Introduction

The huge volume of cement globally consumed accounts for 7% of the carbon dioxide emissions (CO₂). This emission mainly occurs during the production phases of clinker, a fundamental component of Portland Cement. The calcination of both limestone and clay itself and the use of fossil fuels contribute to the high CO₂ emissions of the cement industries. The amount in kilograms of carbon dioxide generated per ton of cement produced varies depending on the type of cement to be obtained. It is common knowledge that cement with a higher percentage of clinker replaced by other materials is also the cement that emits less amount of CO₂ (for example, CPIII). These materials can be blast furnace slag, limestone or pozzolanic materials, which aim to meet the requirements of concrete and mortar to be produced (SNIC, 2019; WBCSD, 2016).

Therefore, the most widely used strategy to reduce the environmental impacts of CO₂ emissions is the reduction of clinker content. Thus, active mineral additives are used in order to replace a certain percentage of cement. These additives, such as blast furnace slag and fly ash supply and even improve the physical and chemical properties of the reduced amount of cement. In this way, a certain type of concrete or mortar can be obtained with characteristics that will depend on the additive used and its replacement content. The reuse of waste from other production processes as additives and the reduction of the amount of clinker produced, and, consequently of CO₂ emissions, make explicit the environmental and economic advantages of this strategy, as well as the existing potential of other materials to replace cement.

It is acknowledged that mineral additives have two types of effects on cement hydration, improving the properties of the concrete or mortar to be obtained, with compressive strength being one of the most important along with durability (Noaman; Karim; Islam, 2019). These are: the pozzolanic effect and the filler effect. The pozzolanic effect is related to the formation of compounds with cement properties (C-S-H), through the reaction between silica (silicon dioxide) and calcium hydroxide present in the material. This calcium hydroxide comes from the incomplete initial hydration of cement and its presence limits the development of compressive strength. Alternatively, the filler effect comprises the filling of the voids by the ground particles of the additives, the refinement of the pores and the reduction of the water accumulated under the aggregates. The result is a more compact mortar or concrete, which favors the workability in fresh concrete, at the moment of preparation, and the density and permeability in hardened concrete.

Commonly, the relevance of each effect on the properties of the final product varies depending on several factors: the pre-treatment of the additive, the percentage replaced of the cement, the curing, the age of the test specimens, among others. Therefore, it cannot be stated that, for example, the compressive strength varies linearly as a function of the percentage of a given additive. In fact, normally, this property reaches, after 28 days, the most appropriate value with a certain amount, but this value decreases if this percentage is higher.

In addition to the materials already used as mineral additives (fly ash, silica fume, metakaolin, etc.), several studies have been conducted to evaluate the contribution of pozzolanic and filler effects of new materials, considered waste from other industries, on the characteristics of Portland Cement.

One of these studies, conducted in 2019, compared the contribution of the effects of rice husk ash as a mineral additive to the mechanical properties of concrete blocks (Noaman; Karim; Islam, 2019). In this research, as in many others, the effects were differentiated by comparing the properties of concrete containing the additive studied, of concrete containing a non-reactive material, but with similar granulometric composition; and of concrete without any substitution (reference or control). They used rice husk ash ground at different times to increase and compare fineness, and, as non-reactive filler, natural sand ground for 150 minutes. The results showed that both effects were significant in all used percentages for substitution, but that at 28 days, both effects showed better compressive strength results for the 15% ratio. In all cases, the development of compressive strength due to the filler effect was clearly lower than that of the pozzolanic effect.

Another study evaluated the influence on the compressive strength of concrete after substituting cement with elephant grass ash at a 20% ratio (Cordeiro; Sales, 2015). The use of this plant is justified by its rapid growth, its enormous production of 40 tons/ha/yr, and the considerable amount of ash generated by direct combustion due to its calorific value of 4200 kcal/kg. According to the results obtained, elephant grass ash is suitable as a pozzolanic material since there were no significant changes in compressive strength when compared to the theoretical value of 35 Mpa at 28 days. In tropical and subtropical areas, where elephant grass production is significant, its potential as renewable energy is evident.

