Microwave reflection and transmission measurements for evaluating water reaction within geopolymers with different precursors

Abstract

In addition to their mechanical properties, geopolymer structures are considered to be environmentally and economically friendly structures due to the non-requirement of calcination in their binding process, as compared with ordinary Portland cement structures. To take advantage of these properties, the water reaction occurring inside the precursors, which are used to supplement sufficient alkali activation during the binding process (geopolimerization), needs to be analyzed. In this study, microwave reflection and transmission measurements are applied as a non-destructive tool in the detection of water reaction with geopolymer paste samples with fly ash (FA), ground granulated blast furnace slag (GGBFS), and calcinated clay or metakaolin (MK) precursors having different sodium silicate/sodium hydroxide ratios. Our measurements indicate the following results. First, microwave transmission properties are more useful than and preferable to microwave reflection properties for the analysis of water reaction within all test geopolymer paste samples (FA, GGBFS, and MK). Second, reflection measurements () exhibited minimal sensitivity to specimen length, with differences below 0.45 dB for thicknesses between 10 and 20 mm. Third, conversely, transmission responses () showed pronounced variations of up to 35 dB, underscoring their enhanced sensitivity to internal water reaction mechanisms. Fourth, dielectric characterization indicated that FA powder possessed substantially higher dielectric loss () relative to GGBFS () and MK (), which directly governed the magnitude of signal attenuation. Finally, the negligible mass loss observed across all prepared samples (<2.5%) confirms that the evolution of was dominated by the transformation of free water into bound water within the matrix rather than by evaporative losses. Scanning electron microscope micrographs and energy dispersive spectroscopy tests were also performed to examine the effect of water reaction within the prepared samples.

Introduction

Ordinary Portland cement (OPC) concrete structures are widespread in the construction industry. Nonetheless, they are fragile, which in turn limits their usage for areas with potential earthquakes. For such applications, Engineered Cementitious Composites (ECCs), as a reinforced OPC concrete with fibers, could be suited¹ mainly due to their higher ductility compared with that of OPC concretes. In addition, ECCs have superior tensile and flexural strengths as a result of interaction of fibers with cement matrix, which are the main parameters in minimizing crack size². On the other hand, high amount of heat is required in the fabrication of cement, which constitutes a main input in the overall cost of cement production. Aside from economical point of view, OPC production rises the amount of pollution in the world. According to a recent analysis, it has been estimated that such production contributes to approximately 7% of total in the world³. To eliminate the aforementioned adverse effects of cement used in the construction of OPC concretes, geopolymer concretes (GPCs), as a new form of concrete structure, are devised. GPCs do not require high temperature calcination step because they do not utilize cement as a binder material. Therefore, they are considered as economically friendly structures thanks to significant reduction of (up to 80%) in comparison with that of OPC concretes in their synthesis process⁴. In addition to these advantages, GPCs have low creep and shrinkage and superior mechanical properties⁵.

In the fabrication process, aluminosilicate and high-calcium powders, which operate as precursors, are utilized to provide sufficient alkali-activation. In the literature, different sorts of precursors such as fly ash (FA), ground granulated blast furnace slag (GGBFS), and calcinated clay or metakaolin (MK), are utilized to initiate such geopolimerization in the binding process within sodium and potassium hydoxite or silicate solutions. A geopolymer is a 3D zeolite-like network structure including silicon–oxygen and aluminium–oxygen (Al-O) tetrahedral. The polymerization reaction processes are more complex than the hydration reaction of OPC. Davidovits proposed that the geopolymer structure results from the following chemical reaction⁶:

1

where M shows the alkali cation, n shows the degree of polycondensation, w shows the number of chemically bound water molecules, z shows the Si/Al ratio that can be 1, 2, or 3. The structures of geopolymers can be single silicon-aluminium (Si/Al:1 - Gel 1), double silicon-aluminium (Si/Al:2 - Gel 2), and triple silicon-aluminium (Si/Al:3 - Gel 3) as shown in Fig. 1⁷.

Fig. 1.

Geopolymer structures according to Davidovits⁷.

The reaction procedure basically includes these four steps.

1. Dissolution stage The Si-O and Al-O bonds are broken in the alumina-silicate precursor by alkali activators, and the Si-O and Al-O tetrahedral monomers are released.

2. Diffusion stage The dissolved Si-O and Al-O tetrahedral monomers diffuse into the reaction system. Based on the principle of chemical equilibrium, the Si and Al concentrations reduce on the particle surface due to the diffusion, and the dissolution process continues.

3. Polycondensation stage Si-O and Al-O tetrahedra form amorphous -Si-O-Al-O- structures or zeolite crystals through polymerization.

4. Hardening stage A dehydration reaction occurs, forming a geopolymer with higher mechanical strengths.

These steps are performed simultaneously. However, only the leading step is different during the reaction process. For geopolymer materials having various raw materials, the polymerization mechanisms are not exactly the same, but they essentially conform to the foregoing reaction process^7,8.

To gain more insight into and benefit from geopolymer structures, the interaction of precursors within these solutions should be well-understood. For example, reaction of precursors with water may change with the type of the precursor. While mass majority of water has a state translation from free-water to bound-water for the precursors with a higher calcium content (e.g., GGBFS precursor) in the binding process⁹, significant amount of water is assumed to be initially bound in the activator solution and then has a state transformation from bound-water to free-water for precursors with a lower calcium content¹⁰. In this study, our main concern is to examine the water reaction occurring inside geopolymers with FA, GGBFS, and MK precursors, which have different calcium contents¹¹, over the first hours after the geopolymer samples are prepared.

