Alkali-activated binary concrete based on a natural pozzolan: physical, mechanical and microstructural characterization

R. Robayo-Salazar^a, R. Mejía de Gutiérrez^a,*, F. Puertas^b

a. Composites Materials Group (CENM), School of Materials Engineering, Universidad del Valle, (Cali, Colombia)

b. Eduardo Torroja Institute for Construction Science (IETcc-CSIC), (Madrid, Spain)

*: ruby.mejia@correounivalle.edu.co

1. INTRODUCTIONTOP

From an environmental perspective, alkali-activated materials (AAMs), which are sometimes also called geopolymers (1), are considered a sustainable alternative to Portland cement (OPC) in some applications in the construction industry because their CO₂ footprint is lower than that of OPC production (2). However, these materials (AAMs) are not expected to be utilized in all OPC applications due to the limited availability of some of the raw materials and the requirement for some rigorous control protocols during their manufacturing and curing (3). However, AAMs are currently considered key materials for the future sustainability of this industry (4) and are included in some standards, such as ASTM 1157 (EUA), NTC 121 (Colombia), PAS 8820 (United Kingdom), SIA 2049 (Switzerland), GB/T 29423 (China), and other prescriptive standards (Ukraine, Russia, Australia). Moreover, ASTM and RILEM (TC 244-AAM) technical committees (5) have focused their efforts on establishing regulations and standards for controlling and promoting their use in the construction industry worldwide.

Since these materials were first fabricated, they have been shown to exhibit high mechanical performance and chemical resistance, making their use attractive (6–8). However, some barriers to exploiting these advances on an industrial scale currently exist (9). It is believed that the lack of studies focused on using readily available aluminosilicates (precursors) is the main barrier. Therefore, some authors (3) argue that studies should focus on the use of resources such as natural pozzolans (NPs) of volcanic origin, which comprise 0.84% of the world’s soils (124 million hectares) (10). Indeed, volcanic soils represent an important resource for countries such as Iran, Cameroon, Japan, China, Saudi Arabia, Turkey, Jordan, Italy, Greece, the United States of America, Mexico, Chile, Ecuador and Colombia (11), where their deposits are generally concentrated in areas of high demographic and economic growth. For example, NPs constitute 11% of Colombia’s territory and are concentrated in the central and southwest zones of the country. In Chile, they represent 50% of the total arable land of the country and are concentrated in the metropolitan regions, while they represent 18% of Japan’s territory (12, 13). Moreover, Southern Europe is the original source of pozzolans.

Little information about the use of NPs in the field of alkaline activation has been published in the literature (14). However, Allahverdi et al. (15, 16), Kani et al. (17–21), Lemougna et al. (22–25), Bondar et al. (26–34), Djobo et al. (35–40) and Tchakoute et al. (41–44) reported promising results demonstrating their potential as a geopolymeric precursor. These researchers state that NPs generally have a certain degree of amorphous content (≤ 30%) that makes them reactive in the presence of strongly alkaline solutions. However, due to their predominantly semi-crystalline nature and low CaO and reactive Al₂O₃ contents, hydrothermal curing treatments (40-90°C) are commonly employed to improve the mechanical performance of early age AAMs. Such treatments undoubtedly limit the use of these materials in some applications. Therefore, it has been demonstrated that modifiers or secondary sources of calcium (CaO) and reactive Al₂O₃ can be incorporated into these binders to render the thermal curing process unnecessary. Indeed, adding a small amount (≤ 30%) of granulated blast furnace slag (GBFS) to NPs to obtain alkali-activated binary cements allows these materials to harden at room temperature (25°C) (16, 21, 45–48), thus enabling their use in “on-site” applications. Incorporating GBFS into NPs is advantageous because its dissolution during the alkaline activation process produces many Ca²⁺ ions that participate in the formation of calcium silicate hydrate (C-S-H) and calcium aluminosilicate hydrate (C-A-S-H) gels (49–51).

It should be noted that the above-mentioned results are not validated at the concrete level. Moreover, most of the published studies have been focused on determining the properties of pastes and mortars based on NPs. Therefore, studies based on concrete are very limited. Bondar et al. (33, 34) reported that concretes with compressive strengths of close to 25 MPa after 28 days of curing (20 ± 2°C) were obtained from Iranian pozzolans activated with KOH and Na₂SiO₃; however, compressive strengths of up to 50 MPa were observed after curing at 60°C. Haddad and Alshbuol (52) reported compressive strengths of 8.9-24.1 MPa for concrete obtained by curing a Jordanian NP activated with NaOH and Na₂SiO₃ under laboratory conditions for 28 days. They also found that thermally treating the material at 80°C for 24 hours resulted in a compressive strength of up to 30.8 MPa at 28 days. In his doctoral thesis, Najimi (53) studied the properties of concretes based on alkali-activated NP/slag using combinations of 70/30, 50/50 and 30/70; from the results, the author concluded that 50/50 is the optimum composition because the resultant concrete featured the highest compressive strength, i.e., 20 to 45 MPa at a curing age of 28 days, depending on the ratio of hydroxide to sodium silicate used in the mixture.

