Effect of the SiO₂/Al₂O₃ Molar Ratio on the Microstructure and Properties of Clay-based Geopolymers: A Comparative Study of Kaolinite-based and Halloysite-based Geopolymers

Abstract

As 1:1 dioctahedral clay minerals, kaolinite and halloysite have similar chemical compositions. However, halloysite often possesses a nanotubular structure and special surface reactivity compared to platy kaolinite. The objective of this current work was to determine the effect of the SiO₂/Al₂O₃ ratio on the microstructure and properties of geopolymers derived from two kinds of kaolin: platy kaolinite and nanotubular halloysite. The chemical structures and compositions of the geopolymers obtained were characterized through X-ray diffraction (XRD), Fourier-transform infrared (FTIR) spectroscopy, and nuclear magnetic resonance (NMR), whereas the microstructural analysis was performed by scanning electron microscopy (SEM), the Brunauer–Emmett–Teller (BET) method, and N₂ physisorption analysis. The results indicated that calcined halloysite showed greater geopolymerization reactivity than calcined kaolinite. In addition, the mechanical properties of the clay-based geopolymers depended not only on the SiO₂/Al₂O₃ ratio but also on the morphology of the clay. Crystalline zeolite A and geopolymer were produced after alkali-activation of kaolin with a SiO₂/Al₂O₃ ratio of 2.5; these products possessed porous and heterogeneous microstructures having poor compressive strength. As SiO₂/Al₂O₃ ratios increased to >2.5, geopolymers with compact microstructure and high compressive strength were produced after alkali-activation of kaolin. Notably, at a given condition, halloysite-based geopolymers exhibited greater early compressive strength, more compactness, and more homogeneous microstructure than kaolinite-based geopolymers. This can be attributed to the nanotubular microstructure of halloysite, which can release more Si and Al during alkali activation than platy kaolinite. These results indicated that the various morphologies and microstructures among clays have significant impact on the microstructure and compressive strength of geopolymers.

Introduction

Alkali-based geopolymer is a kind of inorganic polymeric material comprising cross-linked tetrahedral [AlO₄] and [SiO₄] units, having a three-dimensional network structure, where hydrated alkali metal cations (e.g. Na⁺, K⁺, etc.) are distributed throughout the network to balance the negative charges of [AlO₄]^– (Amran et al., 2020; Najafi et al., 2020). As a new type of cementitious material, geopolymer has been regarded as a potential alternative to ordinary Portland cement (OPC) because it not only consumes less energy and emits less CO₂ during manufacturing but also exhibits better properties, including greater mechanical strength, better fire resistance, and greater durability (Farhan et al., 2019; Tognonvi et al., 2020; Chen et al., 2021). Normally, alkali-based geopolymer is synthesized by the dissolution of aluminosilicates into an alkaline activator, followed by a curing process (Zhang et al., 2021). In general, raw materials, such as clays (Liew et al., 2016; Werling et al., 2022), industrial waste (Shekhovtsova et al., 2018; Zhang et al., 2022), and other aluminosilicate-rich material with sufficient reactive alumina and silica (Nana et al., 2019; Wang et al., 2022), can be used for geopolymer preparation.

Clay minerals are naturally occurring minerals found around the world which are important raw materials in various domains (Mbey et al., 2019). Kaolinite (Al₂(OH)₄Si₂O₅, Kln) is a typical 1:1 dioctahedral clay mineral, which usually has a hexagonal flake or plate-like shape. Each layer of Kln contains a silica tetrahedral sheet and an alumina octahedral sheet and these sheets are bound by sharing oxygen atoms, while adjacent sheets are bound by hydrogen bonds (Deng et al., 2017; Li et al., 2020). Kln has been applied in the stabilization of contaminated soil (Liu et al., 2020), supplementary cementitious materials (Haw et al., 2020), and ceramic preparations (Prasad et al., 1991). Kln is also widely used for geopolymer preparation and geopolymerization mechanism analysis due to its high pozzolanic activity and simple chemical composition (Lolli et al., 2018; Wang et al., 2020). Compared to other precursors, such as red mud (He et al., 2012), fly ash (Cai et al., 2020), or illite (Hu et al., 2016), geopolymer derived from alkali activation of Kln shows greater compressive strength, shorter setting time, and better thermoelectrical properties. Many factors (e.g. SiO₂/Al₂O₃ molar ratio, liquid/solid ratio, and curing conditions etc.) have been reported previously to impact on the microstructure and mechanical properties of Kln-based geopolymers (Duxson et al., 2007; Heah et al., 2012; Qian et al., 2017; Yuan et al., 2016). Among these, the SiO₂/Al₂O₃ molar ratio is one of the most important factors in geopolymer preparation, which controls the microstructure, chemical composition, and properties of as-obtained products (Liew et al., 2016; Tian et al., 2021; Liu et al., 2022). Qian et al. (2017) found that the compressive strength of calcined Kln-based geopolymer increased from 2.1 to 31.2 MPa when the Si/Al ratio increased from 1.0 to 2.0, while it decreased to 5.5 MPa when the Si/Al ratio increased to 5.0. Khalifa et al. (2020) also concluded that during alkali-activation, geopolymer was formed when Si/Al > 1.5, and zeolite was formed when Si/Al < 1.5.

