Characterisation of Industrial Side Streams and Their Application for the Production of Geopolymer Composites

This study focuses on characterisation of side streams including biomass fly ash, biomass bottom ash, coal fly ash, green liquor dregs, limestone mine tailings, and electric arc furnace steel slag from different industrial locations in Finland. It was found that the fly ash samples contained the highest Al2O3 and SiO2 concentrations, a large number of spherical particles of small sizes and high specific surface areas. Fly ashes and steel slag were observed to contain higher amounts of amorphous phases compared to the other side streams. The high loss on ignition value of the coal fly ash and green liquor dregs was found to exceed the limitations for their application in geopolymer composites. Despite their relatively high concentrations in ashes and steel slag, the leaching tests have shown that no hazardous metal leached out from the streams. Finally, test specimens of geopolymer composites (GP2) were prepared by the application of biomass fly ash, bottom ash, and limestone mine tailings without any pre-treatment process, in addition to the ordinary Portland cement-(R) and metakaolin-based geopolymer composites (GP1). The measured compressive (14.1 MPa) and flexural strength (3.5 MPa) of GP2 suggest that it could be used in concrete kerbs and paving flags. The data has also shown that over 500% of the compressive strength was developed between 7 and 28 days in GP2, whereas in the case of reference concrete (R) and the metakaolin-based geopolymer composite (GP1) it was developed in the first 7 days.


Introduction
According to the 2030 climate and energy framework adopted by the European Council in October 2014, the target for 2030 is to reduce the GHG (greenhouse gas) emissions by 40% (from 1990 levels), to increase the share of renewable energy to at least 32%, and to improve the energy efficiency to 32.5% [1]. Therefore, continuous efforts are required; for instance, the cement industry, on a global scale, is responsible for 8% of world's CO 2 emissions despite improvements in terms of energy efficiency [2]. As part of the attempts to address issues related to waste management and greenhouse gas emissions in the construction industry, numerous studies have been conducted that consider the use of geopolymer composites as an alternative to ordinary Portland cement (OPC) [3][4][5].
Geopolymer composites are inorganic materials that have been produced by polymerisation of materials with high aluminosilicate contents. The polymerisation is a result of the fast chemical reaction of silica and aluminosilicate under strong alkaline conditions with, e.g., NaOH, Ca(OH) 2 , Na 2 O, LiOH, and Na 2 SiO 3 [6]. Different polymeric chains may be generated, such as poly ( [7]. steel slag from different industrial locations in Finland. To demonstrate the findings in practice, a geopolymer composite was produced using different residues without any pre-treatment, and preliminary results on the strength of the material were compared with those of OPC concrete. For this purpose, a number of analytical methods (PSD, BET, Digital image analysis, SEM, EDS, XRF, XRD, ICP and TGA) were used to investigate the composition and structure of the side streams in addition to the leaching study to evaluate their chemical stability in aqueous environment. Finally, the compressive and flexural strength tests were performed to determine the mechanical properties of the produced geopolymer composites.

Origins of the Samples
The total amount of samples considered in this study was eight, and they were all collected from industrial plants in Finland. The origins of the samples and the notations used are presented in Table 1.