In a previous study (47), the feasibility of obtaining a binary binder (paste) using a 70% NP:30% GBFS mixture activated with a NaOH and waterglass (Na₂SiO₃·5H₂O) solution was explored. In that study, the alkaline binder was classified as a GU (general use) and LH (low heat of reaction) cement type based on the NTC 121 standard (equivalent to ASTM 1157 (54)). The objective of the present study is to obtain an alkali-activated binary concrete (AABC) using the same binary binder and determine its physical (water absorption, density and porosity), mechanical (compressive, flexural, and splitting tensile strengths and modulus of elasticity) and microstructural (characterized by scanning electron microscopy and energy-dispersive X-ray spectroscopy (SEM-EDS)) properties in order to demonstrate the feasibility of using this material as an alternative to OPC concrete in countries rich in volcanic soils, such as Colombia.

2. MATERIALS AND METHODOLOGYTOP

2.1. MaterialsTOP

A mixture of 70% natural volcanic pozzolan (NP) of Colombian origin and 30% granulated blast furnace slag (GBFS) was used as the AABC precursor. The chemical compositions of these materials were determined by X-ray fluorescence (XRF) and are listed in Table 1. The high SiO₂/Al₂O₃ molar ratio (6.79) and low alkaline content (CaO, Na₂O and K₂O) of the NP are notable. The GBFS has a SiO₂/Al₂O₃ molar ratio of 4.93 and a CaO content of 40.3%. The NP and GBFS mean particle sizes were determined to be 20.63 and 26.44 μm, respectively, by laser granulometry. Adding GBFS to the mixture as a calcium source enables the AABC to be cured at room temperature (25°C). The alkaline activator consisted of a dissolution of industrial grade sodium hydroxide (NaOH) and commercial sodium silicate or waterglass (Na₂SiO₃·5H₂O) (SiO₂ = 32.09%, Na₂O = 11.92%, H₂O = 55.99%) (SiO₂/Na₂O modulus = 1.1). Ordinary Portland cement (OPC) (GU type-NTC 121, equivalent to ASTM C1157 (54)) was used to produce the reference concrete. Its chemical composition is included in Table 1. It should be noted that in a previous study (47), the binder obtained by the alkaline activation of the 70% NP-30% GBFS mixture was classified as a GU and LH type cement based on its performance and the NTC 121 standard.

The natural aggregates used to produce the concretes (AABC and OPC) were crushed gravel with a maximum particle size of 12.7 mm and siliceous sand extracted from a local river with a fineness modulus of 1.85. These aggregates meet the specifications for use in concrete mixtures according to ASTM standards. Their main properties are listed in Table 2, and their granulometric distributions are shown in Figure 1.

2.2. Mixture design and specimen preparationTOP

The design specifications for the AABC mixtures were a minimum compressive strength of 21 MPa at 28 days and a slump of greater than 18 cm (fluid consistency). The amount of the 70%NP-30%GBFS precursor and the L/S ratio required to meet these design specifications were 400 kg/m³ and 0.35, respectively. The proportions of the mixture components were determined using a modified version of the “absolute volume method” in the ACI 211.1 (55) guidelines for OPC concretes. In this method, the absolute volumes displaced by all the mixture components and thus the densities of each component, including the alkaline activator (sodium hydroxide, waterglass and mixing water solution), must be known. The volume of the naturally trapped air was not considered. The optimal granulometric combination of the aggregates (45% sand-55% gravel) was determined by the “Fuller-Thompson” method (56). The design specifications and methodology used to determine the proportions for the OPC concrete were the same as those used for the AABC to ensure that the results are comparable. Table 3 shows the dry weights of the components used to prepare 1 m³ of both types of concrete (amount of cement (OPC) or precursor (NP-GBFS): 400 kg/m³). In addition, the binder content (NP-GBFS) used in the AABC preparation was varied between 300 and 600 kg/m³, and the compressive strength, water absorption, density and porosity of the concrete samples were measured. The proportions of these mixtures (Table 4) were determined following the same procedure (absolute volume method) using a constant L/S ratio (0.35).