Halloysite (Al₂(OH)₄Si₂O₅·2H₂O, Hly) is also a 1:1 dioctahedral clay mineral, showing some similarities in chemical composition to Kln. However, due to an additional monolayer of water molecules between the unit layers of Hly, the mismatch between the tetrahedral SiO₄ and octahedral AlO₆ sheets of adjacent layers causes the wrapping of the 1:1 aluminosilicate layers, thus forming a nanosized tubular structure (Singh, 1996; Yuan et al., 2008; Yuan, 2016). Owing to its nanotubular structure, Hly normally possesses a larger specific surface area, less chemical stability, and lower ordering degree than Kln (White et al., 2012; Yuan et al., 2015). These different features between Hly and Kln might lead to a different geopolymerization process.

In fact, many previous studies have proposed that the geopolymerization behavior is also affected significantly by the morphologies and structure of Kln. For example, through different calcination technologies, Medri et al. (2010) and Nicolas et al. (2013) prepared various morphologies of metakaolinite. They found that metakaolinite powder with rounded agglomerates was more sensitive to the geopolymerization condition and required less water during geopolymerization than lamellar ones. Hollanders et al. (2016) compared the pozzolanic activity of Klns from different regions. They found that the raw Kln with a lower degree of ordering required a lower activation temperature, which had greater pozzolanic activity and a faster pozzolanic reaction rate. Zhang et al. (2016) noted different reactivity and geopolymerization behavior between Kln and Hly. They reported that the leaching rate of Si and Al from uncalcined Longyan kaolin (containing 31 wt.% Hly and 52 wt.% Kln) in strong alkaline solution was very close to that of the Suzhou kaolin (containing 91 wt.% Kln) calcined at 700°C for 1 h. Moreover, they also found that geopolymer prepared from Longyan kaolin exhibited greater geopolymerization reactivity (e.g. faster setting, faster development of compressive strength) than that from Suzhou kaolin (Zhang et al., 2012a).

Therefore, the use of Hly as a geopolymer precursor has attracted increasing attention. Kaze et al. (2018) used Cameroon-metahalloysite for geopolymer preparation and received products with a maximum compressive strength of ~27.5 MPa. They also found that by increasing the calcination temperature from 600 to 750℃, the reactive phase content of Hly increased, which shortened the setting time and improved the rheological behavior of geopolymer pastes (Kaze et al., 2020a). Zhang et al (2020b) studied the effects of calcination temperatures on the microstructure and compressive strength of Hly-based geopolymer and found that when calcination temperature reached 750℃, the Hly-based geopolymer exhibited a compact microstructure and the greatest compressive strength. All these results indicated that Hly is a promising precursor for geopolymer preparation.

The results above also demonstrated that the geopolymerization process of Hly differs from that of Kln (Izadifar et al., 2020; Zhang et al., 2020c). Although a recent study from Tchakouté et al. (2020) compared the microstructural and mechanical properties of Kln-based geopolymers with Hly-based ones, the Hly used in that study was spherical, while the dominant morphology of naturally occurring Hly is tubular (Yuan et al., 2015). In addition, both Kln and Hly samples contain many impurities. This may limit the widespread applicability of conclusions from any specific study. Moreover, the studies of Hly-based geopolymer are far less reported than those for other precursors (e.g. Kln, illite, and fly ash, etc.). The effects of the SiO₂/Al₂O₃ ratio on the microstructure and mechanical properties of Hly-based geopolymer remain unknown even though it can control the phase formation.

The objective of the current study was to compare the effect of the SiO₂/Al₂O₃ ratio on the microstructure and mechanical properties of platy Kln-based and nanotubular Hly-based geopolymers to enable a further understanding of the effect of morphology of 1:1 dioctahedral clay minerals on their geopolymerization process and on the geopolymers obtained. This work will provide a better insight into the various geopolymerization behaviors between Kln and Hly as well as the properties of as-obtained geopolymers.

Experimental

Materials

Two kinds of precursors, i.e. platy kaolinite (Kln) and nanotubular halloysite (Hly), were used for the preparation of geopolymers by alkali activation. Kln was collected from Maoming, Guangdong, and Hly was provided by I-Minerals, Inc., Vancouver, Canada. The chemical compositions of Kln and Hly measured by X-ray fluorescence (XRF) are presented in Table 1. The Kln and Hly powders were calcined in a muffle furnace at 750°C in air for 2 h (heating/cooling rate of 5°C/min) and were denoted as Kln_750℃ and Hly_750℃, respectively.

Table 1 Chemical composition of Kln and Hly (wt.%)

L.O.I.: loss on ignition

The alkaline activator was prepared by mixing analytical-grade NaOH pellets (purity ≥ 96%) with commercial sodium silicate (original SiO₂ 26.5 wt.%, Na₂O 8.5 wt.%, and water 65.0 wt.%) and Milli-Q water. The solution used in leaching was 10 mol/L NaOH solution. Solutions were stored for 24 h prior to use.

Preparation of Clay-based Geopolymers

Kln_750℃ or Hly_750℃ powders were mixed with the alkaline activator to prepare the geopolymers. The molar ratio of SiO₂/Al₂O₃ was set as 2.5, 3.0, or 3.5 while the molar ratio of Na₂O/Al₂O₃ was set to 1.0. However, due to the various minimum water requirements for workability of Kln_750℃ and Hly_750℃, the molar ratio of H₂O/Na₂O was set to 10.0 and 11.5 for Kln_750℃ and Hly_750℃, respectively. The resulting geopolymeric paste was poured into 20 mm × 20 mm × 12 mm plastic molds and covered with a thin film of polyethylene. The molded specimens were cured at 50℃ for 48 h and then were further cured at 80℃ for another 48 h. The hardened samples were demolded and sealed in plastic bags at ambient temperature. The products obtained were labeled Kln_750℃-X or Hly_750℃-X, where X represents the molar ratio of SiO₂/Al₂O₃ of the raw materials. For example, Kln_750℃-2.5 is the alkali-activation product with a SiO₂/Al₂O₃ molar ratio of 2.5.