Analyses
Particle size distributions (PSD) of the samples were measured with a laser diffraction particle size analyser with a Hydro EV particle dispersing unit (Malvern Mastersizer 3000, Malvern Instruments, Malvern, UK). The PSD of bottom ash (BA) was not measured as the particle size of the sample exceeded the operating range of the equipment (3 mm). Image analyses were performed with Morphologi G3 particle characterisation equipment (Malvern Instruments, UK). The original steel slag consisted of large aggregates; therefore, it was necessary to homogenise it for the study. Due to the major change in the structure of the steel slag following crushing, PSD measurements and image analysis were not performed for this sample. The specific surface areas (SSA) were measured with a BET surface area analyser (Micromeritics 3Flex, Norcross, GA, USA) under nitrogen atmosphere. The SSA of the steel slag sample was not measured for the above-mentioned reason. A scanning electron microscope (SEM) was used to investigate the surface topography of the samples; elemental compositions were obtained by energy-dispersive X-ray spectroscopy (EDS) mapping (Hitachi SU 3500 scanning electron microscope, Tokyo, Japan). The BA and MT samples were manually crushed for SEM and EDS analyses. EDS analysis was performed for at least 3 points for each sample and the average value was taken. Elemental compositions were obtained by a sequential X-ray fluorescence spectrometer (XRF) (Bruker AXS S4 Pioneer, Billerica, MA, USA). Carbon and sulphur contents were determined by a LECO CS844 Analyser (LECO, St. Joseph, MI, USA). Bottom ash (BA) and FA2 (biomass fly ash 2) were ground prior to the XRF and carbon and sulphur analyses. All samples underwent X-ray diffraction (XRD) analysis to investigate their mineral compositions. The samples were homogenised with a pestle and mortar prior to the analyses. The mineral compositions were determined with an X-ray diffractometer (Bruker D8 Advance X-ray diffractometer, Billerica, MA, USA, Cu source-Kα = 1.5406 Å). Metal concentrations were determined by inductively coupled plasma mass spectroscopy (ICP-MS). ICP-MS analyses were performed after complete digestion using standardized 4-acid digestion and peroxide smelt digestion methods. The leaching tests were performed according to EN-12457-2 (shaking for 24 h, L/S = 10), which was followed by filtration of the leachates and their analysis by ICP-MS (Agilent 799 inductively coupled plasma mass spectrometer, Santa Clara, CA, USA). The thermogravimetric analyses (TGA) were performed with a thermomicrobalance (Netzsch STA 449 C Jupiter, Netzsch-Gerätebau, Selb, Germany). The original solid samples were heated from 25 to 1200 • C at a constant heating rate of 10 • C/min under air atmosphere. An online mass spectrometer was used for identification of the gaseous products (QMS Netzsch, Selb, Germany).

Preparation and Testing of the Geopolymer Composites
Two different geopolymer composites were prepared: 1) Metakaolin-based geopolymer composite (GP1) (Argical M1000, Imerys S.A.) and 2) Biomass fly ash-based geopolymer composite (GP2) by Apila Group Ltd. In addition, an OPC (Plussementti CEM II/B-M (S-LL), Finnsementti Oy) concrete was prepared and used as a reference material. A commercial coal fly ash (Ecofax M20, Fatec Oy) was incorporated as the binder material in addition to FA1 in GP2. The mixture design of the geopolymer composites is presented in Table 2. The preparation of the materials was performed as follows. A premixed Na 2 SiO 3 /NaOH/H 2 O solution with a solids content of 44% and SiO 2 /Na 2 O weight ratio of 1.7 (Geosil 34417, Woellner GmbH, Ludwigshafen, Germany) was cooled to room temperature and poured into a 20-L capacity Hobart-mixer followed by the addition of the aluminosilicate precursors and aggregates. The mixture was blended for 3 min at low speed (107 rpm), and the fresh pastes were then placed inside the moulds of sizes 100 × 100 × 100 mm 3 for the compressive strength tests, and of 40 × 40 × 160 mm 3 for the flexural strength tests. The moulds were vibrated for 15-45 s to level the cast surface and to release the air bubbles inside. A plastic sheet was utilised to cover the samples to prevent the moisture loss. The samples were cured while wrapped in an air-tight plastic at room temperature (20 ± 1 • C) for 7 and 28 days for mechanical testing. Sample cubes and prisms were tested according to standards SFS-

Particle Size Distributions and Specific Surface Areas
Particle size distributions and specific surface areas of the samples are shown in Table 3 and Figure 1. MTF contains the finest particles with FA3, while MT were found to consist of coarser particles leading to a smaller specific surface area. Interestingly, FA2 has a smaller surface area than FA1 considering its fine particle size distribution. Similarly, the specific surface area of MTF was found to be very small despite its fine particle size distribution. A possible explanation for this might be the smaller amount of carbon present in this sample compared to the others, as the correlation between the specific surface area and unburnt carbon content is widely acknowledged [30]. It seems also possible that, FA1 might contain higher number of porous, or generally more porous, particles compared to FA2, resulting a higher specific surface area than FA2.