Research on water interaction with geopolymers is a rapidly growing field, especially in the context of durability and sustainability. Different studies have recently been performed to examine such a reaction mechanism in geopolymers. The study¹² employs a non-destructive neutron radiography methodology to visualize how water moves through the pores of a geopolymer in real-time, providing more accuracy than traditional weight-gain methods. Besides, the method¹³ uses a standardized leaching test to measure how many ions (Si, Al, and Na) are released when a geopolymer sample is submerged into water. Likewise, the study¹⁴ compares how different activators such as sodium hydroxide and potassium hydroxide change the pore structure and, consequently, the water absorption rate of the geopolymer concrete. Different from previous studies^12–14, microwave signals can be utilized, as a non-destructive tool for the field site, to detect and monitor water reactions inside geopolymers, since these are very sensitive to state (free-water versus bound-water) and degree (level of humidity) of free water within materials^15,16. Previous and recent microwave measurements have demonstrated the great potential of these signals for inspection of physical, mechanical, and chemical properties of OPC structures; e.g., evaluation of free-water evaporation and hydration formation¹⁷; prediction of water-to-cement (w/c) and sand-to-cement (s/c) ratios¹⁸ and coarse aggregate-to-cement (ca/c) ratio¹⁹; determination of compressive strength²⁰; monitoring chloride ingress²¹; and examination and quantification of alkali-silica-reaction (ASR)^22–27. Microwave signals also have demonstrated the capability of interrogation of mechanical, and chemical properties of ECC structures; e.g., evaluation of compressive strength, modulus of rupture, and modulus of elasticity^28–31; sample curing conditions^32,33; and their shielding effectiveness and absorbance characteristics with some additional additives^34–37.

Aside from the applicability of microwave signals for examining various properties of OPC and ECC structures, they have recently been applied for evaluating absorbance characteristics of geopolymers with additional additives^38,39 and improvement on curing process in the binding of geopolymers^40–42. In a recent study, microwave signals are utilized to evaluate water reaction within geopolymers with FA as a precursor⁴³. This study reveals the effect of two different solutions of sodium hidroxite in the preparation of FA geopolymer on microwave reflection properties. It was demonstrated that free-water is transformed to bound-water at early ages of FA geopolymers in the same manner of the hydration process within CBMs. However, this study is restricted to geopolymers with a FA precursor only and examines just microwave reflection properties using an open-ended rectangular waveguide. It is known that while reflection properties mainly present surface information of lossy samples, especially for geopolymers in fresh state, transmission properties present internal information of these samples, because transmitted signals must travel through a sample^22,44. Besides, the open-ended waveguide used in this study⁴³, although nondestructive and non-invasive, necessitates a close contact between the waveguide aperture and the sample surface, which might be difficult to fully satisfy for geopolymer samples in their fresh states (at their early ages) where significant amount of geopolymerization is considered to be taking place⁴³. Furthermore, this measurement technique requires a wider transverse plane to perform accurate reflection property measurements, if the sample does not have sufficient attenuation^43,45,46. On the other hand, the authors of the study in⁴⁷ utilized both reflection and transmission measurements of blast furnace slag and fly ash samples over X-band (8.2-12.4 GHz) microwave frequencies to evaluate their electromagnetic properties. This study was limited to electromagnetic property evaluation of blast furnace slag and fly ash samples at a given time but not their water reactions which take longer time to examine water reactions occurring inside these samples.

In this study, results of both microwave reflection and tranmission properties of F-type FA, GGBFS, and MK geopolymers with different sodium hidroxite solutions are presented using a traditional rectangular waveguide measurement setup (a fundamental microwave setup) over a longer period of time (over 30 hours) so that temporal variation occurring inside F-type FA, GGBFS, and MK geopolymers with different solutions could be monitored to get more information. It is shown by measurements of this setup that microwave transmission properties are more preferable to reflection properties for inspection of water reaction mechanism within geopolymers, just like transmission properties are more appropriate to detection of ASR within CBMs^22,23.

Mixture and sample preparation and curing condition

Different geopolymer paste samples (two sets for FA, GGBFS, and MK binders) using two different alkaline solutions were produced to demonstrate the potential of microwave signals for examining water reaction mechanism and phase transitions. In line with the study⁴³, these alkaline solutions had sodium silicate ()/sodium hydroxide (NaOH) ratios of 1 and 2.5. In the rest of the manuscript, they are, respectively, referred to as the Na-1 and Na-2 solutions.