Characterization of Clay-based Geopolymers

1 g of Kln_750℃ or Hly_750℃ was mixed with 30 g of 10 mol/L NaOH solutions at 25 ± 2°C in a shaker for various times. Then, after centrifugation and filtration, the leachate was separated. The concentrations of Si and Al dissolved from raw materials were determined by an iCAP 7000 Series inductively coupled plasma optical emission spectrometry instrument (Thermo Fisher Scientific Inc., Waltham, Massachusetts, USA).

The setting times of pastes prepared by mixing the Kln_750℃ or Hly_750℃ with alkaline activator were tested by a Vicat apparatus. 7- and 28-day compressive strengths of geopolymers were performed on a YAW-300D Compression Resistance Tester (Schlikör, LiXian, Zhejiang, China) with a loading rate of 500 N/s.

Thermogravimetric (TG) analyses were performed on a Netzsch STA 409PC instrument (Selb, Bayern, Germany) for Kln and Hly. Powdered samples (~ 10 mg) were analyzed by heating in a corundum crucible from 30 to 1000°C at a rate of 10°C/min.

The XRD patterns of clays, calcined clays, and 28-day geopolymers were collected on a Bruker D8 Advance diffractometer (Mannheim, Karlsruhe, Germany), operating at 40 kV and 40 mA using CuKα radiation. The specimens were powdered finely to investigate their diffraction from 3 to 70°2θ with a scanning speed of 3°2θ/min.

The FTIR spectra of clays, calcined clays, and 28-day geopolymers were recorded on a Bruker Vertex 70 spectrometer (Mannheim, Karlsruhe, Germany). The products were mixed with 0.9 mg of sample and 90 mg of KBr and powdered finely before the mixture was pressed into a disk for analysis. FTIR spectra of the disks were collected at 4000 to 400 cm^–1 with 64 scans, at a resolution of 4 cm^–1.

The SEM micrographs and energy-dispersive X-ray (EDX) spectroscopy results of clays, calcined clays, and 28-day geopolymers were obtained using an SU8010 field-emission scanning electron microscope (Hitachi, Tokyo, Japan) at an accelerating voltage of 15 kV. Specimens were prepared for analysis by being anchored tightly on the surface of conducting tape, coated with a layer of platinum, and then transferred to the microscope.

Measurements of pore-size distribution and the total pore volume of 28-day geopolymers’ powder were carried out by N₂ physisorption analysis in a Micromeritics ASAP 2020 instrument (Micromeritics Co., Norcross, Georgia, USA) at liquid-nitrogen temperature (–196°C). Before measurements, the products were outgassed at defined temperatures for 24 h at 393 K at the degassing port. The total pore volume was obtained from N₂ uptake at a relative pressure of 0.97.

Solid-state ²⁹Si cross-polarization magic-angle-spinning (CP/MAS) NMR spectra and ²⁷Al MAS NMR spectra of 28-day geopolymers’ powder were recorded on a Bruker Avance III600 spectrometer (Mannheim, Germany). For ²⁹Si CP/MAS NMR spectra, the magnetic field strength was 14.1 T with a resonance frequency of 119.2 MHz and was recorded using a contact time of 6 ms, a π/2 pulse length of 2.3 μs, a recycle delay of 2 s and a spinning rate of 10 kHz. For ²⁷Al MAS NMR spectra, the magnetic field strength was 14.1 T with a resonance frequency of 156.4 MHz and was recorded using a small-flip angle technique with a pulse length of 0.5 μs (< π/12), a recycle delay of 1 s and a spinning rate of 14 kHz. The chemical shifts of ²⁷Al and ²⁹Si were given in ppm referenced to 1 mol/L Al(NO₃)₃ and tetramethylsilane (TMS), respectively.

Results and Discussion

Characterization of Kln, Hly, and their Heating Products

TG-DSC curve

TG curves (Fig. 1) showed mass losses of 1.42% for Kln and 2.78% for Hly below 450℃, which were ascribed to the dehydration of Kln and Hly (loss of physical adsorbed water and/or interlayer water). Another major mass loss occurred in the temperature range from 450 to 650℃ and was associated with the dehydroxylation of Kln and Hly. Above 650℃, the mass loss continued but to a small extent up to 1000°C. The total mass losses for Kln and Hly were ~14.21% and 15.38%, respectively.

Fig. 1 TG-DSC curves of Kln and Hly

In the DSC curve, broad endothermic peaks at ~520 and 507℃ for Kln and Hly, respectively, corresponded to dehydroxylation of Al–OH and Si–OH. This result indicated that, compared to Kln, the dehydroxylation of Hly starts at a lower temperature. In addition, a narrow exotherm appeared at 998°C for Kln, which should be due to the formation of spinel-like γ-Al₂O₃ or an alumina-rich mullite (Okada et al., 1986; Sonuparlak et al., 2005). However, this exothermic event (990°C) of Hly during the calcination should be attributed to the formation of nanosized γ-Al₂O₃ based on the ²⁹Si and ²⁷Al MAS NMR results (Smith et al., 1993). By using high-resolution TEM, Yuan (2016) also identified the formation of nanosized γ-Al₂O₃ when Hly was calcined at 1000°C.

XRD results

For Kln, most of the XRD reflections (Fig. 2) belonged to kaolinite with the strongest reflection at 12.2°2θ (with a d ₀₀₁ of 7.25 Å). A small peak at 26.7°2θ indicated the presence of a small amount of quartz. When calcined at 750℃, a new broad reflection emerged at ~21.1°2θ due to the dehydroxylation of Kln. This result indicated that the crystal structure of Kln had been destroyed and the Kln transformed into metakaolinite after heat treatment at 750℃.

