Thermal Evolution of Geopolymer in the Process of High-Temperature Treatment

Abstract

More and more attention had been given to geopolymers (GPs) over the last decades because of an increasing urgency to search for high-performance and/or environment-friendly alternatives to traditional Portland cement. In addition, geopolymer technology could also provide an innovative approach to prepare advanced ceramic products by overcoming problems faced in the conventional preparation technology. With only the need to go through appropriate thermal treatment procedure, geopolymers could be directly in situ transformed into advanced ceramics such as leucite or pollucite with adjustable microstructures, mechanical properties, coefficient of thermal expansion, and melting points. In the process of high-temperature treatment, multiple parameters, such as the composition of geopolymer, treatment temperature, thermal insulation, etc., would affect the phase composition and microstructure of the resulting products. In the present chapter, two kinds of mixed-alkali metal ion-activated geopolymer systems, Cs(1-x)LixGP (where x = 0, 0.1, 0.2, and 0.3) and Cs(1-x)NaxGP (where x = 0, 0.1, 0.2, 0.3, and 0.4), respectively, were designed and prepared. Phase composition, microstructure evolution, and thermal expansion behaviors of the ceramics derived from the geopolymers were characterized and the effects of ion substitution on the thermal evolution of geopolymer were evaluated.

1. Introduction

More and more attention had been given to geopolymers over the last decades because of an increasing urgency to search for high-performance and/or environment-friendly alternatives to traditional Portland cement [1, 2, 3, 4, 5, 6]. They were typically synthesized at ambient or elevated temperature by alkali-activated process between alkali-activated solutions and aluminosilicates. And the source of aluminosilicates was very extensive and inexpensive, such as metakaolin, fly ash, blast furnace slag, etc. [7, 8, 9, 10, 11]. Most notably, there was almost no emission of greenhouse gases throughout the preparation process, which made geopolymer technology more competitive in terms of environmental and economic advantages.

From their composition, geopolymers were a class of amorphous materials consisting of cross-linked [AlO₄]⁻ and [SiO₄] tetrahedra units, and the negative charge aluminum in fourfold coordination was balanced by alkali metal or alkaline earth metal cation, such as Na⁺, K⁺, Mg²⁺_, and Ca²⁺ [12, 13, 14, 15]. As for geopolymers, alkali metal hydroxides were involved in many processes during the alkali-activated reaction or geopolymerization reaction, including accelerating the dissolution of aluminosilicates, stabilizing the solution species and colloids, reducing the electrostatic repulsion between the anions, and promoting gel formation and rearrangement [16, 17, 18]. Therefore, as for alkali metal hydroxide, the OH⁻ group in it mainly accounted for creating the reactive precursors, while the alkali cations would play an important role in catalyzing gel formation, acting as a structure-directing agent. Since the alkali metal was one of the most easily alterable constituents in the framework of geopolymer, it was very valuable to investigate the effects of alkali cation species on the evolution of microstructure and properties of the geopolymer system [19]. Fernández-Jiménez et al. proved that Na⁺ was much more conducive to the coagulation and precipitation process than K⁺, and the resulting geopolymer gels would consist of a much wider diversity of Qⁿ species in the case of Na⁺ [20, 21, 22]. Duxson et al. suggested that larger cations were more inclined to bind by the gel in mixed-alkali-activated systems that contained Na⁺ and K⁺ [23]. Ikuo et al. reported that the thermal expansion coefficient (CTE) of the cubic cesium leucite-based compounds decreased with much smaller ions occupying/substituting the crystallographic sites of Cs⁺, such as Na⁺ or Li⁺ substitutional ions [24, 25, 26].

Pollucite (CsAlSi₂O₆) exhibits a unique thermal expansion curve that has two stages: the average CTE is ~12.5 × 10⁻⁶ K⁻¹ from 298 to 473 K and ~2.2 × 10⁻⁶ K⁻¹ from 473 to 1473 K [24, 27]. In recent years, many reports have given evidence that the thermal expansion of pollucite decreased with ionic substitution at the Cs⁺ sites by smaller alkali metal ions, such as Na⁺ and Li⁺ [24, 25, 26, 27, 28, 29, 30, 31, 32, 33]. But, the mechanisms about how thermal expansion properties of pollucite are affected have not yet been studied clearly. In general, there were three main methods to prepare pollucite, including the solid-state reaction method, sol-gel method, and ion exchange method from leucite [34]. However, solid-state reaction always required high sintering temperatures and easily suffered from problems of cesium volatilization, agglomeration, furnace contamination, etc. Ion exchange and sol-gel techniques were also uncompetitive due to their high cost and severe preparation conditions. Moreover, it was almost impossible to get fully densified pollucite ceramic products because the volatilization of cesium was inevitable in the high-temperature environment during its preparation process. Therefore, only limited researches have been made on pollucite glass ceramic until now. In contrast, geopolymer technology could be an excellent alternative to prepare pollucite ceramic products through in situ conversion due to its ability to from crystalline ceramic phase after proper thermal treatment [35].

