Introduction

Cement is the primary component in the global civil construction sector. However, its production is associated with high energy consumption and significant greenhouse gas emissions. Given the cement industry’s inherent environmental commitment to mitigate these emissions, partially replacing clinker with supplementary cementitious materials (SCM) has emerged as a key strategy for short-term decarbonization (Scrivener; John; Gartner, 2018). Indeed, a variety of alternative materials have been adopted as SCM, including rice husk ash (Thiedeitz; Ostermaier; Kränkel, 2022), calcined clays (Schulze; Rickert, 2019), and natural pozzolans (Sierra et al., 2010), which have been used throughout history, but are currently being optimized (Mota dos Santos; Cordeiro, 2021). Nevertheless, SCM availability varies considerably between countries, underscoring the importance of seeking locally abundant materials with suitable properties for cement replacement. In short, it is imperative to seek regional resources for partial clinker replacement in cement production, thus promoting circular economy principles and fostering sustainable development in the sector (GCCA, 2021).

In this context, sugarcane bagasse ash stands out as a promising material due to its pozzolanic potential and widespread availability in different countries, especially in tropical zones. Sugarcane is one of the main crops in the world, grown in more than half of the countries, with Brazil, India, China, Thailand, and Pakistan being the largest producers (Abdalla et al., 2024). In 2022, global sugarcane production reached about 1.9 billion tons (FAO, 2023), which corresponds to a generation of around 12 million tons of bagasse ash, assuming full utilization of bagasse for cogeneration. The large quantity of this material is very attractive for application in the construction sector. Furthermore, using bagasse ash as SCM contributes to reducing environmental impacts by partially replacing cement or clinker and by valorizing an agro-industrial by-product (Klathae et al., 2021; Alvarenga; Cordeiro, 2024). Several studies have confirmed the high reactivity of bagasse ash and its benefits in different cementitious material properties (Bahurudeen; Santhanam, 2015; Andreão et al., 2020; Mali; Nanthagopalan, 2021). Indeed, the pozzolanic activity of bagasse ash can contribute to cement hydration (Barbosa; Cordeiro, 2021; De Siqueira; Cordeiro, 2022a) and strength gains of concrete, especially at later ages (Cordeiro et al., 2018). Additionally, different works have reported that concrete and mortar with bagasse ash presented better mechanical performance, lower water absorption, and higher resistance to chloride ions and gas penetration, enhancing the durability of cementitious systems (Bahurudeen et al., 2015; Rajasekar et al., 2018; Memon et al., 2022).

In addition to the use of SCM, extending the durability of cementitious products is critically important in promoting the sustainability of structures. In general, concretes and mortars can be exposed to a variety of environments and corrosive agents, such as sulfated soils, acid precipitation, and sewage systems. Specifically, acidic environments are highly corrosive due to the high alkalinity of the cementitious matrix (Glasser; Marchand; Samson, 2008). When exposed to solutions with low mineral content or when the pH of the solution decreases, the equilibrium of the cement matrix is disrupted due to the hydrolytic decomposition of the hydrated cement compounds (Duchesne; Bertron, 2013). Therefore, the physical properties and pozzolanic activity of SCM enhance durability by reducing cement consumption and refining the pore structure (Juenger; Snellings; Bernal, 2019). Portlandite consumption by pozzolanic reaction can increase durability since this compound is the most vulnerable to acid attack in the cementitious matrix.

In the continuous pursuit of sustainable development within the construction sector, it is crucial to understand the effect of emerging SCMs on the different attributes of cementitious systems, particularly their durability against different aggressive agents. As such, this study aimed to evaluate the effect of sugarcane bagasse ash on the durability of cementitious mortars against sulfuric acid attack, comparing its performance with that of mortars containing limestone filler (used as a reference material) and ordinary Portland cement. Previous studies have reported that the use of bagasse ash in cement-based materials results in better resistance against sulfuric acid attacks in terms of mass loss and strength loss (Arif; Clark; Lake, 2016; De Siqueira; Cordeiro, 2022b; Memon et al., 2022). Accordingly, the present work intends to contribute to this discussion by evaluating the microstructure of the degraded mortars and, at the same time, proposing the use of bagasse ash in the production of pozzolanic blended cements.

