FA and GBFS were used as precursors to manufacture alkaline-activated binders and mortars. The FA and GBFS used in this study were industrial by-products supplied by a thermoelectric plant and a Colombian steel factory, respectively. FA and GBFS were subjected to a conditioning process by grinding in a ball mill to reduce its particle size. Table 1 shows the chemical composition of the FA, GBFS, and OPC used to produce the binary and hybrid binders. The chemical composition of the precursors was determined using X-ray fluorescence analysis by a Malvern Panalytical Axios^mAX spectrometer. It is worth noting the high content of unburned carbon (14.96%) contained in the fly ash (FA). The particle size distribution was carried out in a laser particle size analyzer (Malvern Mastersizer, 2000). The particle size distribution of the raw materials is shown in Figure 1. The average particle size D[4,3] of FA and GBFS was 19.50 μm and 17.79 μm, respectively.

A commercial sodium silicate (SS) and an alternative activator based on RHA as silica source were used as the activators in this study; in both cases, sodium hydroxide (NaOH: analytical grade, 99% purity, Merck) was used to adjust the proportion of sodium in the alkaline solution. The chemical composition of the industrial SS was 32.4% SiO₂, 13.5% Na₂O, and 54.1% H₂O. The RHA used in this study is an agro-industrial by-product obtained from the burning of rice husks used as biomass fuel. RHA has a high silica content (96%), as shown in Table 1; however, applying the method developed by Mehta (19), only 36% of the silica in RHA is considered reactive or soluble in an alkaline medium (NaOH solution). The low percentage of soluble silica is due to the lack of control in the combustion process. The average particle size of RHA (D[4,3]) after the grinding treatment for about 15 minutes in a planetary ball mill, was of 5 μm; this material shows a bimodal curve (Figure 1).

The alternative alkaline activator (NaOH+RHA) was obtained mixing the silica source (i.e., RHA) with a sodium hydroxide solution, in order to obtain a molar ratio SiO₂/Na₂O or solution modulus (Ms) equal to 1.5 in the FA/OPC-RN geopolymer and 1.2 in the FA/GBFS-RN geopolymer (Table 2). These molar ratios were also used in mixtures activated with SN. This means that the purpose was to replace the total amount of SiO₂ from commercial sodium silicate with silica from RHA. It should be noted that the contribution of SiO₂ from RHA was considered as 96%, taking into account the chemical composition listed in Table 1. The solution modulus (Ms) used in this study are in accordance with those used by Tchakoute et al. (13) to produce MK-based geopolymers and activated with waterglass from RHA. The alkali activators RN and SN were stored at room temperature for one day before its use. The pH of the activator solutions SN and RN was in the range of 12.2 to 12.7.

Two types of binders were produced, a binary binder FA/GBFS with a 70/30 proportion and a hybrid binder FA/OPC with an 80/20 proportion; each mixture was activated with the two alkaline activators (SN and RN). The SiO₂/Al₂O₃ and Na₂O/SiO₂ molar ratios and solution modulus M_S (SiO₂/Na₂O) used to prepare each binder are presented in Table 2, where RN corresponds to the mixture of RHA and NaOH, and SN corresponds to the solution mixture of commercial SS and NaOH. The mix proportions were calculated using the chemical composition of precursors and activators in order to obtain the molar ratios specified in Table 2. The liquid/solid ratio (L/S) where L represents the water content added and the provided by the activator, and S includes the precursors and the activator anhydrous, varies depending on the workability of the paste. FA/GBFS-SN, FA/OPC SN and RN type were prepared using a L/S of 0.35, FA/GBFS-RN was prepared with L/S of 0.42.

The cementitious paste was elaborated, mixing all the components in a Hobart mixer for approximately 10 minutes; the resulting paste was poured into cubic moulds of 20 mm per side. The specimens were cured in a chamber at room temperature with a relative humidity greater than 90%.