Digital Image Analysis
Aspect ratio, circularity (roundness), and convexity values of the samples are presented in Table 4. The aspect ratio is defined as Dmax/Dmin, where Dmax and Dmin are the largest and smallest dimensions of the projected particle [32]. Circularity indicates the similarity degree of the particle to a circle, and it is given with the equation 4 / where S is the area and P is the perimeter [32]. Convexity is the ratio of the hull perimeter to the perimeter, and it is a measure of the overall roundness of a particle [32]. In a study investigating the influence of aspect ratio, convexity and circularity, it was reported that increasing the aspect ratio has a major negative influence on the packing density of sand grains, while the strongly correlated parameters convexity and circularity have a minor positive effect on the packing performance of the same materials [32]. In a comprehensive study investigating the impact of particle circularity on the packing density of natural and crushed sands, it was found that with more circular particles, it is possible to reduce the void ratio, and thereby to produce a matrix with high packing density [33]. The convexity was reported to be proportional to the packing density [14], and it was attributed to the difficulty of filling the voids between concave particles [34]. The highest circularity values obtained for fly ashes are consistent with the SEM analysis results pre- In geopolymeric reactions, the reactivity of the raw materials usually increases with particle fineness due to the increased surface area, which subsequently improves the mechanical properties of the geopolymer composite [4]. This improvement could also be attributed to the increased SiO 2 /Al 2 O 3 ratio, which is known to provide the optimal binding properties to the geopolymer composite when it is within the range of 2.0 to 3.5; and reduced contents of hematite (Fe 2 O 3 ), lime (CaO) and loss on ignition [12,31]. In addition, finer particles are known to improve the microstructure of the final composite by decreasing the capillary pore size in the structure [8]. When the four different types of ashes investigated in this study are assessed from this point of view, FA1, FA2 and FA3 could potentially produce geopolymer composites with better microstructural characteristics and mechanical properties, compared to the ones produced from BA. The elimination of coarse particles in BA must be performed by milling. From the aggregates point of view, MTF and GLD seem to be promising fine aggregates as they do not contain particles larger than 4 mm. Li et al. (2017) observed that the durability and compressive strength of the concrete increased with the addition of fine limestone fillers (D50 = 13.6 µm), and concluded that different fillers with similar particle sizes would have the same effect on concrete through increased packing density [13]. BA and MT, on the other hand, could be used as a coarse aggregate due to their coarse particle size distribution.

Digital Image Analysis
Aspect ratio, circularity (roundness), and convexity values of the samples are presented in Table 4. The aspect ratio is defined as D max /D min , where D max and D min are the largest and smallest dimensions of the projected particle [32]. Circularity indicates the similarity degree of the particle to a circle, and it is given with the equation 4πS/P 2 where S is the area and P is the perimeter [32]. Convexity is the ratio of the hull perimeter to the perimeter, and it is a measure of the overall roundness of a particle [32]. In a study investigating the influence of aspect ratio, convexity and circularity, it was reported that increasing the aspect ratio has a major negative influence on the packing density of sand grains, while the strongly correlated parameters convexity and circularity have a minor positive effect on the packing performance of the same materials [32]. In a comprehensive study investigating the impact of particle circularity on the packing density of natural and crushed sands, it was found that with more circular particles, it is possible to reduce the void ratio, and thereby to produce a matrix with high packing density [33]. The convexity was reported to be proportional to the packing density [14], and it was attributed to the difficulty of filling the voids between concave particles [34]. The highest circularity values obtained for fly ashes are consistent with the SEM analysis results presented in Section 3.3, where also high contents of cenospheres can be observed. In this study, the side streams investigated with respect to their potential utilisation as aggregates are limestone tailings, green liquor dregs and bottom ash. According to Table 4, the convexities of the side streams do not vary significantly. However, the lowest circularity (0.859) and highest aspect ratio (1.543) measured for the bottom indicate that it would impair the packing density and mechanical characteristics of the final material in case of its application without milling.