Methodology

The sugarcane bagasse ash was collected from the dust reactor of an ethanol distillery in Rio de Janeiro state (RJ), Brazil. The ash exhibited significant contamination by quartz and organic compounds due to sand that adhered to the sugarcane during harvesting, and inefficient burning in the boilers, respectively. To reduce contamination, the ash was separated by densimetric fractionation (Andreão et al., 2020) and underwent autogenous re-burning in a rudimentary furnace, with a maximum temperature of approximately 640 °C and heating rate of approximately 6.5 °C/min (De Siqueira; Cordeiro, 2022a). Meanwhile, the limestone filler was supplied by a mining company in the state of Rio de Janeiro, Brazil, and contained about 94% CaCO₃. It should be noted that the limestone was used as a filler-control material. Moreover, limestone was chosen because it is widely used in the production of blended cements around the world, enabling a realistic evaluation of the cements produced in this study.

Both materials were dry ground in a ball mill with a capacity of 50 L, operating at 30 rpm in open circuit, according to the procedures described by De Siqueira and Cordeiro (2022a). The volumes of grinding media (steel balls with diameters between 20 and 38 mm) and feed were 20 L and 10 L, respectively. The objective of grinding was to obtain ultrafine particles to improve bagasse ash reactivity and, at the same time, contribute to the filler effect of both materials. Additionally, a similar particle size distribution was established for both materials to minimize differences in physical effects and ensure adequate assessment of the pozzolanic effects of the ash. In this case, grinding time was 2 h for ash (denominated SCBA) and 1 h for limestone (LF), which resulted in a D₅₀ of 4.2 μm for both materials.

Figure 1 shows the X-ray diffraction patterns of SCBA and LF, revealing the presence of quartz and calcite as the main minerals, respectively. An amorphous halo between 15 and 30° is indicative of the presence of amorphous silica in SCBA, as previously reported by De Siqueira and Cordeiro (2022a). Table 1 shows the chemical composition, obtained by X-ray spectroscopy (Shimadzu EDX-720), of SCBA and LF, and their main physical characteristics. It is noteworthy, in this case, the sum of SiO₂, Al₂O₃, and Fe₂O₃ equal to about 83% and the high specific surface area of SCBA, which resulted in a performance index (ABNT, 2014a) of 127%. SEM image in Figure 2 clearly reveals the original cellular-based structure (Cordeiro; Toledo Filho; De Almeida, 2011) of the SCBA used in this study, justifying its high BET specific surface area. These characteristics made it possible to classify the ash as a pozzolan according to NBR 12653 (ABNT, 2014b).

Three mortars (M-PC1, M-PC2, and M-PC3) were produced with each of the cements to evaluate the influence of the mineral admixtures on durability against sulfuric acid (H₂SO₄) attack. To that end, deionized water and standard sand (ABNT, 2015) were used in the mix design, at water-to-cement and sand-to-cement ratios of 0.48 and 3.0, respectively. The superplasticizer (modified carboxylic ether with a density of 1.12 g/cm³ and oven-dried residue of 28.9%) content was fixed at 0.012% of cement mass for all mixes. Mortar mixing was carried out according to the procedures established in NBR 7215 (ABNT, 2019).

Next, SCBA and LF were used to produce blended cements, in line with the requirements of NBR 16697 (ABNT, 2018). Portland clinker and gypsum were supplied by a cement factory in the state of Espírito Santo, Brazil. The materials were dry interground for 2 h, using the same parameters described earlier. Thus, the three cements evaluated were PC1, an ordinary Portland cement, PC2, a cement with 14% clinker replaced by SCBA (equivalent to the Brazilian cement CP II-Z), and PC3, a cement with 14% clinker replaced by LF (equivalent to the Brazilian cement CP II-F). The same clinker replacement content was used to ensure a proper comparison between the two cements, although the different nature of LF (filler) and SCBA (pozzolan) is known. Details about the production of blended cements can be obtained in De Siqueira and Cordeiro (2022a).