Research conducted in 2008 points to the benefits of using sugarcane bagasse ash in mortar as a mineral additive due to its significant production and its high silica content (Cordeiro et al., 2008). In order to differentiate the effects, they compared the pozzolanic effect of mortar with sugarcane bagasse ash and with ground Quartz, considered non-reactive and of similar granulometric composition. In this case, the ratio mix for the substitution with pozzolanic material was of 35%. It is worth noting that the grinding time (reduction of particle size) was analyzed, leading to the conclusion that this positively influences the compaction and pozzolanicity of the material when it is longer. The results showed that the values for compressive strength met the theoretical ones and, therefore, substitution at such a high ratio was possible.

The use of materials with high silica content and with the possibility of being ground to smaller particle sizes than the cement itself represents a great opportunity for the research of the production of concrete and mortar, in order to evaluate the physical and chemical effects of the final product. The objective of these is to evaluate how the results of the tests vary depending on the ratio, and, thus, determine the ideal ratio mix depending on the type of concrete or mortar to be produced. In addition to the advantage previously mentioned regarding the reduction of clinker, the reuse of waste represents another, possibly greater, advantage of reuse in places with a significant amount generated. In this context, we come to the material that meets the required conditions and that is the central theme of this study: glass.

Glass has a high recycling potential, which allows for the possibility of recycling it almost completely and indefinitely in some cases (Dhir et al., 2018). For example, glass waste shows an average recovery rate of 79% in the 28 countries of the European Union. In some of them, the amount recovered is greater than that produced (Finland, Spain, Germany, Czech Republic, and the Netherlands). This is probably due to additional efforts to recover glass waste produced years ago (EU, 2016). However, in Brazil, glass has a low reuse, with only 25% of glass materials being recycled (Abividro, 2023). According to the Brazilian Association of Glass Industries (Abividro, 2023) 325,000 tons of glass were destined for recycling in 2022. The main market for the reuse of glass is the glassmakers themselves, where the materials return to the production of packaging. However, in the construction industry, several recent studies indicate that glass powder has a satisfactory performance as a pozzolanic material, as a substitute for cement in obtaining concrete and mortar.

Generally, it contains inorganic elements, being silica the main one. Soda-lime glass is the most used (glass containers, bottles, household glass, etc.) as it exhibits inorganic elements, and, consequently, pozzolanic properties when finely ground and reacting with calcium hydroxide in the presence of water.

Studying this material as a mineral additive in cement for the production of concrete and mortar is relatively recent since initial studies tried to use it as a substitute for aggregate. The alkali-silica reaction was the impediment to use glass powder as an aggregate since the size and geometry of particles generated deleterious expansion. However, studies have confirmed that there is no expansion if the particles are smaller than 75µm (Freitas, 2019; Soares; Ferreira; Salvador Filho, 2018). Therefore, it is natural to assume that the function of glass powder is more suitable as cementitious.

Thus, the objective of this work was to evaluate the physical and mechanical properties of mortar with the incorporation of glass powder of mesh size #200 (75 µm), seeking to differentiate the filler and pozzolanic effects of the material. We conducted chemical and mineralogical analysis for the glass powder to be used in the production of mortar, and determined its ideal ratio mix of in relation to cement (10%, 20%, 30% and 50%). The effects of glass powder in mortar were compared with the incorporation of limestone filler (material with a similar particle size distribution curve) in the same ratios.

Materials and methods

To produce mortar, the following materials were used: Portland cement CP V ARI, glass powder of mesh size #200 (75 µm), fine aggregate with maximum size of 2.4 mm and limestone filler 325 (non-reactive) with a particle-size distribution similar to glass powder to quantify the strength gain by filler effect (FE) of the pozzolanic material studied (finely-ground glass). Figure 1 shows images of the material used in the manufacture of mortar.