The prefabricated elements (blocks and pavers) were manufactured following the same steps mentioned above, the solid precursors FA/GBFS were mixed with the alkaline activator (RN) for about five (5) minutes to make a binder. Afterward, river sand was added to the binder to make a mortar; the mixing process continued to complete 10 minutes. The blocks were cast in a manual compression machine “Cinva-Ram machine” used to make soil-cement blocks, and the pavers were manufactured in a vibratory compactor machine. Both elements were kept covered with a plastic film to avoid water release from the material; all the elements were cured in a climatic chamber at controlled conditions (25°C and 90% H.R). The sand used was from the river with a maximum particle size of 1.18 mm (mesh No 16). The L/S ratio for blocks and pavers was 0.45 and 0.38, respectively.

Microstructural analysis, physical, and mechanical tests were performed as follow:

Figure 4 shows the microstructure of the four studied systems: FA/GBFS-SN, FA/GBFS-RN, FA/OPC-SN, and FA/OPC-RN, after 28 and 360 days of curing. Based on the SEM images, at the age of 28 days of curing, all samples exhibit a non-uniform surface; there were found dense and uneven areas, microcracks from drying shrinkage process, and some unreacted particles (marked with a star) from FA which are characterized by having a spherical shape. Based on the observations, it is assumed that many of the unreacted particles are dissolved at the age of 360 days, becoming part of the reaction gels.

At the age of 28 days, RN systems have a more porous and uneven microstructure compare to SN type systems, which have a smoother and denser appearance. These results indicate that the dissolution of the particles in SN binders has a higher reaction degree than RN binders at an early age. At the age of 360 days, all the systems show a denser and a similar surface. It is worth noticing RN systems show a low reaction degree at an early age; at a more extended time, it reaches a comparable microstructure than SN. This performance agrees with the mechanical behaviour shown in Figure 6.

Elements such as Si, Al, Na, Ca, and O, were detected by EDX (area marked in red) on both systems FA/GBFS and FA/OPC. Independent of the activator RN or SN, the chemical elements with the higher participation on the selected area were O (28-43%), Si (14-29%) and Ca (3-33%); the elements with the lower participation were Na and Al. The presence of the above-mentioned elements allowed authors to infer the co-existence of reaction products with Si-O-Al-Na bonds and Si-O-Ca-Al-Na bonds arrangements. On the other hand, EDX analysis also indicates the presence of carbon (C) in a high percentage from 8-23%; it is assumed to come from the high content of unburnt material in FA, which has a LOI of 14. 96% (shown in table 1).

Figure 5 shows the ²⁹Si MAS NMR spectra of FA/GBFS (SN (a) and RN (b)) and FA/OPC (SN (c) and RN (d)) alkali-activated binders. Table 5 includes the percentages of participation of each silicate unit for each curing time 28 and 360 days. The values were calculated from the deconvolution of each spectrum following the Gaussian adjustment function; the regression coefficient (R²) varies between 0.96 to 1.

Based on the results from SEM images and NMR spectra, it is confirmed the co-existence of two types of gels in the binary and hybrid systems. These results are in agreement with Gao et al. (21), Wang and Scrivener (22), Mejía et al. (23) and García-Lodeiro et al. (24), who also observed a sodium aluminum silicate hydrate gel (N-A-S-H), calcium sodium aluminum silicate hydrate gel (C,N)-A-S-H), and a modified calcium (aluminum) silicate hydrate gel (C-(A)-S-H) in alkali-activated Ca-rich systems based on fly ash, ground blast furnace slag and Portland cement. N-A-S-H and (C,N)-A-S-H gels results from the activation of FA, the latter in presence of a calcium source; and (C-(A)-S-H) is generated from the activation of GBFS and OPC in the presence of FA. In the traditional hydration of OPC, the final product is a C-S-H gel with a 2D structure that does not contain aluminium.

The spectra on Figure 5 and the values included in table 5 show that SN activator promotes the formation in similar proportions of N-A-S-H, (C,N)-A-S-H and calcium gels and C-(A)-S-H at an early age (28 days), the sum of all Q⁴ (1Al, 2Al, 3Al, and 4Al) resulted in the participation of 50% and 55%, where Q⁴(4Al) units have the lower participation; units Q²(1Al) represents 45 and 41% in FA/GBFS-SN and FA/OPC-SN respectively. On the other hand, the activator RN leads to the formation of C-(A)-S-H gels represented by Q²(1Al) units over a N-A-S-H or (C,N)-A-S-H gels at the age of 28 days, the calculated percentage fluctuated from 76% to 43% for FA/GBFS-RN and FA/OPC-RN respectively. Probably due to the low reactivity of silica in rice husk ash, the alkaline solution initially attacks secondary sources rich in calcium, promoting the formation of calcium gel. From the results, it is proposed that silica from RHA reacts slowly compared to the soluble silica present in sodium silicate.