Considering molar concentration in the studies^48,49, an 8-molar sodium hydroxide solution was chosen in order to control the water content more easily and to let geopolymer samples have longer setting times. Besides, sodium contents in two solutions were equalized to the water-to-binder ratio, which was calculated from alkaline solution, superplasticizer, and extra water. Polycarboxylate ether-based superplasticizer was added to each mixture while producing geopolymer pastes to adjust proper workability. The same amount of superplasticizer was also added to all six mixes to keep the same water-to-binder ratio. Table 1 presents quantity of the constituents used to produce geopolymer samples, while Table 2 gives the chemical content of the sodium hydroxide. Since 8-molar sodium hydroxide was used in both Na-1 and Na-2 solutions to ensure geopolymerization, the missing water ratio in the Na-1 solution was equalized by adding extra water. As a result, the ratio of water-to-sodium in two different Na solutions was equalized. Tables 3, 4 and 5 show, respectively, the chemical content of the NaOH, , the properties of superplasticizers, and the chemical composition of the F-type FA, GGBFS, and MK powders with a maximum grain size of m. The prepared sodium hydroxide solutions were left to stand for 24 hours and allowed to cool down in the laboratory to reduce the high heat generated during mixing with powders. Each powder (1000 g) was gradually poured into the prepared hydroxide solutions and then stirred approximately 10 minutes to produce homogeneous FA, GGBFS, and MK geopolymer paste samples. Immediately after the stirring process, some amount of fresh FA, GGBFS, and MK paste samples were poured into a metallic hollow rectangular waveguide cell with a cross section of 72.136 mm 36.068 mm (see Fig. 2) to produce rectangular prism samples with lengths of mm. Likewise, the same procedure was exercised to prepare two additional (test) FA geopolymer paste samples with lengths of mm and 20 mm. It is noted that a tiny nylon bag (see the inset in Fig. 2) was used to eliminate direct contact of poured geopolymer samples with metallic waveguide walls, and a thin Scotch tape was pressed between the bottom section of the cell and upper flange of the additional waveguide section to firmly hold the geopolymer sample mixtures or power materials in vertical position. Furthermore, immediately after a prepared mixture sample was poured into the cell, the cell with this thin Scotch tape⁵⁰ was vibrated not only to have a sample with an even top surface but also have a homogeneous sample mixture (sample integrity) (see the inset in Fig. 2). The effects of a thin Scotch tape and a tiny nylon bag are assumed to be negligible on microwave reflection and transmission properties⁵⁰. Microwave measurements, to be discussed in Section 3, were performed over 0–30 hours with a 5-hour interval at ordinary room conditions ( 23 °C 1°C) for each prepared sample. Microwave measurements were repeated four times (requiring preparation of a new sample) for each FA, GGBFS, and MK geopolymer paste type, resulting a total of 8 geopolymer samples for each GGBFS and MK binder type and a total of 24 geopolymer samples for the FA binder type. Table 6 presents the dimensions and numbers of FA, GGBFS, and MK geopolymer samples.

Table 1.

Mix proportions in the preparation of different geopolymers with the Na-1 and the Na-2 solutions.

Solution	Binder	NaOH		Superplas.	Water
	(g)	(g)	(g)	(g)	(g)
Na-1	1000	500	500	50	40
Na-2	1000	320	800	50	–

Table 2.

Chemical content of the sodium hydroxide (NaOH).

Property	Value
NaCl (%)	max – 0.10
(%)	max – 0.40
Fe (%)	max – 0.15
Ar (%)
Density ( °C) ()	2.13
Molecular weight (mol/g)	40.01

Table 3.

Chemical content of the sodium silicate ().

Property	Value
(%)	12.89
(%)	25.42
/	2
Water (%)	61.69
Bome ( °C) (°Be)	44.45
Density ( °C) ()	1.442
Module	2.11

Table 4.

Properties superplasticizer.

Property	Value
Material structure	Polycarboxylate ether
Color	Green
Density ( °C) (kg/lt)	1.05
Chloride ion content (%) (EN 480-10)
Alkaline content (%) (EN 480-12)
pH value
Corrosion behavior	Contains only the compounds
	compatible with EN 934-1:2018
	Annex A.1 standard

Table 5.

Chemical composition (%) of FA, GGBFS, and MK powders.

Component	CaO	O + O				MgO
FA	2.24	3.75	57.20	7.10	24.40	2.40
GGBFS	34.12	1.32	36.40	0.68	10.39	10.30
MK	0.50	0.55	53.00	0.50	43.00	0.40

Fig. 2.

Rectangular waveguide setup used for reflection and transmission measurements.

Table 6.

Dimensions and numbers of FA, GGBFS, and MK geopolymer samples.

Geopolymer binder type	Binder solution	Dimensions	Number of samples
FA	Na-1	72.136 mm 36.068 mm mm	4
	Na-2		4
	Na-1	72.136 mm 36.068 mm mm	4
	Na-2		4
	Na-1	72.136 mm 36.068 mm mm	4
	Na-2		4
GGBFS	Na-1	72.136 mm 36.068 mm mm	4
	Na-2		4
MK	Na-1	72.136 mm 36.068 mm mm	4
	Na-2		4

Microwave measurement setup and its calibration

Waveguide measurements were chosen to examine the water reaction within prepared FA, GGBFS, and MK geopolymer samples with Na-1 and Na-2 solutions. Aside from the feasibility and utility that waveguide measurements provide information about both reflection and transmission properties of a sample, it is a fundamental measurement setup widely utilized in the literature. The cross section of 72.136 mm 36.068 mm was selected, resulting in a frequency range of 2.60–3.95 GHz (S-band or WR-284) for the dominant mode propagation. The same frequency band was also used for examining ASR occurring inside CBMs through waveguide reflection and transmission properties^22,23. However, different from these studies which utilized a setup in horizontal direction to examine hardened CBMs positioned into a rectangular cross section, a typical rectangular waveguide positioned vertically, as shown in Fig. 2, was employed to monitor water reaction within fresh geopolymer samples.