Dihydrate calcium sulfate (CaH₄O₆S, gypsum) content was kept constant at 5% to maintain the same formulation among the cements and prevent undersulfation (Avet; Scrivener, 2018). In this case, a previous calorimetry study confirmed the appropriate dosage of gypsum (De Siqueira; Cordeiro, 2022a). Table 2 presents the physical parameters and chemical composition of the cements in this study. All cements met the Brazilian standard requirements (ABNT, 2018). Also, the similar particle size distribution of the three cements (Figure 3) allowed the results to be properly compared with minimal interference from packing density and enabled a better assessment of the pozzolanic reactivity of SCBA. Further information on the production process and complete characterization of the materials used and the produced cements are described in De Siqueira and Cordeiro (2022a).

The packing density of all mixes was assessed using Betonlab Pro 3 software, based on the Compressible Packing Model (CPM). According to De Larrard (1999), the CPM enables the calculation of the virtual packing density (γ) of a granular mix, representing the maximum packing density achieved by sequentially placing the grains while maintaining their original shapes. The actual packing density (ϕ) can then be derived from this virtual value, taking into account the specific compaction method used, which is characterized by a scalar parameter denominated compaction index (K). Thus, within this framework, as K tends to infinity, ϕ converges towards γ, as shown in Equation (1). To apply the CPM, the packing parameters for sand and cements must be determined as described by De Larrard (1999). For sand, the vibration and compaction test (K = 9) was utilized. The experimental packing densities for cement samples in wet conditions with optimum superplasticizer content were evaluated using the water demand test (K = 6.7) also described by De Larrard (1999).

Considering n is the number of particle classes; y_i the volume fraction; β_i the virtual packing density of the i^th class; and γ⁽ⁱ⁾ the virtual packing density when i is the dominant class. The index K assumes a value of 4.5 when compaction occurs by simple pouring, 6.7 for water demand, and 9.0 when the placing process is vibration followed by compression with 10 kPa of pressure.

After molding, the specimens were cured in lime-saturated water for 28 days. Next, half of the specimens were kept in lime-saturated water, and the remainder were immersed in acidic solution with 1.5% H₂SO₄ for 56 days, following the procedures of Khan et al. (2019). The specimen-to-solution volume ratio was kept at 0.25 and solution pH was maintained at the desired level (0.5 to 1.5) by sporadic solution exchange. The tests were conducted after the exposure period, i.e., at 84 days of age, for both attacked and nonattacked mortars.

Compressive strength was determined by the failure of three cubic specimens (50 mm edge) of each cement mortar in a Shimadzu UHI-500kN universal testing machine, at a loading rate of 0.5 mm/min. Total water absorption was established by immersing cylindrical specimens (25 mm diameter and 50 mm height) in water, according to NBR 9778 (ABNT, 2009). Three specimens were used for each mix. For the capillary absorption test, conducted in accordance with ASTM C1585-2 (ASTM, 2020), cylindrical specimens (25 mm diameter and 50 mm height) were dried in an oven at 60 °C until constant mass. Three specimens from the same mix were placed on a 5 mm water film, with epoxy resin on the lateral surface and a plastic sheet on top to ensure absorption only through the base of the specimens, and mass was monitored periodically. The obtained results were statistically compared using one-way analysis of variance (ANOVA) and Duncan’s multiple-range test (p ≤ 0.05).