Cement

The cement used in this research was Portland cement CP V ARI, manufactured by Holcim Brasil S/A, the analysis provided by the manufacturer was used as a reference, with its specifications standardized by NBR 16697 (ABNT, 2018a).

The granulometric analysis of the cement was conducted with a FRITSCH Laser Particle Sizer, Germany, model ANALYSETTE 22 NanoTec, which uses laser diffraction to determine particle size through calculations based on Mie Theory. It is noteworthy that, to avoid particle agglomeration in the laser particle sizer, potable water at room temperature was used as dispersion medium together with a dispersing agent, Darvan C-N from Vanderbilt, an ammonium polymethacrylate.

The mineralogical analysis of the cement used X-ray powder diffraction (XRD). The objective of the test was to analyze the mineralogical phases of the material. To perform the test, an X-ray unit was used: Rigaku Rotaflex, model RU200B, with nominal scan 2θ of 3° to 120°, pitch 0.02°, scanning 2°/min, voltage 40 kV and 60 mA, copper anode.

For the chemical analysis of the material, the X-ray fluorescence (XRF) test was used, a non-destructive technique to identify chemical elements. The determination of the specific mass of cement was determined by the Le Chatelier’s Flask Method, according to NBR 16605 (ABNT, 2017) and the determination of the fineness and specific surface area of this material was determined by the air permeability method (Blaine’s method), according to the NBR 16372 (ABNT, 2015).

Glass powder

The glass powder used is classified as soda-lime glass, which comes from amber-colored bottles. The bottles were washed, had their labels and glue removed, then crushed into shards in a concrete mixer with steel balls, and the obtained shard was ground in a ball mill. For grinding, a mill coated with flint (siliceous rock) and balls of the same material were used. The steps of the process of obtaining the material can be seen in Figure 2.

The product obtained is a glass powder that was fractionated through sieves into particles of size <75 µm (mesh size #200). The final sifting took place at the materials and components Laboratory (LMC), Federal University of São Carlos (UFSCar). Initially, the glass powder was dried in an oven at 110 ± 5 °C for 24 h, then sieved in the three established meshes through a shaker, the Ro-Tap W S. TYLER.

The granulometry, specific gravity, fineness determination, XRD analysis and chemical composition of the glass powder used the same procedures presented for Portland cement. The morphology of the samples was characterized by the scanning electron microscopy (SEM) of a sample of dry particles. SEM analyses used an Inspect F50 scanning electron microscope, from the brand FEI.

The pozzolanic activity of this glass powder was verified by Borges et al. (2021) using the methods Performance index with Portland cement according to NBR 5752 (ABNT, 2014), Determination of the fixed calcium hydroxide content – Modified Chapelle method, according to NBR 15895 (ABNT, 2010) and Pozzolanicity by electrical conductivity, according to Luxán, Madruga and Saavedra (1989).

Limestone filler

The limestone filler was used as an inert material, with a particle size distribution curve close to that of glass powder, being an auxiliary material for differentiation between the filler and pozzolanic effects of the glass powder. Thus, to characterize it, we used a laser to determine the particle-size distribution of the material and compared it to the glass powder of mesh size #200 (75 µm).

For this study, we used the filler limestone 325, sold by BRASIL MINAS.

Fine aggregate

It was used natural fine aggregate, sieved river sand available in the region of São Carlos, BR, with a maximum dimension of 2.4 mm, analyzed according to the tests shown in Table 1.

Material analysis

The aggregate presented fineness modulus of 2.18, and the maximum dimension found was 2.4 mm. The granulometric composition of the sand was carried out in accordance with NBR 17054 (ABNT, 2022b). The aggregate particle size distribution curve is between the upper optimal zone and the lower usable zone, as established by NBR 7211 (ABNT, 2022a). Figure 3 shows the particle-size distribution graph of the aggregate used.

The analysis of other physical parameters of the aggregate can be seen in Table 2.

The chemical composition of cement, glass powder and limestone filler are presented in Table 3.

