Materials

The muscovite used in the present study was obtained from Anhui Henghao Technology Co., Ltd., Anhui Province, China. The chemical composition of the muscovite consisted mainly of Fe₂O₃, SiO₂, and Al₂O₃ (Table 1). The X-ray diffraction (XRD) pattern (Fig. 2) showed that the main mineral phase in the muscovite sample was muscovite with high crystallinity. The muscovite (~150 kg) was dried at 105°C until it reached a constant mass, and was then used for the subsequent experiments.

Table 1 Chemical composition of the raw materials (wt.%)

The Portland cement (PC 42.5) used in the study was obtained from China United Cement Corporation, Shandong Province, China. The major constituents of the Portland cement were SiO₂, Al₂O₃, CaO, and Fe₂O₃ (Table 1).

Analytical reagent-grade calcium oxide (AR, ≥98%), obtained from Shanghai Macklin Biochemical Co., Ltd., Shanghai Province, China, was used as an activator.

Particle-size distribution and specific surface area

The particle-size distribution curves (Fig. 1) of the initial and mechanically activated muscovite milled for various durations in a planetary mill revealed that the curve corresponding to the initial muscovite was unimodal. The curve later became bimodal or multimodal, depending on the grinding. After 60 min of grinding, the particle-size distribution curve transformed to a broad bimodal curve with two distinct particle size ranges of 0.5–6.0 μm and 8–30 μm. After 120 min of grinding, particles larger than 80 μm disappeared, and the number of fine particles <2 μm in size increased significantly as the grinding time increased. The characteristic diameters D₁₀, D₅₀, and D₉₀ were used to characterize the effects of the mechanical activation on the particle-size distribution. These indices represent the particle size at which 10%, 50%, or 90% of the weight is finer. The variation in the characteristic diameter and specific surface area with grinding time (Table 2) indicated that the characteristic diameters D₁₀, D₅₀, and D₉₀ decreased significantly over 20 min of grinding, from 17.22 μm, 95.24 μm, and 253.90 μm to 5.69 μm, 47.76 μm, and 169.23 μm, respectively. Thereafter, a gradual change was observed. The size reduction appeared to reach a limit at 160 min of grinding, with particle size increasing with further grinding. Similar change trends have been reported by other researchers using the mechanical activation method (Alex et al. Reference Alex, Kumar, Roy and Mehrotra2014; Andrić et al. Reference Andrić, Terzić and Aćimović2014; Ilić et al. Reference Ilić, Radonjanin, Malešev, Zdujić and Mitrović2016). The specific surface area increased significantly, from 0.81 m²/g to 28.14 m²/g, with grinding time of 120 min, after which the changes in the specific surface area were insignificant. With further grinding, the specific surface area of the 200-min mechanically ground muscovite was less than that of the 160-min ground muscovite. Combined with the particle size distribution analysis, this result was inconsistent with previously reported findings, where the reduction in the particle-size and the increase in the specific surface area occurred in the initial stages, while a prolonged grinding time resulted in decreased specific surface area (Mitrović and Zdujić Reference Mitrović and Zdujić2014; Perrin-Sarazin et al. Reference Perrin-Sarazin, Sepehr, Bouaricha and Denault2010).

Fig. 1 Particle size distribution curves of the muscovite samples ground for different durations

Table 2 Variation in the characteristic diameters (D₁₀, D₅₀, and D₉₀) and specific surface area with grinding time

Based on the analysis above, the following changes occurred during mechanical activation: the particle size reduction appeared to reach a limit at 160 min of grinding. However, the tendency was toward increased particle size with a further increase in grinding time, probably due to aggregation of the ground fine particles. This tendency was also somewhat more apparent from the significant decrease in surface area at 200 min of grinding (28.82 m²/g) as compared to 160 min of grinding (23.44 m²/g).

XRD analysis

The XRD patterns of the initial and mechanically activated (for various grinding times) muscovite with (Fig. 2) revealed that the muscovite reflections decreased significantly with grinding time. The decrease in reflection intensity is a manifestation of the amorphization (Kitamura & Senna Reference Kitamura and Senna2001), and the loss of reflection intensity illustrates the breaking of the bonds between the muscovite layers. Such an observation implies that significant amorphization was mechanically induced in the muscovite structures during grinding (Mitrović & Zdujić Reference Mitrović and Zdujić2014). The reduction in the reflection intensity was insignificant in the initial stages—especially within the first 40 min—but significant changes in the particle size and specific surface area were observed during this period. With prolonged grinding, further changes in particle size and specific surface area were insignificant compared with the significantly decreased intensity of the muscovite reflections. Regarding the energy entering the muscovite particle that was obtained from mechanical activation, some of it was released rapidly and some of it may have been stored (Dandurand et al. Reference Dandurand, Gout and Schott1982). In the initial stages, energy storage existed mainly in the form of new surfaces, but with prolonged grinding, energy storage may have caused the formation of internal defects in the muscovite which was the main reason for the amorphization. Crystalline muscovite continued to be observed, indicating that the mechanically induced amorphization was incomplete.