The concretes were prepared in a CreteAngle horizontal mixer with mixing times of 8 and 15 minutes for the OPC concrete and AABC, respectively. The slump (≥18 cm) of the mixtures was measured according to the ASTM C143 standard (57) (Figure 2). The AABC exhibits notably high fluidity and resistance to segregation (cohesiveness). The mixtures were cast and vibrated for 30 seconds on an electric vibrating table to remove trapped air. The moulds were covered for 24 hours with a polymer film, and then the specimens were removed from the mould. The AABC specimens were cured at room temperature (± 25°C) and a relative humidity of greater than 80% until the test age was reached. The OPC was cured in water (immersion). The final distribution of the aggregates and the homogeneity of the mixtures in the hardened state were determined by examining a longitudinal cut of the cylindrical specimen (Figure 3).

2.3. Instrumental techniques and testsTOP

To analyse the raw and concrete materials, the following instruments and test methods were used:

The data reported for all physical and mechanical tests are the averages of three test cylinders.

3. RESULTS AND DISCUSSIONTOP

3.1. Mechanical characterizationTOP

Figure 4 compares the evolution of the compressive strengths of the AABC and OPC concretes. In general, both concretes met the design strength specifications (f’c = 21 MPa) and exhibited equivalent performance (~ 23 MPa) after 28 days of curing. It should be noted that the strength of the AABC increased more than that of the OPC concrete at longer curing times, reaching a value of up to 33.2 MPa at 360 days, which was 10.8% higher than that reached by the OPC concrete. However, its early age compressive strength (7 days) was 17.5% lower than that of the reference concrete (OPC). This behaviour was also observed at the mortar level in a previous study (47) and is attributed to differences in the nature and kinetics of the reactions that give rise to the cementitious gels. In contrast, Ibrahim et al. (63) demonstrated the possibility of reaching 75% of the final strength (28 days) in the first 3 days of curing using an NP-based (400 kg/m³) concrete activated with NaOH and waterglass but cured at 60°C. The mechanical performance of the AABC (25°C) (Figure 4) is similar to those reported by Bondar et al. (33) for a concrete based on an Iranian NP cured at 20 ± 2°C (~ 25 MPa at 28 days) (NP content of 344-417 kg/m³; KOH + Na₂SiO₃ alkaline activator) (34). Bondar et al. (33) also reported a compressive strength of up to 50 MPa after 28 days when this same concrete was cured at 60°C. In another study, Haddad and Alshbuol (52) measured compressive strengths of 8.9-24.1 MPa for a Jordanian NP-based (410 kg/m³) concrete activated with NaOH and Na₂SiO₃ after 28 days of curing under laboratory conditions (in air). Similarly, the authors also emphasized that thermally treating the concrete at 80°C for 24 hours resulted in performance of up to 30.8 MPa at 28 days.

Some authors (33, 52, 63) clearly support the practice of heat-treating NP-based concretes to increase their early age mechanical performance, although it is believed that doing so limits their real application. Therefore, in this study, the mixtures were designed to achieve an acceptable early age mechanical performance using a curing temperature of 25°C. Figure 5 shows the effect of the binder (NP-GBFS) content (kg/m³) on the compressive strength of the AABC. In general, the performance of the AABC increased with increasing binder content (300 < 450 < 600 kg/m³). This behaviour is due to the increase in the phase responsible for the mechanical strength, i.e., the binder, and the decrease in the volume of aggregates in the mixture. When the binder content was 600 kg/m³, compressive strengths of up to 31.5 and 43.4 MPa were measured after 28 and 360 days of curing, respectively. Because of this behaviour, the curing time (at 25°C) required to achieve a higher design strength (f’c) can be optimized by adjusting the binder content. Furthermore, the effects of the binder content on the strength increased with increasing curing time. For example, the mixture with a binder content of 600 kg/m³ was 1.81 and 2.41 times stronger than the mixture with a binder content of 300 kg/m³ after 7 and 360 days, respectively.

The modulus of elasticity results are presented in Table 5. These results are consistent with the compressive strength (90 days) results: the AABC had a slightly higher modulus of elasticity (by 4.2%) than the OPC (29.14 vs. 27.97 GPa). These values are consistent with those reported by Bondar et al. (33) for a concrete based on an Iranian NP (391 kg/m³) activated with KOH and Na₂SiO₃. These researchers reported values of modulus of elasticity of 32.7 and 29 GPa for the NP-based and reference (OPC) concretes, respectively, after 28 days of curing at 20 ± 2°C. In this work, the ratios of transverse unitary deformation to longitudinal or axial unitary deformation (Poisson’s ratio) obtained for the AABC and OPC concretes were 0.16 and 0.19, respectively (Table 5). These results reflect that the AABC is slightly more rigid than the OPC.