Therefore, as a part of our continuing research, a series of ion-substituted cesium-based geopolymer samples, Cs_(1-x)Li_xGP (where x = 0, 0.1, 0.2, and 0.3) and Cs_(1-x)Na_xGP (where x = 0, 0.1, 0.2, 0.3, and 0.4), were prepared. Phase composition, microstructure evolution, and thermal expansion behavior of ceramics derived from Cs_(1-x)Li_xGP and Cs_(1-x)Na_xGP were characterized, with the aim of investigating the effect of ion substitution on the thermal evolution of the geopolymer.

2. Experiments and characterization

Simultaneous thermogravimetry (TG) and differential thermal analysis (DTA) (Netzsch STA 409, Germany) were carried out under Ar gas flow (20 mL/min) with temperature up to 1200°C at a heating rate of 5°C/min in alumina crucibles. Sample powders were sintered without the occurrence of any reaction between the samples and the crucible.

The phase composition of samples was analyzed using X-ray diffraction (XRD, 40 KV/100 mA, D/max-γB CuKa, Rigaku Corp., Japan) method to obtain the X-ray diffraction spectra at 2θ = 10°–90° with a scanning speed of 4°/min. Slow step-scans with a step width of 0.02 and a step time of 3 s were carried out to determine the shifts of X-ray spectrum.

The microstructure of geopolymers before and after heat treatment was investigated by a scanning electron microscope (SEM, 30 KV, Helios Nanolab600i, FEI Co., USA). Energy dispersive spectrometer (EDS) was also adopted to study the elemental arrangement and phase distribution. Transmission electron microscope (TEM, 300 KV, Talos, FEI Company) was also employed to analyze its microstructure. For ceramics derived from geopolymer, TEM samples were ion-milled at low temperature, using a Fischione ion mill (Model 2; Export, PA).

3. Results and discussions

3.1 Thermal analysis

Figure 1 displays the results of thermogravimetry (TG) analysis as to Cs_(1-x)Li_xGP and Cs_(1-x)Na_xGP systems, respectively. Both systems showed similar TG curves. There was significant mass loss as the temperature increased until 800°C. The escape of free water, including water adsorbed on the surface and pore solution, accounted for the mass loss in the interval below 300°C. The mass loss between 300 and 800°C could be mainly attributed to the condensation and polymerization of hydroxyl connected with Si/Al atom, which resulted in the formation of oxo-bridge in the framework of geopolymer [13, 36]. As the temperature rose further, the TG curves remained essentially unchanged, suggesting almost no mass loss in this stage. The total mass loss increased regularly with the increase of Li⁺/Na⁺ content. Take Cs_(1-x)Na_xGPs for example, Na⁺ had a smaller ionic radius than that of Cs⁺, which means Na⁺ ion had a higher hydration energy than that of Cs⁺ ion [37]. Therefore, the amount of hydration water associated with Na⁺ was much more than that with Cs⁺ ion and, so, it was easy to see why the mass losses increased with increase in Na⁺ ion substitution amount in the Cs_(1-x)Na_xGP system.

Figure 1.

Thermal gravimetric curves of Cs_(1-x)Li_xGPs (a) and Cs_(1-x)Na_xGPs (b).

Figure 2 displays thermal shrinkage curves of these two geopolymer systems. The process of thermal shrinkage could be divided into four stages [14, 38, 39]: (I) Structural resilience (RT ~100°C): due to only free water lost in this interval, the dimensions of the tested samples could be maintained substantially, and the corresponding densification degree of the tested samples did not change significantly synchronously; (II) Dehydration (100–300°C): the causes of the shrinkage and deformation in this interval could be attributed to the capillary contraction induced by the escape of water from micro- and nano-pore solutions; (III) Dehydroxylation (300–800°C): the condensation and polymerization between T–OH groups (T = Al/Si) caused the water escape and shrinkage in the interval, with more weight damage and shrinkage at this stage than those in stage (II); (IV) Viscous sintering (above 800°C): the shrinkage in this stage was caused by the generation of molten amorphous glass phase, and the presence of molten amorphous glass phase would facilitate sintering and densification, which also means the maximum shrinkage behavior would occur in this stage [13, 19, 40, 41]. When the treatment temperature climbed to 1200°C, almost no shrinkage was observed, suggesting that the shrinkage process was fully completed.

