Non-traditional and Natural Pozzolans as Precursors for Sustainable Alkali-Activated Binders: Reactivity, Phase Assemblage, and Composition Analysis

Engineering ›› DOI: 10.1016/j.eng.2025.05.003

Article

Author information +

History +

PDF

Abstract

This study evaluates four groups of non-traditional aluminosilicate industrial coal byproducts and natural pozzolans—ground bottom ashes (GBAs), low-purity calcined clays (CCs), volcanic ashes (VAs), and fluidized bed combustion ashes (FBCAs)—as potential precursors for alkali-activated binders. Each group was activated using a mixture of sodium silicate and sodium hydroxide solution under optimized solution parameters. Their reaction behavior, pore solution changes, phase assemblage, and composition were analyzed using various material characterization tools. Results indicate that the reaction rate and total heat are similar to those of conventional fly ash precursors, with total heat release 38%–82% lower than that of Portland cement. Pore solution analyses reveal the formation of typical alkali-activated gel phases, and nuclear magnetic resonance (NMR) revealed calcium sodium aluminosilicate hydrates with aluminum/silicon (Al/Si) ranging 0.07–0.36. A novel reactivity index was proposed using Al-NMR. CCs and GBAs exhibited superior performance compared to other two materials used.

Graphical abstract

Keywords

Alkali-activated ground bottom ashes / Low-purity calcined clays / Volcanic ashes / Fluidized combustion bed ashes / Microstructure characterization

Cite this article

Download citation ▾

Roshan Muththa Arachchige, Shubham Mishra, Jan Olek, Farshad Rajabipour, Sulapha Peethamparan. Non-traditional and Natural Pozzolans as Precursors for Sustainable Alkali-Activated Binders: Reactivity, Phase Assemblage, and Composition Analysis. Engineering DOI:10.1016/j.eng.2025.05.003

登录浏览全文

4963

注册一个新账户忘记密码

1. Introduction

1.1. Alkali-activated materials

Alkali-activated materials (AAMs) are produced by reacting aluminosilicate precursors with an alkaline activator and have been recognized as an eco-friendly alternative to ordinary Portland cement (OPC) due to their low CO₂ emissions (40%–75% less) and minimal natural resource usage [1], [2], [3]. While the adoption of alkali-activated cements in engineering applications remains limited, their scalability is increasingly acknowledged due to their superior durability and sustainability compared to OPC. Fly ash and slag are the most commonly used precursors in AAMs. However, the increased recycling of steel and the decommissioning of coal power plants in favor of greener energy sources have created uncertainty regarding the future availability of these materials [4], [5]. This challenge calls for exploring alternative aluminosilicate sources as sustainable precursors for AAMs. Beyond addressing precursor shortages, the development of AAMs aligns with global sustainability objectives, including waste beneficiation initiatives and net-zero carbon goals in the construction industry. This study explores the potential of non-traditional AAM precursors by evaluating their material availability, reactivity, and broader feasibility in practical applications.

1.2. Non-traditional precursors for alkali activation

Several non-traditional aluminosilicate materials, including ground bottom ashes (GBAs), low-purity calcined clays (CCs), volcanic ashes (VAs), and fluidized bed combustion ashes (FBCAs) have been identified as promising candidates for alkali activation. The availability of other waste streams byproducts, such as high-quality municipal waste incineration and plant-based ashes, is geographically limited, their composition varies significantly, and they are used for non-structural concrete applications [6], [7], [8], [9]. This highlights the necessity of utilizing locally available non-traditional pozzolanic materials as precursors in formulating AAMs.

Non-traditional and natural pozzolans (NNPs) such as CCs, VAs, GBAs, and FBCAs are more widely accessible. Annually, the United States produces approximately seven million tonnes of kaolinitic clay, 14 million tonnes of FBCA ash, 24 million tonnes of bottom ash, and over 600 million tonnes of VAs [10], [11], [12]. Existing literature indicates that these four groups of non-traditional materials exhibit mechanical performance comparable to or better than conventional materials, both as supplementary cementitious materials in systems and as precursors in AAMs [13], [14], [15], [16], [17], [18]. These non-traditional precursor-based AAMs address the growing shortage of conventional precursors by utilizing locally available minerals and aluminosilicate sources.

1.3. Reaction kinetics and gel formation

Pore solution analyses of conventional fly ash, slag, and metakaolin-based AAMs revealed many findings about the reaction mechanism. The pore solution of AAMs is rich in anions like OH⁻, SiO₃²⁻, AlO₂⁻, SO₄²⁻, and cations like Ca²⁺, Na⁺, and K⁺, depending on the type of binder and activating solutions. The high alkaline medium of the pore solution facilitates the cleavage of Si–O–T (T = Si/Al) bonds on the surface of aluminosilicate precursors [1]. Higher alkalinity of the pore solution accelerates the dissolution kinetics. At a given temperature and pressure, the pore solution composition controls the ongoing reaction process and determines the reaction products [19], [20], [21], [22], [23]. In the pore solutions of alkali-activated fly ash and slag pastes, the soluble silicates in the activator result in an increase of Si, Al, calcium (Ca), potassium (K), iron (Fe), and magnesium (Mg) concentrations. However, an increase in temperature leads to a decrease in those concentrations due to precipitation of reaction products [21]. Fig. 1 describes the formation mechanism of the above reaction products. Si–O–T (T = Si/Al) band shift towards low wavenumbers in Fourier transform infrared spectroscopy (FTIR) provides evidence for the polycondensation reaction during alkali activation [24], [25], [26]. This mechanism is exothermic and often evaluated using isothermal calorimetry. The total heat of hydration of AAMs are lower compared to OPC due to the formation of different phases than calcium silicate hydrate (C-S-H) [22].

1.4. Reaction products and their composition

Studies [27], [28], [29], [30], [31] on conventional fly ash, slag, and metakaolin revealed that the reaction products of alkali activation comprise of calcium aluminosilicate hydrates (C-A-S-H), sodium aluminosilicate hydrates (N-A-S-H), and calcium sodium aluminosilicate hydrates (C-N-A-S-H). The Al/Si, Na/Si, and Ca/Si ratios can vary based on the composition of the raw materials. Table 1 describes the properties and differences of the three main reaction products [27], [28], [29], [30], [31].

²⁷Al magic angle spinning-nuclear magnetic resonance (MAS-NMR) investigations of metakaolin-based products determined that reacted samples contain predominantly tetrahedral-Al(IV) (-60 ppm) with trace amounts of octahedral-Al(VI) (-0 ppm). ²⁹Si-NMR suggests that reacted metakaolin contains all five types of Q⁴(mAl) (m implies the number of substitutions of Al to the neighboring Si tetrahedra) present in a three-dimensional (3D) network of N-A-S-H [4]. Studies [32], [33] on binary mixtures of slag and metakaolin suggest the formation of a heterogeneous phase involved in the transformation of the 3D network to C-A-S-H. Aluminosilicate polymers are generally characterized by their Al/Si determined using solid-state Si²⁹-NMR. To provide an accurate and meaningful interpretation, Si-NMR spectra are deconvoluted by fitting the minimum possible number of Gaussian peaks by maintaining consistency with structural constraints for a C-(N)-(A)-S-H gel described by the cross-linked substituted Tobermorite model and Q⁴ aluminosilicate model [34], [35]. Those two approaches suggest three different equations to calculate the Al/Si ratio involving the Qⁿ groups (Table 2 [34], [35]), where n indicates the number of adjacent tetrahedral SiO₄ linked to a specified SiO₄ tetrahedron. The Qⁿ groups are assigned based on the silicate and aluminate environment around a particular silicate unit (Fig. 2).

Thermodynamic models coupled with pore solution analysis suggest various reaction products for AAM systems. High-Ca systems produce C-(N)-A-S-H, hydrotalcite, calcium alumino ferro hydrates, and Mg–Al layered double hydroxide, while low-Ca systems produce N-A-S-H, zeolite, sodalite, calcium carbo/sulfo-aluminate hydrates, hydrogarnets, and stratlingite [6]. In AAMs, the gel structure of C-(N)-A-S-H differs from OPC systems due to lower Ca/Si. This results in longer mean chain lengths (MCL) and the presence of both cross-linked C-(N)-A-S-H gels and non-cross-linked N-A-S-H gels. Molecular simulation (MS) studies show that Al influences the gel structure, enhancing cross-linking, and improving mechanical properties, particularly at high Al/Ca ratios. According to MS, the range of Al/Si in C-A-S-H lies between 0.1 and 0.3 [36], [37], [38]. The transformation mechanism of C-A-S-H to C-N-A-S-H can be elucidated via MSs [7].

1.5. Research significance and innovation

While existing literature provides insights into the mechanical properties of non-traditional AAMs, there is a lack of research dedicated to the microstructural characterization of AAMs using blended NNP precursors. Limited research was conducted on chemistry and microstructure characterization of Ca-blended low purity CCs, Ca-blended GBAs, ternary blends containing VAs, CCs, calcium hydroxide, and FBCAs, CCs, and calcium hydroxide. Thus, a comprehensive exploration of the chemistry inherent in various alkali-activated binary and ternary blends of NNPs is imperative. Such an investigation holds the potential to unveil critical insights for screening novel materials, enhancing mechanical and durability performance, and mitigating costs and CO₂ emissions associated with resultant products.