Surface Morphologies and Chemical Compositions
Sample morphologies determined with SEM are shown in Figure 2. The relative Al and Si concentrations of the samples are presented in the mappings ( Figure S1). Figure 2A-C present biomass fly ash and coal fly ash samples. These samples consist of irregularlyshaped, hollow and spherical particles (cenospheres), which have been observed in other studies [5,35]. The dark and irregularly shaped particles seen in coal fly ash, for instance in the centre of Figure 2C, are likely related to unburnt carbon, which is an inference supported by the high carbon concentrations obtained by the EDS analysis (Table 5). Figure 2 shows that fly ashes ( Figure 2H), the lighter zone is considered to contain high concentrations of iron, while calcium is thought to be present at high concentrations in the darker zone [27].
The chemical compositions of the industrial side streams determined with EDS and XRF are presented in Tables 5 and 6, respectively. It is apparent from these tables that GLD contain a very small amount of SiO 2 and Al 2 O 3 and MT have a very low Al 2 O 3 content resulting in a SiO 2 /Al 2 O 3 ratio of 26.3. The SiO 2 content is very high in the case of BA; however, its coarse particle size distribution and low specific surface area would hinder its dissolution during the geopolymerisation. Even though the SiO 2 /Al 2 O 3 ratios are similar among the other samples, the highest concentrations of these components can be found in coal fly ash (FA3). Similar results have been previously reported for coal fly ash [17], biomass fly ash [36] and electric arc furnace steel slag [28]. ence supported by the high carbon concentrations obtained by the EDS analysis (Table 5).  Figure 2G) contain large, square and irregularly-shaped particles. Of the two different zones of the SS (Figure 2H), the lighter zone is considered to contain high concentrations of iron, while calcium is thought to be present at high concentrations in the darker zone [27]. The chemical compositions of the industrial side streams determined with EDS and XRF are presented in Tables 5 and 6, respectively. It is apparent from these tables that GLD contain a very small amount of SiO2 and Al2O3 and MT have a very low Al2O3 content resulting in a SiO2/Al2O3 ratio of 26.3. The SiO2 content is very high in the case of BA; however, its coarse particle size distribution and low specific surface area would hinder its dissolution during the geopolymerisation. Even though the SiO2/Al2O3 ratios are similar among the other samples, the highest concentrations of these components can be found in coal fly ash (FA3). Similar results have been previously reported for coal fly ash [17], biomass fly ash [36] and electric arc furnace steel slag [28].
CaO improves the pozzolanic properties of aluminosilicate precursors by its conversion to Ca(OH)2 in the presence of water. Ca(OH)2 forms C-S-H (calcium silicate hydrate)   CaO improves the pozzolanic properties of aluminosilicate precursors by its conversion to Ca(OH) 2 in the presence of water. Ca(OH) 2 forms C-S-H (calcium silicate hydrate) when reacted with Si, which is known to enhance the compressive strength of the composite material in addition to the geopolymeric gel [37]. However, excessive amounts of CaO reduces the setting time of the composite significantly, which results in poor workability properties to the composite. Furthermore, it may interfere the geopolymerisation reaction between sodium and the aluminosilicates. As a result of the incomplete reaction, the compressive strength of the material decreases significantly.
The influence of excess iron oxide is similar to that of calcium oxide: it consumes the alkalis in the matrix and limits the geopolymer formation [26]. On the other hand, when incorporated with fly ash, iron oxides may fill large pore volumes of the ash, leading to a reduction in water demand and an increased compressive strength [29]. Finally, it increases the stability of the material against sulphate attack, which is known to deteriorate the concrete properties [26].
The reactivity of a fly ash is closely related to its unburnt material content [31]. High specific surface areas of non-reactive porous and rough particles, such as unburnt carbon, are responsible for increasing the water demand of a workable mixture. As a result of increasing the amount of water added, the alkalinity of the solution decreases, which might lead to increased reaction times for geopolymerisation, or decreased compressive strength of the final material in the case of insufficient alkali concentrations [38]. Finally, high unburnt carbon (>10 wt.%) is responsible for changing the characteristic colour of the geopolymer composite darker and increasing the concentrations of water-soluble chlorides, which is known to increase corrosion [39]. Therefore, the EU limits [40] the highest content of loss on ignition in fly ash to 5-9% for its applications in concrete. High loss on ignition obtained in the case of FA3 and GLD exceed the maximum LOI allowed and hamper their application as raw materials.