Immediately at the end of acid attack, samples were removed from the specimens for X-ray diffraction and SEM analyses to evaluate the degraded layer. For X-ray diffraction analyses, the degraded layer (easily identified visually) was scraped from the specimens using a stainless steel spatula. The material was then sieved through a 75-µm mesh and subsequently ground manually using a mortar and pestle. The analyses were conducted using a Rigaku Miniflex 600 diffractometer, operating at 40 kV and 15 mA (Cu K-alpha radiation). Data were collected at angles between 10 and 60°, with an angular step of 0.02° and a rotation speed of 5°/min. For SEM analyses, 5-mm thick slices were extracted from the cubic specimens and vacuum-impregnated with epoxy resin. Then, the impregnated samples were sanded at 125 and 6 µm with isopropyl alcohol, polished with 3 and 1 µm diamond suspensions, using an automatic polishing machine, and coated with a thin layer of conductive carbon. The SEM analyses were performed on an FEI Quanta 400 MLA scanning electron microscope, with a backscattered electron (BSE) detector and energy-dispersive spectroscopy (EDS), operating at 20 kV and a working distance of 11 mm. The objective of the test was to observe the layer degraded by acid attack and gypsum formation in the samples. The average thickness of the degraded layer adhered to the specimen surface was estimated after cutting the cubic specimens, which were photographed and analyzed using Image J software.

Results and discussion

Table 3 presents the packing density of each mortar mix, along with the experimental parameters obtained for the three cements and the sand used in CPM. The results of the actual packing showed minimal differences among the three systems studied. M-PC1 presented a packing density of 0.794, which was approximately the same as that observed for M-PC2 (0.805) and M-PC3 (0.798). In this case, the inclusion of 14% SCBA and LF in the cements caused marginal interference in the particle packing density. This slight variation was expected due to the predominant granular nature of the sand (sand-cement ratio of 3.0) and the low incorporation of SCM in the cements. Small changes in the packing density of mortars with the inclusion of bagasse ash were also reported by Alvarenga and Cordeiro (2024). Although the optimization of the granular skeleton was not the objective in this study, the results indicate that the packing of the three mortars was similar. This similarity is important for distinguishing the effects of SCBA reactivity. Further research on the maximum packing of SCM-cement-mix could help determine adequate contents of pozzolan or filler to obtain optimized cementitious systems.

Figure 4 presents the compressive strength results at 84 days for mortars cured in lime-saturated solution throughout this period (nonattacked specimens) and in lime-saturated solution for 28 days followed by exposure to sulfuric acid for the remaining time (attacked specimens). According to the ANOVA, there were significant differences in compressive strength among the mixes. Table 4 summarizes the obtained results, where different letters in the last column mean that there was statistical difference between the mortars at the 5% level of probability based on Duncan's multiple range test. Table 5 shows the ANOVA results for the compressive strength of attacked and nonattacked mortars. Assessment of the nonattacked mortars showed significantly superior performance for M-PC2 in relation to the other mixes. In this case, the properties of SCBA, especially its high specific surface area and pozzolanic activity (Table 1), contributed to better mechanical performance, reaching 62.1 MPa. Previous studies also reported the good behavior of bagasse ash on the mechanical properties of cement-based materials (Bahurudeen et al., 2015; Cordeiro et al., 2018; Mali; Nanthagopalan, 2021). On the other hand, LF did not improve the strength of M-PC3, whose average value (47.9 MPa) was significantly lower than that of M-PC1 and M-PC2. This is attributed to the dilution effect of adding a material with primarily physical effects (Millán-Corrales et al., 2020; Senhadji et al., 2014), which were not significant in the studied mortars (Table 3). Thus, since the packing of M-PC2 and M-PC3 was similar, the good performance of SCBA can be attributed to its high pozzolanic activity. Interestingly, 14% clinker replacement by SCBA resulted in an approximate 14% increase in compressive strength at 84 days when compared to the reference mortar.

Comparison of the attacked and nonattacked mortars demonstrated that the sulfuric acid attack was quite aggressive, resulting in significant compressive strength losses for all mixes, as expected. Figure 5 shows the aspect of mortars at 84 days. Indeed, sulfuric acid attack consumes portlandite from the cementitious matrix and forms gypsum, an expansive compound that causes degradation in the hardened mortar. Besides this degradation, the decalcification of C-S-H by acid exposure causes strength loss in cementitious systems (Bassuoni; Nehdi, 2007; Senhadji et al., 2014). Strength declined by 46, 59, and 58% for attacked mortars M-PC1, M-PC2, and M-PC3, respectively, compared to nonattacked mortars. It should be noted that compressive strength was calculated considering the original area of the specimen, simulating a service condition. Loss of compressive strength was lower for the reference mortar, and M-PC3 exhibited the lowest strength after the acid attack. The superior performance of M-PC2 cured in lime-saturated solution was not replicated after the acid attack, since compressive strength loss was more pronounced in this mortar after attack. Nevertheless, there was no statistically significant difference between the compressive strength of M-PC1 and M-PC2 after the attack. Also, M-PC2 showed significantly better performance than M-PC3 in terms of durability against sulfuric acid. Although limestone is widely used as SCM in the production of blended cements, its effect was detrimental to the compressive strength of the mortars, either before or after acid attack.

