New Metakaolin-Based Geopolymers with the Addition of Different Types of Waste Stone Powder

The search for new alternative raw materials and their subsequent use supports the sustainability of natural resources. This article describes the use of waste stone powder (WSP) in geopolymer synthesis. Five different types of WSP (feldspar, limestone, marlstone, dolomite, marble) were comprehensively characterized and their influence on the resulting properties (especially mechanical and textural properties, setting time and structure) of metakaolin-based geopolymer composites was investigated. Their comparison with a geopolymer composite containing only quartz sand has revealed that WSP additions have a small but positive effect on the mechanical or textural properties of geopolymers. Moreover, setting time measurements have shown that waste stone powders can be used as an accelerator of geopolymer reaction solidification. The results demonstrated that the mentioned types of WSP, previously landfilled, can be used for the preparation of geopolymers. This can help reduce the environmental burden.


Introduction
Sustainability is becoming a keyword in all aspects of human life. Research often focuses on the development of new and more efficient methods and technologies for using natural resources as well as on the possibilities of exploiting hitherto neglected resources such as waste materials or raw materials that cannot be used classically. Not only does the use of these waste materials save natural resources, but it also has ecological and economic benefits. Organic and inorganic waste materials are meaningfully reused in other ways than landfilling, e.g., by thermal recovery [1,2], in construction, alternative binder and composite production [3][4][5][6][7], or as potentially effective adsorption materials [8,9].
This article focuses on waste materials from natural stone processing. The extraction and processing of natural stones are among the earliest human activities. Stonemasons and sculptors in all cultures have participated in the creation of buildings, monuments, statues and other artifacts. During the extraction of natural stone in quarries, its handling, transportation and especially final processing, waste materials are generated. Larger broken pieces or cuttings can often be used for other applications, e.g., for decorative or horticultural purposes. On the contrary, fine powder generated during stone cutting and grinding is used minimally. The waste stone powder (WSP) slurry is usually filter-pressed and subsequently landfilled. This material has recently begun to be partially used in mortar and concrete or in the stabilization of clayey soils [10][11][12].
One of the ways to reuse and recycle very fine waste stone powder is its application to geopolymer materials, known since 1979. Since then, these innovative materials, their synthesis and properties have been studied worldwide [13][14][15][16]. Their structure is predominantly amorphous and consists of aluminate and silicate units (-Si-O-Al-O-) in different Si/Al ratios [14,[17][18][19]. An appropriate choice of raw materials and fillers and/or additives significantly affects the resulting properties of geopolymer composites, in particular heat and fire resistance, high resistance to aggressive media, high early strength, and mechanical properties comparable with other inorganic binders. The presence of pores of different sizes in the final products could affect the mentioned properties both negatively and positively. Therefore, textural analyses should be provided. Additionally, geopolymers are able to accept different types of materials into their network regardless of granulometry.
The influence of various types of active or inert fillers in the different matrices (involved or not in the geopolymer reaction, respectively) has been studied by many researchers [20][21][22]. Perez-Cortes et al. [23] investigated the addition of limestone in metakaolin-based geopolymers. According to Ref. [24], geopolymer composites were prepared with dolomite quarry dust using fly ash-based geopolymer. The properties of marble waste in connection with the metakaolin via alkali activation were also studied [25]. However, the comparison of five different WSP (feldspar, limestone, marlstone, dolomite, marble) has not been reported so far.
This article presents a characterization of five different types of WSP (feldspar, limestone, marlstone, dolomite, marble) and investigates their influence on the properties of metakaolin-based geopolymer composites. It may be assumed that the addition of WSP will affect the resulting properties of geopolymers, especially setting time, structure, mechanical and textural properties.