Fig. 2 XRD patterns of Kln, Hly, and their calcination products

The interlayer distance, d ₀₀₁, of Hly (Fig. 2) was similar to that of Kln. This also demonstrated that the Hly used in this study is 7 Å-halloysite, from which the water monolayer was lost. The reflections belonging to halloysite disappeared and the existence of a broad reflection emerged at ~21.1°2θ when Hly was calcined at 750℃. This also indicated that the structure of Hly was destroyed and an amorphous phase formed.

SEM images

The SEM images (Fig. 3a,b) showed significantly different morphologies between Kln and Hly. The morphology of Kln was lamellar while that of Hly was nanotubular. When calcined at 750℃, they kept their original shapes. This result demonstrated that heating at 750℃ did not change the morphology of Kln and Hly. Accordingly, their specific surface areas did not change significantly either. Determined by BET, the specific surface areas of the Kln and Kln_750℃ were 14.73 and 15.53 m²/g, respectively, while Hly and Hly_750℃ were 28.29 and 27.62 m²/g, respectively. Specifically, the specific surface area of Hly was nearly twice as large as that of Kln.

Fig. 3 SEM images of a Kln, b Hly, c Kln_750℃, and d Hly_750℃

FTIR results

The spectrum of Kln (Fig. 4) exhibited four absorption bands ranging from 3700 to 3600 cm^–1, where the peaks at 3694, 3669, and 3651 cm^–1 were attributed to the O–H stretching of inner-surface hydroxyl groups and 3620 cm^–1 was attributed to O–H stretching of inner hydroxyl groups (Madejová & Komadel, 2001). In addition, the absorption at 913 cm^–1 was also attributed to O–H vibration of inner hydroxyl groups. However, the absorption bands at 3669 and 3651 cm^–1 did not appear in the FTIR spectrum of Hly, demonstrating the lower ordered structure of Hly compared to Kln (Parker, 1969). The peak at 1103 cm^–1 was assigned to perpendicular Si–O stretching while bands at 1032 and 1009 cm^–1 were related to the in-plane Si–O stretching vibration. Moreover, the absorptions of 795, 754, 697, and 432 cm^–1 were all attributed to Si–O vibrations. The band at 539 cm^–1 was assigned to Al–O–Si deformation and 471 cm^–1 was attributed to Si–O–Si deformation (Madejová & Komadel, 2001). The peaks ranging from 1110 to 430 cm^–1 in the spectrum of Kln were similar to those of Hly, indicating the similar structures of Kln and Hly at the molecular scale.

Fig. 4 FTIR spectra of Kln, Hly, and their calcination products

After calcination at 750℃, the vibration corresponding to the hydroxyl group of Kln disappeared, along with the broad band at 3433 cm^–1 which was attributed to the O–H of adsorbed water. A broad and asymmetric band centered at 1082 cm^–1 was assigned to the symmetrical stretching vibration of Si–O–Si and Si–O–Al bonds. In addition, the broad bands at 800 and 466 cm^–1 were both ascribed to the Si–O–Si stretching vibration from the amorphous silica. These results demonstrated that the structure of Kln was destroyed and amorphous SiO₂ was formed (Yuan et al., 2012). Similar results can also be found in the FTIR spectrum of Hly_750℃.

Dissolution behavior of Kln_750℃ and Hly_750℃

The concentrations of Si and Al leached from the clays in NaOH solutions increased with leaching time up to 48 h and then remained nearly unchanged as leaching time increased (Table 2 and Fig. 5). The concentration of Si and Al leached from Kln_750℃ increased from 205.2 to 8080.5 ppm and 114.7 to 4188.7 ppm, respectively. The concentration of Si and Al leached from Hly_750℃ increased from 281.4 to 9253.3 ppm and 189.9 to 4274.0 ppm, respectively.

Table 2 Concentration of Al and Si leached from Kln_750℃ and Hly_750℃ in NaOH solutions

Fig. 5 Concentrations of Si and Al dissolved from clays in NaOH solution

The concentrations of Si and Al leached from Hly_750℃ were greater than those from Kln_750℃ under the same leaching conditions; the leaching times from Hly_750℃ were shorter than from Kln_750℃. This result confirmed that the pozzolanic activity of Hly is greater than that of Kln, which may favor geopolymerization. The dissolution of the aluminosilicate precursor is the first step of geopolymerization, which occurs mainly in the early stages, followed by polycondensation of oligomers. This was also evidenced by isothermal calorimetry measurement in previous studies (Sun & Vollpracht, 2017; Zhang et al., 2012b, 2013), which found two obvious exothermic peaks 5 h after geopolymerization. Therefore, the early stage of dissolution plays an important role in the formation of geopolymeric gels. In the first 3 h, the Si and Al concentrations leached from Hly_750℃ were 2690.5 and 1526.7 ppm (Table 2), which was nearly three times greater than those from Kln_750℃. This may lead to the improvement in the degree of geopolymerization.

Properties of Clay-based Geopolymers

Setting Time

The setting times shown in Fig. 6 illustrated that as the SiO₂/Al₂O₃ ratio increased, the initial setting time increased from 93 to 212 min and final setting time was increased from 221 to 324 min for Kln_750℃-based geopolymeric pastes; and the initial setting time was increased from 52 to 146 min and final setting time was increased from 198 to 263 min for Hly_50℃-based geopolymers. The alkalinity of the solution decreased gradually with the increase in the SiO₂/Al₂O₃ ratio, which reduced the dissolution degree of Si and Al from precursors and thus prolonged the setting time. This was also evidenced by Zhang et al. (2013), who reported that the greater the alkali concentration, the more rapid the geopolymerization.