Mineral Compositions
The XRD patterns of the samples are shown in , whose presence has been reported previously [17]. As can be seen in Figure 3d, the biomass bottom ash consists of quartz (SiO 2 ) and albite (NaAlSi 3 O 8 ) while the main crystalline phase detected in dregs (Figure 3e) was calcite, which agrees with the previous investigations [5,40] and the high content of CaCO 3 observed in the XRF analysis (Table 6) [11]. According to Figure 3, the order of the crystallinity was BA > GLD > MT > FA3 > FA1 > FA2 > SS. The crystalline and amorphous contents of aluminosilicate sources play a significant role in the production of geopolymer composites. The crystalline phase in coal fly ash consists of quartz and mullite primarily, and quartz is only partially involved in the geopolymerisation reactions, which result in the production of composites with low mechanical strength. The vitreous phase, on the contrary, is easily dissolved in the alkaline solution, and therefore the rate of geopolymerisation reaction is proportional to the vitreous content [19,31]. According to the correlation between the amorphous content and the mechanical properties of a geopolymer composite, the high level of crystallinity observed in BA, GLD, MTF and MT might indicate that these streams are not well suitable as binders for geopolymer composites. and quartz is only partially involved in the geopolymerisation reactions, which result in the production of composites with low mechanical strength. The vitreous phase, on the contrary, is easily dissolved in the alkaline solution, and therefore the rate of geopolymerisation reaction is proportional to the vitreous content [19,31]. According to the correlation between the amorphous content and the mechanical properties of a geopolymer composite, the high level of crystallinity observed in BA, GLD, MTF and MT might indicate that these streams are not well suitable as binders for geopolymer composites.

Hazardous Elements' Concentrations and Leaching Tests
The ICP-MS analysis was performed after complete dissolution of the solid samples, and the concentrations of potentially hazardous elements in the industrial side streams are presented in Table 7. When a comparison is made between FA2 and BA that originates from the same boiler, the concentrations of all the hazardous metals are higher in the fly ash than the bottom ash, which agrees with the literature [40]. High Cr (7690 mg/kg) concentrations in electric arc furnace steel slag were also found in other studies [44]. When all samples are compared with respect to their hazardous elements content, limestone tailings contain the lowest concentrations within the sample group. The limit values for harmful elements content in fly and bottom ashes in the Government Decree concerning the Recovery of Certain Wastes in Earth Construction (Decree 591/2006) issued by the Finnish Ministry of the Environment are given in Table 7 [45]. With respect to these values, FA2 exceeds the limits for Cu, Zn and Pb, the concentrations of Cu and Zn exceed given

Hazardous Elements' Concentrations and Leaching Tests
The ICP-MS analysis was performed after complete dissolution of the solid samples, and the concentrations of potentially hazardous elements in the industrial side streams are presented in Table 7. When a comparison is made between FA2 and BA that originates from the same boiler, the concentrations of all the hazardous metals are higher in the fly ash than the bottom ash, which agrees with the literature [40]. High Cr (7690 mg/kg) concentrations in electric arc furnace steel slag were also found in other studies [44]. When all samples are compared with respect to their hazardous elements content, limestone tailings contain the lowest concentrations within the sample group. The limit values for harmful elements content in fly and bottom ashes in the Government Decree concerning the Recovery of Certain Wastes in Earth Construction (Decree 591/2006) issued by the Finnish Ministry of the Environment are given in Table 7 [45]. With respect to these values, FA2 exceeds the limits for Cu, Zn and Pb, the concentrations of Cu and Zn exceed given limits in BA, SS exceeds the limit for Cr, while Zn concentration exceeds the limit value in case of GLD. Table 7. Concentrations of hazardous elements in samples (after the complete dissolution of the solids), quantification limits (LOQ) and limit values for the content of harmful metals in fly and bottom ashes generated from coal, peat and wood according to the Finnish Government Decree 591/2006, values expressed in mg kg −1 dry matter [45]. The Government Decree concerning the Recovery of Certain Wastes in Earth Construction (Decree 843/2017) also sets limits for the solubility of these elements, as their leaching could hinder the application of secondary raw materials in earth construction [46]. These limits and the leaching results (EN-12457-2, shaking for 24 h, L/S = 10) for each industrial residue are presented in Table 8 [47]. As can be seen in Table 8, no element exceeds the maximum allowable leaching limit, indicating that with respect to the leachable hazardous element contents, these samples could be suitable for the production of geopolymer composites. Table 8. Concentrations of different hazardous elements in samples after leaching, quantification limits (LOQ) and limit values for solubility of harmful substances (mg/kg, L/S = 10 L/kg) in fly ashes and bottom ashes generated from coal, peat, and wood according to the Finnish Government Decree 843/2017 [46], n.d.: not determined.