Table 4 shows some physical properties of the cement and mineral admixtures.

The mineralogical characteristics, obtained by XRD, of the cement and glass powder of mesh size #200 are presented by the diffractograms in Figure 4.

Figure 4(a) shows that the clinker is mainly composed of tricalcium silicate or Alite-C3S (Ca₃SiO₅) and Dicalcium silicate or belite - C2S (Ca₂SiO₄) identified by their characteristic peaks in the diffractogram. Allite (C₃S) shows peaks in typical positions, mainly between 29° and 34° (2θ), while belite (C₂S) shows peaks in nearby regions, generally between 31° and 33° (2θ) (Scrivener; Snellings; Lothenbach, 2016; Stutzman; Feng; Bullard, 2016).

In Figure 4 (b), it is possible to observe that the glass powder is predominantly amorphous, as indicated by the broad halo between 10° and 40°. This amorphous halo is consistent with findings reported by (Alves et al., 2014; Dobiszewska et al., 2023). However, unlike the observations made by these authors, a distinct peak corresponding to SiO₂ was identified within the 2θ range of 25° to 30°. This SiO₂ peak can be attributed to contamination during the dry grinding process, likely caused by the wear of flint balls used in the milling process, which are rich in SiO₂.

A similar SiO₂ peak within the 25° to 35° range was also observed by Elaqra and Rustom (2018) when analyzing glass powder. The authors associated this peak with impurities introduced during the crushing process, further supporting the interpretation of contamination from the milling environment.

Figure 5 shows the particle size distribution curves of cement, glass powder of mesh size 200 and limestone filler.

Analyzing the particle size distribution, in d50, the particle size was 7.37 µm for cement, 14.19 µm for glass powder, and 13.40 µm for limestone filler, which shows the great similarity between glass powder and limestone filler regarding grain size distribution. Therefore, the analysis satisfactorily validates limestone filler as a non-reactive material to replace glass powder, enabling its use in the present study. The glass powder was analyzed by scanning electron microscopy (SEM).

Another important observation revealed by the grain size analysis concerns the product commercially known as “Fíler Calcário 325.” The name reflects the mesh size of a 325 Mesh sieve, which has a standard opening of 45µm. However, during the analysis of the collected sample, it was found that the largest particle present in the sample measured 75µm. This indicates a discrepancy between the commercial specification and the results of the conducted analysis. Figure 6 shows the micrographs obtained.

SEM analysis of the glass powder reveals angular particles with well-defined edges and irregular shapes, characteristics that are commonly observed in milled materials. The rough surfaces of these particles exhibit the presence of fine particles, which are adhered to the surfaces. Additionally, a non-uniform particle size distribution is evident, with a tendency towards agglomeration. Figure 6 provides a visual representation of the morphology of the glass particles, which is consistent with the findings reported in studies such as Li et al. (2022). These studies also document the presence of angular and rough particles in glass powders obtained through milling processes.

Test program

The produced mortar was molded in cylindrical specimens (50 x 100 mm), and some parameters were established: mix ratio of 1:3 (cement:fine aggregate) and 0.55 water/binder ratio, aiming to obtain a consistency of 230±10 mm. For the substitution of cement for finely ground glass, the ratios were 10%, 20%, 30% and 50% by volume, and the curing of the specimens was through immersion in lime-saturated water.

To analyze the changes in the properties due to the filler effect of the glass powder, the same number of specimens (Spec.) was molded. However, at this time, the different ratios of cement were substituted for limestone filler. For these Spec., the same mix ratio and water/binder ratio were maintained as in the Spec. with finely-ground glass, so it was possible to analytically study the increase in each property analyzed by pozzolanic action and by filler effect.

The parameters that remained fixed throughout the study were: the type of cement, the fine aggregate, and the water. The variables were the ratios of glass powder and limestone filler used to substitute cement and, consequently, the consumption of cement in the mixes. The consumption of material in kg/m³ of the mortar used in this study can be seen in Table 5.

