The effect of activating solution on the mechanical strength, reaction rate, mineralogy, and microstructure of alkali-activated fly ash

The fly ash utilized in the present work was generated in a coal-fired power plant in The Netherlands (type F according to the ASTM classification). The chemical composition determined by X-ray fluorescence (XRF) is presented in Table 2 and the particle size distribution is shown in Fig. 1. The mean ash particle diameter of fly ash is 21.46 μm and surface area measured by nitrogen adsorption is 1,600 m²/kg. XRD (Fig. 2) revealed that the major crystalline phases in fly ash are quartz (SiO₂) and mullite (3Al₂O₃, 2 SiO₂).

The alkaline solutions were prepared by mixing sodium hydroxide (analytical grade, >98% purity) with distilled water and sodium silicate solution (Na₂O: 8.25 wt%, SiO₂: 27.50 wt%). A series of activating solutions with different SiO₂ and Na₂O content were prepared in this study. The SiO₂ content was in the range of 0–1.5 mol/kg fly ash, and Na₂O content was between 1.0 and 1.5 mol/kg fly ash. The composition of alkali solution for each mixture is shown in Table 3. The water-to-fly ash mass ratio was kept constantly at 0.35. Other mixtures, e.g., with higher SiO₂ (2 mol) or lower alkali Na₂O (0.5 mol) were also tried for comparison. However, some solutions (SiO₂ = 2.0 mol) were not stable with precipitation formed; other samples (Na₂O = 0.5 mol) could not set due to the low alkali content, which were not studied further.

AAF were synthesized by mixing fly ash with activating solutions. For mechanical tests, specimens were cast in 40 × 40 × 160 mm³ molds. Samples for MIP and XRD tests were prepared in small plastic bottles. After casting, all the samples were vibrated for 2 min to remove the entrapped air. The samples were then sealed by plastic foil and cured in an oven at 40 °C before tests. All the tests were performed at the curing ages of 7 and 28 days, respectively, at room temperature.

Figures 3 and 4 show the influence of SiO₂ and Na₂O content on the compressive and flexural strength of AAF at the curing age of 7 and 28 days, respectively. With more Na₂O content in the activator, the compressive strength of AAF increased at both curing ages. Bakharev [21] also found higher strength when Na₂O addition increased from 2 to 8 wt% of fly ash. However, further increase in alkali content (till 14.90 wt% of fly ash) was reported to result in a decrease of compressive strength [23]. In this study, higher alkali content (2 mol/kg fly ash) was not used due to the precipitation of alkali solutions when silica was present. Increasing the amount of SiO₂ only led to a higher compressive strength at the age of 7 days (Fig. 3). However, at 28 days curing age, the highest strength appeared at different silica content with respect to the Na₂O amount. For example, when the Na₂O content was 1.0, 1.3, 1.5 mol/kg fly ash, the highest compressive strength was obtained with the silica content of 1.0, 1.3, and 1.3 mol/kg fly ash, respectively (Fig. 4). Therefore, it could be stated that in the presence of silica in activator, higher amount of alkali (Na₂O) was beneficial for the compressive strength development. However, there was an optimal content of soluble silica (SiO₂) with respect to the Na₂O content. The existence of the optimal silica content in activator agrees well with the findings by Fernández-Jiménez et al. [20] and Duxson et al. [30] in previous studies.

For flexural strength, a different trend was observed compared to the compressive strength. When the SiO₂ content varied from 1.0 to 1.5 mol/kg of fly ash, more SiO₂ and Na₂O content led to a decrease of flexural strength both at 7 and 28 days curing ages. The reduction of flexural strength is thought to relate to the structural ordering of the reaction products of AAF. One plausible reason is that increasing the silica and sodium contents in activating solution may resulted in reduced level of long-range structural ordering products [31]. The less chemically connected products lead to the decrease of flexural strength.

The heat evolution and cumulative heat of the AAF (at 40 °C) with different contents of activator during the first 168 h is presented in Fig. 5. In general, the released heat of AAF was considerably lower compared to ordinary Portland cement, which was reported to have a total heat release around 350 J/g after 7 days [32]. When the alkali content was fixed (three gray lines or three black lines in Fig. 5), a higher SiO₂ content led to a lower heat release, whereas an increase of alkali content resulted in a higher total heat release (line 1.5–1.5 and 1.5–1, or 1.3–1.5 and 1.3–1, or 1–1.5 and 1–1 in Fig. 5). The result revealed that the addition of SiO₂ retarded the reaction, while Na₂O was favorable to the reaction. By using acid attack method, Criado et al. [25] also found lower reaction degree with increased silica content of AAF.