This setup includes a vector network analyzer (VNA), two high-precision coaxial cables each with a length of 100 cm, two coaxial-to-waveguide adapters, a measurement cell within which samples are poured, and two additional waveguide straights or sections between the measurement cell and the adaptors. The VNA purchased from Keysight Technologies (N9918A) has an operating frequency range between 30 kHz and 26.5 GHz, a dynamic range of approximately 90 dB, and a directivity larger than 32 dB⁵¹. While coaxial cables were used for transferring electromagnetic signals from the VNA output to the beginning of the coax-to-waveguide adaptors, these adapters were employed to transform a TEM signal into a dominant-mode signal. Besides, additional waveguide sections with sufficient lengths (more than 200 mm) (at least 2 times larger than the free-space wavelength corresponding to the mid-frequency (3.275 GHz) of the entire band) were used in charge of eliminating the effect of high-order mode wave propagation through the waveguide, if any, so that only the reflection and transmission properties of the sample corresponding to the dominant mode propagation would be measured.

The measurement setup used for microwave measurements was calibrated up to the ends of the measurement cell by the thru-reflect-line (TRL) calibration procedure—a highly accurate, versatile, and simple microwave calibration procedure used frequently for waveguide measurements (waveguide flanges are sexless connectors). While a waveguide section with a length of approximately 30.46 mm was used as the line standard, two metals in planar form with a transverse dimension sufficiently greater than the waveguide cross section were operated as the reflect standards. After the calibration, its accuracy was tested by performing reflection and transmission measurements of the line standard. It was noted that magnitudes of reflection measurements were recorded to be less than dB, whereas transmission measurements had a magnitude of around dB over the entire frequency band. Besides, has a phase value of nearly ° at the mid-frequency. Additionally, another line standard with a length of nearly 80 mm was utilized in measurements, and it was observed that, just as the line standard have, magnitudes of reflection and transmission measurements were recorded to be less than dB and around dB over the entire frequency band. These results clearly indicate the accuracy of the TRL procedure before starting to microwave reflection and transmission properties of prepared geopolymer samples⁵².

Some geopolymer sample mixtures or power materials, to be discussed shortly, were poured into the line standard (with a length of approximately 30.46 mm) used in the calibration procedure so that different lengths of FA, GGBFS, and MK geopolymer samples were set as mm, 15 mm, and 10 mm.

Complex reflection and transmission scattering (S-) parameters ( and ) of each sample were conducted by the measurement setup after its calibration and sample pouring into the cell. While shows the ratio of complex reflected amplitude to the incidence amplitude , shows the ratio of complex transmitted amplitude (signal through the sample and empty section of the cell) to . These parameters can be written as

2

3

where and P are the reflection coefficient at air-sample interface and the complex propagation factor through the sample. Their expressions are

4

5

Here, ( GHz for our measurements) and f are the cutoff and operating frequencies; is the free-space (unbounded) wavenumber; and is the complex relative permittivity of the sample with its real part () showing polarization capability and its imaginary part () showing absorption characteristics of a lossy dielectric sample (e.g., geopolymer samples).

Results and discussion

Microwave reflection and transmission measurements

Dielectric properties of constituent materials

Before discussing the results of reflection and transmission properties of FA, GGBFS, and MK geopolymer samples with Na-1 and Na-2 solutions, dielectric properties of FA, GGBFS, and MK powders and Na-1 and Na-2 solutions were extracted. Figure 3a–c illustrate, respectively, the average real and imaginary parts of the extracted of the FA, GGBFS, and MK powders with three different lengths of mm, mm, and mm (four independent measurements were carried out for each length of sample) using microwave measurements by our microwave setup in Fig. 2 (without a thinner nylon bag). We applied our method⁵³ for a stable extraction of these powders. In addition, Fig. 3d–f present, respectively, the average real and imaginary parts of the extracted of distilled water, Na-1 solution, and Na-2 solution (four independent measurements were conducted for each sample) using a slim coaxial probe (N1501A from Keysight Instruments with materials measurement software) with an averaging of 16 successive measurements. The probe measurements were calibrated by open, short, and distilled water references. It should be noted that measured of the FA, GGBFS, and MK powders correspond to effective values which take into account of the effect of void presence between powders particles⁴⁷.

Fig. 3.

Real and imaginary parts of the extracted of (a) the FA powder, (b) GGBFS powder, (c) the MK powder, (d) distiller water (DW), (e) the Na-1 solution, and (f) the Na-2 solution.

The following points are noted from the results in Fig. 3a–f. First, measured real and imaginary parts of the effective of the FA, GGBFS, and MK powders at S-band (2.6-3.96 GHz) are close to those in the studies^{43,47,54–56}. Second, measured real part of the effective of the MK powder is much smaller than measured real parts of the effective of the FA and GGBFS powders. This indicates that polarization capability of the MK powder is smaller than polarization capabilities of the FA and GGBFS powders. Third, measured imaginary parts of the GGBFS and MK powders are greatly smaller than measured imaginary part of the FA powder. Aside from intrinsic loss behavior (larger ) of these powders, such results could be partly associated with the point that the specific surface of used MK is higher than those of GGBFS and F-type FA. Due to the fineness and compact microstructure, absorption capacity of MK becomes low, that of GGBFS is moderate, and that of FA is high. Fourth, real and imaginary parts of DW are in good agreement with the well-known data in the literature¹⁶, indicating accurateness of the coaxial probe measurements. Fifth, and of the effective of the Na-1 solution are, respectively, greater than and of the effective of the Na-2 solution, because water content of the Na-1 solution is greater than the water content of the Na-2 solution.