Materials
For the production of the standard geopolymer samples, industrially supplied raw materials were selected, including clay material L05 (ČLUZ a.s., Nové Strašecí, Czech Republic), quartz sand STJ 25 with the grain-size range of 0.063-0.315 mm (Sklopísek Střeleč, a.s., Střeleč, Czech Republic), potassium silicate solution 3.0 (Vodní sklo, a.s., Prague, Czech Republic), and potassium hydroxide (KOH; Penta, s.r.o., Prague, Czech Republic). These products have their parameters (e.g., chemical composition, purity, and particle size) guaranteed by their producers. The chemical compositions of the clay material and quartz sand are presented in Table 1, where they are complemented by loss on ignition (LOI). The types of waste stone powder used in the experiments (feldspar, limestone, dolomite, marble, marlstone) come from Czech stone-processing plants. The dried powder was used in the original granulometry, i.e., it was not further pulverized. The chemical compositions and particle size analyses of the various types are summarized in Table 2 and  Table S1 (Supplementary Materials), respectively. = 0.50). The geopolymer matrix was then transferred to a laboratory mixer with a planetary mixing action, where quartz sand (42.5 wt. %) or a mixture of quartz sand and stone powder was added to the geopolymer matrix. The A addition of WSP was 8 g per 50 g of the clay material L05 (5.6 wt. %), whereas the B addition was 16 g of WSP per 50 g of the same clay material (10.6 wt. %). After homogenization (10 min), the geopolymer mass was poured into 40 × 40 × 160 mm molds, which were vibrated (5 min) to remove unwanted air bubbles. The molds were protected from water evaporation by being covered by a glass panel for 24 h. After that, the solid samples were de-molded and stored for 28 days in plastic bags and further in the air under laboratory conditions. The designation of geopolymer samples and mix design are summarized in Table S2 (Supplementary Materials).

Methods
Chemical composition was determined using a non-destructive X-ray fluorescence (XRF) analyzer (Spectro IQ, Kleve, Germany) equipped with a HOPG Barkla crystal. The instrument has the following parameters: helium atmosphere, the palladium target material, the target angle from the central ray: −90 • , the focal point: a 1 mm × 1 mm square, the maximum anode dissipation of 50 watts, and 10-cfm forced-air cooling.
The particle-size distribution was measured by a CILAS 920L analyzer equipped with laser-beam reflection. The measurement was performed in an aqueous suspension with the simultaneous use of ultrasound to avoid an agglomeration of the tested particles.
Samples for X-ray diffraction (XRD) analysis were ground to analytical fineness using a McCrone Mill in an ethanol suspension. XRD analyses were performed with a Bruker D8 Advance X-ray diffractometer (Bruker, Billerica, MA, USA) using Cu Kα radiation and a Lynx Exe XE detector. The powder sample was placed on a flat surface (non-diffractive silicon). Diffraction was recorded in the range of 4-80 • , in steps of 0.015 • and with a reading time of 0.8 sec per step. The obtained X-ray diffraction records were qualitatively evaluated in the Diffrac.Eva program (Bruker AXS, 2015) using the ICDD PDF-2 database (ICDD 2018).
The setting time was determined by a fully automatic testing device, Vicatronic (Matest, Viale Mantegna, Italy), according to the Czech StandardČSN EN 196-3 (Methods for Testing Concrete -Part 3, 2017). The data were evaluated using the VicatWin program (Matest, Viale Mantegna, Italy). The measurement was repeated twice for each geopolymer mixture.
After 28 and 90 days, mechanical properties were measured using prisms with a standard size of 40 × 40 × 160 mm according to the Czech StandardČSN EN 196-1 (Methods for Testing Concrete-Part 1, 2016). The tests were performed on an E152 compression and flexural testing machine with an E170 compression and E172 flexural devices (Matest, Viale Mantegna, Italy). The resulting values of flexural and compressive strength were determined as the arithmetic average of three and six specimens, respectively.
The infrared (FTIR) spectra were measured using an iS50 spectrometer (Nicolet Instruments, Madison, WI, USA) by the attenuated total reflection (ATR) method using a diamond crystal. All the spectra were recorded in the absorption mode at the resolution of 4 cm −1 with 64 scans in the middle infrared range of 4000-400 cm −1 . The obtained spectra were processed using the OMNIC software, version 9.
Indirect physicochemical methods, such as gas sorption or mercury intrusion/extrusion porosimetry, are usually used for the complex textural characterization of solid materials. The texture of porous solid materials is usually characterized by the surfaces of micropores in selected size ranges, by pore-size distribution and porosity [26].
A volumetric sorption analyzer (Thermo Fisher Scientific, Waltham, MA, USA) was used for the determination of the surface properties of mesopores and macropores, with the main parameter being specific surface area (S BET ). The measurement was based on nitrogen physisorption at the temperature of −196 • C and under pressurization to 100 kPa, and the calculation was performed using the BET equation [27].
PASCAL 140 Evo and PASCAL 440 Evo mercury porosimeters (Thermo Fisher Scientific, Waltham, MA, USA) were used for the measurements of the pore-size distributions of meso-and macropores in the size range of 3.5 nm-60 µm, with porosity (P) as the main parameter, and of the surface of meso-and macropores (S Hg ) [28]. The measurements were based on the intrusion of mercury into pores under increasing pressure and extrusion under decreasing pressure.
An IGA-100 gravimetric sorption analyzer (Hiden Isochema, Warrington, UK) serves for the determination of the surface of micropores (S m ) from the CO 2 adsorption isotherm, where n is the real adsorption capacity at 25 • C and under the pressure of 0.1 MPa. After the degassing and evacuation of the materials at 105 • C, CO 2 adsorption was measured at the temperature of 25 • C [29,30].
XRF, XRD and IR analysis are based on one measurement of a representative sample obtained by quartering. The resulting values of textural analyses proceed from two measurements of representative samples. Table 2 shows the chemical analyses of the WSP. The highest amount of alumina is visible for feldspar and marlstone (17.42 wt. % and 4.45 wt. %, respectively). Major calcium oxide content was identified for limestone and marble (55.04 wt. % and 51.24 wt. %, respectively). In contrast, the lowest CaO content was identified in feldspar (0.55 wt. %). This, together with the quantity of potassium oxide (7.47 wt. %), indicates that it is probably potassium or sodium-potassium feldspar. The lowest value of the loss on ignition (LOI) was observed for feldspar (0.61 wt. %), followed by marlstone (13.88 wt. %). The LOI values of the remaining types of WSP are very similar, which results from their carbonate nature.