The water absorption results for attacked and nonattacked mortars are presented in Figure 6 and Table 6. Also, Table 7 indicates the one-way ANOVA data. For nonattacked mortars, M-PC2 exhibited less water absorption than the reference, and M-PC3 the highest among the mixes studied. In this case, the pozzolanic effect of SCBA allowed the gradual pore closing and a greater microstructure densification (Ganesan; Rajagopal; Thangavel, 2007; McCarthy; Dyer, 2019), in contrast to the predominantly physical effect of LF on M-PC3. The addition of fine limestone particles may have increased the number of interconnected pores of smaller diameter, which led to an increase in water absorption compared to M-PC1, as observed by Souza et al. (2020). The same behavior could be observed for M-PC2, but it was compensated for by the pozzolanic reactions of the SCBA. After exposure to acid, absorption increased by 18, 14, and 12% in M-PC1, M-PC2, and M-PC3, respectively, in relation to nonattacked mortars of the same age. The overall increase in absorption in attacked specimens is due to greater porosity caused by degradation of the cementitious matrix exposed to sulfuric acid. It is important to underscore that there were no statistically significant differences in total absorption between all attacked mortars. Therefore, absorption performance of the mortars was equivalent regardless of the mix, due to the significant degradation of the cementitious matrix.

Capillary absorption data (Figure 7) showed two distinct behaviors, with high rates for M-PC1 and M-PC3 and lower values for M-PC2 over the 7-day assessment period. In regard to initial absorption, recorded in the first 24 h and represented by dotted linear regression lines, the sorptivity of M-PC2 (3 × 10^-3 mm/s^0.5) was 3.0 and 3.7 times lower than that of M-PC1 and M-PC3, respectively. This disparity persisted in secondary absorption (represented by dashed lines), whereby the absorption rate of M-PC3 was twice that observed for M-PC2 between the first and seventh day of testing. The pozzolanic nature of SCBA was noteworthy but did not significantly impact compressive strength, which is evident in the results presented in Figure 4. Ganesan, Rajagopal, and Thangavel (2007) also noted reduced absorption in concretes with up to 15% bagasse ash replacement. In turn, Bahurudeen et al. (2015) observed the same behavior with up to 25% of cement replacement by bagasse ash. The correlation between sorptivity and total water absorption in acid-attacked mortars (Figure 8) revealed a good coefficient of determination (R² = 99%) when initial absorption was considered. In this case, the dashed lines represent linear regression. The same was not observed for the correlation between total absorption and sorptivity in secondary capillary absorption.

As previously discussed, gypsum formed and accumulated on the outer surface of the mortar as a layer of white precipitate, resulting from the reaction between sulfuric acid and portlandite. This gypsum was easily identified, and its formation was confirmed by X-ray diffraction analyses of the precipitate in all mortars. Figure 9 shows the X-ray diffraction pattern of the precipitate for the M-REF specimen (the other mixes presented the same result). In addition to gypsum, quartz was also identified, due to the detachment of sand following the dissolution of the mortar surface. Gypsum formation was also assessed by SEM images of the mortars through sulfur (S) mapping, aided by EDS, as shown in Figure 10. The outermost layer of all the mortars contained a higher gypsum concentration due to direct contact with the acidic solution. However, significant gypsum levels were also observed inside M-PC1 (Figure 10a), but not in M-PC2 (Figure 10b) and M-PC3 (Figure 10c). In this case, the greater microstructure densification caused by the SCM contributed to reducing acid penetration into the mortars, minimizing gypsum formation within the specimens. This effect was more significant for M-PC2 since SCBA exhibits pozzolanic activity, resulting in a thinner degraded layer than that observed in M-PC3. This result explains the lower water absorption in M-PC2.