The consistency of all mortar mix ratios was checked, adopting an established consistency of 230±10 mm. The NBR 13276 (ABNT, 2016) was used for the consistency test.

The mixing procedure adopted for the mortar is the one presented by NBR 7215 (ABNT, 2019), a standard for determining the compressive strength of cylindrical specimens. Before performing the mechanical mixing process, the additives used in this study (glass powder or limestone filler) and Portland cement were previously mixed and homogenized in a closed container for approximately two minutes.

The analysis of the mortar was conducted through the evaluation of the physical and mechanical properties of the specimens. Table 6 details the mortar tests to evaluate these properties, as well as the methodology, ages, quantity, and size of the Spec.

In addition, pastes with reference trait (REF) and with the traits related to the replacement contents of 20% and 50% of cement by glass powder (GP20 and GP50) and by limestone filler (LF20 and LF50) were made.

To make the samples of the pastes, the same mortar mix was maintained, removing only the fine aggregate from the mixture. The same mixing procedure presented was also maintained. For molding the pastes, plastic molds measuring 40 mm in diameter and 50 mm in length were made.

For the pastes, the X-Ray Diffraction (XRD) test was carried out at the age of 28 days. To prepare the samples for the test, firstly, the specimens were broken into smaller pieces to remove some flakes of the hardened paste. These smaller pieces were taken to the mortar-pistil type mill to be reduced into powder grains that were dried in an oven and passed through the #200 mesh in the Ro-Tap sieve shaker. Only the passing material was stored for XRD.

With this test, we try to identify the crystalline phases of the substances present in each sample. It is a qualitative analysis and indicates the compounds present in the peaks that appear in the diffractograms.

According to Scrivener, Snellings and Lothenbach (2016), the application of the X-ray diffraction technique in samples of cementitious composites can be used to quantify the degree of hydration of anhydrous cement and provides information on the formation of individual phases of hydrates.

Statistical analysis

The data from the mortar tests were analyzed using Analysis of Variance (ANOVA), a methodology used to compare three or more treatments, one of the most flexible and practical techniques for comparing multiple means (Miller, 1985).

For the ANOVA method, a statistical test was conducted to verify if there is a difference in the distribution of a value among three or more groups. This is done through the decomposition of the sum of squares for each source of variation in the model. Based on Fisher-Snedecor’s F-test, the hypothesis that there is no source of variation among the samples is tested (Miller, 1985).

The hypotheses were as follows: H0 – There is no statistically significant difference; H1 – There is at least one value with a statistical difference.

The p-value is also compared with α (adopted as 5% in this study). If p-value ≥ α, H0 is accepted (there are no significant differences); If p-value < α, H0 is rejected (there are significant differences). Subsequently, the data that showed significant differences were subjected to the Tukey test using the PAST statistical analysis software to determine which sampling values have statistically significant differences. If the H0 hypothesis is rejected, the Tukey procedure was used to determine which pairs of means have statistically significant differences (Driscoll, 1996).

Thus, in this study, ANOVA checks for the existence of statistically significant differences between the average values of the evaluated samples, while Tukey’s test was used to identify which means are different.

Results and discussion

Mortar tests

Figure 7 compares the reference mortar’s results of axial compressive strength with the different ratios of glass powder and limestone filler. A theoretical value is also presented for each ratio without substitution in relation to cement (ex: theoretical value of REF with 10% = REF × 0.9). In addition, strength increases are differentiated by pozzolanic effect (PE) and filler effect (FE).

This methodology presented by Noaman, Karim, and Islam (2019) to obtain a theoretical value, calculated based on the percentage of results obtained from different properties of reference test specimens, is uncommon and not widely used. This is due to the existence of many influential variables in the process, such as the water-cement ratio and particle packing. For this reason, it does not accurately represent what is really happening. Nonetheless, this study offers a new perspective and serves as a starting point for differentiating between the physical and chemical effects of supplementary cementitious materials.