Waste Stone Powder Characteristics
The particle-size distributions of the waste stone powder, presented in Table S1, have shown differences between the individual types. The highest content of particles smaller than 1 µm was observed for marble (12.6 wt. %) and dolomite (8.06 wt. %) and the lowest for feldspar (2.81 wt. %). As evidenced by the D90 value in Table S1, marble has the finest particles of all the WSP studied-90% of the particles are smaller than 24.45 µm. Despite the low value of the particles below 1 µm, feldspar is the second finest waste stone powder (D90 = 63.56 µm). In contrast, the highest value of D90 was observed for marlstone and dolomite (118.63 µm and 105.26 µm, respectively).
The mineralogical analyses of the stone powder (Table S3-Supplementary Materials) correspond to the chemical analyses presented in Table 2. They have confirmed the authenticity of the stone powder. In the case of feldspar, the presence of potassium and sodium-potassium feldspar in the form of microcline and albite was confirmed. At the same time, feldspar also contains quartz, kaolinite and muscovite, the products of its weathering. In the case of limestone powder, the analysis has determined that the material is not pure calcite but one with magnesium admixture (magnesium calcite). Marlstone is predominantly composed of magnesium calcite and quartz, followed by microcline and muscovite. Dolomite powder is composed of dolomite and calcite magnesian with traces of quartz and biotite. Table 3 shows the results of the initial, final and real setting times (IST, FST and RST, respectively) of geopolymer composites with different types of waste stone powder when RST = FST − IST. As a reference, the IST and FST values of geopolymer with quartz sand (STJ 25) were considered (535 and 660 min, respectively). From the results, it is clear that the addition of WSP and mainly the particle size (D50) influence the IST and FST for all types of stone powder except for dolomite, which was removed from the correlation because of its higher MgO content. For the remaining types of WSP, the smaller WSP particle size (D50) has led to shorter IST and FST (R 2 = 0.9310 and 0.9854, respectively), see Figure S1 (Supplementary Materials). The rate of the geopolymer reaction appears to depend on the particle size of the added WSP as in the case of clay materials [31]. The shortest IST was observed for marble, B addition (290 min). The Marble-B sample also has the shortest FST (375 min) and thus the RST (85 min). Although Feldspar-B and Limestone-B have similar RST (85 min and 95 min, respectively) to Marble-B, their solidification began and ended later (see Table 3). In the case of marlstone, it is remarkable that although solidification in Marlstone-A began one hour later than in Marlstone-B, the resulting total setting time is almost identical (115 min and 125 min, respectively). In addition, marlstone is also the only waste stone powder for which the RST is shorter for the A addition than for the B addition.