Figure 2.

Thermal shrinkage curves of Cs_(1-x)Li_xGPs (a) and Cs_(1-x)Na_xGPs (b).

However, after Cs⁺ ion was substituted, significant differences are observed in Stage IV. For Cs_(1-x)Li_xGP samples, they all showed two sintering steps in Stage IV. Taking Cs_0.9Li_0.1GP for example, as depicted in Figure 3, there were two sintering steps: Region i: 850–1050°C and Region ii: 1050–1300°C. On raising Li⁺ ion content from 0 to 30 mol%, the onset temperature of two steps decreased gradually, from 850 to 730°C for Region i and from 1050 to 970°C for Region ii, respectively. In general, the melting points of the MAlSi₂O₆ decline with decreasing ionic radius of M⁺ ion (M = Li, Na, K, Cs). Meanwhile, onset sintering temperatures for NaGP, KGP, and CsGP were 650°C [36], 850°C [14], and 1200°C [13], respectively, indicating that geopolymer containing M+ ion (MGP) also showed a similar trend with MAlSi₂O₆. Therefore, it was reasonable to deduce that LiGP would exhibit the lowest onset sintering temperature, resulting in the lower temperature in Region i of Stage IV. Region ii might be due to melting of Li-based aluminosilicates. It should be pointed out that after doping with Li, all the samples show lower overall thermal shrinkage than pure CsGP, implying that the presence of Li facilitated sintering of Cs_(1-x)Li_xGPs to dense microstructure. The same trend could also be observed in the Cs_(1-x)Na_xGP system.

Figure 3.

Partial thermal shrinkage curves of Cs_0.9Li_0.1GP.

3.2 Phase composition

Figure 4 provides XRD patterns of Cs_(1-x)Li_xGPs and corresponding ceramic products derived from it. For Cs_(1-x)Li_xGPs, all the samples show a typical amorphous character with a broad amorphous hump from 20° to 30° 2θ, and almost no changes in phase composition had been observed based the XRD results of Cs_(1-x)Li_xGPs with different Li⁺ content. However, a substantial amount of pollucite was observed after pure CsGP was treated at 1300°C for 2 h. In addition, a small amount of spodumene was also observed after Cs⁺ ion was partially substituted by Li⁺ ion, and the content of spodumene was proportional to the Li⁺ ion substitution in Cs_(1-x)Li_xGPs system. The results of quantitative analysis indicated that the content of spodumene was 8.8 and 14.2 wt.% corresponding to x = 0.2 and 0.3 in Cs_(1-x)Li_xGPs system, respectively. However, the spodumene content was too small to calculate using XRD data when Li⁺ ion content was 10 mol%.

Figure 4.

XRD patterns of Cs_(1-x)Li_xGPs (a) and ceramics derived from it (b).

Unlike Cs_(1-x)Li_xGPs (x > 0), the phase composition of Cs_(1-x)Na_xGPs (x > 0) was not completely amorphous after the introduction of Na⁺ ion, and the crystallinity of Cs_(1-x)Na_xGPs gradually increased as more Cs⁺ ion was substituted. The XRD results indicated the presence of pollucite in the unheated samples partially substituted by Na⁺ ion (Figure 5a), which suggested that the presence of Na⁺ ion would have contributed to the crystallization of pollucite. As for the crystallization of pollucite, a possible reason could be attributed to the formation of a zeolite containing sodium, aluminum, and silicon, such as analcime (NaAlSi₂O₆·H₂O). The crystallization temperature of zeolite was very low, about 300°C, and the crystallization temperature will become lower (~120°C) for the samples with low Si/Al ratio under the hydrothermal condition [42, 43]. The geopolymer was nanoporous together with numerous capillaries, filled with pore solution. This structure was like a hydrothermal reaction environment. So, it was possible to produce some fine zeolite nucleus if the reaction time was long enough, although the reaction temperature was low. Analcime and pollucite have similar crystal structure and both were tetragonal and belong to I41/acd (space group). Meanwhile, lattice constant of analcime was a~13.727 and c~13.686, which was close to that of pollucite, a~13.677 and c~13.691 Å. Therefore, it was highly possible that the zeolite nucleus could have served as the nucleation site for pollucite, which contributed to pollucite formation. So, crystalline pollucite could be present in Cs_(1-x)Na_xGPs when x > 0, which was not the same as the phenomenon of Cs_(1-x)Li_xGPs system.