Fig. 2 XRD patterns of the initial and ground (for various amounts of time) muscovite

Based on the XRD pattern data for muscovite, analysis of the relative crystallinity was conducted using the intensity of the strongest characteristic peaks (Makó et al. Reference Makó, Frost, Kristóf and Horváth2001), namely, the intensity of the muscovite (009) reflection. The calculated results (Fig. 3) according to Eq. (1) revealed that the main decrease in the relative crystallinity occurred after 40–160 min of grinding, and it was reduced to 14.99% at 160 min.

(1) $\alpha =\frac{I_{\text{ground}}}{I_{\text{initial}}}\times 100$

Fig. 3 Variation in the intensity of the muscovite (009) reflection and the relative crystallinity with grinding time

In Eq. (1), α is the relative crystallinity of the muscovite, I_ground is the intensity of the muscovite (009) reflection of the mechanically ground muscovite, and I_initial is the intensity of the initial muscovite (009) reflection (assuming an integral intensity of 100%). Thereafter, the variations were insignificant. However, a continuous decrease in relative crystallinity occurred despite the fact that the specific surface area decreased when the grinding time was prolonged, which was consistent with a previous study (Mitrović & Zdujić Reference Mitrović and Zdujić2014).

Thermal behavior

In order to study the thermal characteristics of the initial and mechanically activated muscovite during thermal treatment from 50 to 1100°C, TG and derivative thermogravimetric (DTG) analyses of typical samples (Fig. 4) were carried out. Chemically, the crystal structure of the muscovite dehydroxylation at high temperatures can be summarized as 2(OH)⁻¹ → H₂O + O²⁻. The weight loss of initial muscovite at temperatures up to 400°C was only 0.39 wt.% (Fig. 4a), which was attributed to the physically adsorbed water. The total weight loss at 1000°C was 4.45%, and dehydroxylation began to form a dehydroxylated phase at ~600°C, which means that most of the hydroxyls were removed at temperatures between 600 and 1000°C. This result is inconsistent with previously reported findings, where most of the hydroxyls were removed at temperatures below 850°C (Gridi Reference Gridi2008). In their work on the thermal behavior of muscovite, Kodama & Brydon (Reference Kodama and Brydon1968) reported a lower dehydroxylation temperature for mechanically induced muscovite with smaller particle sizes. With prolonged grinding, the dehydroxylation temperature decreased from ~900 to 770°C at 80 min of grinding time (Fig. 4b). Between 600 and 900°C, a stepwise weight-loss with a large and asymmetric peak occurred, which may be due to the formation of a disordered or partially amorphous structure as a result of the mechanical activation. In the sample mechanically activated for 160 min of grinding, the dehydroxylation process was no longer detected, indicating that mechanical activation transformed the crystal structure of the muscovite into a metastable dehydroxylated phase. Several studies have discussed the dehydroxylation reaction for muscovite (Gridi Reference Gridi2008; Mazzucato et al. Reference Mazzucato, Artioli and Gualtieri1999; Tokiwai & Nakashima Reference Tokiwai and Nakashima2010). Guggenheim et al. (Reference Guggenheim, Chang and Koster van Groos1987) reported that dehydroxylation occurred through the loss of hydroxyl groups, where the residual oxygen moved to the z-coordinate position of the Al cation, and the Al-O polyhedra was in fivefold coordination in the dehydroxylated phase of the muscovite. The weight-loss peak at 235°C was attributed to the adsorbed water of the mechanically ground muscovite. On the basis of the TG data, the amount of adsorbed water (under 400°C) showed an increasing trend (0.39%, 2.90%, and 4.30% for 0, 80, and 160 min, respectively) with grinding, attributed mainly to the increase in the specific surface area. In contrast, the weight loss of the samples corresponding to dehydroxylation exhibited a decreasing trend with grinding, reflecting the dehydroxylation reaction that occurred during the grinding process.

Fig. 4 (a) TG and DTG curves of the initial (KAl₂(Si₃Al)O₁₀(OH,F)₂) and mechanically activated muscovite after grinding for 80 min (b) and 160 min (c)

FTIR spectra

The structural changes in the initial and mechanically activated muscovite (ground for 80 min and 160 min) were determined using FTIR spectroscopy (Fig. 5). The spectra for the initial muscovite showed the following characteristic bands: 3623, 687, and 412 cm⁻¹ (O–H vibrations); 1026, 748, 527, and 472 cm⁻¹ (Si (Al)–O and Si–O–Si (Al) vibrations); 826 cm⁻¹ (Al–OH); and 3447 and 1646 cm⁻¹ (OH and HOH vibrations of the adsorbed water) (Ilić et al. Reference Ilić, Radonjanin, Malešev, Zdujić and Mitrović2016; Tang et al. Reference Tang, Zhang and Bao2016; Vizcayno et al. Reference Vizcayno, Gutiérrez, Castello, Rodriguez and Guerrero2010; Zhang et al. Reference Zhang, Redfern, Salje, Carpenter and Hayward2010).