In this document, starting from Figure 7, the theoretical values are displayed in bar charts. These values are represented by bars in the same color used for the values related to the properties of the reference test specimens, but differentiated by a hatch pattern.

The pozzolanic and filler effects of the glass powder in the mortar were quantified to compare both regarding the evaluation of the mortar’s axial compressive strength. The strength of the mortar, in the three ages assessed, were influenced by PE and FE, in which the increases of compressive strength observed in the ratio with limestone filler refers only to the FE of the glass powder, also observed in the research of Noaman, Karim and Islam (2019).

Thus, the strength gain by the pozzolanic action of the glass powder is calculated from the difference between the values obtained in GP and LF for each ratio at each age. For example, at 91 days the strength of GP10 was 41.70 MPa, and for LF10, 39.60 MPa. Therefore, the increase in strength due to the pozzolanic reaction of the glass powder is 2.10 MPa (41.70 – 39.60). All PE values are presented in the figure for each curing age.

It is possible to observe that, at 91 days, there is an increase in PE as the substitution ratio increases, while at other ages it is not possible to see this uniformity. A possible interpretation for this result may be the relatively slower chemical activity of glass powder as a pozzolan, also verified in other studies (Du; Tan, 2017). The authors suggest that is with a longer curing age than the benefits of the pozzolanic reaction of the glass powder begin to appear. Also at 91 days, there is the highest increase by PE (5.23 MPa) in GP50.

Regarding the comparison between the pozzolanic and filler effects of the glass powder, it is possible to observe that, at 7 and 28 days, there is no comparable relationship between the two effects. However, at 91 days, the behaviors are comparable, because as the substitution ratios increase, both effects tend to increase as well. Evaluating the differentiation between EP and EF of rice husk ash, Noaman, Karim and Slam (2019) showed a similar advance in the two effects (while one effect increases or decreases, the other follows), but they obtained an optimal ratio of 15% substituting cement for rice husk ash. The ratios above this (20% and 25%) show a reduction in both effects in the evaluation of axial compressive strength.

Regarding the statistical analysis for 7 days, there were no significant differences between the same contents of glass powder and limestone filler replacement. In relation to the reference mix, only the GP10 mix did not show a significant difference. For 28 days, in relation to the reference, only the 10% levels (for both GP and LF) did not present significant differences. At the age of 91 days, only the GP10 and GP20 mixes are statistically equivalent to the reference mix.

In Figure 8, it is possible to compare the results for all ratios studying tensile strength by diametral compression, and to see the differentiation between the filler and pozzolanic effects of the glass powder for the different ratios.

For tensile strength by diametral compression, it is possible to observe that the highest strengths between different ratios refer to 10% and 20% substitution with glass powder (GP10 = 4.22 MPa and GP20 = 4.14 MPa). Such values are higher than the strength for the reference ratio REF (4.01 MPa), however, through statistical analysis, it is possible to observe that the ratios GP10, GP20, GP30, LF10, LF20 and LF360 were statistically equivalent to the reference ratio.

Still in statistical analysis, only the highest substitution ratio (50%), both for glass powder and limestone filler, showed significant difference when compared to the reference ratio. For GP50, the difference in relation to REF was a decrease in tensile strength by diametral compression of 29.17%, and for LF50, the decrease in relation to REF was 48.38%.

Another interesting result is that, up to the 30% ratio, the behaviors of PE and FE are relatively similar (greater and lesser increases in strength occur together between two effects, showing lines that accompany each other on the graph). However, in the 50% ratio, this agreement does not occur, showing an analogous behavior in relation to the rest of the graph.

Sharifi et al. (2016) also studied the tensile strength by diametral compression of concrete with mix ratios (0, 5, 10, 15, 20, 25 and 30%), substituting cement for glass powder (grain sizes lower than 100 µm). The authors found, at the age of 28 days, that the 10% ratio presented a higher value of tensile strength by diametral compression in relation to all other ratios analyzed, exhibiting an increase in strength of 6.62% in relation to the reference ratio (0%). For ratios greater than 15%, the strength measured by the authors was lower than that recorded in the reference ratio.