MIP is widely used to measure both the porosity and pore size distribution of cement-based materials [33, 34]. However, the application of MIP on AAF was rarely reported. Though it was argued that MIP method cannot reflect the true pore size distribution of porous materials due to the different pore shapes and “ink-bottle” effect [35], the comparison between the MIP results itself could still provide valuable information on the pore structure development of AAF.

For ordinary Portland cement, in general, there are two peaks in the differential curves, corresponding to two pore systems: the gel pores of cement paste are formed by the C–S–H gels, with pore size diameter ranging from 0.5 nm to 0.01 μm [36]; the capillary pores refers to the spaces left by the water that do not filled with reaction products, with pore size range of 0.01–10 μm [36].

For the differential curves of AAF (Figs. 6, 7), despite two peaks were also detected for most of the samples, corresponding to two pore systems in AAF, the formation of the pores was very different compared to Portland cement. The “gel pores” of AAF, with the diameter in the range of several nanometers to 0.02 μm are formed by the aluminosilicate gels. Unlike the formation of C–S–H gels in Portland cement paste, which grow around the surface of cement particles [27], the gels of fly ash activated with high silica content were observed evenly distributed in the matrix (Fig. 8c–f). As for the “capillary pores” in AAF, the capillary pores that are typical in Portland cement were not observed. On the other hand, a number of pores were formed by the dissolved fly ash particles (Fig. 8c–f). These pores were in the range of several microns to larger than 10 microns in diameter, constituting the “capillary pore” system as shown in Fig. 6b, d, f. Based on the information discussed above, the simplified processes of microstructure formation in AAF and cement paste are illustrated in Fig. 9. As demonstrated in Fig. 9a, because the capillary pores of AAF are blocked by the homogeneous gels, they can only be accessed through the gel pores when mercury is intruded in the matrix, which would cause a significant “ink-bottle” effect. Therefore, the capillary pores in AAF measured by MIP (Figs. 6, 7) were in the range of 0.1–1 microns, about one order of magnitude smaller than the real diameter observed from ESEM images (Fig. 8). As for specimen with no or very low soluble silica content (0–1.5 and 0.5–1.5), the formation of gel phase was limited and only grew around the un-reacted fly ash particles, typical capillary pores as in Portland cement were observed in Fig. 8a, b resulting in the only peak of capillary pore in the differential curve in Fig. 6b for specimen 0.5–1.5. The poorly reacted fly ash and the high porosity of the matrix led to the low mechanical properties of AAF (Fig. 4). Similar findings on the pore structure of AAF were also reported by Lloyd et al. [37] using wood’s metal intrusion method and scanning electron microscopy [38, 39].

As illustrated from Table 4, the total porosity obtained from MIP decreased as silica and alkali content increased both at 7 and 28 days. Increasing the SiO₂ or Na₂O content from 1.0 to 1.5 mol, the critical pore diameter (the diameter corresponding to the peak in differential curve) of gel pores of AAF shifted to a lower pore diameter (Table 4; Figs. 6, 7), indicating a denser pore structure formed.

According to Glukhovsky [40], the alkali activation process of aluminosilicate materials was constituted of three stages: destruction–coagulation, coagulation–condensation, and condensation–crystallization. Other authors [31] extended Glukhovsky’s theory and proposed a more detailed description of the model for the alkali activation process. Five steps were involved in the reaction process: namely dissolution, speciation equilibrium, gelation, reorganization and polymerization and hardening. In general, in the initial stage, when alkali solution comes in contact with fly ash particles, it dissolves the vitreous phase in fly ash, releasing monomers aluminates and silicates species into the medium [19]. These small ionic species contact with each other, forming a coagulated structure, giving rise to a three-dimensional aluminosilicate gel, also called zeolite precursors [11]. This gel is responsible for the cementitious properties of alkali activated material.

The addition of alkali cations into the activating solution is considered to have two effects: first, it increases the pH in the solution; high amount of OH⁻ ions helps to sever the Si–O–Si and Si–O–Al bonds of the reactive fly ash into small silicon and aluminum species [19]; second, the alkali cation balances the electric charge in the tetrahedral Si and Al structure [25]. Therefore, in this study, when more sodium oxide presents in the activator, specimens exhibit a higher compressive strength, higher extent of reaction, lower porosity, and finer pore size distribution. Higher alkali content dissolves more fly ash particles into the medium, ready for the next steps of reactions (gelation and polymerization).