Effect of sample length on transmission measurements

Before considering microwave reflection and transmission properties as a nondestructive and noninvasive tool for evaluating water reaction occurring inside FA, GGBFS, and MK geopolymers with Na-1 and Na-2 solutions, the effect of sample length on these properties was examined. For example, Fig. 4a,b illustrate, respectively, the average (out of four measurements for each set) magnitudes of reflection properties () and transmission properties () over frequency of FA geopolymers with Na-2 solution for mm, 15 mm, and 20 mm, for a time corresponding to immediately after each FA geopolymer sample was prepared and poured into the measurement cell in Fig. 2 (referred to as 0 hour in the rest of the manuscript). For conciseness, magnitudes and phases of reflection and transmission properties of FA geopolymer with Na-1, and of GGBFS and MK geopolymers with Na-1 and Na-2 solutions with different lengths are not presented. It is noted that magnitudes of and (in dB) are related to their linear ones by the expression

6

Fig. 4.

Measured average magnitudes (dB) of (a) reflection properties () and (b) transmission properties () over frequency of FA geopolymers with the Na-2 solution for mm, 15 mm, and 20 mm.

The following points are observed from the results in Fig. 4a,b. First, while the difference between values of FA geopolymers with Na-2 solution for mm, 15 mm, and 20 mm are small (less than 0.45 dB over the entire band), the difference between of the same geopolymers reaches as high as 35 dB. Such a difference arises due to the effect of both volumetric attenuation, which is highly dependent on the sample thickness, within any geopolymer sample and reflection from its surface (for transmission properties), as compared with reflection from any high-loss geopolymer sample (showing at least 10 dB attenuation) for reflection properties^44,57. For understanding this point clearer, one can refer to the expressions of and in (2) and (3). It is well-known that if a sample possesses at least 10 dB attenuation , then and in (2) and (3) can be approximated to^44,57

7

It is seen from (7) that changes with (reflection from sample surface), whereas varies with both and P (a term related to attenuation through the sample).

Second, values of FA geopolymers with Na-2 solution for 15 mm and 20 mm are less than the widely-accepted nominal value of dB⁵², below which accuracy of (or ) measurements significantly decrease. Besides, as seen from Fig. 3a–c, the tested FA powder has the highest value among the tested powders in geopolimerization in our study. Therefore, for an accurate water reaction, reflection properties and transmission properties (over frequency and time) of FA, GGBFS, and MK geopolymers for mm were monitored in our study.

Reflection and transmission properties and extracted permittivities of geopolymers over frequency and time

Figures 5, 6 and 7 illustrate, respectively, and FA, GGBFS, and MK geopolymers ( mm) with Na-1 and Na-2 solutions versus frequency for curing times of 0–32 hours with a 5-hours interval. Because all prepared samples after 32 hours of curing were hardened, the samples were removed from the waveguide cell, which otherwise might be stuck within the measurement cell and therefore limit the usage of this cell for a future analysis of other samples. It is seen from the results in Figs. 5, 6 and 7 that transmission properties of FA, GGBFS, and MK geopolymers with the Na-1 and Na-2 solutions considerably changes in reference to reflection properties . Therefore, transmission properties could be much preferable for analysis of water reactions within FA, GGBFS, and MK geopolymers, similar to the ASR analysis within OPC samples using transmission properties^22,23.

Fig. 5.

Measured average (a) and (b) , and (c) and (d) of the FA geopolymers ( mm) with the Na-1 and Na-2 solutions for curing times of hours with a 5-hour interval.

Fig. 6.

Measured average (a) and (b) , and (c) and (d) of the GGBFS geopolymers ( mm) with the Na-1 and Na-2 solutions for curing times of 0–30 hours with a 5-hour interval.

Fig. 7.

Measured average (a) and (b) , and (c) and (d) of the MK geopolymers ( mm) with the Na-1 and Na-2 solutions for curing times of 0–30 hours with a 5-hour interval.

Figure 8a–c demonstrate, respectively, the average (out of four measurements for each set) transmission properties of FA, GGBFS, and MK geopolymers ( mm) with the Na-1 and Na-2 solutions over time at the mid-frequency (3.275 GHz). It is noteworthy that the sweeping mode of the VNA was set active during each session of measurements (lasting less than 15 minutes for all samples) at each one-hour intervals during a 30-hours overall measurement time. Aside from measurements, the sweeping mode of the VNA was set passive to eliminate any drifting which could introduce measurement errors influencing the final results. Therefore, one-time calibration process was sufficient for all microwave reflection and transmission measurements. The following points are noted from Fig. 8a–c. First, measured of FA geopolymers ( mm) with the Na-1 and Na-2 solutions are, respectively, less than measured of GGBFS and MK geopolymers ( mm) with Na-1 and Na-2 solutions. It is thought that such a difference mainly comes with loss behavior of the FA used as a precursor in the geopolymerization. As seen from Fig. 3a–c that (around 0.50) of the FA powder is much greater than those of the GGBFS powder (nearly 0.08) and MK powder (nearly 0.03). Second, similar to the results in Fig. 3 of the study⁴³, of the FA geopolymer with the Na-1 solution at a given curing time is much smaller than of the FA geopolymer with the Na-2 solution. The same is valid for the GGBFS geopolymer samples. This difference can be explained by the water content within the Na-1 and Na-2 solutions. In the preparation of Na-1 and Na-2 solutions, for a preassigned liter of a solution, while the amounts of the sodium silicate and sodium hydroxide will be identical for the Na-1 solution, the amount of sodium silicate will be 2.5 times the amount of the sodium hydroxide for the Na-2 solution. For instance, assuming a 35 lt of solution, the Na-1 solution has equal amounts of sodium silicate and sodium hydroxide (17.5 lt) whereas the Na-2 solution has 25 lt sodium silicate and 10 lt sodium hydroxide content. For the considered 8-molar sodium hydroxide solution in our study, the sodium hydroxide has 68% water content (depends on molar value^48,49) and the sodium silicate has 56% water content. Then, the amount of water in the Na-1 solution will be higher than the amount of water in the Na-2 solution. Therefore, measured of the FA with Na-1 solution will be smaller than measured of the FA with Na-2 solution, provided that water in the solutions is in free-water state, which will be linked with our third point to be discussed shortly. To further examine this, effective of FA, GGBFS, and MK goepolymers ( mm) with the Na-1 and Na-2 solutions were retrieved versus time using the method⁵³, as shown in Fig. 9a–f. Additionally, loss tangent () values of the FA geopolymer with the Na-1 and Na-2 solutions are also drawn in Fig. 10a. It is noted from Fig. 9a,b that of the FA geopolymer with the Na-1 solution ( on the average) is greater than of the FA geopolymer with the Na-2 solution (1.38 on the average).