This study aimed to determine the reaction kinetics, phase assemblages, and microstructural characteristics of NNP-based inorganic polymers. Reaction kinetics analysis was facilitated through isothermal calorimetry, FTIR spectroscopy, and pore solution measurements. Microstructural characterization of the resulting reaction products was analyzed using solid-state NMR, and energy-dispersive X-ray spectroscopy (EDS). By systematically comparing reaction pathways across different precursor groups, this study offers a refined understanding of the evolution of gel structures and phase transformations in non-traditional AAMs.

Unlike previous studies that primarily rely on fundamental composition analysis, this work critically evaluates the local nano/micro-scale interactions of reactive calcium, alumina, and silica in governing phase transitions in blended precursor systems. The mechanisms governing reaction product formation and identifying the range and composition of these products were achieved, attributing specific causes to their occurrence. The present study pioneers a novel Al-NMR-based reactivity index, offering a quantitative framework for evaluating and optimizing precursor performance in alkali-activated systems. This innovation provides a more systematic method for screening sustainable precursor materials. The integration of advanced material characterization techniques facilitated a comprehensive understanding of the chemical and microstructural complexities governing the behavior of these inorganic polymers. Thus, this study highlights the potential of various non-traditional precursors to produce reaction products and microstructures comparable to those of widely used fly ash and slag.

2. Methodology

2.1. Materials

The eleven pozzolans used in this study fall under four major groups namely GBAs, low-purity CCs, VAs, and FBCAs. They are collectively termed NNPs. The X-ray fluorescence spectroscopy (XRF) oxide composition and X-ray diffraction (XRD) mineralogy of the raw materials are presented in Table 3, Table 4. Fig. 3 contains the particle size distribution of the raw materials. Other physicochemical properties can be found in Refs. [17], [18]. A reagent-grade sodium silicate (10.6% of Na₂O and 26.5% SiO₂), sodium hydroxide (99% pure pellets), and calcium hydroxide (99% purity) are used as activators.

Fig.3 illustrates the particle size distribution (PSD) curves of NNPs compared to OPC based on volume percentage as a function of particle diameter. In Fig.3(a), GBAs exhibit D₅₀ values comparable to the D₅₀ of OPC. These GBAs display relatively narrow and uniform distributions, indicating suitable fineness for use as reactive binders. In Fig.3(b), CCs show a wider variation in PSD. CC1 and CC3 are finer than OPC, with D₅₀ values of 9.4 and 9.7 µm, respectively, while CC2 is significantly coarser at 20.8 µm. The finer particle sizes of CC1 and CC3 suggest enhanced potential for pozzolanic activity, whereas the broader distribution of CC2 may reflect variability in calcination or grinding. Fig.3(c) presents the PSDs of volcanic ashes, where VA2 and VA3 are notably finer than OPC, while VA1 closely matches OPC. In contrast, Fig.3(d) shows that FBCAs are considerably coarser than OPC. The particle size characteristics of these alternative precursors vary widely, and their suitability for binder applications depends significantly on their fineness and uniformity relative to OPC.

2.2. Sample preparation

The activating solution used is characterized by three solution parameters: silica modulus, the mass ratio between equivalent SiO₂ and Na₂O, sodium oxide content, the mass percentage of equivalent Na₂O per binder mass, and solution to binder ratio (s/b). The binder compositions and solution parameters mentioned in Table 5 were the results of an optimization process to achieve mortar strength of at least 25 MPa in 28 d [16]. The 7 and 28 d compressive strength of 50 mm × 50 mm × 50 mm alkali-activated mortar cubes are included in Table 5. Calcium hydroxide (CH) was used as a strength enhancer as most of NNPs lack Ca which is a major component of C-A-S-H. CC2 was added to increase Al content of VAs and FBCAs as they showed less reactivity.

Paste samples were prepared in 5 cm × 10 cm cylinders following ASTM C305 procedures. The specimens were then cured at (25 ± 1) °C and relative humidity greater than 90%. After corresponding curing periods, the hardened paste samples were cut into 3 mm and 2.5 cm thick specimens (used in scanning electron microscope (SEM)) using a slow-speed glycol-lubricated diamond precision saw. Then, the specimens were quenched in propanol for at least a week to stop further hydration by removing water. The 3 mm thick samples were dried in an oven at (40 ± 1) °C for 24 h then ground and sieved through a 75 μm sieve to acquire a fine powder. The 2.5 cm thick samples were impregnated in epoxy and polished to obtain a smooth surface suitable for SEM studies.

2.3. Experimental methods

2.3.1. Isothermal calorimetry

The heat evolution rate and the total heat were determined using a isothermal conduction calorimeter (TA instruments, New Castle, DE, USA) as per ASTM C1679–17. Measurements of heat evolution were performed on paste specimens activated at 25 °C. The activating solution was injected to 5 g of dry precursors in an admix ampoule for insitu mixing. The mixture was manually stirred for 1 min followed by 30 s rest and again stirred for 1 min. Mixing of the precursor and the activating solutions was done using the same procedure for all the materials. The heat evolution data collection started after reaching the thermal equilibrium and continued for 14 d. For each NNPs used in the experiment, only a single sample was tested. This decision was based on prior research and data that showed minimal variation between duplicate samples of the same material.

2.3.2. Diluted pore solution analysis

The precursors and the solutions were mixed at a solution to binder ratio of four. The high s/b paste was prepared in a hard polypropylene test tube by mixing 2.5 g of binder with 10 g of solution. The test tubes were sealed promptly and cured at 50 °C until sampling time. The sampling times were 6, 12, 24, and 72 h. After the specific curing times, the solution was vacuum filtered, and ionic concentrations were measured using inductively coupled plasma mass spectroscopy (ICP-MS). The method produced a solution analogous to the actual pore solution of activated pastes, which helped better understand the reaction kinetics. However, the ionic concentration could be much higher in low solution to binders ratio actual systems. Two replicates were tested from each NNP.

2.3.3. Fourier transform infrared spectroscopy

The FTIR technique used in this study was consistent with ASTM C494/C494M-05a (ASTM 2005). The sample preparation was similar to that of thermogravimetric analysis (TGA), except that the paste samples were cured for 6 h, and 1, 3, 7, and 28 d before testing. One milligram of the solid powdered sample was combined with approximately 300 mg of KBr and then thoroughly ground in a mortar and pestle. Sufficient ground sample (-10 mg) was placed into the pellet die to cover the bottom and pressed at 35–70 MPa pressure. FTIR analysis was performed using a Thermo-Fisher Nicolet iS20 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). Spectral analysis was performed over the 400–4000 cm⁻¹ with a spectral resolution of 4 cm⁻¹. Five spectra were collected for each sample, of which three replicates were used so that 15 spectra were obtained from one sample and averaged.

2.3.4. Solid-state nuclear magnetic resonance spectroscopy

The sample preparation was similar to that of TGA and 28 d old samples were used. Solid-state ²⁹Si and ²⁷Al MAS-NMR experiments were conducted using an AVIII 500 MHz solid-state NMR spectrometer with 4 mm Bruker 4.0 mm HX probe (Bruker Biospin GmbH, Rheinstetten, Germany) at 12 kHz spinning speed. The ²⁹Si chemical shifts were referenced to neat tetramethylsilane (TMS) at 0 ppm. All solid-state ²⁹Si NMR spectra were acquired with a 5 s acquisition delay and 256 scans. The ²⁷Al chemical shifts were referenced to Al(NO₃)₃, 1 mol·L⁻¹ at 0 ppm. All solid-state ²⁷Al NMR spectra were acquired with a 0.1 s acquisition delay and 1024 scans. One replicate weighing 20–50 mg was used for acquiring data.

2.3.5. Energy dispersive X-ray spectroscopy

The hardened 28 d old paste samples used for SEM/EDS analyses were 2 cm × 1 cm × 1 cm in size, soaked in isopropanol to stop further hydration, then epoxy-impregnated, and polished. They were sputter-coated with 60% gold and 40% palladium. EDS analyses were conducted under a JEOL JSM-7900F SEM (JEOL Ltd, Peabody, MA, USA) in backscattered electron mode with acceleration voltage of 15 kV, probe current of 10 nA, and a working distance of -10 mm. The elemental ratios were calculated using at least 200 EDS data points. One replicate was used from each NNP for the analysis.

3. Results and discussion

3.1. Heat flow rate and total heat evolved

Fig. 4 illustrates the heat evolution rates and cumulative heat for the activated NNP pastes. In contrast to the typical five-stage heat evolution profile observed in OPC systems, the heat evolution for activated NNPs exhibited a single prominent peak, which formed immediately upon mixing the precursors. The dissolution and induction periods were indistinguishable due to the rapid early onset of reaction product formation. This behavior closely resembles that of activated fly ash systems [33].

Due to their slow reaction kinetics, low s/b, and reduced activator concentration, GBAs had the smallest first peak in heat flow compared to other NNPs (Fig. 4(a)). The total heat emission of GBAs was 74%–82% lower than OPC (Fig. 4(b)). This change among GBAs can be attributed to the CaO content of the raw materials. GBA3 with the highest CaO content manifested the highest total heat among GBAs.