Thermogravimetric Analysis
Thermogravimetry (TG) and the normalised MS curves for H 2 O and CO 2 released from the samples are presented in Figures 4 and 5, respectively. The first slight weight loss, corresponding to temperatures below 200 • C in Figure 4, may be attributed to the release of surface water from the particles, which can also be seen as a weight increase in the m/z 18 line in Figure 5a [21]. According to Figure 5a, it can be said that only GLD contains a small amount of adsorbed water at its surface, as no peak was observed in the m/z 18 line for the other samples.
contents, which agrees with the presence of calcite (Figure 3). BA shows no weight loss, suggesting very low contents of CaCO3.
The weight losses of the samples during the entire thermogravimetric analysis were 4% (FA1), 5.5% (FA2), 8.5% (FA3), 38% (GLD) and 7.5% (MT). These values are consistent with the LOI values presented in Table 5. With respect to the standard specifications and previous studies, prior to their incorporation in geopolymer composites as binders, the unburnt carbon contents in the FA3 and GLD must be reduced.

Characteristics of the Prepared Geopolymer Composites
The side stream selection to produce geopolymer composites was determined according to the analysis performed. FA1 was applied as the binder due to its smaller amount of toxic metals compared to the ones present in FA2 and SS. GLD was not applied The weight losses between 400 and 700 • C are attributed to the oxidation of the unburnt material content. These reactions can also be observed in the increase of the ion current representing the release of CO 2 (m/z = 44) in Figure 5b. In the case of FA1, the weight loss could also be due to the dehydration of Ca(OH) 2 , whose presence was shown in Figure 3 above. The increase in m/z 18 line supports the dehydration of the hydrated lime for this sample [35,36,48]. No weight loss was observed within this temperature range in the case of bottom ash, limestone tailings, and steel slag, suggesting that these side streams do not contain unburnt carbon. The weight losses observed among the samples were as following: GLD > FA3 > FA1 > FA2. In the case of GLD, the weight loss of 9% is attributed to the unburnt carbon and the organic matter [21]. The weight loss in FA3 (8%) is due to its high unburnt carbon content [35].
The weight losses occurring after 700 • C are likely to be related to the CaCO 3 decomposition reactions which produce CaO and CO 2 [21,36,48,49]. With respect to their weight loss in this range, the samples can be ordered as following GLD > MT > FA2 > FA1 > SS. The extraordinarily high weight loss of GLD and MT may be attributed to the high CaCO 3 contents, which agrees with the presence of calcite ( Figure 3). BA shows no weight loss, suggesting very low contents of CaCO 3 .
The weight losses of the samples during the entire thermogravimetric analysis were 4% (FA1), 5.5% (FA2), 8.5% (FA3), 38% (GLD) and 7.5% (MT). These values are consistent with the LOI values presented in Table 5. With respect to the standard specifications and previous studies, prior to their incorporation in geopolymer composites as binders, the unburnt carbon contents in the FA3 and GLD must be reduced.