In Figure 9, it is possible to compare the results for all ratios studying static modulus of elasticity, and to see the differentiation between the filler and pozzolanic effects of the glass powder for the different ratios.

The analysis of the graph shown in Figure 9(a) shows a decrease in the modulus of elasticity as the substitution of cement increases. Thus, the highest modulus value is 33.42 GPa, the reference ratio. However, through statistical analysis, it is possible to verify that, in relation to the reference ratio, there is no significant difference for GP10 and GP20.

For GP30 and GP50, there was a significant difference in relation to the reference ratio identified by statistical analysis. The static modulus of elasticity was 15.41% lower for GP30 in relation to REF, and, in GP50, the decrease was 25.91%. For the ratios with limestone filler, they were all statistically different in relation to REF.

Analyzing Figure 9(b), it is possible to observe that in the lower ratios (10% and 20%) the increase of the modulus, in relation to the theoretical value, occurs mainly due to the PE of the glass powder, however, in the ratios 30% and 50%, the filler effect is more significant. It is interesting to note that, in the 10% ratio, the FE shows an increase in modulus of 0.36 GPa, while the 50% ratio shows an increase of 7.03 GPa, a considerable increase between them.

He et al. (2019) also analyzed the modulus of elasticity of cementitious composites with different ratios substituting cement for glass powder (0, 10, 20 and 30%) at the ages of 7, 28 and 91 days. Up to 28 days, the authors also found that the reference mix had the highest value for modulus of elasticity among all ratios, but at 91 days, the mixes with glass powder in ratios of 10% and 20% of presented equivalent modulus to the reference mix. The authors attribute this result to the lower pozzolanic reaction of the glass powder in the beginning, showing effective result only after a longer curing time.

Figure 10 shows the graph comparing the behavior of the different ratios of the mortar studied throughout the test (72 hours) of water absorption by capillarity at 28 days.

Observing Figure 10, it is evident that, in the whole process, the 50% ratio presented the lowest values for water absorption. At some points, only the ratios GP10 and GP20 appear above the curve of the reference ratio.

Through statistical analysis, at the end of the 72 hours of testing, only the ratios with 50% of substitution (GP50 and LF50), and LF30 showed significant difference compared to the reference ratio. In GP50 and LF50, water absorption was 33.33% lower than reference. And the mortar with LF30 had a water absorption 28.57% lower than reference.

Ultra fine glass particles physically fill capillary pores, reducing capillary absorption (Matos; Sousa-Coutinho, 2012). The behavior reported by the authors was observed in our evaluation of water absorption by capillarity in the mortar, because as we increased the ratio of substituted cement for fine material, either glass powder or limestone filler, water absorption decreased. The reduction can be observed by comparing the reference ratio (0%), that presented an absorption of 1.35 g/cm², with the highest substitution ratios, GP50 and LF50, in which both presented a water absorption of 0.90 g/cm² (a decrease of 33.33%).

An interesting remark to be made concerning the results of the present study on water absorption compared to the axial compressive strength results is that, as the glass ratios in the mixes increased, the compressive strengths decreased and so did the water absorption. However, it is expected that the results for these characteristics were the opposite.

According to Mehta and Monteiro (2014), in concrete, permeation (a term used to describe the transport of liquids induced by capillary forces) is influenced by the volume and connectivity of capillary pores in the cement matrix. Thus, in the mortar here studied, the incorporation of fine material may have led to changes in pore sizes, connectivity and tortuosity that achieved this result.

In literature, studies with glass powder in which water absorption and axial compressive strength were investigated also show similar results for both properties. Matos and Sousa Coutinho (2012) a slight decrease in sorptivity (1.26%) of mortar mix ratio of 10% glass powder in relation to the reference mortar, while at the same ratio, there was a decrease in axial compressive strength in all ages analyzed by the authors. Patel et al. (2019) even finding a decrease in water absorption with higher ratios of glass powder, through the analysis of water absorption by immersion, they also found a decrease in the mortar’s compressive strength as the ratios increased.