At 160–240 J·g⁻¹, CCs showed the highest overall heat release (Fig. 4(c)) among all activated NNPs (38%–58% lower than OPC). This suggests that the most reactive class of materials is CC. CC3 with lower average particle size and higher amorphous content was the most reactive precursor of the CC group. Eventhough CC2 had the highest amorphous content out of the CCs, due to low activator parameters it showed low total heat. The heat flow peak for CC1 paste was the highest during the rapid precipitation and condensation period probably due to higher activator parameters (Fig. 4 (a)).

Compared to the other two VAs, VA3 exhibited a slightly higher initial heat flow peak (Fig. 4(c)). Among FBCAs, FBCA2 emitted the highest heat after 14 d due to its higher CaO content and lower average particle size. The total heat releases of VAs and FBCAs were about 80–125 J·g⁻¹, a 68%–79% reduction compared to the OPC (Fig. 4(d)). Among VAs, VA1 paste had the highest cumulative heat, probably due to its higher alkali content in raw material form.

CCs exhibited total heat comparable to alkali-activated slag and metakaolin [39], [40], [41], [42], [43]. The total heat of VAs and FBCAs resembled that of activated Class C fly ash, while GBAs demonstrated similarities with Class F fly ash in total heat emission [11]. AAM systems exhibit distinct reaction mechanisms involving dissolution, gelation, condensation, polymerization, and growth. Unlike the sequential hydration stages of OPC, alkali activation processes occur concurrently, resulting in a single main heat flow peak. Both the activating solution parameters and the reactivity of the precursor material influence the total heat of hydration of AAMs.

Johnson–Mehl–Avrami–Kolmogorov (JMAK) equation is widely used for modeling of early-age nucleation and growth in cement hydration and alkali activation [44].

(4)

X (t) = 1 - e x p (- K_{a} t^{m})

where X(t) is the degree of product formation, K_a is the rate of product formation, t is the time, and m is the Avrami exponent. By differentiating the above equation with respect to t, following equation is obtained.

(5)

R (t) = A m K_{a} t^{m - 1} e x p (- K_{a} t^{m})

where R(t) is the rate of heat evolution, and A is a scaling factor to bridge between the hydration degree X(t) and rate of heat evolution R(t). Early crystallization kinetics of NNP pastes were modeled using isothermal calorimetry heat flow rate data following Eq. (5). Table 6 contains the early crystallization parameters determined using the JMAK model. The simulated heat flow curve compared to the experimental heat flow curves are given in Fig. S1 in Appendix A. Avrami equation has achieved a very satisfactory data approximation for all samples with higher coefficient of determination (R²) values. Although GBAs have the lowest rate of product formation among the materials, NNPs have higher hydration rates than OPC, fly ash, and slag [45], [46], [47]. Avrami exponents of the NNPs are closer to two (similar to conventional materials), which indicates two-dimensional (2D) nucleation and growth of products [48], [49].

Fig. 5 shows a proportional relationship between the bound water at 14 d calculated using TGA [50] and the total heat of hydration. The total heat indicates the amount of reaction products, while the bound water percentage is the water chemically incorporated into the reaction products. This correlation is also common in other traditional AAMs [51], [52], [53], [54]. The results indicate that water incorporation into NNP pastes ranges from approximately 7.5% to 12% by mass. The optimal solution-to-binder ratio directly affects the bound water in the reaction products. Higher optimal s/b ratio would increase the bound water which is evident among CCs. Moreover, C-A-S-H gels tend to incorporate more water than N-A-S-H gels due to their layered structure, the presence of calcium ions, and their ability to retain more interlayer water [29]. The structural differences between these two gel types are fundamental to their differing water incorporation capacities. This is clear in GBAs and FBCAs as they have varying CaO content.

3.2. Pore solution analysis

This section covers the variation of ionic concentrations of the solution extracted from NNP paste activated with high solution-to-binder ratio. The activating solution is rich in sodium cations, silicate anions, and hydroxyl anions. Thus, the pore solution is initially rich in those ions. The variation of concentartions of sodium ions([Na]) and silicon ions ([Si]) were observed to be similar, although the magnitudes of the concentrations were different (Figs. 6(a) and (b)) due to the charge balancing effect. The initial [Na] and [Si] of the NNPs were different in the activated NNPs as different solution parameters were used (Table 5). The [Na] and [Si] of activated GBAs decreased slowly, implying a low rate of product formation, while the [Na] and [Si] of CCs reduced rapidly, indicating rapid product formation in CCs. In VAs and FBCAs, the [Na] and [Si] reduced moderately. After 3 d, the [Si] of both CCs and GBAs approached zero. VAs and FBCAs exhibited slower reactivity, resulting in the retention of [Na] and [Si] in the solution.

The [Al] of the solutions increased with time for 24 h followed by a subsequent decline as shown in Fig. 6(c). CCs and GBA1 had higher [Al]. The aluminates are slowly incorporated into the product formation because of their low solubility compared to the silicates. The concentration of calcium ions ([Ca]) in the pore solution of NNPs remained near zero as they act as nucleation sites and promote the rapid growth of C-A-S-H [55], [24], [56], [57], [58]. The [Ca] of GBAs and CC2 increased up to 24 h, then, decreased, reaching near zero values by the end of 3 d (Fig. 6(d)).

In comparison to Na and Si, Mg, Ca, Fe, Al, K, and sulfur (S) were present as trace elements. The [Fe] of the NNPs was minimal (Fig. 7(a)). GBA1 had the highest [Fe] due to the presence of haematite. The iron-containing compounds generally do not participate in alkali activation reactions. However, due to the high pH medium, iron hydroxides tend to precipitate with time, eventually decreasing the [Fe]. The [K] of the solutions followed the same trend as [Al] as shown in Figs. 7(b). The temporal increase suggests that Mg²⁺ ions do not actively participate in the formation of reaction products within the 72 h (Fig. 7(c)). As far as sulfur, usually found in the form of sulfates, only FBCA exhibited some in the product, while amounts were negligible in the other NNPs (Fig. 7(d)). Sulfates do not participate in alkali activation reactions but remain in the solution, balancing the charge with other cations like Na⁺, K⁺, and Mg²⁺.

Solution analysis revealed several important details about how different ionic concentrations varied with time during alkali activation. Na, Si, and Ca were very reactive, and their concentrations in the solution were reduced rapidly. Silicates are slowly released into the solution in less reactive materials like VAs and FBCAs. K⁺ and aluminates from the binder dissolved slowly into the solution and started reacting with alkali-activated gels. Traces of Mg, S, and Fe were released into the solution but did not participate in any reactions. However, the reaction kinetics related to the original pore solution could be different with respect to the time domain due to the solution dilution. The pore solution chemistry of fly ash and slag AAMs follows the same trend as in NNPs. However, K⁺ incorporation and aluminate dissolution are not very significant [21].

Fig. 8 shows the variation in pH of the pore solution with time. The zeroth reading represents the pH of the activating solution. The pH increases as soon as the precursor comes in contact with the solution, probably due to the rapid dissolution of added CH and alkalis in the binders. All materials continued to maintain the achieved pH beyond 7 d. The activating solution parameters of VAs and FBCAs were higher than those of CCs and GBAs. However, the pH of the activating solution of VAs and FBCAs were lower, implying that higher the silicate concentration, the fewer hydroxide ions are released into the solution. VAs and FBCAs had lower overall pH values. Conventional fly ash and slag systems do not show a significant short-term increase in the pH but a slow pH decrease with time [21], [59].

3.3. Gel formation and reaction kinetics

Fig. 9 depicts the FTIR spectra of raw and 28 d old activated NNPs. The FTIR transmittance peaks were identified based on their corresponding wavenumbers (Table 7). Similar trends in peaks are visible in other aluminosilicates, but their wavenumbers and intensities differ based on reactive content [60], [61], [62]. The major peak visible in all the materials corresponding to wavenumbers between 1010–1080 cm⁻¹ is identified as polymerized silica and alumina. This peak results from the vibrational modes of Si–O–T (T = Si/Al) bonds. Si–O–T (T = Si/Al) bonds are present both in reacted and unreacted aluminosilicates. The reaction products of alkali activation in NNPs are primarily polymerized silica and alumina, which are binding agents in the AAM paste, resulting in the strength metrics. The width of a Gaussian peak commonly described by the parameter full width at half maximum (FWHM) reveals the crystallinity of the compound identified by the peak. Lower FWHM indicates higher crystallinity and vice versa.