Fig. 8.

Measured average of (a) the FA, (b) the GGBFS, and (c) the MK geopolymers ( mm) with the Na-1 and Na-2 solutions over curing time (hours) at the mid-frequency (3.275 GHz).

Fig. 9.

Measured (a) and (b) of the FA geopolymer, (c) and (d) of the GGBFS geopolymer, and (e) and (f) of the MK geopolymer over curing time (hours) using average S-parameters (obtained from four samples for each set).

Fig. 10.

(a) Measured of the FA geopolymer with Na-1 and Na-2 solutions over curing time (hours) using average S-parameters (obtained from four samples for each set) and (b) measured temporal percentage (%) change in mass of the FA, GGBFS, and MK samples with Na-1 and Na-2 solutions.

As a third point that could be drawn from Fig. 8a–c, an increase in of the FA and GGBFS geopolymers with the Na-1 and/or Na-2 solutions with time can come from two possible mechanisms⁴³: a) evaporation of water from surfaces of the prepared samples and b) transformation of free-water to bound-water. Because all prepared samples were poured into a tiny nylon bag within a vertical measurement cell and because each prepared sample was kept within the measurement cell during all curing times, evaporation of free-water was not considered as a main mechanism for the differences of of the FA and GGBFS geopolymers with the Na-1 and/or Na-2 solutions. To validate this assumption, variation of masses of all prepared samples were measured by a highly accurate mass scale (Sf-400C - Digital Scale with 0.01 g sensitivity and 600 g capacity) before and after entire experiments in Fig. 10b. It is seen from Fig. 10b that all prepared samples have less than 2.5% mass decrease over the entire experiment. Because most of water within each sample was remained during entire experiment and because all prepared samples can be considered as lossy samples (demonstrating at least 10 dB attenuation^44,57 – see (7)), an increase in (highly related to ) could be associated with the transformation of free-water to bound-water within samples. It is known that free-water has higher and than those of bound-water¹⁶. Therefore, FA and GGBFS geopolymers with the Na-1 and Na-2 solutions seem to indicate such transformation and ongoing geopolymerization process (similar to OPCs) during hardening, which we think are mainly related to relatively high calcium content of the FA and GGBFS powders¹¹ compared with the calcium content of the MK, as seen in Table 5.

As a fourth point that could be drawn from Fig. 8a–c, of the MK geopolymer with the Na-1 solution at a given curing time is much greater than of the FA geopolymer with the Na-2 solution. Because the Na-1 solution has higher amount of free-water than the Na-2 solution has and because evaporation of water is not an effective agent in changing (as discussed before and validated in Fig. 10b, the reason for higher of the MK geopolymer with the Na-1 solution can be explained by the absorption capability of free-water (opposite of the hydration-like process occurring inside the FA and GGBFS samples). It was noted that for precursors having low calcium contents (such as MK powder in our analysis), water bound initially in the solution is assumed to be consumed by the alkaline hydrolysis¹⁰. The fifth point, related to the fourth point, is that while of the MK geopolymer with the Na-1 solution decreases with time, of the MK geopolymer with the Na-2 solution increases with time. This can be explained as follows. The MK geopolymer with the Na-1 solution seems to adsorb water over whole curing time (over 32 hours), since bound-water has a lower than the free-water has and a decrease in is mainly associated with a higher value for lossy samples producing at least 10 dB attenuation^44,57. On the other hand, in accordance with the aforementioned points, it can be stated that the MK geopolymer with the Na-2 solution seems to have a transformation of free-water to bound-water (geopolymerization) over curing time of the stored free-water.

Results of SEM and EDS tests

In order to make a comprehensive analysis, together with microwave non-destructive measurements, scanning electron microscope (SEM) micrograph and energy dispersive spectroscopy (EDS) tests of prepared FA, GGBFS, MK geopolymer samples were also performed. It is noted that these results are obtained. It is noted that data for the SEM and EDS results were monitored from various sides prepared samples, only those representative ones differentiating different properties of the prepared samples were recorded.