Fig. 6 Initial and final setting time of clay-based geopolymeric pastes

As was observed, the initial and final setting times were longer for Kln-based geopolymer than Hly-based geopolymer at a given SiO₂/Al₂O₃ ratio. This may be due to the fact that the greater and faster dissolution of Si and Al from Hly_750℃ than from Kln_750℃ resulted in the faster polycondensation of oligomers when mixing the clay and activator. Thus, the geopolymeric paste was hardened more quickly for alkali-activated Hly_750℃ than Kln_750℃ (Zhang et al., 2012a).

Compressive Strength of Clay-based Geopolymers

Clay-based geopolymers with a SiO₂/Al₂O₃ molar ratio of 2.5 showed rather low 7- and 28-day compressive strengths (Fig. 7). When the molar ratio of SiO₂/Al₂O₃ increased to 3.0, the compressive strengths were improved significantly. The 7-day compressive strength increased to 27.7 MPa for Kln_750℃-based geopolymers, and increased to 58.7 MPa for Hly_750℃-based geopolymers. This result demonstrated that increasing the SiO₂/Al₂O₃ ratio could improve significantly the mechanical properties of geopolymers. When the SiO₂/Al₂O₃ ratio increased to 3.5, the compressive strength of Kln_750℃-based geopolymers increased slightly, whereas that of Hly_750℃-based geopolymers decreased to 51.0 MPa, which may be attributed to the decreased alkalinity of the activator. Hly_750℃-based geopolymers showed greater compressive strength than Kln_750℃-based geopolymers, resulting from the greater geopolymerization reactivity for Hly than Kln.

Fig. 7 Compressive strength of Kln- and Hly-based geopolymers

Note that, after 28 days of ageing, the compressive strength of Kln_750℃-based geopolymer increased slightly; however, the compressive strength of Hly_750℃-based geopolymers decreased, possibly because of the curing conditions. When cured at high temperature (80℃), Hly_750℃-based geopolymeric paste hardened too fast to dissolve sufficiently, accompanied with fast evaporation of water, leading to the destruction of the geopolymer matrix. This process, therefore, may play a detrimental role in the development of compressive strength (Zhang et al., 2009, 2021). In comparison, Kln_750℃-based geopolymers reacted continuously in the later stage of geopolymerization with improvement of compressive strength.

Structure of Clay-based Geopolymers

XRD Results

The XRD patterns of Kln_750℃-2.5 and Hly_750℃-2.5 showed some reflections indexed to zeolite A with a d ₀₀₁ of 12.3 Å, which indicated that zeolite was formed (Fig. 8). In fact, previous reports found that geopolymers contained nanocrystalline zeolites, which can be crystallized under certain conditions, including high curing temperature, high humidity, etc. (Provis et al., 2005; Wang et al., 2017). Alkali-activation of Kln with a small Si/Al ratio induced the formation of zeolite easily. Qian et al. (2017) also found that zeolite A was formed after alkali-activation of metakaolinite at a Si/Al ratio of 1.0. Hounsi and Lecomte (2013) studied alkali-activated Kln using 8 M NaOH solution and found that reyerite was formed. Similarly, Hly_750℃-2.5 also contained zeolite A, but a smaller amount, according to the lower intensity of corresponding characteristic peaks (Fig. 8b). In addition, a broad reflection centered at 27.7°2θ was observed in the XRD patterns of Kln_750℃-2.5 and Hly_750℃-2.5. Compared to XRD patterns of Kln_750℃ and Hly_750℃ (Fig. 2), the center of the broad reflection shifted to higher regions after alkali-activation, which indicated that the network was reorganized and geopolymer had formed (Kaze et al., 2020a; Nkwaju et al., 2019).

Fig. 8 XRD patterns of a Kln- and b Hly-based geopolymers

When the SiO₂/Al₂O₃ ratio increased to >3.0, the zeolite crystal disappeared, accompanied by an increase in intensity of the broad peak centered at ~27.7°2θ, indicating that the amount of amorphous geopolymer was increased. The quartz, however, remained intact after geopolymerization for Kln-based geopolymer due to its chemical stability.

These results demonstrated that the SiO₂/Al₂O₃ ratio affected the phase composition, which controlled the formation of zeolite or geopolymer for alkali-activated Kln_750℃ or Hly_750℃. Their products showed no significant differences in the mineralogical phase under the same SiO₂/Al₂O₃ ratio, however.

FTIR Results

The absence of significant differences in local environments of the OH and other FTIR-active modes indicated that no significant differences transpired at the molecular scale between Kln_750℃- and Hly_750℃-based geopolymers (Fig. 9). Furthermore, the peaks of the geopolymers were broad, which demonstrated that the alkali-activation products were highly amorphous.

Fig. 9 FTIR spectra of Kln- and Hly-based geopolymers

The broad peaks in the range 3444 to 3464 and at 1650 cm^–1 were attributed to the stretching and bending vibrations of O–H from physically adsorbed water (Madejová & Komadel, 2001). The absorption at ~703 cm^–1 corresponded to the bending vibration of Si–O–Al^IV, while those at ~581 and ~439 cm^–1 should be associated with the stretching vibration of Al–O and the bending vibration of Si–O–Si, respectively. In addition, a shoulder observed at ~882 cm^–1 was related to the presence of non-bridging-oxygens (e.g. Al–O–, Si–O–) resulting from the geopolymerization process.