NNPs revealed interesting differences in their characteristics. The polymerized silica peaks of GBAs showed a clear difference between raw NNPs and the activated NNPs (Fig. 9(a)). The activated GBAs had narrower peaks, implying that the silica and alumina of the reacted GBAs are polymerized into a less amorphous structure than the raw GBAs. The next prominent peak (3440 cm⁻¹) corresponds to bound water. The activated GBAs had higher bound water content than raw GBAs, implying the incorporation of water in alkali activation reaction. The other identified minor peaks correspond to various vibrational modes of silicates, aluminates, carbonates, and water. The polymerized silica and alumina peaks of raw and reacted CCs showed some material-specific changes (Fig. 9(b)). Raw CC3, the clay with the highest percentage of kaolinite, exhibited sharp peaks corresponding to alumina and silica vibrational modes in kaolinite. Raw CC3 had the highest peak, while raw CC1 had the lowest. The intensity of the major peak of CC1 increased after alkali activation. No drastic changes were seen in CC2 after the reaction. The size of the peaks corresponding to bound water increased after the reaction. The raw VAs had the highest peaks corresponding to polymerized silica and alumina compared to the activated VAs (Fig. 9(c)). Considering the higher XRF oxide composition of silica in VAs, it can be assumed that the peak with the wavenumbers from 1010–1080 cm⁻¹ represents mainly the Si–O–Si bond vibrations. As they contain less reactive alumina, VAs are not very reactive precursors despite the high amorphous contents of raw materials. Activated VAs had less intense silica peaks due to the ternary blending of the VAs with CC2 and CH to produce compressive strength without heat curing. Nonetheless, activated FBCAs had higher polymerized silica and alumina peaks (Fig. 9(d)). In addition to the other minor peaks corresponding to aluminate, silicate, and water, the carbonate peaks were more visible in FBCAs.

The visual inspection of FTIR spectra of both raw and reacted NNPs provides several insights into the chemistry and reaction kinetics of the materials. When the raw materials were alkali-activated, their FWHM and peak intensity changed at the major peak, corresponding to the vibration of Si–O–T (T = Si/Al) bonds. This prominent peak rolled towards the left (decreasing wavenumber) once the raw materials were activated, implying that the bond strength of the reaction products was less than that of the raw materials containing unreactive minerals. The peaks corresponding to bond vibrations of water molecules were prominent in all activated NNPs. Raw CC3 exhibited a sharp peak between 3500–3750 cm⁻¹, corresponding to the bound water of kaolinite minerals. After the activation, the bound water of other reaction products increased, and the relevant peak became larger. Several other minor peaks were visible in the FTIR spectra of all the materials, which can be traced back to the bond vibrations of other compounds.

The temporal changes of the intensity and wavenumbers of the peak corresponding to the Si–O–T (T = Si/Al) vibration (1010–1080 cm⁻¹) of the alkali-activated NNPs are given in Fig. 10. The general trend of the change of peak intensity is that the higher peak intensity of the Si–O–T (T = Si/Al) of raw materials drops once the NNPs are introduced to the activator and then gradually increases with time. The peak intensity of GBAs is very low at the beginning compared to CCs because of the abundance of Si–O–T (T = Si/Al) bonds in clay minerals available in CCs (Fig. 10(a)). The 28 d peak intensities of CC1 and CC3 were the highest, suggesting that those two materials produced more products with Si–O–T (T = Si/Al) bonds. The fact that CC1 and CC3 were activated with an alkaline solution with higher solution parameters made them more reactive and produced more reaction products. The same trend for peak intensity change was also observed in VAs and FBCAs (Fig. 10(b)). VAs, with higher silica XRF oxide composition, had the highest 28 d peak intensity out of all the NNPs.

The wavenumber of the FTIR peaks measures the frequency of the infrared radiation absorbed/transmitted by the molecular bonds. It tells how much energy is needed to cause a vibration in a molecular bond. The peak position of the FTIR spectra was plotted as a function of curing time, as displayed in Figs. 10(a) and (b), respectively. The peak of the primary Si–O–T (T = Si/Al) band carries vital information on the formation and structure growth in alkali-activated NNPs. Raw NNPs have higher wavenumbers. Once the raw materials are dissolved in the alkaline medium, the wavenumber dropped drastically and slightly decreased with time. Well-polymerized structures require IR waves with higher wavenumbers to cause bond vibration. Raw NNPs have highly polymerized aluminosilicates, which dissolve in a highly alkaline medium and restructure into Na⁺ and Ca²⁺ containing aluminosilicate polymers that are not densely polymerized. GBAs and CCs had a higher change in wavenumbers, indicating their higher dissolution, while VAs and FBCAs exhibited lower change in wavenumbers even after dissolution, proving their less reactivity.

The above results imply that the Si–O–T (T = Si/Al) bonds in the raw materials break down once an alkaline medium is introduced, leading to the formation of aluminosilicate reaction products. The dissolution process was fast during the first 6 h and product formation started immediately. Similar kinetics were seen in CH-blended metakaolin AAMs [16]. The aluminosilicate structure of the reaction product was completely different from the reactant materials. The polymerization of the structure progressed over time and the quantity of the reaction product increased. The polymeric structure of the reaction product was not densely packed compared to the minerals in the raw material. However, the degree of polymerization increased while the unreacted grains dissolved with time.

The Si–O–T (T = Si/Al) vibration peak can further be deconvoluted into sub peaks representing residual vibration modes corresponding to Si-Qⁿ groups and the area under the sub peaks can be used to calculate Al/Si ratio of the reaction products (Fig. S2 and Table S1 in Appendix A). However, Q-FTIR is not an accurate and reliable method of calculating Al/Si due to the low intensity of the peaks representing amorphous reaction products.

Fig. 11 indicates the temporal changes of the peak intensity and wavenumbers of the FTIR spectral peak corresponding to the vibration of bound water. The same variation pattern was seen in majority of the materials. During the first 6 h, the peak intensity and wavenumber showed an initial increase, followed by a drop, and eventually a gradual recovery. The peak intensity is an indication of the quantity of water. The raw materials had very minimal bound water content. The peak intensity increased once the dissolution with the aqueous alkaline medium was initiated. With the hydration, as the free water molecules turn into bound water, the peak intensity decreases and becomes constant (Figs. 11(a) and (b)).

The FTIR peak with wavenumber 3445–3457 cm⁻¹ represents the capillary water trapped in the pores (Fig. 11(b)). After 1–6 h of dissolution, all the materials exhibited FTIR peak 3445–3457 cm⁻¹. Then, most of the capillary water was converted to chemically bound water. The alkali-activated products consume capillary water in the alkaline medium for charge-balancing purposes [69]. Further polymerization of the reaction products restricts the degree of freedom of water molecules. The stretching and bending of bound water correspond to the peaks in the range 3430–3440 cm⁻¹. The increase of the wavenumber from 3–28 d implies that the degree of freedom of bound water reduces due to polymerization and condensation of the reaction products.

3.4. Aluminum sites and degree of reaction

The coordinated aluminum sites in an aluminosilicate gel provide information about the degree of reaction of the material. ²⁷Al solid-state NMR spectroscopy is used to quantify the aluminum sites (Fig. 12). Due to the paramagnetic effects caused by hematite, a certain percentage of error can be associated with the calculations related to GBAs, CC1, and FBCA2. GBAs had smaller peaks than other NNPs (Fig. 12(a)). The peaks of reacted GBAs appeared more intense than those of raw GBAs.

Pure kaolinite has predominantly VI-coordinated aluminum sites (Al(VI)). Due to calcination, those sites turn into metastable Al(V) sites [70]. The coordination number of alumina decreased from 6 to 5 to 4 from raw kaolinite to calcined kaolinite to activated CC [32]. The intensity of peaks of activated CC was sharp compared to the other AAMs, as shown in Fig. 12(b). The peaks corresponding to Al(VI) and Al(V) were quite visible and more prominent in CCs compared to other materials. CC3 had the highest peaks out of all NNPs due to a higher percentage of kaolinite.

The peak intensities of the reacted VAs were weaker than the raw VAs (Fig. 12(c)). Raw VAs mostly consist of Al(IV) sites rather than reactive Al(V) sites, revealing why VAs are difficult to activate. Raw VA1 had the highest Al(IV) peak out of all VAs. The peaks observed in FBCAs were moderately high compared to other materials (Fig. 12(d)). Activated FBCAs showed more intense peaks of Al(IV) sites than those of the raw materials.

The broad Al NMR peaks in raw NNPs resulted from the highly disrupted geometry of all three aluminum sites. During the alkali activation reaction, pentahedral (Al(V)) and octahedral (Al(VI)) sites were converted to tetrahedral sites (Al(IV)). The reacted NNPs contained predominantly Al(IV) (-60 ppm) with trace amounts of Al(VI) (-0 ppm) and Al(V) (-30 ppm). Al NMR indicated that CC is the highest reactive of all the NNPs. The abundance of more Al(IV) sites instead of Al(V) sites in the raw materials is one of the reasons for the lower reactivity of VAs and FBCAs.

The area under the Al(IV) peaks of raw and activated NNPs was calculated. The ratio between the change that occurred in Al(IV) area after alkali activation and the Al(IV) area of the raw NNP is termed reactivity index and presented in Fig. 13(a). In the GBA series, raw GBA1 had the least amount of Al(IV) while raw GBA3 had the highest. Activated CCs produced the highest amount of Al(IV) sites. In the VA series, the raw VAs had the highest Al(IV) amount while activated VAs produced relatively small amounts. The FBCA series exhibited moderate amounts of Al(IV) comparable to GBAs. In comparison, CCs recorded the highest reactivity; the GBA and FBCA series were much lower, and the VAs were the lowest. Fig. 13(b) depicts the correlation between the reactivity calculated using Al-NMR vs. the 28 d compressive strength of the NNP mortars. The two variables correlates well with correlation coefficient (R) of 0.79 and R² of 0.69. The F-statistic (Fisher’s statistic, used to determine if there is a statistically significant relationship between the dependent and independent variables) is large enough that probabiity value (p-value) is almost zero. These results clearly ascertain that the reactivity with respect to the compressive strength can be determined through Al-NMR.