Figure 5.

XRD patterns of Cs_(1-x)Na_xGPs (a) and ceramics derived from it (b).

For the resulting ceramics derived from Cs_(1-x)Na_xGPs (x > 0), only crystalline phases of pollucite were observed in the corresponding XRD pattern, which were different with the Cs_(1-x)Li_xGPs system. So, the difference between Cs_(1-x)Li_xGPs and Cs_(1-x)Na_xGPs also indicated that Na⁺ may have only existed in the form of amorphous glass phase.

Based on the result of slow step-scans of ceramics derived from Cs_(1-x)Li_xGPs, it was possible to find that with increase in Li⁺ ion substitution, almost no deviation occurred for the characteristic peaks (4 0 0) and (3 2 1) corresponding to pollucite (Figure 6), which proved that the lattice parameters of pollucite did not change as more and more Li⁺ ions were introduced [33, 44]. This implied that Li⁺ ions did not occupy Cs crystallographic sites of the pollucite framework and phase separation occurred during heating. Diphasic compositions of product ceramics derived from Cs_(1-x)Li_xGPs also conform to the aforementioned two-step sintering behavior observed in thermal shrinkage results.

Figure 6.

Slow step-scan XRD patterns of ceramics derived from Cs_(1-x)Li_xGPs.

In contrast, for the heated samples derived from Cs_(1-x)Na_xGPs, the peaks (4 0 0) corresponding to pollucite shifted to the high-angle region with increase in sodium substitution (Figure 7), which suggested that the lattice parameters of pollucite decreased with increases in sodium content [33, 44]. The decline of pollucite’s lattice parameters proved that Na⁺ partially occupied/substituted the crystallographic sites of Cs⁺ in the pollucite crystal structure during high-temperature processing. The difference between Cs_(1-x)Na_xGP and Cs_(1-x)Li_xGP systems could be attributed to the difference between the size of the ions and the form of the substituted ions. Compared with Li⁺ ion, the size of Na⁺ ion was closer to that of Cs⁺ ion. In the Cs_(1-x)Li_x system, Li⁺ ion was only in the form of spodumene in the heated samples. The different forms of Na⁺ and Li⁺ ions also suggested that the Na⁺ ion had a higher degree of freedom than Li⁺ ion in the mass transfer process. So, it was much easier to occupy/substitute the crystallographic sites of Cs⁺ ion in the pollucite crystal structure with Na⁺ ion.

Figure 7.

(a) The (4 0 0) peak in a 2θ of 25–27 (b) Lattice parameters of pollucite.

3.3 Microstructure evolution

Figure 8 displays the microstructure of as prepared Cs_(1-x)Li_xGPs. The particles are too small to be observed clearly for pure CsGP. For other Cs_(1-x)Li_xGPs (x ≥ 0.1), the average particle sizes (APSs) are close to ~120, ~200, and ~250 nm corresponding to x = 0.1, 0.2, and 0.3, respectively, which could also be attributed to the difference in hydration energy between Li⁺ ion and Cs⁺ ions. By comparison, the hydration energy of Li⁺ was higher than that of Cs⁺, which means that there were more [AlO₄] and [SiO₄] associated with Li⁺ ion than that around Cs⁺ ion in the framework of the geopolymer. Therefore, higher Li⁺ ion contents in the geopolymer always resulted in particles with larger APSs [37]. The same variation trend was also observed in Cs_(1-x)Na_xGPs system, as shown in Figure 9.

Figure 8.

Microstructure of Cs_(1-x)Li_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3 (observed from fracture surface).

Figure 9.

Microstructure of Cs_(1-x)Na_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3, (e) x = 0.4 (observed from fracture surface).