Fig. 5 FTIR spectra of the initial and mechanically ground (for 80 and 160 min) muscovite

Mechanical activation can destroy the crystal structure of muscovite by breaking or weakening the O–H, Al–OH, Al–O–Si, and Si–O bonds (Kodama & Brydon Reference Kodama and Brydon1968; Papirer et al. Reference Papirer, Eckhardt, Muller and Yvon1990; Vizcayno et al. Reference Vizcayno, Gutiérrez, Castello, Rodriguez and Guerrero2010). Prolonged grinding gradually decreased the absorption intensity of the hydroxyl vibration bands at ~3623, 687, and 412 cm⁻¹, and the bands at 687 and 412 cm⁻¹ disappeared after 160 min of grinding, indicating that structural modifications occurred due to the partial dehydroxylation. The existence of bands arising from hydroxyl vibrations at 3622 cm⁻¹ after 160 min of grinding suggested that the dehydroxylation of muscovite was not completed. The decrease in absorption intensity of the bands at approximately 826, 748, 527, and 472 cm⁻¹ was attributed to the increase in the structural disorder of muscovite with prolonged grinding. In addition, the HOH bending vibration of the adsorbed water at ~3434–3447 cm⁻¹ became wider due to an increase in the amount of adsorbed water connected to the particle surface following grinding, and this finding was consistent with the results of the TG analysis.

Pozzolanic activity

According to the analysis above, the mechanical activation process decreased particle size, increased specific surface area, and caused the dehydroxylation reaction of muscovite powders, which increased the amorphous structure. Changes in the pozzolanic activity index were observed as a function of the grinding time, increasing from 36.98% to 95.02% with 200 min of grinding (Fig. 6). During the initial stages—within the first 40 min—the corresponding relative crystallinity of the mechanically activated muscovite was 93.39% according to XRD analysis, and the pozzolanic activity index increased to 61.24%, which still did not qualify it as an active pozzolan. This finding was mainly due to the high relative crystallinity of muscovite. At 60 min of grinding, the relative crystallinity decreased sharply decreased to 66.14%, and the pozzolanic activity index increased to 75.57%, which qualified the material for use as an active pozzolan (≥65% at 28 days according to the Chinese National Standard GB/T 2847-2005). After 160 min of grinding, the changes in the pozzolanic activity index were insignificant, and the result was consistent with the change in the relative crystallinity. Therefore, it could be inferred that the increase in the pozzolanic activity was the consequence of the increased specific surface area, reduced particle size, decreased relative crystallinity, and dehydroxylation caused by mechanical activation.

Fig. 6 Changes in the pozzolanic activity index for the initial and mechanically ground muscovite after 28 days of curing

The present study indicated that the pozzolanic activity index of muscovite could be improved by using mechanical activation because of the amorphous structure produced by the dehydroxylation reaction after grinding. The pozzolanic activity is the capacity of pozzolan to react with calcium hydroxide to form calcium silicate hydrates (Cheng et al. Reference Cheng, Huang, Li, Liu, Li and Wei2016). To further verify the pozzolanic activity of activated muscovite, the amount of calcium hydroxide consumed by the initial and mechanically activated muscovite (typical samples ground for 80 min and 160 min) at various hydration times was measured using the TG technique. Three different types of paste samples were prepared by mechanical mixing of lime, muscovite (ground for 0 min, 80 min, and 160 min), and water at a mass ratio of 3:7:7, and the paste samples were then cured in a curing chamber with relative humidity ≥95% and temperature of 20 ± 1°C for different lengths of time. The calcium hydroxide content in the pastes cured for 1, 3, 7, 14, and 28 days was detected using TG analysis. The results (Fig. 7) demonstrated that the Ca(OH)₂ content in mechanically activated muscovite–lime systems decreased significantly with curing time, especially at the early hydration age; the consumption of Ca(OH)₂ revealed that the mechanical activation caused a pozzolanic reaction between Ca(OH)₂ and activated muscovite to yield hydration products. However, the change in the Ca(OH)₂ content in the initial muscovite–lime system was insignificant, indicating that the initial muscovite was a relatively stable material with no pozzolanic activity. In addition, the consumption of Ca(OH)₂ presented an increasing trend with grinding and curing time, indicating that the pozzolanic activity of muscovite was improved via mechanical activation.

Fig. 7 Ca(OH)₂ contents in the paste samples vs the hydration time as determined by TG

In order to determine the morphology and chemical composition of the hydration products formed by pozzolanic reaction between mechanically activated muscovite (ground for 160 min) and lime, the activated muscovite–lime paste cured for 28 days was selected for SEM-EDS analyses. Based on the EDS analysis (Fig. 8, region A), the amorphous reticulate structural morphology (Fig. 8) in the activated muscovite–lime paste was identified as a C–S–H gel that solidified Al³⁺ and K⁺ during the hydration process. The SEM-EDS analysis results again demonstrated that mechanical activation could improve the pozzolanic activity of muscovite, and the major hydration product in the activated muscovite–lime system was an amorphous C–S–H gel.

Fig. 8 SEM images and EDS analysis of the activated muscovite–lime paste sample cured for 28 days