Reactivity with respect to the area under Al(IV) peak has significant implications for understanding the role of aluminate forms in alkali activation. The tetrahedral form is very stable, but the pentahedral form is very reactive. The reactivity of the CCs indicated that calcination is responsible for turning Al(IV) to Al(V). As a result, using NMR to test the reactivity of aluminates in NNPs or other materials in the future will prove to be a robust screening criterion for choosing a potential new precursor.

3.5. Type and composition the gel

Fig.14 shows the ²⁹Si NMR spectra of both raw and activated NNPs. In this study, the Qⁿ(mAl) notation is used to describe the chemical bonds in the local structure of the resonating Si nuclei, where the subscript n indicates the number of adjacent tetrahedral SiO₄ linked to a specified SiO₄ tetrahedron, and the m implies the number of substitutions of Al to the neighboring Si tetrahedra [31], [32], [33], [34], [35]. The NMR signals appear weak in the GBA series due to the paramagnetic effects caused by hematite present in the raw materials (Fig. 14(a)). GBA1, with the highest haematite content, produced a spectrum with the lowest resolution, indicating the possibility of a higher error percentage. Compared to the spectra of unreacted NNPs except VAs, the spectra of activated materials possess more intense signals. This can be taken as evidence that alkali activation rearranged the aluminate and silicate into a well-ordered polymerized structure. The major peaks moved towards increasing chemical shifts in all the NNPs after alkali activation. Municipal waste ash and fly ash AAM systems demonstrated the similar weak NMR signals and chemical shift change after activation [6], [71], [72].

CCs exhibited strong NMR signals compared to GBAs due to the presence of aluminosilicate networks in clay minerals (Fig. 14(b)). Raw VAs had the highest intensity of NMR signals due to the presence of more than 70% SiO₂ oxide composition (Fig. 14(c)). FBCAs had the same common trends as GBAs and CCs but stronger signals than GBAs (Fig. 14(d)). The presence of various minerals in FBCAs and GBAs is a potential reason for the overall weak NMR signals.

NMR spectra of each NNP were deconvoluted into sub-peaks, and the sub-peaks were attributed to corresponding Qⁿ groups based on the chemical shift. A sample deconvoluted NMR spectrum is given in Fig. 15. All the raw NNPs show evidence of having Q⁴ and Q³ silicate networks of aluminosilicate minerals. In addition, amorphous silica has Q⁴ environments as well. The crystalline mineral forms do not react easily. However, when an amorphous silica gets activated by a highly alkaline medium, it starts reorganizing, incorporating aluminates into its structure. The shift towards the left after the reaction seen in Fig. 15 indicates that the products with Q²(1Al) and Q³(nAl) became prominent in the alkali-activated NNPs.

The area under the deconvoluted peaks is attributed to characteristic Qⁿ groups, which determine the silicate and aluminate environment prevailing in the alkali-activated gels (Table 8). C-A-S-H has a single chain, non-cross-linked structure with Q¹, Q², and Q²(1Al) aluminosilicate environment. Aluminosilicate hydrates with both Ca²⁺ and Na⁺ cations tend to form a cross-linked/branched polymeric structure with Q³(nAl) groups, while N-A-S-H has Al-rich Q⁴(nAl) polymer network. These Qⁿ groups were used to calculate the Al/Si ratio of the inorganic polymers identified in alkali-activated NNPs.

Fig. 16 graphically represents the Al/Si ratio of the reaction products of NNP-based AAMs. The Al/Si varies throughout the compound, heterogenous AAM gel due to the presence of mainly two different cations. Energy dispersive X-ray spectroscopy data discussed in the next section provides a range of values for Al/Si. Minimum and maximum values of Al/Si determined using EDS are used as a benchmark to verify the Al/Si calculated using NMR. The main reaction product in reacted GBA3 is C-A-S-H due to the higher abundance of CaO (Fig. 16(a)). It has the highest Al/Si ratio, revealing that the most reactive alumina is incorporated in the C-A-S-H gel, resulting in an Al-deficient N-A-S-H gel form in reacted GBA3. In GBA1 and GBA2 precursors, most reactive alumina is consumed to form N-A-S-H during activation, whereas a minor fraction contributes to the cross-linked product (C-N-A-S-H) and C-A-S-H in the resulting AAMs. This product formation was governed by the CaO composition present in the raw GBAs, making Ca content in a precursor an important parameter that decides the type and composition of the gel.

The Al/Si of N-A-S-H is high in CCs (Fig. 16(b)). Due to the lower solution parameters used in activation, CC2 has lower Al/Si in N-A-S-H. In CC2, most of the aluminates were invested in forming C-N-A-S-H. The Al/Si of C-A-S-H in CC1 and CC2 were the lowest.

N-A-S-H and C-N-A-S-H had higher Al/Si compared to C-A-S-H in VAs (Fig. 16(c)). The Al/Si of N-A-S-H formed in alkali-activated FBCAs were high (Fig. 16(d)). FBCA2 with higher CaO content exhibited the highest Al/Si in C-A-S-H between FBCAs. C-N-A-S-H of FBCAs had a lower Al/Si ratio.

The Al/Si of C-A-S-H (0.05–0.4), C-N-A-S-H (0.1–1.0), and N-A-S-H (0.08–1.5) differs considerably in conventional AAMs [62], [73], [74], [75]. The Al/Si ratios observed in NNPs fell within these ranges.The type of gel formed in NNPs depends on the chemical composition of the raw material and the solution parameters. Once activated Ca-rich NNPs like GBA3 and FBCA2 formed more C-A-S-H. The rest of the materials formed more N-A-S-H and C-N-A-S-H. The Al/Si ratio of C-A-S-H gel components of most of the materials except high calcium GBA3 and FBCA2 is low. Due to the rapid formation of calcium silicate hydrate, aluminate incorporation into the polymer network is slow. Thus, C-A-S-H is deficient in aluminates.

3.6. Elemental composition and gel microstructure

Fig.17 depicts box and whisker plots of the EDS elemental ratios of the alkali-activated gel of NNPs. GBAs had a higher Na/Si ratio than the other two elemental ratios, an almost similar Al/Si ratio, and an increasing Ca/Si ratio from GBA1 to GBA3. The higher Ca/Si ratio of GBA3 suggests that it produced Ca-rich C-A-S-H and C-N-A-S-H. The Ca/Si ratio of GBA2 is lower than 0.2, suggesting that the reaction products of GBA2 are rich in N-A-S-H and C-N-A-S-H. The lower Ca/Si ratio and higher Na/Si ratio of GBA1 suggest that it produced more N-A-S-H.

The highest Al/Si ratio was recorded in CCs, as they contain Al-rich clay minerals. CCs had comparable Na/Si ratios with GBAs and Ca/Si ratios less than 0.2. The higher Na/Si ratios suggest that CCs prominently contain N-A-S-H. CC3 had the least variance in the data set out of the CCs. This implies that CC3 has more consistent gel chemistry.

The elemental ratios of the alkali-activated VAs were similar across different materials. The highest ratios of the VAs were Na/Si ratios, which approached 0.3; the lowest were the Ca/Si ratios, and the intermediate Al/Si ratios were around 0.2. This suggests that VAs produced a majority of N-A-S-H. The two FBCAs showed different trends in their Ca/Si ratios and Na/Si ratios, despite having similar Al/Si ratios. FBCA1 had a higher Na/Si ratio and a lower Ca/Si, indicating a higher proportion of N-A-S-H, while FBCA2 had a higher Ca/Si ratio than Na/Si ratio, suggesting dominant C-A-S-H formation.

The elemental ratios of the alkali-activated gels depend on the material chemistry and the solution parameters of the activating solution. Materials with higher Ca and Al contents had higher Ca/Si and Al/Si ratios. The activating solution primarily controls Na/Si of the gels. A study on the same NNP gels revealed that the higher the elemental ratios, the higher the micromechanical properties will be. The reduced modulus and the microhardness of NNP gels range from 12–34 and 0.7–2.0 GPa, respectively [76]. Another study on alkali-aggregate reaction (AAR) susceptibility of NNP mortars and concrete revealed that NNPs exhibit superior AAR resistance probably due to higher elemental ratios observed [77], [78]. Furthermore, a proportional relationship exists between the elemental ratio and the compressive strength of NNP, and it depends on the major reaction product (Fig. 18).