Unlike the effect of alkali, the addition of soluble silica influences the properties of AAF in two ways. On one hand, the soluble silica species in activating solution cause a lower total heat release (Fig. 5), corresponding to a lower extent of reaction. It is reported [41] that with more silica amount, the fraction of monomer and dimer form of silicate species in the alkaline solutions decreased, while the trimer unit content grew. The higher degree of silicate ion polymerization in the medium retards the further dissolution of fly ash particles, leading to a lower reaction degree. On the other hand, these silicate species react with the aluminum species dissolved from the fly ash particles to form the aluminosilicate gels. Unlike the gels formed without soluble silica (Fig. 8a), which mainly precipitate on the surface of fly ash particles, they were evenly distributed in the matrix, forming a more homogeneous and denser microstructure (Fig. 8c–f). Therefore, lower total porosity and finer gel pore system were observed for specimens with more SiO₂ content at early age (Table 4). This is the reason why the specimens with higher SiO₂ content exhibit a lower degree of reaction but developed a lower total porosity. The compressive strength, which was closely related to its reaction degree and microstructure, as a result, increases with higher SiO₂ content in early age (Fig. 3), and exhibited an optimal value at different silica content with respect to the alkali content.

In this study, a series of activating solutions, with the SiO₂ content varied from 0 to 9% and Na₂O between 6.2 and 9.3% (by the weight of fly ash), were applied to activated fly ash. The result demonstrated that the highest strength was obtained when SiO₂ content was 6% (1.0 mol/kg fly ash) and Na₂O content at 9.3% (1.5 mol/kg fly ash). Though the optimal value would vary substantially with the chemical composition and reactivity of fly ash (Table 1), similar amount of activator was reported in previous studies, where 7 wt% of SiO₂ and 7 wt% Na₂O (weight ratio of fly ash) were used in the study by Lloyd et al. [37], and 7.5 and 8 wt% Na₂O (weight ratio of fly ash) were reported by Škvára et al. [24] and Bakharev [21], respectively.

The mineralogy of AAF under different conditions (activator, temperature, or curing condition) was widely reported [42, 43]. Figures 10 and 11 show that the reaction products of AAF were predominantly amorphous compound, regardless of the alkali solutions. Quartz and mullite are the main crystalline phases remained in the un-reacted fly ash. The vitreous phase in original fly ash exhibited a bump between 20 and 25° 2θ. After the alkali activation, the bump shifted to a higher degree in the range of 27–30° 2θ.

Different types of zeolite crystalline phases were observed in Fig. 10 for the specimen with medium amount of soluble silica (faujasite-Na for 0.5–1.5 and 1–1.5) and without silica (hydroxysodalite for 0–1.5) presented in activating solution. However, no clear crystalline was detected from the XRD pattern when the silica content further increased (mixture 1.3–1.5 and 1.5–1.5). These results are consistent with the findings by other authors [23, 25]. From the observations, it is assumed that the higher degree of silica polymerization in solutions with higher soluble silica content requires more energy for transferring the aluminosilicate gel into well-organized crystalline [25].

The sodium oxide content also had a prominent effect on the mineralogy. When the silica content was 1.0 mol, an increase of the alkali content resulted in the formation of zeolite phase faujasite-Na (Fig. 11, pattern 1–1.5); while the silica content was kept constant at 1.5 mol/kg fly ash, new crystalline phase was not detected in high sodium oxide samples. Similar result was also reported by Lloyd [19] in the aging of AAF.

One of the most important reasons to study the mineralogy of AAF is to evaluate the stability of AAF in a long timescale [19]. In previous studies of alkali activated metakaolin [19, 44], which has similar composition of raw material and reaction products as AAF, the loss of strength was detected accompanying with the formation of zeolite crystalline phases. In AAF, similar results were also reported [25], indicating that a higher content of zeolites in AAF prevented the mechanical strength development. However, it was found in this study that specimen 1-1.5, which had crystalline zeolite formed at 28 days, exhibited the highest compressive strength. In the aging test of AAF by Lloyd et al. [19], the drop of compressive strength was also not observed with the formation of zeolite crystalline phases. As for the other two mixtures, specimens 0-1.5 and 0.5-1.5, which also had zeolite crystalline phases but showed lower mechanical strength, it is assumed that the lower strength is mainly related to the porous and loosen microstructure formed in the matrix (Figs. 6a, 8a, b), rather than the zeolite.