Geopolymer Composite Characteristics
The geopolymer composites with WSP added were also monitored in terms of mechanical properties. Table 4  Furthermore, only samples with the B addition of WSP were used for X-ray and IR analyses and for the textural analyses (pore surface areas, porosity, real sorption capacity) because of the higher effect of WSP.
The XRD analyses of geopolymer composites with the B addition have not revealed (Table S4-Supplementary Materials) any formation of new crystalline phases. The presence of an amorphous phase, which is typical of geopolymer materials, has newly been detected in all samples in large amounts in comparison with the results of the original WSP. As expected, after the dilution of the crystalline phases of pure stone powder with geopolymer, the contents of these phases were reduced to minor or even trace. The FTIR-ATR spectra of all the studied materials are shown in Figure 1. The FTIR method is able to evaluate the geopolymer reaction by the identification of geopolymer bonds. The formation of geopolymer bonds (Si-O-Al) can be proven by a spectral shift of the principal band at 1080 cm −1 to lower wavenumbers [32]. This is evident in the case of the spectrum of the STJ 25 sample, in which the wavelength value decreases to 1003 cm −1 . The reactions of this material with WSP do not shift the Si-O-Al band much, which means that the WSP does not engage in the geopolymer structure. The shifts are negligible with respect to resolution (4 cm −1 ).  AcC-activated carbon, SBET-specific surface area, Sm-surface of micropores, SHg-surface of meso-and macropores, Por-porosity, n-real adsorption capacity, Esorp-adsorption energy. The textural parameters presented in Table 5 as the specific surface area and the surface of micropores provide information about pores up to 50 nm. Activated carbon serves as the referential material only for a comparison of textural parameters. The variability of pore surfaces in the studied materials is low. Nevertheless, the relationship between the parameters n and S m , shown in Table 5, is evident (R 2 = 0.9482). Their behavior is thus similar to the most solid natural materials [33,34]. AcC-activated carbon, S BET -specific surface area, S m -surface of micropores, S Hg -surface of meso-and macropores, P-porosity, n-real adsorption capacity, E sorp -adsorption energy.
Micropores are the main contributor to the total surface area and a key factor for enhancing sorption capacities. On the other hand, the meso-and macropores strongly contribute to material instability. S Hg positively correlates with IST and FST (R 2 = 0.7544, R 2 = 0.6977, respectively). There is a strong positive relationship (R 2 = 0.9102) between S Hg and the flexural strength after 28 days; on the other hand, the trend with the flexural strength after 90 days is the opposite (R 2 = 0.5505), as can be seen in Figure S2A,B. The larger the surface of the meso-and macropores, the higher the flexural strength at 28 days of maturation. A possible explanation is that in a material with low maturation when the humidity is higher, the water is adsorbed in mesopores, which have space for deformation and cause some elasticity of the material. On the contrary, a higher content of micropores (see Figure S2C,D) decreases the flexural strength after 28 d but causes better flexural strength in longer, 90-day maturation time (R 2 = 0.8421) and better comprehensive strength (28 d) (R 2 = 0.7886). It was found that the comprehensive strength after 90 days is not much affected by the presence of any type of pores.
The pore-size distribution obtained by mercury porosimetry (Figure 2) shows the prevalence of the pores below 10 nm. It is evident that the basic geopolymer composite STJ 25 has a relatively higher volume of pores of about 250 nm, which are, however, eliminated with the addition of WSP. The presence of larger pores is about six times lower in comparison with AcC (see Table 5). The average content of micropores in the STJ 25 is lower than in geopolymers with the addition of feldspar, marlstone and dolomite and higher than with marble and limestone added. The low porosity (P) means that the material has a low content of all pores smaller than 60 µm. In the case of the STJ 25, the P-value is similar to the P-values of the composite materials with WSP. It is evident that the addition of WSP has not affected the textural properties of the studied composites much-they mostly depend on the properties of the basic material STJ 25. N 2 adsorption isotherms ( Figure 3) confirm that the materials are microporous (the presence of pores <2 nm) and mesoporous (the presence of pores of 2-50 nm). The hysteresis curve indicates the prevailing shapes of pores, which fall into a mixed group of H2-inkbottle pores and H4-slit-and wedge-shape pores [35,36]. The first type, H2, is usually a wide hysteresis with the inflection point at the pressure of 300-400 kPa on the desorption curve which is an indicator of ink-bottle pores with a narrow or almost closed neck. In the second type, H4, the shapes of pores are more open than the cylindrical pores present, for example, in AcC [35,36]. The types and courses of hysteresis depend on many factors, in this case on the presence of carbonates (as shown in Figure 1). Feldspar, marble, limestone and dolomite have shifted the curves to higher pressures ( Figure 3A-C, which implies the presence of more open pores and a smaller share of micropores. The amount of the adsorbed N 2 increases with decreasing Si/Al ratio. Marlstone