphate cements. Materials Research, v. 23, n. 2, p. 1–10, 2020. RIBEIRO D. V. DE PAULA G. R. MORELLI M. R. Effect of MgO/NH4H2PO4 ratio on the properties of magnesium phosphate cements Materials Research 23 2 1 10 2020 RIBEIRO, D. V.; PAULA, G. R.; MORELLI, M. R. Effect of boric acid content on the properties of magnesium phosphate cement. Construction and Building Materials, v. 214, p. 557–564, 2019. RIBEIRO D. V. PAULA G. R. MORELLI M. R. Effect of boric acid content on the properties of magnesium phosphate cement Construction and Building Materials 214 557 564 2019 SARKAR, A. K. Hydration/dehydration characteristics of struvite and dittmarite pertaining to magnesium ammonium phosphate cement systems. Journal of Materials Science, v. 2, p. 2514–2518, 1991. SARKAR A. K Hydration/dehydration characteristics of struvite and dittmarite pertaining to magnesium ammonium phosphate cement systems Journal of Materials Science 2 2514 2518 1991 SOUDÉE, E.; PÉRA, J. Mechanism of setting reaction in magnesia-phosphate cements. Cement and Concrete Research, v. 30, n. 2, p. 315–321, Feb. 2000. SOUDÉE E PÉRA J Mechanism of setting reaction in magnesia-phosphate cements Cement and Concrete Research 30 2 315 321 02 2000 TANG, H. et al. The protective effect of magnesium phosphate cement on steel corrosion. Construction and Building Materials, v. 255, p. 119422, 2020. TANG H The protective effect of magnesium phosphate cement on steel corrosion Construction and Building Materials 255 119422 119422 2020 THOMAS, M. The effect of supplementary cementing materials on alkali-silica reaction: A review. Cement and Concrete Research, v. 41, n. 12, p. 1224–1231, 2011. THOMAS M. The effect of supplementary cementing materials on alkali-silica reaction: A review Cement and Concrete Research 41 12 1224 1231 2011 VENKATACHALAM, S. et al. The ASR mechanism in concrete and the influence of lithium in mitigating it: A critical review. Materials Today: Proceedings, v. 65, p. A1-A6, 2022. VENKATACHALAM S. The ASR mechanism in concrete and the influence of lithium in mitigating it: A critical review Materials Today: Proceedings 65 A1 A6 2022 WAGH, A. S.; JEONG, S. Y. Chemically bonded phosphate ceramics: I, a dissolution model of formation. Journal of the American Ceramic Society, vol. 86, no. 11, p. 1838–1844, Nov. 2003. WAGH A. S. JEONG S. Y. Chemically bonded phosphate ceramics: I, a dissolution model of formation Journal of the American Ceramic Society 86 11 1838 1844 11 2003 WALLING, S. A.; PROVIS, J. L. Magnesia-based cements: a journey of 150 years, and cements for the Future? Chemical Reviews, v. 116, n. 7, p. 4170–4204, Apr. 2016. WALLING S. A. PROVIS J. L. Magnesia-based cements: a journey of 150 years, and cements for the Future? Chemical Reviews 116 7 4170 4204 04 2016 ZHENG, X. et al. Effect of albite on shrinkage and carbonation resistance of alkali-activated slag. Construction and Building Materials, v. 409, p. 134141, 2023. ZHENG X Effect of albite on shrinkage and carbonation resistance of alkali-activated slag Construction and Building Materials 409 134141 134141 2023