Gypsum formation in the specimens, as occurred for M-PC1, did not result in densification of mortar microstructure, as reported by Schmidt et al. (2009), although less compressive strength loss after acid attack was observed for this mix. Thus, the microstructure of the reference mortar was compromised by the attack on the cementitious matrix, with clear evidence of gypsum formation in regions up to approximately 4 mm from specimen surface. The average thickness of the degraded layer for M-PC1 was 0.62 ± 0.01 mm, whereas degradation in M-PC2 and M-PC3 occurred primarily on the surface, also leading to considerable post-attack strength losses. The thicker gypsum layer in M-PC3 (1.67 ± 0.03 mm) may have contributed to increased absorption after the attack, while for M-PC2, the thinner layer (0.82 ± 0.02 mm) coupled with visible mortar degradation indicate surface delamination, which may have contributed to greater compressive strength loss. The analyses conducted in this study demonstrate that using SCBA reduced acid penetration into the cementitious matrix; however, this was not sufficient to improve the compressive strength of mortars after the 56-day sulfuric acid attack. Future research should evaluate the performance of mortars with SCBA-cement against prolonged acid attacks and investigate how gypsum formation in the reference mortar may affect its long-term performance.

Conclusions

Based on the findings of this study, it can be concluded that SCBA is a suitable SCM for the production of blended Portland cement. Indeed, mortars containing SCBA showed superior mechanical performance when compared to the other mixes before the attack. As the packing density values of all mortars were similar, the better performance of M-PC2 can be attributed to the pozzolanic properties of SCBA. On the other hand, LF did not improve compressive strength in M-PC3, with a significant decrease in relation to the reference. For nonattacked mortars, M-PC2 compressive strength was 14% higher than M-PC1, while M-PC3 compressive strength decreased by 12% compared to reference mortar. The sulfuric acid attack was highly aggressive, causing considerable strength losses in all the mortars, especially M-PC2. Strength decreased 46%, 59%, and 58% for M-PC1, M-PC2, and M-PC3 attacked mortars, respectively, compared to nonattacked specimens. Gypsum was formed in the mortars due to the reaction with sulfuric acid, but was more significant on the specimen outer surface. The presence of gypsum inside M-PC1 was not observed in the other mortars, especially M-PC2. Gypsum formation inside the samples may have contributed to the lower compressive strength loss of the reference, while surface degradation significantly affected the post-attack performance of M-PC2 and M-PC3. Capillary absorption results indicated distinct patterns between mortars M-PC1, M-PC2, and M-PC3, with higher rates for M-PC1 and M-PC3 and lower values for M-PC2 over a 7-day assessment period. The same behavior was observed for total water absorption, with significantly lower absorption for M-PC2 in relation to the other mixes before the attack. Indeed, M-PC2 presented 6% lower water absorption than the reference nonattacked mortar, while M-PC3 showed 8% higher absorption than M-PC1. However, no significant differences were found in total water absorption of attacked mortars, as the degradation of cement matrix increased porosity of all mortars. The results of this study confirmed the great potential of SCBA for application as a SCM in the production of blended cements. As such, using SCBA in blended cement production can contribute to the decarbonization of the cement industry while maintaining the good performance of cementitious systems. Future research should investigate higher levels of clinker substitution with SCBA in the production of pozzolanic Portland cements, equivalent to Brazilian CP IV. Also, long-term durability studies that include other aggressive environments and quantify the environmental performance of these cements through life cycle assessment should be performed.

Acknowledgments

This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. The authors would like to thank the Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) and the Fundação de Amparo à Pesquisa do Estado do Rio de Janeiro (FAPERJ) for the additional funding provided. The authors also thank the Mineral Technology Center (CETEM-MCT, Brazil) for their collaboration in the SEM analyses.

References

Edited by

Publication Dates

History