SEM micrograph results

Figure 11a–f illustrate SEM micrograph test results of FA, GGBFS, MK geopolymer samples with Na-1 and Na-2 solutions at the curing day of 5 under ambient environment. It should be noted that there is no curing method (i.e., heat-curing) applied on the specimens; therefore, these SEM micrographs may change at the days of 28 or even more depending on the rate of dissolution in the raw FA, GGBFS, and MK materials. In general, elevated-temperature curing accelerates the geopolymerization process, resulting in an enhanced early strength performance, especially for FA and MK based GPC^58,59. Due to the higher CaO content in the slag particles, there is no need to have an elevated temperature curing for an early strength gain. It is noted from the SEM micrographs in Fig. 11a–d that the microstructure of FA samples with Na-1 and Na-2 solutions consists of spherical vitreous particles of different sizes, whereas the microstructure of GGBFS samples with Na-1 and Na-2 solutions consists of irregular-shaped and sharped-edged with rough surface. Meanwhile, it is observed from the SEM micrographs in Fig. 11e,f that particles of MK samples with Na-1 and Na-2 solutions are irregular in shape. The micrographs of the FA samples with different Na solutions in Fig. 11a,b showed that large quantities of unreacted spherical FA particles having different particle sizes with smooth texture, indicating the incomplete geopolimerization process. The heterogeneously distributed cavities also exist in the binding paste, resulting in a more porous microstructure. In addition, the GPC microstructure includes great numbers of voids and microcracks, which would spread further and form a porous microstructure. A similar finding was also reported in an earlier investigation⁶⁰. The micrographs of GGBFS samples in Fig. 11c,d revealed that large quantities of irregular-shaped unreacted slag grains exist with narrower microcracks and voids, indicating the dense microstructure as compared to the microstructure of FA samples. On the other hand, a few amount of irregular-shaped unreacted MK grains exist with great numbers of reaction products, indicating high geopolymerization reactions for the MK samples’ microstructure. In addition, the available crack number and size are very low compared to those of the FA and GGBFS microstructures. This can be attributed to the high reactivity and fineness of MK grains, reducing porosity and permeability, while enhancing strength and durability. The crack number, length, and size are significant since cracks can cause a weakening of the mechanical performance and durability of a GPC. The MK samples showed the best, GGBFS samples performed the moderate, and FA samples exhibited the worst microstructure at the day of 5 considering the availability of cavities, voids, and microcracks.

Fig. 11.

Measured SEM micrographs at the curing day of 5 under ambient environment: (a) The FA sample with Na-1 solution, (b) the FA sample with Na-2 solution, (c) the GGBFS sample with Na-1 solution, (d) the GGBFS sample with Na-2 solution, (e) the MK sample with Na-1 solution, and (f) the MK sample with Na-2 solution.

EDS results

The EDS analyses were also performed on SEM images for the chemical characterization of the products of FA, GGBFS, MK samples with Na-1 and Na-2 solutions. The main elements identified in the EDS of FA samples in Fig. 12a,b are oxygen (O), silicon (Si), carbon (C), aluminum (Al), and sodium (Na), and calcium (Ca). Also, some traces of Mg (magnesium), sulfur (S), and Fe (Iron) were also detected, since these elements are available in FA particles. The main elements identified in the EDS of the GGBFS samples in Fig. 12c,d are Si, O, Na, C, Ca, and Al. Also, some traces of Mg, potassium (K), S, and manganese (Mn) were also found, all of which are available in an ordinary GGBFS particle. Moreover, the fundamental elements identified in the EDS of the MK samples in Fig. 12e,f are O, Si, C, Al, and Na. In addition, some traces of Ca and Fe were observed, all of which are present in a typical MK particle.

Fig. 12.

Measured EDS plots at the curing day of 5 under ambient environment: (a) The FA sample with Na-1 solution, (b) the FA sample with Na-2 solution, (c) the GGBFS sample with Na-1 solution, (d) the GGBFS sample with Na-2 solution, (e) the MK sample with Na-1 solution, and (f) the MK sample with Na-2 solution.

It should be noted that MK particles are the richest in silica (53%) and alumina (43%) but lack of calcium content (0), while FA particles have high silica (60%) and alumina (20%) but weak in calcium content (3%), whereas GGBFS particles have moderate silica (35%) and alumina (10%) but rich in calcium content (35%). Table 7 presents percentage values of Na, Al, Ca, and Si and Na/Al, Al/Si, Ca/Si, and Na/Si ratios of FA, GGBFS, and MK geopolymer paste samples. While the Si/Al ratio is mainly linked with controlling the geopolymer network structure, the Ca/Si ratio defines the binder phase (the glue), manifesting a N-A-S-H or C-A-S-H transformation. EDS results showed that with an increase in SS/SH ratio, Na/Al and Na/Si ratios decrease, while Si/Al and Ca/Si ratio generally increase as shown in Table 7. For FA samples, N-A-S-H type of gel is formed, with low Ca/Si ratio and high Al/Si ratios. The Na/Al, Si/Al, Ca/Si, and Na/Si ratios were 2.8 and 2.68, 3.15 and 3.81, 0.26 and 0.42, and 0.89 and 0.70 for the FA binder gels having SS/SH ratios of 1 and 2.5, respectively. For GGBFS samples, C-A-S-H type gel is created with high Ca/Si and Na/Al ratios. The Na/Al, Si/Al, Ca/Si, and Na/Si ratios were 5.46 and 2.81, 4.62 and 5.13, 0.69 and 1.15, and 1.18 and 0.55 for the GGBFS binder gels having SS/SH ratios of 1 and 2.5, respectively. For MK samples, N-A-S-H type of gel is produced with low Ca/Si and Na/Si ratios. The Na/Al, Si/Al, Ca/Si, and Na/Si ratios were 0.69 and 0.64, 1.26 and 1.49, 0.02 and 0,01, and 0.55 and 0.43 for the MK binder gels having SS/SH ratios of 1 and 2.5, respectively. Si and Al the main components of N-A-S-H gel in the MK samples⁶¹, and the geopolymerization proceeds as the atomic ratios of Si/Al and Na/Si increase⁶². The results showed that Si/Al atomic ratios of FA and GGBFS samples are greater than those of MK samples, indicating that geopolymerization occurred more in MK samples, in parallel with their SEM micrographs. SEM/EDS analysis results also showed that the atomic ratios of the Na/Al, Si/Al, Ca/Si, and Na/Si were similar to the literature findings^63–65.