The NMR spectra associated with binary and hybrid SN and RN type systems show at the age of 360 days, a decrease of the intensity and area of the signals related to Q⁴(4Al) units; also it is observed increase of Q⁴(3Al) percentage compared to the binders at 28 days. This behaviour states that gel (C,N)-A-S-H is enriched in silica over time and decreases the amount of aluminium tetrahedrons in the structure of the tectosilicate. It is inferred that the aluminium species are probably bound in the 2D structure of gel C-(A)-S-H, increasing the cross-linking and the central units Q²(1Al) of the chain. Additionally, it was found that and age of 360 days of curing, Q²(1Al) units in all the binders (FA/GBFS-SN and RN; FA/OPC-SN and RN) increase its participation occupying around of 60% to 70%, these results allow to assume that C-(A)-S-H gel, at extended ages has the majority occupation in the cementitious matrices.

On the other hand, at the age of 360 days, a dissolution of the unreacted species Q⁴(0Al) from FA and RHA, Q⁰ and Q¹ units from slag and Portland cement, was observed. Wang and Scrivener (22) suggest that the dissolution of the species increases the length of the C-(A)-S-H gel chains and densify the structure.

Figure 6 shows that, at an early curing age (7 days), the compressive strength of the pastes is between 20 and 38 MPa. For systems with added GBFS, the type of silica source did not have a significant effect on the mechanical performance of the resulting materials for short aging durations; the difference between the compressive strength of SN and RN was only 2 MPa; for the pastes with added OPC, the difference between the compressive strength of the SN and RN systems was approximately 9 MPa. For both FA/GBFS and FA/OPC systems, the use of RN as alkali activator led to achieved low compressive strength compared to SN systems. This effect can be attributed to the low content of soluble silica present in RN. Previous studies have reported that FA/GBFS binary systems prepared using RHA as a source of silica and higher quality ash have compressive strengths similar to those reported in this study (i.e., 35 MPa for materials after 7 days of curing) (23).

After 28 days, the pastes made with the SN activator (SS + NaOH) had higher compressive strengths than the pastes made with the RN activator (RHA + NaOH). The FA/GBFS-SN system cured for 28 days had a strength that was 96% of the compressive strength of the FA/GBFS-SN pastes cured for 360 days. On the other hand, the compressive strength of FA/OPC-SN at 28 days was 64% of the final strength at 360 days. This behavior indicates that GBFS reacts more quickly in the alkaline medium than OPC, even though the proportion of OPC is only 20% compared to the 30% of GBFS in the systems studied. In the RN systems, the reactive silica present in the RHA (36%) participates in the reaction in the early stages of aging (i.e., 7 and 28 days).

For extended periods of aging (i.e., 360 days), the FA/GBFS-RN and FA/OPC-RN systems have strengths of 65 MPa and 61 MPa, respectively, which are up to twice the strength reported by these mixtures at 28 days. Geraldo et al. (25) reported a significant evolution of the compressive strength from 28 to 90 days when MK-geopolymers are activated using sodium silicate based on RHA. It should be noted that in the SN systems, the high content of soluble silica imparts the material with high strengths of up to 85 MPa at the same curing age. The strengths of FA/GBFS-RN and FA/OPC-RN at 360 days were 13% and 28% lower than SN-activated specimens. A study using silica sources as RHA (88% SiO₂) and microsilica was developed by Luukkoven et al. (12), the authors demonstrated that silica availability in the alkali-activated blast-furnace slag mortar affects compressive strength over time. However, their results were 32-34 MPa at 28 d; these values are comparable to those obtained in the present study.