Notably, for all geopolymers, a major band which appeared between 1200 and 900 cm^–1 was normally denoted the “main band,” which was attributed to an asymmetric stretching vibration of Si–O–T (T: Si or Al). This band has been used widely to investigate the chemical properties of geopolymers (Rees et al., 2007; Zhang et al., 2020a). Compared to FTIR spectra of Kln_750℃ and Hly_750℃ (Fig. 2), the main band shifted significantly to a lower wavenumber after alkali-activation, owing to the increased amount of Si–OH and the substitution of Al in the silicate networks, leading to the decreased molecular vibrational force-constant (Rees et al., 2007; Zhang et al., 2020c). This displacement also reflected that Kln or Hly had dissolved and the geopolymers were formed. With the increase in the SiO₂/Al₂O₃ ratio, the wavenumber of main bands for geopolymers increased slightly. This may be attributed to the incorporation of more Si into the geopolymer matrix or the more unreacted raw materials, which warrants further study (Bewa et al., 2020).

NMR Results

The ²⁷Al MAS NMR spectra of Kln_750℃ and Hly_750℃ (Fig. 10a) both showed three resonances centered at approximately 6, 31, and 59 ppm, which were assigned to Al^VI, Al^V, and Al^IV, respectively. After reacting with alkaline solution, the ²⁷Al MAS NMR spectra of all samples had a strong resonance at ~61 ppm, accompanied by the disappearance of Al^V and Al^VI. Therefore, Al was mainly in four-coordination (Al^IV) in the geopolymer despite the SiO₂/Al₂O₃ ratios, indicating that increasing the SiO₂/Al₂O₃ ratio did not significantly change the local structure of the Al environment in the geopolymer framework.

Fig. 10 ²⁷Al MAS NMR spectra of a Kln_750℃ and its alkali-activation products and b Hly_750℃ and its alkali-activation products; ²⁹Si CP/MAS NMR spectra of c Kln_750℃ and its alkali-activation products; and d Hly_750℃ and its alkali-activation products

The ²⁹Si CP/MAS NMR spectra of Kln_750℃ and Hly_750℃ showed one broad resonance centered at approximately –101 ppm (Fig. 10b), which was attributed to a Q³ Si environment, namely, Si was linked to three other Si atoms through oxygens (Maia et al., 2014). After geopolymerization, this peak shifted to a higher field position and decreased in intensity. This Si chemical shift was attributed to the Q⁴(mAl) environments, which was affected by Al substitution (Skibsted & Andersen, 2013). Previous studies have reported that for replacement of Si neighbors, as Q⁴(mAl), each additional [AlO₄] would increase the chemical shift by approximately 5 ppm (Walkley & Provis, 2019). Therefore, with the increase of SiO₂/Al₂O₃ ratios, this peak shifted to lower field position.

Due to the fact that each peak centered between –86 and 90 ppm was not unequivocally assignable to a single Si environment, Gaussian peak deconvolution of the ²⁹Si NMR spectra for all geopolymers was conducted to analyze Q⁴(mAl) (0 ≤ m ≤ 4) Si environments and obtain more specific information about the Si coordination environments. Deconvolution of geopolymers produced five peaks located at approximately –108, –100, –93, –87, and –80 ppm (Fig. 11), which were attributed to Q⁴(0Al), Q⁴(1Al), Q⁴(2Al), Q⁴(3Al), and Q⁴(4Al), respectively.

Fig. 11 Deconvolution results for ²⁹Si CP/MAS NMR spectra of clay-based geopolymers: a Kln_750℃-3.0; b Kln_750℃-3.5; c Hly_750℃-3.0; and d Hly_750℃-3.5

All geopolymers consisted mainly of Q⁴(2Al) and Q⁴(3Al) (Table 3), which demonstrated that one [SiO₄] tetrahedron connected mainly to two or three [AlO₄] tetrahedra in this geopolymer network. In addition, the fraction of Q⁴(mAl) structural units rich in Si (i.e. Q⁴(0Al), Q⁴(1Al), and Q⁴(2Al)) increased as the SiO₂/Al₂O₃ ratios increased. When the SiO₂/Al₂O₃ ratios increased from 3.0 to 3.5, the value of (Q⁴(0Al) + Q⁴(1Al) + Q⁴(2Al))/(Q⁴(4Al) + Q⁴(3Al)) increased from 0.77 to 1.29 and from 1.17 to 2.43 for Kln-based and Hly-based geopolymers, respectively. At the same SiO₂/Al₂O₃ ratio, the geopolymers derived from Hly contained more Si than those derived from Kln, which may be one reason why Hly-based geopolymers exhibited better properties than Kln-based geopolymers (He et al., 2016; Qian et al., 2017).

Table 3 Fractional areas of the constituent NMR peaks

SEM Results

Scanning electron micrographs (Fig. 12a) revealed that Hly_750℃-2.5 exhibited a rather loosely bonded microstructure with some macropores and cracks, which corresponded to the low compressive strength. At high magnification, some nanotubular species were loosely stacked together without the formation of continuous integrity (Fig. 12d). These nanotubular species should be the unreacted Hly_750℃ resulting from incomplete dissolution. In addition, some prismatic particles formed, which were comprised mainly of Na and O with traces of C, Si, and Al elements according to EDX analysis (Fig. 12g, #1). This particle should be attributed mainly to the NaOH, while a small amount of Na₂CO₃ formed through the reaction between NaOH and atmospheric CO₂. Similarly, Kln_750℃-2.5 also had a rather uncompacted and inhomogeneous microstructure (Fig. 12h,k), which thus exhibited low compressive strength.