Characteristics of the Prepared Geopolymer Composites
The side stream selection to produce geopolymer composites was determined according to the analysis performed. FA1 was applied as the binder due to its smaller amount of toxic metals compared to the ones present in FA2 and SS. GLD was not applied as filler since they contain a considerable amount of organics and unburnt material. The compressive and flexural strengths of the prepared composites are presented in Figure 6, while the initial setting times and abrasion resistance results are shown in Table 9. Metakaolin-based geopolymer composite (GP1) developed 25.1 MPa of compressive strength and 8.1 MPa flexural strength, whereas they were measured as 50.0 and 6.0 MPa in the case of reference composite after 28 days. The compressive strength of GP2 was measured as 2.3 MPa after 7 days. Despite its significant increase to 14.1 MPa (by 513%) after 28 days, this value is below the minimum requirement for a structural construction material. The drastic increase in the compressive strength of GP2 between 7 and 28 days suggests that the most significant geopolymerisation reactions occur after 7 days, which is different from the strength development of the reference product and GP1, which produce over 80% of the maximum compressive strength in the first 7 days. This retardation may be attributed due to the slower kinetics of aluminosilicates dissolution and polycondensation reactions [18]. The relatively low compressive strength of GP2 may be attributed to the high concentration of calcium in FA1, which is known to reduce the compressive strength of the material by shortening the setting time of the fresh paste so much that the sequential geopolymerisation reactions do not occur appropriately [18]. Even though the maximum compressive strength reached in this study (14.1 MPa) is lower than that obtained in previous studies, see e.g., [6,19,22], the purpose of this work was to prove the applicability of more challenging industrial side streams (biomass fly ash and limestone tailings) with no pre-treatment, rather than aiming at achieving the best mechanical properties. Further improvements can be obtained by reduction of the calcium and iron content, particle size reduction, or by increasing the temperature, and the solution alkalinity. Nevertheless, GP2 can find applications in paving flags (EN 1339) and in concrete kerb units (EN 1340).
compressive strength reached in this study (14.1 MPa) is lower than that obtained in previous studies, see e.g., [6,19,22], the purpose of this work was to prove the applicability of more challenging industrial side streams (biomass fly ash and limestone tailings) with no pre-treatment, rather than aiming at achieving the best mechanical properties. Further improvements can be obtained by reduction of the calcium and iron content, particle size reduction, or by increasing the temperature, and the solution alkalinity. Nevertheless, GP2 can find applications in paving flags (EN 1339) and in concrete kerb units (EN 1340).

Conclusions
In this work, side streams obtained from the forest, pulp, energy and mining industries were investigated to assess their suitability without any pre-treatment as raw materials for geopolymer composites in the construction industry. The combination of the findings provides support for the conceptual premise that FA1 and FA3 are the most suitable side streams to produce geopolymer composites due to their high amorphous content, small toxic metal concentrations, large content of Al and Si and high specific surface area with correspondingly small particle sizes. FA1 was chosen over FA3 since it contains a significantly smaller loss on ignition value. Bottom ash is not suitable as a binder or coarse

Conclusions
In this work, side streams obtained from the forest, pulp, energy and mining industries were investigated to assess their suitability without any pre-treatment as raw materials for geopolymer composites in the construction industry. The combination of the findings provides support for the conceptual premise that FA1 and FA3 are the most suitable side streams to produce geopolymer composites due to their high amorphous content, small toxic metal concentrations, large content of Al and Si and high specific surface area with correspondingly small particle sizes. FA1 was chosen over FA3 since it contains a significantly smaller loss on ignition value. Bottom ash is not suitable as a binder or coarse filler without milling due to its small aluminium concentration, very coarse particles and high concentrations of toxic metals. Fine and coarse limestone tailings can be used as aggregates due to their thermal stability, smaller toxic metals concentration and acceptable particle size distributions. The results of the mechanical properties tests show that the reference composite (R) has the highest compressive (50 MPa) and flexural strength (6.0 MPa). The metakaolin-based geopolymer composite (GP1) exhibits sufficient compressive and flexural strength for its application in structural construction, while GP2 can be utilised in concrete kerb units and paving flags with a 14.1 and 3.3 MPa compressive and flexural strength, respectively. These findings provide a deeper insight into the use of secondary raw materials in geopolymer composites as binders and fine and coarse aggregates. Nevertheless, more experimental studies are required on the production of geopolymer composites to improve the characteristics of the final product.