Fig. 19 contains representative back-scattered SEM images of some selected alkali-activated samples with marked EDS points (Points 1–7). Light gray color particles represent the unreacted grains and minerals, while the dark gray matrix represents the reacted alkali-activated gel. The SEM images of reacted GBA3 and VA2 exhibited a higher percentage of unreacted mineral grains than those of CC3 and FBCA2. The SEM image of CC3 showed evidence of unreacted kaolinite with sheet-like morphology. The white areas represent the oxides of the d-block element such as anatase, hematite, and magnetite. No morphological differences pertaining to C-A-S-H, N-A-S-H or C-N-A-S-H could be seen in the alkali-activated gel areas, even though EDS showed different elemental compositions. Thus, the elemental composition calculated using EDS points was used to identify the differences in the gel area. The elemental composition of the marked EDS points is given in Fig. 20. The major difference was noted to be in the composition of the two cation species Ca and Na. Point 1 in GBA3 and Point 7 in FBCA2 had a higher Ca composition, showing the presence of a Ca-dominant gel. The Point 5 in VA2 had the highest Na composition. Most of the points exhibited a higher composition of Na than Ca. The points in CC3 had a higher Al composition. The EDS point data provided evidence for the existence of most C-N-A-S-H gels in reacted NNPs. The elemental composition of the materials varied significantly across different points, suggesting that the resulting gel matrix is a complex and heterogeneous silicate hydrate, dominated by either calcium or sodium.

The pore structure is part of the microstructure and pore diameters of NNPs ranged from 20 Å (2 nm) to 200 Å (20 nm). Pores with a diameter of around 90 Å (9 nm) were notably abundant. Nevertheless, a clear trend indicated an increase in pores with smaller diameters over curing time, reflecting pore refinement [13].

3.7. The role of Ca in the reaction mechanism

One of the main objectives of this study was to identify the role of Ca in NNP AAM systems. Due to the higher charge, Ca²⁺ is more reactive than Na⁺ in the alkali activation process. Isothermal calorimetry proves that high-Ca systems produce more heat. The added CH dissolution increased the pH of the pore solution. ICP-MS, FTIR, and TGA data proved that all the added CH reacted rapidly. A study proposes that prenucleation clusters/precursor particles absorb OH⁻, Na⁺, and Ca²⁺ ions and desorb aluminates and silicates due to the high pH of the activator. These ions transit through the CH layer (transformed into a complex) and initially generate C-(A)-S-H type gels. As the network-forming elements diffuse out of the CH layer, using the bridging units of the C-(A)-S-H, C-(N)-A-S-H are formed [79]. This same mechanism is proved by the Al/Si of reaction products calculated using NMR and EDS. High-Ca materials attract more aluminates, increasing the Al/Si of C-(N)-A-S-H. JMAK model suggests the immediate 2D nucleation and growth which is characteristic to C-A-S-H. Ca²⁺ ions are essential as nucleation sites, fostering the growth of C-A-S-H through ionic interactions, condensation, and stabilization of amorphous phases. Ca²⁺ also participates in charge balancing by integrating into the C-A-S-H lattice through coordination with oxygen atoms [45], [46], [80]. Once Ca concentration is fully depleted, dissolved ions arrange into N-A-S-H, and a 3D aluminosilicate network growth occurs. This step is strongly inferred by the increase in wavenumber and peak intensity of Si–O–T (T = Si/Al) peaks in FTIR. While gel formation exhibits traits of both classic and non-classic growth pathways, the involvement of amorphous precursors and intermediate phases strongly supports a non-classic growth mechanism [12].

3.8. Reaction product characterization

The three final main aluminosilicate hydrate gels (i.e., C-A-S-H, N-A-S-H, and C-N-A-S-H) possess different elemental ratios; hence, a definite stoichiometry could not be determined. The type and composition of the gel is governed by the reactive oxides of the binder and the solution parameters. C-A-S-H has a linear nanostructure, and Na⁺ gets attached to C-A-S-H, turning it into a branched, cross-linked C-N-A-S-H gel. The Ca-deficient systems produce more N-A-S-H gels. Q-XRD provided some evidence for the formation of new alkali-activated minerals in the materials (Fig. S3 and Table S2 in Appendix A). Since the main reaction products of alkali activation are amorphous inorganic polymers, very small peaks corresponding to N-A-S-H were visible in all the materials. However, N-A-S-H gel structures can transform into zeolites with time, about 1% of zeolite was formed in CC2 and CC3, which is also present in metakaolin geopolymers [45], [46], [80], [81], [82]. The amorphous content calculated from Q-XRD ranges from approximately 70% to 95%, which almost matches the results of nanoindentation done on the same materials that are published elsewhere [18].

The only source of alumina is the solid NNP. Thus, the dissolution of aluminates is slower. The aluminates are incorporated with the most prominent gel; therefore, the Al/Si raio of the particular gel type is high. The percentage of reactive alumina calculated using Al-NMR is a novel index for a precursor’s reactivity.

3.9. Practical application, cost, sustainability, and limitations

Alkali-activated concretes (AACs) have demonstrated broad applicability in various sectors, including residential construction, hydraulic engineering, transportation infrastructure, waste management, and environmental engineering [37], [83], [84], [85], [86]. Despite the higher cost of NNP concrete—approximately seven times that of conventional OPC concrete—its significant reduction in CO₂ emissions, by 37% to 50%, positions it as a more sustainable option. Strategies to reduce costs and optimize mixtures have been explored in several studies. However, handling AACs, which exhibit a high pH range of 13–14, requires strict adherence to safety protocols, including the use of protective gear such as gloves, safety glasses, and long-sleeve clothing. Only trained personnel should handle AACs, as outlined by occupational safety guidelines. The alkali activation process also presents challenges, including the need for skilled labor during casting, large-scale ex-situ casting studies, elevated temperature curing for specific materials, and the high cost of activators. Nonetheless, the findings of this research are impactful, promoting sustainable material development, enhancing the performance and durability of construction materials, and encouraging environmentally friendly practices within the construction industry. This study provides a foundation for optimizing alkali-activated binder formulations and encourages further exploration of alternative binders that can reduce dependence on traditional precursors and OPC systems.

4. Conclusions

This study advances the fundamental understanding of reaction mechanisms in non-traditional precursors for AAMs, providing critical insights for optimizing their design and ensuring their viability as sustainable alternatives to OPC-based systems. The findings establish a strong foundation for the next generation of low-carbon construction materials, facilitating their adoption in structural applications using locally available natural resources or underutilized industrial by-products. This research promotes waste beneficiation, reduces carbon emissions, and supports a circular economy, enhancing both environmental and economic feasibility.

A key innovation of this study is the introduction of an Al-NMR-based reactivity index, which provides a quantitative framework for evaluating precursor performance and optimizing mix designs. By integrating solid-state NMR, FTIR, and Avrami-based simulations, this work offers a more comprehensive understanding of gel formation, structural evolution, and multi-stage reaction pathways of low-purity CCs, VAs, GBAs, and FBCA ashes. The results highlight the role of calcium in modifying activation mechanisms, the dynamic interplay between dissolution and polycondensation, and the evolution of C-A-S-H, N-A-S-H, and hybrid C-N-A-S-H gels. Al-NMR analysis confirms the stabilizing effects of aluminum incorporation on gel cross-linking.

Among the studied precursors, CCs exhibited the highest reactivity, forming denser gel networks, while GBAs demonstrated moderate reactivity due to their varying calcium content. In contrast, VAs and FBCA ashes showed lower intrinsic reactivity but benefited from ternary blending strategies. These findings emphasize the potential of tailored precursor combinations to enhance performance and expand the applicability of AAMs.

Future research should focus on long-term durability assessments and large-scale field validation to predict phase stability across diverse environmental conditions. Expanding the Al-NMR reactivity index to encompass a broader range of precursor chemistries will be essential for standardizing alternative binder formulations, ultimately accelerating the transition toward structural applications of concrete containing these precursors.

CRediT authorship contribution statement

Roshan Muththa Arachchige: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Formal analysis. Shubham Mishra: Writing – review & editing, Validation, Methodology, Formal analysis. Jan Olek: Writing – review & editing, Supervision, Project administration, Funding acquisition. Farshad Rajabipour: Writing – review & editing, Supervision, Project administration, Funding acquisition. Sulapha Peethamparan: Writing – review & editing, Writing – original draft, Supervision, Project administration, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: [Jan Olek, Sulapha Peethamparn and Farshad Rajabipour reports financial support was provided by Federal Highway Administration. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.].

Acknowledgment

The authors gratefully acknowledge the financial support from the Federal Highway Administration, USA (693JJ31950019) to a team of researchers at Purdue University, The Pennsylvania State University, and Clarkson University.

References

Publishing order | Descend order by publishing year | Descend order by cited within

[1]	Zhuang XY, Chen L, Komarneni S, Zhou CH, Tong DS, Yang HM, et al. Fly ash-based geopolymer: clean production, properties and applications. J Clean Prod 2016; 125:253-267.

[2]	Provis JL. Alkali-activated materials. Cem Concr Res 2018; 114:40-48.

[3]	Shi C, Jim AFénez, Palomo A. New cements for the 21st century: the pursuit of an alternative to Portland cement. Cem Concr Res 2011; 41(7):750-763.

[4]	Scrivener KL, John VM, Gartner EM. Eco-efficient cements: potential economically viable solutions for a low-CO₂ cement-based materials industry. Cem Concr Res 2018; 114:2-26.