Fig. 12 SEM images of Hly- and Kln-based geopolymers: a, d Hly_750℃-2.5; b, e Hly_750℃-2.0; c, f Hly_750℃-3.5; g EDX result of Hly-based geopolymer; h, l Kln_750℃-2.5; i, m Kln_750℃-3.0; k, n Kln_750℃-3.5

Hly_750℃-3.0 had significant continuity texture with an homogeneous and compact microstructure (Fig. 12b,e). The densification of alkali-activated products was consistent with the improvements in compressive strength of the geopolymers. The EDX analysis showed that the geopolymer matrix was composed mainly of Na, Si, Al, and O elements, indicating the formation of sodium aluminate silicate hydrate (N–A–S–H). In Fig. 12c, Hly_750℃-3.5 still showed a compact microstructure with micropores. However, unreacted Hly_750℃ was still embedded in the geopolymer matrix (Fig. 12f), causing the less homogeneous structure of Hly_750℃-3.5 and decreased the compressive strength.

Notably, the microstructure of geopolymers derived from Kln_750℃ was obviously less homogeneous and less compact than from Hly_750℃ under the same conditions. Many unreacted Kln_750℃ crystallites were easily found to be embedded in the geopolymeric matrix, which caused the formation of many cracks and destroyed significantly the homogenous structure of the geopolymer (Fig. 12l,m). This may account for the lower 7-day compressive strength of Kln_750℃- compared to Hly_750℃-based geopolymers. However, further studies and more technologies are required to research the similar 28-day strengths between Kln_750℃- and Hly_750℃-based geopolymers.

These results also confirmed that the soluble SiO₂ was very important for the geopolymer formation. The lack of soluble SiO₂ would inhibit the dissolution of minerals and impede the polycondensation of [SiO₄] and [AlO₄] oligomers even using a very highly alkaline solution for alkali-activation (Liew et al., 2016; Singh et al., 2005). On the other hand, the morphological differences between Kln and Hly also affect the microstructure of the geopolymers. Nanotubular Hly appeared to possess greater geopolymerization reactivity than platy Kln according to the more compact and homogeneous microstructure for Hly_750℃- than for Kln_750℃-based geopolymers.

Pore Distribution

The pore-size distribution curves of all geopolymers in Fig. 13a exhibited a unimodal distribution. A broad peak occurred in the range from 10–100 nm with a center at 35 nm, which was attributed to capillary pores in the geopolymer matrix (Ma et al., 2013). The broader-shaped peak of Kln_750℃-based geopolymer rather than the narrower Hly_750℃-based peak indicated that the pore distribution of Kln_750℃-based geopolymer has a lower degree of homogeneity than that of Hly_750℃-based geopolymer. Furthermore, the SiO₂/Al₂O₃ ratio was poorly related to the pore-size distribution and cumulative pore volume, at least in the mesopore range. This was shown by the similar average pore diameters of all the geopolymers (~23 nm) (Table 4).

Fig. 13 a Pore-size distribution and b cumulative pore volume of clay-based geopolymers

Table 4 Pore parameters and specific surface area of geopolymers

It is of note that although the pore volume of the Kln_750℃-based geopolymer was less than that of the Hly_750℃-based geopolymer (Table 4), Kln_750℃-based geopolymer exhibited lower compressive strength than Hly_750℃-based geopolymer. Table 4 also shows that Kln_750℃-based geopolymers had much lower specific surface area than Hly_750℃-based geopolymers, which demonstrated that Hly_750℃-based geopolymers may possess more pores (micropore and mespore) than Kln_750℃-based ones. This result revealed that, at least in the mesopore range, porosity has no direct relation with compressive strength. As observed from SEM images, many macropores and cracks were present in the Kln_750℃-based geopolymer matrix (Fig. 12e,f), but these macropores or cracks could not be characterized by N₂ physisorption analysis, which was also an important factor influencing the compressive strength. On the other hand, the presence of the unreacted particles (e.g. NaOH, Na₂SiO₄) would cover the surface of a geopolymer, thus reducing the pore volume (Zhang et al., 2020c).

General Discussion

Recently, owing to its special microstructure and high reactivity, the use of Hly for preparation of geopolymers has attracted increasing attention (Kaze et al., 2018; Blaise et al., 2019; Zhang et al., 2021; Martina et al., 2022; Navid et al., 2023). Kaze et al. (2018) investigated the effects of NaOH concentration on the properties of Hly_600℃-based geopolymers. As the NaOH concentration increased, the compressive strengths increased at first and then decreased. When NaOH concentration increased to 10 mol/L, the compressive strength reached 28 MPa. However, the detailed SiO₂/Al₂O₃ ratio of raw materials was unclear. The same authors also compared mechanical and thermal properties of Hly and Kln (Kaze et al., 2022; Tchakouté et al., 2020) and concluded that Hly-based geopolymer showed better properties than Kln-based geopolymer because Hly possessed a smaller particle size and greater specific surface area than Kln. However, the calcined Hly they used was spherical; this may result in different geopolymerization behavior in contrast to Hly with a nanotubular shape. For example, they stated that the spherical shapes of Hly could provide nucleation sites necessary to bolster the geopolymerization process, which sites were not found in the current study. Besides, the clays (three kinds) used in Kaze’s study contained many impurities, such as illite, anatase, and quartz and the calcination of clays was incomplete, which may influence the geopolymerization of Hly. In addition, Zhang et al. (2021) compared various curing conditions on the mechanical properties and microstructure of Hly-based geopolymers and found that as the curing temperature increased to ≥50℃, the compressive strengths decreased with ageing time. This is consistent with the present study, in which the compressive strength of Hly-based geopolymers decreased with ageing. In contrast, the compressive strength of Kln-based geopolymers increased with ageing, which confirmed that the optimal curing condition is related to the reactivity of precursors and high curing temperature is not suitable for highly reactive precursors.