As for Cs_(1-x)Li_xGPs system, as shown in Figure 10, these precipitates coarsened substantially and all the geopolymers developed a smooth, glassy texture after heating to 1300°C. The coarsening was consistent with the considerable shrinkage observed over the sintering temperature range [45]. Closed pore formations were also observed coincident with significant coarsening and surface area reduction. Pollucite crystals could not be directly observed on any of the fracture surfaces of pure CsGP ceramics despite their noticeable presence in the XRD. As for other Cs_(1-x)Li_xGPs (x ≥ 0.1), the corresponding ceramic product derived from it contain large numbers of spherical particles surrounded by a glassy matrix, with the size of coarsening increasing with Li⁺ ion content. The mean diameters of spherical particles were close to ~100 (x = 0.1), ~250 (x = 0.2), and ~400 nm (x = 0.3), respectively. The back-scattered electron (BSE) image suggested the presence of diphasic compositions in corresponding ceramic products derived from Cs_(1-x)Li_xGPs (x ≥ 0.1) (Figure 10e). EDS spectra showed that cesium content at point A (Figure 10f) was much higher than that at point B (Figure 10g). Because EDS was a semi-quantitative analytical measure and Li⁺ ion could not be detected by it, so, the results of EDS suggested that spherical particle (point A) and glassy matrix (point B) corresponded to pollucite and spodumene, respectively.

Figure 10.

Microstructure of ceramics derived from Cs_(1-x)Li_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3, (e) BSE image of (d), (f) EDS spectrum of point A, (g) EDS spectrum of point B.

From the fracture morphology after hydrofluoric acid (HF) corrosion, the existence of pollucite grains could be clearly observed in ceramic derived from pure GsGP (Figure 11a). In contrast, a polydisperse distribution of pollucite crystals and pores left after dissolution of molten glass phase during etching after the introduction of Cs⁺ ion, and the coarsening trend of pollucite grains were also obvious (Figure 11).

Figure 11.

Microstructure of ceramics derived from Cs_(1-x)Li_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3 etched in 4 wt% HF at room temperature for 20 s.

Figure 12 shows TEM analysis of the ceramic derived from Cs_0.7Li_0.3GP. The grain boundaries could be observed clearly, and the selected area electron diffraction (SAED) patterns of area A indicate the existence of spodumene crystallite in the resulting products, which may have arisen from the recrystallization of molten spodumene during cooling. Meanwhile, SAED patterns of area B also proved the corresponding grain should be pollucite, and the spodumene phase, with no fixed shape, mainly distributed among pollucite grains or in the grain junction area.

Figure 12.

TEM images of ceramics derived from Cs_(1-x)Li_xGPs, (a) x = 0.3, (b) and (c): SAED patterns of point A and B, respectively.

The microstructure morphologies of ceramic products derived from Cs_(1-x)Na_xGPs are given in Figure 13. By comparison, the particles coarsened significantly after being treated at 1300°C than that of corresponding unheated Cs_(1-x)Na_xGP samples. It could be observed that the surface of pollucite grains was covered by smooth and glassy texture. The presence of amorphous glass phase would be conducive to the densification of resulting ceramic products [45].

Figure 13.

Microstructure of ceramics derived from Cs_(1-x)Na_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3, (e) x = 0.4.

After etching the samples in 4 wt% HF acid for 20 s, pollucite crystals could be clearly observed in ceramics derived from heated Cs_(1-x)Na_xGPs when x ≤ 0.3, as shown in Figure 14. It could be clearly observed that the particle size gradually decreased with increasing Na⁺ ion substitution. Compared with ceramics derived from pure CsGP, many pores could be observed, which should be attributed to the dissolution of the amorphous glass phase in the process of static etching. Due to the content of amorphous glass phase and the Na⁺ ion substitution showed a proportional relationship. So, the residual amorphous glass phase will be more in the sample contained a higher Na⁺ ion substitution. Take Cs_0.6Na_0.4GPs for example, the vast majority of pollucite grains could not be observed clearly on the corroded surface of corresponding ceramic products due to the higher content of amorphous glass phase.

Figure 14.

Microstructure of ceramics derived from Cs_(1-x)Na_xGPs, (a) x = 0, (b) x = 0.1, (c) x = 0.2, (d) x = 0.3, (e) x = 0.4 etched in 4 wt% HF at room temperature for 20 s.

Figure 15 shows the TEM analysis of the ceramic evolved from treated Cs_0.6Na_0.4GP. Similar to Cs_(1-x)Li_xGPs system, the presence of crystal grains could be clearly observed in this diphase composition system. The SAED patterns of area A (Figure 15b) suggested the crystal grain to be pollucite. The results of the map-scan suggested that Cs⁺ ions were distributed only in the interior of crystal grain, which is consistent with the TEM results, as shown in Figure 16. However, Na⁺ showed a completely different distribution. Map-scan results revealed that although the vast majority of Na⁺ distributed among pollucite grains, there was still a small portion present in the interior of pollucite grains, which further demonstrated that Na⁺ ions partially occupied Cs crystallographic sites of the pollucite framework during high-temperature processing. Combining the SAED patterns of area B (Figure 15c) and the result of the map-scan indicated again that the Na⁺ ion mainly existed in the form of amorphous glass phase and distributed among the pollucite grains.