has the highest Si/Al ratio and the lowest amount of N 2 adsorbed of all the samples ( Figure 3D). is similar to the p-values of the composite materials with WSP. It is evident that the addition of WSP has not affected the textural properties of the studied composites much-they mostly depend on the properties of the basic material STJ 25. N2 adsorption isotherms (Figure 3) confirm that the materials are microporous (the presence of pores < 2 nm) and mesoporous (the presence of pores of 2-50 nm). The hysteresis curve indicates the prevailing shapes of pores, which fall into a mixed group of H2 -ink-bottle pores and H4-slit-and wedge-shape pores [35,36]. The first type, H2, is usually a wide hysteresis with the inflection point at the pressure of 300-400 kPa on the desorption curve which is an indicator of ink-bottle pores with a narrow or almost closed neck. In the second type, H4, the shapes of pores are more open than the cylindrical pores present, for example, in AcC [35,36]. The types and courses of hysteresis depend on many factors, in this case on the presence of carbonates (as shown in Figure 1). Feldspar, marble, limestone and dolomite have shifted the curves to higher pressures ( Figure 3A-C, which implies the presence of more open pores and a smaller share of micropores. The amount of the adsorbed N2 increases with decreasing Si/Al ratio. Marlstone has the highest Si/Al ratio and the lowest amount of N2 adsorbed of all the samples ( Figure 3D).  N2 adsorption isotherms (Figure 3) confirm that the materials are microporous (the presence of pores < 2 nm) and mesoporous (the presence of pores of 2-50 nm). The hysteresis curve indicates the prevailing shapes of pores, which fall into a mixed group of H2 -ink-bottle pores and H4-slit-and wedge-shape pores [35,36]. The first type, H2, is usually a wide hysteresis with the inflection point at the pressure of 300-400 kPa on the desorption curve which is an indicator of ink-bottle pores with a narrow or almost closed neck. In the second type, H4, the shapes of pores are more open than the cylindrical pores present, for example, in AcC [35,36]. The types and courses of hysteresis depend on many factors, in this case on the presence of carbonates (as shown in Figure 1). Feldspar, marble, limestone and dolomite have shifted the curves to higher pressures ( Figure 3A-C, which implies the presence of more open pores and a smaller share of micropores. The amount of the adsorbed N2 increases with decreasing Si/Al ratio. Marlstone has the highest Si/Al ratio and the lowest amount of N2 adsorbed of all the samples ( Figure 3D). Generally, in the materials studied, the molecules have worse conditions for capture on the pore surface and need to exert more energy for sorption. The worse the sorption, the greater the sorption energy. Materials with high sorption energy have poor sorption ability [27,37]. As can be seen in Table 5, the activated carbon has very high porosity, because the surfaces of the pores are very large. The sorption energy of AcC is approximately two times lower than for composites, which is caused by the easy capture of the CO2 molecules on carbon surfaces. The ability of the studied materials to capture CO2 is low.
The sample with feldspar has a larger pore volume than others. It was observed that Generally, in the materials studied, the molecules have worse conditions for capture on the pore surface and need to exert more energy for sorption. The worse the sorption, the greater the sorption energy. Materials with high sorption energy have poor sorption abil-ity [27,37]. As can be seen in Table 5, the activated carbon has very high porosity, because the surfaces of the pores are very large. The sorption energy of AcC is approximately two times lower than for composites, which is caused by the easy capture of the CO 2 molecules on carbon surfaces. The ability of the studied materials to capture CO 2 is low.
The sample with feldspar has a larger pore volume than others. It was observed that the feldspar content increases the micropore content in natural materials such as sediments [33]. In the case of the geopolymers concerned, the addition of the studied amount of feldspar is not high enough to change the properties of the material much.

Conclusions
The main findings of the presented results can be summarized as follows: Concerning infrared spectroscopy, there is no significant shift in Si-O-Al bands, which means that the WSP is engaged in the geopolymer structure only minimally or not at all.

•
The WSP does not worsen the textural properties, such as pore volumes and surfacesporosity. These remain more or less unchanged, approximately in the values of the porosity of the matrix. Only the addition of feldspar slightly increases the porosity, sorption capacity and content of micropores.

•
The obtained results will make it possible to recycle WSP, which was previously landfilled, as a new source of material for the preparation of geopolymers. The use of alternative sources of raw materials plays a major role in the reduction of the environmental burden and in the pursuit of the sustainability of natural resources.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/cryst11080983/s1, Table S1. Particle-size analyses of the different types of waste stone powder, Table S2. Mix design summary, Table S3. XRD analyses of the stone powder used, Table S4. XRD analyses of geopolymers with different types of waste stone powder, Figure S1. The dependence of solidification time on WSP particle sizes, Figure S2.