Many other studies have been carried out concerning the improved properties of Hly-based geopolymers (Blaise et al., 2019; Zhang et al., 2020b; Kaze et al., 2021), utilization of Hly as an additive in geopolymers (Kaze et al., 2020b; Nemaleu et al., 2021; Navid et al., 2023), and application of Hly-based geopolymers (Barrie et al., 2015; Martina et al., 2022). However, these studies focused more on the properties of Hly-based or Hly-related geopolymers than on the nanotubular morphology of Hly in geopolymerization, which is interesting clay-related research. Although nanotubular Hly shows similar chemical composition to plate-like Kln, their microstructures and morphologies are quite different. These differences between Kln and Hly also cause the different physicochemical properties. Therefore, Hly shows not only similar but also different behavior to Kln during alkali-activation.

The results obtained demonstrated that, as with Kln, the alkali-activation products from Hly were zeolite A and the geopolymer when the SiO₂/Al₂O₃ ratio was 2.5. Owing to the formation of different phases with large amounts of pores, the alkali-activation products derived from Kln and Hly both showed low compressive strength. Likewise, when SiO₂/Al₂O₃ ratio was increased to >3.0, the alkali-activation of Hly produced mainly geopolymers analogous to the alkali-activated Kln-based geopolymers. Sufficient amounts of soluble SiO₂ induced the polycondensation of oligomers to form geopolymers with compact and homogeneous structures (Ferone et al., 2015; Valentini, 2018), which improved the compressive strength.

However, the above findings also indicated that the microstructure of the Hly-based geopolymer differed significantly from the Kln-based geopolymer. This phenomenon can be attributed to the different morphologies of Kln and Hly. As mentioned above, Kln consisted of stacked repeating layers to form a platy morphology. Normally, the particle size of Kln is a few microns, and the lamellar thickness is hundreds or thousands of nanometers. In contrast, Hly usually adopts a rolling of the unit layer to form a tubular morphology. The diameter of the tubular particles is nanometers, while the thickness of the tubular wall is a few nanometers, which is much smaller than that of Kln (Yuan et al., 2015). According to Davidovits (2011), the Kln is attacked by alkaline solution at the edge and basal surfaces and inside the structure, layer by layer, during geopolymerization. Many unreacted Kln particles were present in Kln-based geopolymers due to insufficient dissolution. However, for Hly, both outside and inside surfaces of the Hly nanotube can be attacked by alkaline solution (Fig. 14), the dissolution of Si and Al from Hly was, thus, easier and quicker than from Kln, which was confirmed by the dissolution results (Table 2). This may accelerate the alkali-activation to improve the geopolymerization degree at an early stage (Fig. 6).

Fig. 14 Schematic diagram of alkali-activated Kln- and Hly-based geopolymers

Significantly, the different structures between Kln and Hly might also cause the different geopolymerization reactivity. Compared to Kln, the formation of naturally occurring Hly is prone to form defects under geological effects (Joussein et al., 2005). Therefore, the degree of structural ordering of nanotubular Hly is obviously lower than that of plate-like Kln, resulting in the greater reactivity for Hly than Kln. On the other hand, Hly has a larger specific surface area than Kln, which can provide more contact sites for activators, leading to more release of Si and Al and, thus, a greater degree of geopolymerization than Kln (Zhang et al., 2012a).

Furthermore, the unique properties of Hly might also cause different geopolymerization behavior from Kln. For example, previous reports indicated that the unit mass of the number of surface active points in Hly is greater than that of Kln (Tan et al., 2015; Yuan et al., 2008). In addition, the outer surfaces of the nanotubes of Hly will form hydroxyl groups after calcination at 600–900℃; however, no similar reports have been made about the calcination of Kln (Yuan et al., 2012). These effects on the microstructure and composition of as-synthesized geopolymers should also be considered in future studies.

Conclusions

The dependence of the microstructure and compressive strength of geopolymers derived from platy kaolinite and nanotubular halloysite on the SiO₂/Al₂O₃ ratio was investigated here by means of a combination of spectroscopic and microscopic techniques. When the SiO₂/Al₂O₃ ratio was 2.5, both alkali-activated kaolinite and alkali-activated halloysite exhibited a loosely bound and inhomogeneous microstructure due to the formation of zeolite A. When the SiO₂/Al₂O₃ ratio was 3 or 3.5, the alkali-activation product was a geopolymer which showed compact and dense microstructure with high compressive strength.

In addition, compared to kaolinite-based geopolymers, the microstructure of halloysite-based geopolymers was more compact and homogeneous, and with fewer unreacted particles, which thus exhibited greater early compressive strength. This can be attributed to the different morphologies between halloysite and kaolinite. Halloysite possesses a nanosized tubular morphology, which showed a smaller particle size and larger specific surface area than kaolinite. Therefore, the OH^– can attack readily the halloysite tube from both the inside and outside surfaces, leaching more Si and Al, and forming larger amounts of geopolymer. Moreover, the nanosized particles did not destroy significantly the geopolymer matrix, which results in a better mechanical performance for halloysite-based geopolymers than kaolinite-based geopolymers.

These findings indicated that the mechanical properties and microstructure of clay-based geopolymers are not only influenced by the SiO₂/Al₂O₃ ratio but also by the microstructure and morphology of the raw materials. Nanotubular halloysite is a promising geopolymer precursor, which showed greater geopolymerization reactivity than plate-like kaolinite.