Taking that into consideration, Figure 11 compares the results for all ratios studied after the 72h test of water absorption by capillarity, and it also shows the differentiation between the filler and pozzolanic effects of the glass powder for the different ratios.

In Figure 11, it is possible to observe the graph presented, and see that there was no similarity in the behaviors between filler and pozzolanic effects of glass powder on water absorption. While one effect has an increase, the other has a decrease, and vice versa. The highest increase by PE occurred in the 20% ratio (0.18 g/cm²), and by FE in the 50% ratio (0.22 g/cm²), and in the latter there is no change in behavior by pozzolanic reaction, only by filler effect.

Figure 12 shows the diffractograms of the cement pastes of the REF, GP20, GP50, LF20 and LF50 mixes at the age of 28 days. The analysis of the data obtained in the X-ray diffraction test was performed using the X’PERT HIGHSCORE 2.1 software developed by PANalytical B.V.

Analyzing the diffractograms presented in Figure 12, it is possible to observe that at the age of 28 days there are high crystalline peaks of portlandite and calcite in the reference samples and with replacement of 20% and 50% of cement by glass powder and limestone filler. In the samples of the pastes with the replacement of cement by glass powder, the presence of alite and belite peaks stands out, indicating points where there was no hydration.

It is possible to observe that the reference mixes and those with replacement by finely ground glass have higher portlandite peaks, while the mixes with cement replacement by limestone have higher calcite peaks.

For the portlandite peaks between the angles of 15º to 20º and 30º to 35º in the REF and GP20, it is possible to observe a similarity in the intensities. The peaks between the same angles are also present in GP50, but with a lower intensity. The calcite peak close to the 30º angle is marked in the traces with limestone filler, but it is more intense in the highest substitution content (LF50), as expected.

Conclusions

The results for mortar mixes with a 20% ratio, substituting Portland cement for glass powder, showed statistically equivalent results to the reference mortar in all properties assessed. Therefore, this substitution ratio can be considered the ideal one, since it maintains the characteristics of mechanical strength and water absorption, in addition to decreasing the consumption of Portland cement.

The axial compressive strength of the mortar mixes with cement substituted for glass powder in the three assessed ages were influenced by both pozzolanic effect and filler effect of finely-ground glass. However, at the age of 91 days, there was an increase in PE and FE as the substitution ratio increased, while at other ages it was not possible to see this uniformity. Thus, for the 50% ratio, the increase in compressive strength compared to the reference was 7.82% for FE, and 24.06% for PE. A possible interpretation for this result may be the relatively slower chemical activity of glass powder as a pozzolan, since, at this older age (91 days), the greatest gain by pozzolan effect was obtained for this property (5.23 MPa).

Regarding the tensile strength by diametral compression, the greatest strength gain in relation to the reference by filler effect (an increase of 22.74%) occurred in the 20% ratio, whereas, for pozzolanic effect, it occurred in the 50% ratio, a tensile strength gain of 38.30%.

In relation to the modulus of elasticity, the filler effect increased as glass content was added to the mix, with the 50% ratio presenting an increase of 42.00% only in relation to the FE. Considering the same ratio, due to the pozzolanic effect, the elastic modulus gain compared to the reference was 6.13%. The pozzolanic effect, in this property, stood out in the 20% ratio, with a modulus of elasticity gain of 13.05% when compared to the reference value. Thus, unlike FE, the pozzolanicity for this property did not increase as the glass ratio increased.

Regarding the absorption of water by capillarity, the 50% ratio showed no change in absorption by pozzolanic effect, only by filler effect (an increase of 32.35% in relation to the reference value).

The methodology adopted by this study verifies that, in all properties analyzed, both the pozzolanic effect and the filler effect were present after the incorporation of finely-ground glass in mortar mixes.

References

Edited by

Publication Dates

History