Table 7.

EDS results: percentage values of Na, Al, Ca, and Si and Na/Al, Al/Si, Ca/Si, and Na/Si ratios of FA, GGBFS, and MK geopolymer paste samples.

Sample	Na(%)	Al(%)	Ca(%)	Si(%)	Na/Al	Si/Al	Ca/Si	Na/Si
FA with Na-1	4.6	12.9	3.7	14.5	2.80	3.15	0.26	0.89
FA with Na-2	3.7	9.9	5.9	14.1	2.68	3.81	0.42	0.70
GGBFS with Na-1	2.6	14.2	8.3	12.0	5.46	4.62	0.69	1.18
GGBFS with Na-2	3.1	8.7	18.3	15.9	2.81	5.13	1.15	0.55
MK with Na-1	12.7	8.8	0.32	16.0	0.69	1.26	0.02	0.55
MK with Na-2	14.9	9.6	0.21	22.2	0.64	1.49	0.01	0.43

Potential of microwave measurements

Given that Na-1 solution contains more free-water than Na-2 solution has and considering that water evaporation in microwave measurements was not the main mechanism for a change in , microwave measurements demonstrate the following key results. First, FA and GGBFS geopolymer pastes undergo a transformation of free-water to bound-water, similar to the hydration process in an OPC paste. Second, MK geoppolymer pastes have a transformation of bound-water to free-water, an opposite scenario observed in FA and GGBFS geopolymer pastes. Different transformation mechanisms for FA and GGBFS geopolymer pastes and for MK geoppolymer pastes could be associated with calcium contents of the prepared geopolymer samples. It is noteworthy that while FA and GGBFS geopolymer pastes have very large Ca percentage values with Na-1 and Na-2 solutions, MK geoppolymer pastes have significantly lower Ca percentage values. Although microwave measurements have the potential of examining water reaction within FA, GGBFS, and MK geopolymer pastes, presented results are preliminary and further microwave measurements along with SEM and EDS results are required to fully evaluate the water reaction within geopolymer paste samples with different binder types and binder solutions. Such a study is considered to be within our future research plan.

Finally, in our present study, microwave measurements were utilized to examine water reaction in geopolymer paste samples. However, in-situ thermal monitoring, as a means of multi-physics monitoring, could be implemented for correlating real-time temperature changes (exothermic reaction) with the complex permittivity () would help isolate the effects of heat on molecular mobility. Furthermore, combining microwave data (sensitive to water state) with ultrasonic pulse velocity (sensitive to mechanical stiffness) would allow for a complete mapping of the geopolymerization stages from liquid to solid. In addition, fitting the measured data to these physical models can provide quantitative values for “relaxation time”, which directly relates to how “restricted” the water molecules are within the growing geopolymer network. Using models like the Bruggeman or Maxwell-Wagner theories to predict the volume fraction of the newly forming geopolymer gel (N-A-S-H or C-A-S-H) based on the change in microwave transmission.

Conclusions

Microwave reflection and transmission properties, as a means of non-destructive testing tool, were carried out for evaluating water reaction within the prepared geopolymerv samples with FA, GGBFS, and MK precursors having sodium silicate/sodium hydroxide ratios of 1 and 2.5. It is observed that while reflection properties of geopolymer FA (or GGBFS or MK) samples with Na-1 and Na-2 solutions are slightly different, the corresponding transmission properties are substantially different, meaning that microwave transmission properties are more useful for examining water reaction with the prepared FA, GGBFS, and MK paste samples. It is noted from microwave measurements that the difference between microwave transmission properties of FA (or GGBFS) samples with Na-1 and Na-2 solutions is coming mainly from transformation of free-water to bound-water (just as a hydration-like process within cement-based materials), which is closely related to relatively high amount of calcium content within these samples. On the other hand, it is observed that the difference between microwave transmission properties of MK samples with Na-1 and Na-2 solutions is arising from transformation of bound-water to free-water which is mainly associated with low calcium content within these samples—a scenario opposite of water reaction within FA and GGBFS samples. To support all these conclusions, relative complex permittivity of all prepared samples with Na-1 and Na-2 solutions, together with SEM micrograph and EDS tests, were examined.

Author contributions

U.C.H. investigated the subject; U.C.H. performed conceptualization analysis; U.C.H. and H.K. conducted the experiments; U.C.H. and H.K. validated the results; U.C.H. and H.K. prepared illustrations (visualization); U.C.H. analyzed the results; U.C.H. and H.K. prepared the original draft; and U.C.H. supervised the study.

Funding

U. C. Hasar and H. Korkmaz would like to thank the Scientific and Technological Research Council of Turkey (TUBITAK) under the grant number 120M763 for providing the support to conduct their research.

Data availability

The datasets used and analyzed during the current study are available from the corresponding author (U.C.H) upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.