Passuello et al. (17) evaluated the replacement of SS for RHA in geopolymers based on calcined kaolin sludge and reported comparable compressive strength (<5% difference). These values coincide with the results reported by Bouzon et al. (26) and Villaquirán-Caicedo (9), who evaluated the compressive strength of geopolymer mortars based on fluid catalytic cracking catalyst (FCC) and alkali-activated pastes based on metakaolin, respectively. The results reported for the authors were approximately 41 MPa at 7 days of curing (9, 26). Bernal et al. (27) studied alkali-activated slag binders activated with silicate obtained from the chemical reaction between NaOH and RHA; these pastes reported higher compressive strength than those produced with commercial sodium silicate. It should be noted that all the studies mentioned have used RHA obtained at controlled conditions (600°C), which allowed high percentages of soluble silica (>60%). On the contrary, in the present study, RHA is from an industrial combustor and has 36% of soluble silica in NaOH.

Table 6 shows the five categories analyzed in the LCA study and the impact values associated with each raw material; these values are given for 1 kg of material. Because FA and RHA are classified as “waste,” no inventory of the environmental impacts produced by them exists, and for this reason, FA and RHA are not included in the table. The only values that can be used to calculate the total binder impact of FA and RHA is the energy consumed while preparing FA and RHA; this energy was shown in Table 4.

The values shown in Table 6 are associated with the emission inventories of the raw materials, showing the materials with the most significant individual environmental impact; the results show that NaOH is the component with the highest adverse incidence. In general, the values in Table 6 show that, in the alkaline-activated materials, the alkaline solution is the most polluting component. Other authors, such as Passuello et al. (17), Petrillo et al. (28), and Tempest et al. (29), have also shown the alkaline solution to be the most polluting component of alkaline-activated materials. SS is produced by a liquid furnace process and requires the use of natural resources and high temperatures of approximately 1300°C (Figure 2). On the other hand, sodium hydroxide is produced by electrolysis in a diaphragm cell. Both processes release large amounts of CO₂ into the atmosphere and require large amounts of energy. With regards to calcium sources, the environmental impact of OPC is higher than that of GBFS because of the process used to extract the natural raw materials, the high energy used during the clinkering process, and the CO₂ emissions generated in the same process.

Figure 7 shows the impact of each raw material (represented as a percent) in the four designed binders (Table 3). Because the impact percentages of FA, RHA, and GBFS are lower than that of water, which has an impact percentage between 0.1% and 0.01% (represented by the thin red bar at the top of the columns), the bars assigned to these materials are not visible.

The total impact percentage of the SS in the FA/GBFS-SN and FA/OPC-SN systems for all categories was between 70% and 86%, with the exception being the ozone depletion potential (ODP) category, which contributes to between 29 and 37% of the impact. In this category, sodium hydroxide has a higher impact, between 57% and 70%, due to the raw materials used to produce NaOH; these materials are rich in chlorinated compounds. The emission of chlorofluorocarbon gases (CFCs), which threaten the ozone layer during the manufacturing process of NaOH, also contributes to its impact. In the SN systems, NaOH only contribute to between 8% and 27% of the impact, which depends on the impact category because the amount of NaOH is lower than that of SS in the system (Table 3); however, in the FA/GBFS-RN systems, NaOH has an impact percentage close to 99% due to the absence of a silicate. When OPC is incorporated as the calcium source, the environmental footprint increases.

Figure 8 shows the total results of the impact categories evaluated for each binder. It is observed that, except ODP, the mixes using RN (RHA+NaOH) as activator have lower impacts in AP, GWP, EP, POP and RE. RN systems reduce the environmental impact by 50% to 75% compared to SN systems

However, the ozone depletion potential (ODP) increased up to 34% in RN-activated specimens. In this study, the paste with the greatest impact was FA/OPC-SN, which had a GWP of 514.18 kg∙CO₂∙eq. Specially, GWP is reduced up to 61%; this coincides with Habert et al. (20), who demonstrated that replacing traditional SS with alternative sources of silica substantially reduces the global warming potential of the product. The environmental results reported in this article agree with those reported by Passuello et al. (17), who worked with an alkaline-activated kaolin sludge residue and used RHA as the silica source, authors also concluded that using RHA decreases the environmental impact in several categories. (AP, GWP, EP, and POCP). Villaquiran-Caicedo and Mejía de Gutiérrez (30) reported that the global warming potential emissions were reduced by 47.45% when is replaced 100% SS by RHA in the activator in metakaolin-based geopolymers.