[5]	American Concrete Institute (AC I) Committee 232. Report on the use of fly ash in concrete.Report. Farmington Hills: AC I; 2018.

[6]	Zhu W, Chen X, Struble LJ, Yang EH. Quantitative characterization of aluminosilicate gels in alkali-activated incineration bottom ash through sequential chemical extractions and deconvoluted nuclear magnetic resonance spectra. Cem Concr Compos 2019; 99:175-180.

[7]	Chen B, Perumal P, Aghabeyk F, Adediran A, Illikainen M, Ye G. Advances in using municipal solid waste incineration (MSWI) bottom ash as precursor for alkali-activation materials: a critical review. Resour Conserv Recycl 2024; 204:107516.

[8]	Carvalho R, Silva RV, de J Brito, Pereira MFC. Alkali activation of bottom ash from municipal solid waste incineration: Optimization of NaOH- and Na₂SiO₃-based activators. J Clean Prod 2021; 291:125930.

[9]	Choeycharoen P, Sornlar W, Wannagon A. A sustainable bottom ash-based alkali-activated materials and geopolymers synthesized by using activator solutions from industrial wastes. J Build Eng 2022; 54:104659.

[10]	American Coal Ash Association (ACA A). Coal ash recycling rate increases in 2020.Report. Denver: ACA A; 2021.

[11]	Rajabipour F, Mona F Zahedi, Kaladharan G. Evaluating the performance and feasibility of using recovered fly ash and fluidized bed combustion (FBCA) fly ash as concrete pozzolan [dissertation]. State College: The Pennsylvania State University, University Park; 2020.

[12]	United States Geological Survey (USG S). Mineral commodity summaries 2021.Report. Reston: USG S; 2021.

[13]	Tchakoute HK, Elimbi A, Yanne E, Djangang CN. Utilization of volcanic ashes for the production of geopolymers cured at ambient temperature. Cem Concr Compos 2013; 38:75-81.

[14]	Ghafoori N, Najimi M, Radke B. Natural Pozzolan-based geopolymers for sustainable construction. Environ Earth Sci 2016; 75(14):1110.

[15]	Najimi M, Ghafoori N, Radke B, Sierra K, Sharbaf M. Comparative study of alkali-activated natural pozzolan and fly ash mortars. J Mater Civ Eng 2018; 30(6):04018115.

[16]	Arachchige RM, Olek J, Rajabipour F, Peethamparan S. Non-traditional aluminosilicate based alkali-activated mortars—statistical optimization of solution parameters and processing conditions for optimal compressive strength, workability and setting time. Constr Build Mater 2023; 409:134096.

[17]	Yoon J, Jafari K, Tokpatayeva R, Peethamparan S, Olek J, Rajabipour F. Characterization and quantification of the pozzolanic reactivity of natural and non-conventional pozzolans. Cem Concr Compos 2022; 133:104708.

[18]	Tokpatayeva R, Castillo A, Yoon J, Kaladharan G, Jafari K, Arachchige RM, et al. Comparative study of the reactivity and performance of different nontraditional and natural pozzolans in cementitious system. Adv Civ Eng Mater 2022; 11(2):670-693.

[19]	Fan LF, Zhong WL, Zhang YH. Effect of the composition and concentration of geopolymer pore solution on the passivation characteristics of reinforcement. Constr Build Mater 2022; 319:126128.

[20]	Pouhet R, Cyr M. Carbonation in the pore solution of metakaolin-based geopolymer. Cem Concr Res 2016; 88:227-235.

[21]	Zuo Y, Nedeljkovi Mć, Ye G. Pore solution composition of alkali-activated slag/fly ash pastes. Cem Concr Res 2019; 115:230-250.

[22]	Coleman NJ, Page CL. Aspects of the pore solution chemistry of hydrated cement pastes containing metakaolin. Cem Concr Res 1997; 27(1):147-154.

[23]	Ke X, Bernal SA, Provis JL, Lothenbach B. Thermodynamic modelling of phase evolution in alkali-activated slag cements exposed to carbon dioxide. Cement Concr Res 2020; 136:106158.

[24]	Alventosa KML, White CE. The effects of calcium hydroxide and activator chemistry on alkali-activated metakaolin pastes. Cem Concr Res 2021; 145:106453.

[25]	Alventosa KML, Wild B, White CE. The effects of calcium hydroxide and activator chemistry on alkali-activated metakaolin pastes exposed to high temperatures. Cem Concr Res 2022; 154:106742.

[26]	Aboulayt A, Riahi M, Ouazzani M Touhami, Hannache H, Gomina M, Moussa R. Properties of metakaolin based geopolymer incorporating calcium carbonate. Adv Powder Technol 2017; 28(9):2393-2401.

[27]	Luo Y, Brouwers HJH, Yu Q. Understanding the gel compatibility and thermal behavior of alkali activated Class F fly ash/ladle slag: the underlying role of Ca availability. Cem Concr Res 2023; 170:107198.

[28]	Garbev K, Beuchle G, Bornefeld M, Black L, Stemmermann P. Cell dimensions and composition of nanocrystalline calcium silicate hydrate solid solutions. Part 1: synchrotron-based X-ray diffraction. J Am Ceram Soc 2008; 91(9):3005-3014.

[29]	Liu C, Li Z, Nie S, Skibsted J, Ye G. Structural evolution of calcium sodium aluminosilicate hydrate (C-(N-)A-S-H) gels induced by water exposure: the impact of Na leaching. Cem Concr Res 2024; 178:107432.

[30]	Myers RJ, Bernal SA, Provis JL. A thermodynamic model for C-(N-)A-S-H gel: CNASH_ss. Derivation and validation. Cem Concr Res 2014; 66:27-47.

[31]	Duxson P, Fernández-Jim Aénez, Provis JL, Lukey GC, Palomo A, Van JSJ Deventer. Geopolymer technology: the current state of the art. J Mater Sci 2007; 42(9):2917-2933.

[32]	Souayfan F, Rozi Eère, Paris M, Deneele D, Loukili A, Justino C. ²⁹Si and ²⁷Al MAS NMR spectroscopic studies of activated metakaolin–slag mixtures. Constr Build Mater 2022; 322:126415.

[33]	Abdelrahman O, Garg N. Impact of Na/Al ratio on the extent of alkali-activation reaction: non-linearity and diminishing returns. Front Chem 2022; 9:806532.

[34]	Myers RJ, Bernal SA, San R Nicolas, Provis JL. Generalized structural description of calcium-sodium aluminosilicate hydrate gels: the cross-linked substituted tobermorite model. Langmuir 2013; 29(17):5294-5306.

[35]	Provis JL, Duxson P, Lukey GC, Van JSJ Deventer. Statistical thermodynamic model for Si/Al ordering in amorphous aluminosilicates. Chem Mater 2005; 17(11):2976-2986.

[36]	Peyvandi A, Holmes D, Soroushian P, Balachandra AM. Monitoring of sulfate attack in concrete by ²⁷Al and ²⁹Si MAS NMR spectroscopy. J Mater Civ Eng 2015; 27(8):04014226.

[37]	Zuo Y, Chen Y, Liu C, Gan Y, Göbel L, Ye G, et al. Modeling and simulation of alkali-activated materials (AAMs): a critical review. Cem Concr Res 2025; 189:107769.

[38]	Bahraq AA, Al-Osta MA, Baghabra OS Al-Amoudi, Obot IB, Maslehuddin M, et al. Molecular simulation of cement-based materials and their properties. Engineering 2022; 15:165-178.

[39]	Kim G, Cho S, Im S, Yoon J, Suh H, Kanematsu M, et al. Evaluation of the thermal stability of metakaolin-based geopolymers according to Si/Al ratio and sodium activator. Cem Concr Compos 2024; 150:105562.

[40]	Zeng J, Zhang K, Sun W, Zeng Y, Zou Z. Mechanics and microstructure analysis of geopolymer utilizing ilmenite tailing and metakaolin powder as alkali-activated materials. Case Stud Constr Mater 2024; 21:e03567.

[41]	Ke X, Baki VA. Assessing the suitability of alkali-activated metakaolin geopolymer for thermochemical heat storage. Microporous Mesoporous Mater 2021; 325:111329.

[42]	Zhang SS, Wang S, Chen X. Understanding the role of magnesium ions on setting of metakaolin-based geopolymer. Cem Concr Res 2024; 177:107430.

[43]	Ruiz-Agudo C, Cölfen H. Exploring the potential of nonclassical crystallization pathways to advance cementitious materials. Chem Rev 2024; 124(12):7538-7618.

[44]	Tang ZQ, de FB Souza, Mulder RJ, KwesiSagoe-Crentsil DW. Multistep nucleation and growth mechanism of aluminosilicate gel observed by cryo-electron microscopy. Cem Concr Res 2022; 159:106873.

[45]	Sun Y, Wang ZH, Park DJ, Kim WS, Kim HS, Yan SR, et al. Analysis of the isothermal hydration heat of cement paste containing mechanically activated fly ash. Thermochim Acta 2022; 715:179273.

[46]	Zhang Z, Han F, Yan P. Modelling the dissolution and precipitation process of the early hydration of C₃S. Cem Concr Res 2020; 136:106174.