Table 6 lists the measured splitting tensile strengths and flexural strengths (modulus of rupture) of the AABC and OPC concretes. In general, the two types of concrete exhibited similar performance, and the properties of both concretes increased with increasing curing time (25°C). After 360 days of curing, the AABC had a slightly higher splitting tensile strength than the OPC concrete (3.85 vs. 3.23 MPa), which is consistent with the compressive strength results. As highlighted by Bondar et al. (33), the splitting tensile strength results indicated that an adequate binding force (interface) develops between the aggregates and the paste in the AABC because this property primarily depends on this physical interaction. Figure 6 shows the macro- and micro-interfaces of the AABC and OPC concretes after 28 days of curing. Good cohesion (anchoring) between the paste and aggregates and an adequately homogeneous interfacial transition zone were observed for both concretes. The flexural strength of the AABC appeared to be more sensitive to the test conditions used than that of the OPC concrete: the modulus of rupture of the AABC was 13% lower than that of the reference concrete (OPC) after 360 days of curing (Table 6).

3.2. Physical characterizationTOP

The water absorption, density and porosity results for the AABC and OPC concretes are presented in Table 7. The concretes clearly exhibited similar physical characteristics. Furthermore, a longer curing time (25°C) resulted in a slight decrease in the permeable pore volume and/or the densification of the concretes, which is consistent with the observed increases in the mechanical performance of these concretes with increasing curing time (Figure 4). From the results shown in Table 7, it can be deduced that the AABC has a slightly less permeable porous structure (i.e., it absorbs less water) than the OPC concrete.

Similar results were obtained by the mercury porosimetry technique, as shown in Figure 7. In particular, the total porosities of the OPC concrete and AABC measured by this technique were 15.5 and 15.9%, respectively. However, the results also indicated that the OPC concrete had a slightly more refined pore size distribution than the AABC: the majority of its pores (8.6% of a total of 15.5%) were 0.1–1 μm in size, whereas the highest fraction of pores in the AABC (6.45% of a total of 15.9%) were 1–10 μm in size. The meso-porosity, that is, the fraction of pores smaller than 0.1 μm (100 nm), was 4.99% for the OPC concrete and 3.76% for the AABC.

The effects of the binder content (NP-GBFS) on the water absorption, density and porosity of the AABC are presented in Table 8. These results are consistent with the compressive strength results (Figure 5). The permeable pore volume decreased significantly with increasing binder content (300 > 450 > 600 kg/m³). After 28 days of curing, it decreased from 22.84% to 13.47% when the binder content was increased from 300 kg/m³ to 600 kg/m³, corresponding to a 37% decrease in the water absorption (8.49% for 300 kg/m³ binder vs. 5.33% for 600 kg/m³ binder). Likewise, the densification of the matrix increased with increasing curing age for all the AABC mixtures, which is consistent with the increase in the strength of the mixtures as a function of time (Figure 5).

3.3. Microstructural characterization of the AABCTOP

Figure 8 shows the microstructural composition of the AABC (28 days) using a ternary SiO₂-Al₂O₃-CaO diagram obtained from the SEM-EDS data. The results are grouped into regions representing C-S-H gels (high calcium contents) and intermediate calcium-sodium aluminosilicate hydrate (N,C)-A-S-H and C-A-S-H gels (“dreierketten” structure), which are the products of the interactions between the primary N-A-S-H and C-S-H gels that form over time (64, 65). The points associated with the C-A-S-H and C-S-H gels, which are rich in Ca, have Ca/Si values in the range of 0.7-1.8. Theoretically, sodium aluminosilicate hydrate (N-A-S-H) gels do not contain Ca, unlike the (N,C)-A-S-H gels that have Ca/Si values in the range of 0-0.4, which is consistent with the Ca/Si ranges reported in the literature (38, 66–70). The EDS elemental mapping results shown in Figure 8 provide a visual representation (colours) of the elementary composition of the AABC. In general, the paste that agglomerates the aggregates is rich in calcium (purple), which confirms the formation of C-S-H, (N,C)-A-S-H and C-A-S-H gels due to the chemical interactions between the precursor (NP-GBFS) and alkaline activator (NaOH+waterglass). This compositional analysis of the paste-aggregate interfaces is consistent with the adequate mechanical performance of the AABC.

4. CONCLUSIONSTOP

Based on the experimental results, the following conclusions can be drawn:

The physical, mechanical and microstructural characterization of the AABC shows that NP deposits in countries rich in volcanic soils, such as Colombia, can be used as readily available raw materials for the industrial-scale production and commercialization of alkaline-activated concretes in the foreseeable future.

ACKNOWLEDGEMENTSTOP

This research was financed under the “PUZOGEOH” project in call 592-2012 (Colciencias-Cementos Argos SA-Universidad del Valle), contract No. 0484-2013. The authors thank the Centro de Excelencia en Nuevos Materiales (CENM) and the Instituto de Ciencias de la Construcción Eduardo Torroja (IETcc-CSIC) in Madrid-Spain for the support received during the development of part of this research. R. Robayo-Salazar thanks Colciencias for the support received in call No. 617 of 2013 for doctorate training in Colombia.

REFERENCESTOP