Comparing the sources of calcium in in the system evaluated (OPC and GBFS), the Figure 8 shows that the systems manufactured with FA/GBFS show a reduction between 11% and 42% in all the impact categories, which makes FA/GBFS less impactful than the FA/OPC systems. Therefore, from an environmental perspective, OPC is not a suitable alternative source of calcium. In this case, GBFS is a better alternative source of calcium, which is an industrial by-product.

The mechanical tests and life cycle analysis indicate the feasibility of replacing a traditional silica source with RHA without significantly affecting the mechanical performance of the material. The construction material produced using these systems has added value from the perspective of environmental sustainability and human health.

For construction materials to be deemed sustainable and certified by Cradle to Cradle®, LEED®, among other, waste (including fly ash, slag, and RHA) must be used as the main component of the materials produced, and the impact categories mentioned in this study should be reduced by at least 20% compared to traditional materials. Previous studies show that, in most cases, alkaline-activated materials have less impact in the GWP category compared with OPC-based construction materials; Robayo-Salazar et al. (31) reported a 44.7% reduction in the GWP of concretes made with alkaline-activated natural pozzolan (210.90 kg CO₂ eq/m³) compared to a traditional concrete based on Portland cement (381.17 kg CO₂ eq/m³). Bricks prepared with alkaline-activated materials composed of red clay brick waste had GWP reductions of up to 73% compared to ceramic bricks produced from clay (32).

Based on the results obtained in the life cycle assessment and considering the GWP and RE categories, the FA/GBFS-RN paste was selected as the optimal material for producing the block and paver construction material because it had the lowest GWP (117.06 kg∙CO₂∙eq) and RE (0.30 kg PM_2.5-eq) impact among the four systems studied. The proportions of the mix design of the masonry blocks and pedestrian pavers are included in Table 7.

Figure 9 (a) shows the manufactured block. The masonry element has an excellent finish, the dimensions of the produced block were 30 cm × 15 cm × 11 cm, and the wall thickness between the holes and the surface was 4 cm. Figure 9 (b) shows the geometry of the pedestrian pavers; according to standard NTC 2017, the pedestrian paver is classified as type 2 with an I-shape. The dimensions of the paver were 20 cm × 13 cm × 9 cm, and the inscribed rectangle was 20 × 9 cm (where the nominal length/nominal width ratio was 2.2).

According to the requirements established by Colombian technical standards NTC 4026 and NTC 2017, the masonry block and the pedestrian paver were mechanically and physically evaluated, and the specifications are presented in Table 8. The Colombian standard classifies blocks as either structural or non-structural, depending on the compressive strength. According to this criterion, the masonry material developed is classified as a structural block with a medium weight (1894 kg/cm³) and high compressive strength (16.7 MPa). The paver meets the NTC 2017 requirements. The paver surface showed good wear performance, demonstrating that this material can be placed in high-traffic areas with little risk of wearing. Additionally, the absorption percentage indicates a good response when exposed to the elements.

The results obtained for the block are comparable to studies conducted by Palomo and Fernández-Jiménez (33), who manufactured blocks from an FA/OPC hybrid. Robayo-Salazar et al. (32) developed blocks meeting Colombian technical standards (NTC) from a hybrid system composed of brick waste and 10% Portland cement, and the compressive strength achieved by this hybrid system was equivalent to that of the FA/GBFS-RN binary mixture block, which was 16 MPa. Higher compressive strengths were reported by Kishan and Radhakrishna (7), who activated a binary mixture of FA/GBFS using as activator SS/NaOH to produce masonry blocks and reported a compressive strength of 25 MPa after 28 days of curing. On the other hand, Suksiripattanapong et al. (34) designed a block with municipal sludge and alkaline-activated FA, and the strength was 20 MPa after thermal curing at 75°C.

It is worth noting that few publications have reported manufacturing alkaline-activated pedestrian pavers composed of wastes or alkaline-activated by-products. Robayo-Salazar et al. (32) used a mixture of brick waste and Portland cement and obtained a flexural strength and water absorption of 4.42 MPa and 6.5%, respectively. Kumar and Kumar (35) developed pavers for pavement from an FA/red sludge system and reported a compressive strength of 30 MPa.