[47]	Nazari A, Sanjayan JG. Johnson–Mehl–Avrami–Kolmogorov equation for prediction of compressive strength evolution of geopolymer. Ceram Int 2015; 41(2):3301-3304.

[48]	Han F, Zhang Z, Liu J, Yan P. Hydration kinetics of composite binder containing fly ash at different temperatures. J Therm Anal Calorim 2016; 124(3):1691-1703.

[49]	Thomas JJ. A new approach to modeling the nucleation and growth kinetics of tricalcium silicate hydration. J Am Ceram Soc 2007; 90(10):3282-3288.

[50]	Arachchige RM, Thapa S, Olek J, Rajabipour F, Peethamparan S. Shrinkage behavior of alkali-activated materials using ground bottom ashes, calcined clays, volcanic ashes, and fluidized bed combustion ashes as precursors. Case Stud Constr Mater 2024; 21:e04033.

[51]	Yusslee E, Beskhyroun S. The effect of water-to-binder ratio (W/B) on pore structure of one-part alkali activated mortar. Heliyon 2023; 9(1):e12983.

[52]	Zhao Y, Hu X, Yuan Q, Wu Z, Shi C. Effects of water to binder ratio on the chloride binding behaviour of artificial seawater cement paste blended with metakaolin and silica fume. Constr Build Mater 2022; 353:129110.

[53]	Joseph S, Cizer Ö. Comparative analysis of heat release, bound water content and compressive strength of alkali-activated slag-fly ash. Front Mater 2022; 9:861283.

[54]	Fu Q, Bu M, Zhang Z, Xu W, Yuan Q, Niu D. Hydration characteristics and microstructure of alkali-activated slag concrete: a review. Engineering 2023; 20:162-179.

[55]	L E’Hôpital, Lothenbach B, Le G Saout, Kulik D, Scrivener K. Incorporation of aluminium in calcium-silicate-hydrates. Cem Concr Res 2015; 75:91-103.

[56]	Gartner E. Industrially interesting approaches to ‘low-CO₂’ cements. Cem Concr Res 2004; 34(9):1489-1498.

[57]	Madadi A, Wei J. Characterization of calcium silicate hydrate gels with different calcium to silica ratios and polymer modifications. Gels 2022; 8(2):75.

[58]	Yaphary YL, Lau D, Sanchez F, Poon CS. Effects of sodium/calcium cation exchange on the mechanical properties of calcium silicate hydrate (C-S-H). Constr Build Mater 2020; 243:118283.

[59]	Shehata MH, Thomas MDA, Bleszynski RF. The effects of fly ash composition on the chemistry of pore solution in hydrated cement pastes. Cem Concr Res 1999; 29(12):1915-1920.

[60]	Onutai S, Sato J, Osugi T. Possible pathway of zeolite formation through alkali activation chemistry of metakaolin for geopolymer–zeolite composite materials: aTR-FTIR study. J Solid State Chem 2023; 319:123808.

[61]	Schade T, Middendorf B. Prediction model based on DoE and FTIR data to control fast setting and early shrinkage of alkaline-activated slag/silica fume blended cementitious material. Materials 2023; 16(11):4104.

[62]	Kapeluszna E, ŁKotwica , Ró Ażycka, ŁGo łek. Incorporation of Al in C-A-S-H gels with various Ca/Si and Al/Si ratio: microstructural and structural characteristics with DTA/TG, XRD, FTIR and TEM analysis. Constr Build Mater 2017; 155:643-653.

[63]	Richard T, Mercury L, Poulet F, d L’Hendecourt. Diffuse reflectance infrared Fourier transform spectroscopy as a tool to characterise water in adsorption/confinement situations. J Colloid Interface Sci 2006; 304(1):125-136.

[64]	Mollah MYA, Yu W, Schennach R, Cocke DL. A Fourier transform infrared spectroscopic investigation of the early hydration of Portland cement and the influence of sodium lignosulfonate. Cem Concr Res 2000; 30(2):267-273.

[65]	Mollah MYA, Kesmez M, Cocke DL. An X-ray diffraction (XRD) and Fourier transform infrared spectroscopic (FT-IR) investigation of the long-term effect on the solidification/stabilization (S/S) of arsenic(V) in Portland cement type-V. Sci Total Environ 2004; 325(1–3):255-262.

[66]	Hughes TL, Methven CM, Jones TGJ, Pelham SE, Fletcher P, Hall C. Determining cement composition by Fourier transform infrared spectroscopy. Adv Cem Based Mater 1995; 2(3):91-104.

[67]	Kloprogge JT, Schuiling RD, Ding Z, Hickey L. Wharton D, Frost RL. Vibrational spectroscopic study of syngenite formed during the treatment of liquid manure with sulphuric acid. Vib Spectrosc 2002;28(2):209–21

[68]	Ylm Rén, Jäglid U, Steenari BM, Panas I. Early hydration and setting of Portland cement monitored by IR, SEM and Vicat techniques. Cem Concr Res 2009; 39(5):433-439.

[69]	Khalifa AZ, Cizer Ö, Pontikes Y, Heath A, Patureau P, Bernal SA, et al. Advances in alkali-activation of clay minerals. Cem Concr Compos 2020; 132:106050.

[70]	Rowles MR, Hanna JV, Pike KJ, Smith ME, O BH’Connor. ²⁹Si, ²⁷Al, ¹H and ²³Na MAS NMR study of the bonding character in aluminosilicate inorganic polymers. Appl Magn Reson 2007; 32(4):663-689.

[71]	Cherki A El Idrissi, Paris M, Rozi Eère, Deneele D, Darson S, Loukili A. Alkali-activated grouts with incorporated fly ash: from NMR analysis to mechanical properties. Mater Today Commun 2018; 14:225-232.

[72]	Ma H, Wu C. Mechanical and microstructural properties of alkali-activated fly ash-slag material under sustained moderate temperature effect. Cem Concr Compos 2022; 134:104744.

[73]	Gomez-Zamorano L, Balonis M, Erdemli B, Neithalath N, Sant G. C-(N)-S-H and N-A-S-H gels: compositions and solubility data at 25 °C and 50 °C. J Am Ceram Soc 2017; 100(6):2700-2711.

[74]	Fang G, Zhang M. Multiscale micromechanical analysis of alkali-activated fly ash-slag paste. Cem Concr Res 2020; 135:106141.

[75]	Garcia-Lodeiro I, Palomo A, Fernández-Jim Aénez, MacPhee DE. Compatibility studies between N-A-S-H and C-A-S-H gels. Study in the ternary diagram Na₂O–CaO–Al₂O₃–SiO₂–H₂O. Cem Concr Compos 2011; 41(9):923-931.

[76]	Arachchige RM, Olek J, Rajabipour F, Peethamparan S. Phase identification and micromechanical properties of non-traditional and natural pozzolan based alkali-activated materials. Constr Build Mater 2024; 441:137478.

[77]

Mishra S, Peethamparan S. Alkali-silica reaction resistance of alkali-activated calcined clays using accelerated mortar bar test.In: Proceedings of the 16th International Congress on the Chemistry of Cement 2023—ICCC2023; 2023 Sep 18–22; Bangkok, Thailand, Duesseldorf: ICCC Permanent Secretariat; 2023.

[78]	Mishra S, Peethamparan S. ASR resistance of ground bottom ash-based alkali-activated concrete and the prospect of using MCPT for the evaluation.In: Proceedings of the 17th International Conference on Alkali-Aggregate Reaction in Concrete; 2024 May 18–24; Ottawa, Canada. Berlin: Springer; 2024.

[79]	Hoyos-Montilla AA, Tobón JI, Puertas F. Role of calcium hydroxide in the alkaline activation of coal fly ash. Cem Concr Compos 2023; 137:104925.

[80]	Caron R, Patel RA, Dehn F. Activation kinetic model and mechanisms for alkali-activated slag cements. Constr Build Mater 2022; 323:126577.

[81]	Sarah AS Kareem Mohammed, G Réber, Simon A, Kurovics E, Hamza A. Comparative study of metakaolin-based geopolymer characteristics utilizing different dosages of water glass in the activator solution. Results Eng 2023; 20:101469.

[82]	Williams RP, Hart RD, Van A Riessen. Quantification of the extent of reaction of metakaolin-based geopolymers using X-ray diffraction, scanning electron microscopy, and energy-dispersive spectroscopy. J Am Ceram Soc 2011; 94(8):2663-2670.

[83]	Ji Z, Pei Y. Bibliographic and visualized analysis of geopolymer research and its application in heavy metal immobilization: a review. J Environ Manag 2019; 231:256-267.

[84]	Majidi B. Geopolymer technology, from fundamentals to advanced applications: a review. Mater Technol 2009; 24(2):79-87.

[85]	Zhang J, Shi C, Li N, Zhang Z, Farzadnia N. Carbon dioxide sequestration by alkali-activated materials. Carbon dioxide sequestration in cementitious construction materials, Woodhead Publishing, Cambridge 2018; 279-298.

[86]	Shen W, Zhang C, Yang Z, Cao L, Yang Z, Tian B. Investigation on the safety concrete for highway crash barrier. Constr Build Mater 2014; 70:394-398.

AI Summary AI Mindmap