Figure 15.

TEM images of ceramics derived from Cs_(1-x)Na_xGPs, (a) x = 0.4, (b) and (c): SAED patterns of point 1 and 2, respectively.

Figure 16.

HAADF-STEM and STEM-EDX images illustrating the element distribution.

In heated Cs_(1-x)Li_xGPs system, the particle size of pollucite coarsened with increases in Li substitution. However, Cs_(1-x)Na_xGPs system showed a distinctly different variation tendency, in which the particle size of pollucite grains gradually decreased with increasing Na⁺ ion substitution. The phenomenon of grain refinement in heated Cs_(1-x)Na_xGPs system may have a direct relationship with the presence of zeolite nucleus in corresponding unheated samples. The zeolite nucleus formed in the unheated Cs_(1-x)Na_xGP samples could serve as the nucleation site of pollucite and accelerate the crystallization process of pollucite. As already mentioned, the number of zeolite nucleus was proportional to Na⁺ ion introduction. Therefore, the growth of pollucite grains was more dispersed as the nucleation sites increased.

3.4 Sintering mechanism

A possible sintering mechanism of Cs_(1-x)Li_xGP system can be suggested in light of the theory of reactive liquid-phase sintering [46]. The composition of Cs_(1-x)Li_xGP can be expressed as (1-x)Cs₂O·xLi₂O·Al₂O₃·4SiO₂·wH₂O. In fact, both Li⁺ and Cs⁺ ions were present as silicates, that is, x₁Li₂O·y₁SiO₂ and x₂Cs₂O·y₂SiO₂ (the complexity of composition and structure of silicate indicated that x₁, y₁, x₂, and y₂ did not have a certain value). Therefore, it was reasonable to employ Li₂O and Cs₂O to express the form of Li⁺ and Cs⁺ ions in this system.

It well known that alkaline metal oxides often participated in the sintering process as a sintering aid in many low-temperature sintering systems [46, 47]. In addition, as a typical network-modifying agent, the presence of Li₂O or Li⁺ ion could promote the densification process of sintering [48, 49, 50, 51]. Similarly, the reaction between the alkaline metal oxide dopant and the matrix phase was studied and results showed that the mechanism was exactly the same in many cases. Hence, as for the Cs_(1-x)Li_xGP system, Li⁺ was an intermediate cation with a stronger electric field intensity in the glass-ceramic system than that of Cs⁺ ion [52]. Thus, with increases in temperature of high-temperature processing, the reaction between Li₂O, Al₂O_3, and SiO₂ first occurred at a lower temperature, as in Eq. (1).

$${\mathrm{Li}}_2\mathrm{O}\left(\mathrm{s}\right)+{\mathrm{SiO}}_2\left(\mathrm{s}\right)+{\mathrm{Al}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)\to {\text{LiAlSi}}_2{\mathrm{O}}_6\left(\mathit{l}\right)$$E1

The presence of liquid spodumene (LiAlSi₂O₆) and the simultaneous reaction between Li₂O and the matrix phase decreased the viscosity of the geopolymer system, which doubtlessly accelerated the kinetics of mass-transport processes during the low-temperature sintering. The changes in the sintering processes could reflect from the variation tendency in thermal shrinkage curves with different lithium contents (Figure 3). As the processing temperature rose further, the reaction between Cs₂O, Al₂O_3, and SiO₂ would occur at a higher temperature compared with the onset temperature of spodumene, as in Eq. (2).

$${\mathrm{Cs}}_2\mathrm{O}\left(\mathrm{s}\right)+{\mathrm{SiO}}_2\left(\mathrm{s}\right)+{\mathrm{Al}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)\to {\text{CsAlSi}}_2{\mathrm{O}}_6\left(s\right)$$E2

The presence of liquid spodumene increased the rates of mass transfer and diffusion more easily, conducive to nucleation and crystallization of pollucite (CsAlSi₂O₆). This was consistent with the decrease in crystallization temperature and the grain coarsening behavior observed above with increasing lithium substitution. Meanwhile, the degree of densification was greatly improved as well. Therefore, the presence of Li⁺ ion contributed to the crystallization and sintering densification via aiding the liquid-phase processes.

The sintering process of the Cs_(1-x)Na_xGP system was similar to that of the Cs_(1-x)Li_xGP system. The results above have already proved that Cs_(1-x)Na_xGP ceramics only contained pollucite and amorphous glass phase. Therefore, the main difference in the sintering process between Cs_(1-x)Li_xGP and Cs_(1-x)Na_xGP systems lied in the low-temperature area. The reaction between Na₂O, Al₂O_3, and SiO₂ in the Cs_(1-x)Na_xGP system would occur first at a lower temperature, as in Eq. (3).

$${\mathrm{Na}}_2\mathrm{O}\left(\mathrm{s}\right)+{\mathrm{SiO}}_2\left(\mathrm{s}\right)+{\mathrm{Al}}_2{\mathrm{O}}_3\left(\mathrm{s}\right)\to \text{amorphous glass}\left(\mathit{l}\right)$$E3

In addition to this lower temperature, the sintering process of the two systems also experienced almost the same process at higher temperature intervals.

3.5 Thermal expansion behavior

The results of the dilatometric measurement for Cs_(1-x)Li_xGP and Cs_(1-x)Na_xGP ceramics with correction are shown in Figure 17. The thermal expansion of ceramics derived from Cs_(1-x)Li_xGP and Cs_(1-x)Na_xGP systems showed an increasing trend with increase in temperature, which could be attributed to the larger atomic spacing at elevated temperatures.

Figure 17.

Linear coefficient of thermal expansion of Cs_(1-x)Li_xGP (a) and Cs_(1-x)Na_xGP.

The average thermal expansion coefficient (CTE) was defined as the average value of the relative length change in the temperature range of T₁ < T < T₂, as described by the following Eq. (4):

$${\alpha }_{\text{a}}=\left(L_2-L_1\right)/L_0\left(T_2-T_1\right)=\Delta L/L_0\Delta T$$E4

where L₀, L₁, and L₂ are the lengths of the specimen at temperatures of T₀ (30°C), T₁, and T₂, respectively. From Figure 14, both curves have two very different slopes before and after ~150°C, indicating that there was a phase transition for pollucite from tetragonal to cubic [24, 25]. Thus, the average CTE of Cs_(1-x)Li_xGPs was computed based on data in the interval from 150 to 900°C.

As for Cs_(1-x)Li_xGP ceramics, as depicted in Figure 18, it could be observed that the average CTE decreased from 4.80 × 10⁻⁶ K⁻¹ (x = 0) to 3.61 × 10⁻⁶ K⁻¹ (x = 0.3). The results obtained from our previous study on the K_(1-x)Cs_xGP system proved that increase in Cs⁺ ion introduction would decrease the average CTE of the resulting leucite ceramic. The reason could be attributed to the substitution between Cs⁺ and K⁺ in the leucite framework [33]. However, as for Cs_(1-x)Li_xGP ceramics, the XRD results indicated almost no changes in lattice constants of pollucite, which suggested that the integrity of pollucite crystal cell was unchanged after partial substitution by Li⁺ ion. XRD quantitative analysis suggested that the content of spodumene in resulting ceramic products was proportional to the amount of Li⁺ ion substitution. By comparison, spodumene had a lower CTE (1.0 × 10⁻⁶ K⁻¹) than that of pollucite. Therefore, the presence of spodumene could significantly reduce CTE of the whole system, and the higher content of spodumene meant lower CTE. Another reason could be attributed to the role of molten spodumene, possibly as a buffer phase exists between pollucite crystals. Hence, with the increase in Li substitution or spodumene content, the average CTE of Cs_(1-x)Li_xGP ceramics decreased evidently.

Figure 18.

Thermal expansion properties of Cs_(1-x)Li_xGP ceramics (a) and Cs_(1-x)Na_xGP ceramics (b).

However, the calculated results showed that the average CTE of Cs_(1-x)Na_xGP ceramic products showed a completely different variation pattern compared with Cs_(1-x)Li_xGP ceramic products, increasing from 4.80 × 10⁻⁶ K⁻¹ (x = 0) to 7.26 × 10⁻⁶ K⁻¹ (x = 0.4) with the rise of sodium substitution.

For K_(1-x)Cs_xGPs system involved in our previous studies, the average CTE of the leucite ceramic decreased with increase in the amount of Cs⁺ ion introduced. The reason could be ascribed to Cs⁺ occupying/substituting K⁺ crystallographic sites in the leucite framework, which led to the increase in content of stabilized cubic leucite in resulting ceramic products [33]. However, there was no evidence to suggest any substitution at the lattice level between Li⁺ and Cs⁺ in the Cs_(1-x)Li_xGP system. Besides, as a diphase system, pollucite and spodumene were present as independent components in the resulting products [38]. Therefore, one could conclude that the decline of average CTE was mainly due to the fact that CTE of spodumene was far lower than that of pollucite.

As for Cs_(1-x)Na_xGP ceramics, XRD results suggested that the substitution between Na⁺ and Cs⁺ ions caused the decline of lattice constant of pollucite. According to the Ikuo et al. reports, the increase in Na⁺ in the crystal lattice could decrease the thermal expansion of the pollucite, and with the increase in Na⁺ ion substitution, the average CTE of ceramics derived from heated Cs_(1-x)Na_xGPs decreased [24, 25, 26]. However, compared with Cs_(1-x)Li_xGPs system, the CTE of Cs_(1-x)Na_xGP ceramics showed an opposite trend with increasing Na⁺ ion substitution. The average CTE of Cs_(1-x)Na_xGP ceramics continued to rise with the increase in content of Na⁺ ion introduction. It is well known that amorphous glass phase containing Na⁺ ion always had a higher CTE (9~10 × 10⁻⁶ K⁻¹) than that of pollucite (~4.80 × 10⁻⁶ K⁻¹) [53, 54, 55, 56]. By contrast, the effect on CTE caused by the presence of the amorphous glass phase was much stronger than the impact caused by the substitution of Na⁺ for Cs⁺ in lattice level. So, the presence of amorphous glass phase could account for the rise of average CTE. The content of amorphous glass phase was proportional to the amount of Na⁺ ion substitution. Therefore, the average CTE of ceramics derived from Cs_(1-x)Na_xGPs increased almost linearly with the increasing sodium content.

4. Conclusions

In this chapter, the effects of ion substitution on the thermal evolution of cesium-based geopolymers on heating were studied. The following conclusions could be obtained based on the current results.

1. The introduction of Li⁺ or Na⁺ could favor the formation of pollucite in Cs_(1-x)M_xGPs (x ≥ 0.1), respectively. A possible reason could be attributed to the formation of zeolite nucleus, which could serve as the nucleation sites for pollucite. This also provided a novel method for preparing pollucite at low temperature, even at room temperature.

2. Cs_(1-x)M_xGPs with different content of ionic replacement showed a similar thermal evolution process. The introduction of Li⁺ or Na⁺ changed the sintering mechanism through decreased onset temperature for viscous sintering stage. With the introduction of more Li⁺ or Na⁺, the amorphous glass phase in resulting products would be conducive to its densification process. The introduction of Li⁺ ion led to particle coarsening in part because of the rapid formation of a liquid phase on heating and perhaps because the hydration energy of Li⁺ is greater than that of Cs⁺. However, the particle size of pollucite grain gradually decreased with the increase in Na⁺ contents in Cs_(1-x)Na_xGP ceramics, which had a close relationship with the role of Na⁺ ion in the nucleation process of pollucite.

3. Compared with traditional way, geopolymer technology could be an excellent alternative to fabricate ceramics or ceramic matrix composites in an in-situ convert way. In the Cs_(1-x)Li_xGP ceramics, the introduction of Li⁺ results in the formation of a multiphase system, including pollucite, spodumene, and amorphous glass phase, and Li⁺ only existed inside the spodumene grains. In contrast, two forms of Na⁺ ions were present in the ceramic products derived from Cs_(1-x)Na_xGPs. Therefore, a small portion was present in the pollucite grains due to Na⁺ occupying/substituting Cs⁺ crystallographic sites of the pollucite framework, while the rest were mostly present in the amorphous glass phase and distributed among the pollucite grains.

4. The average CTE of ceramics derived from Cs_(1-x)Li_xGPs and Cs_(1-x)Na_xGPs showed a completely different variation pattern. The reason could be attributed to the difference between two kinds of substituted ion species, which resulted in different phase composition in the corresponding resulting products.

Acknowledgments

The authors acknowledge the support from the National Natural Science Foundation of China (NSFC, Nos. 51372048, 51502052, 51321061 and 51225203). The grant from Fundamental Research Funds for the Central Universities (Grant no. HITNSRIF20165) is also acknowledged.