Waste Slag from Heating Plants as a Partial Replacement for Cement in Mortar and Concrete Production. Part I—Physical–Chemical and Physical–Mechanical Characterization of Slag

Numerous factors influence the complexity of environmental and waste management problems, and the most significant goal is the reuse of materials that have completed their “life cycle” and the reduction in the use of new resources. In order to reduce impact of waste slag on the environment, in the present study, waste slag, generated in heating plants after lignite combustion, was characterized in detail and tested for application as a replacement for cement in mortar or concrete production. For physical–chemical characterization of slag, different experimental and instrumental techniques were used such as chemical composition and determination of the content of heavy metals, investigation of morphological and textural properties, thermal analysis, X-ray, and infrared spectroscopy. Physical–mechanical characterization of slag was also performed and included determination of activity index, water requirement, setting time and soundness. A leaching test was also performed. Presented results show that waste slag may be used in mortar and concrete production as a partial cement replacement, but after additional combustion at 650 ◦C and partial replacement of slag with silica fume in the minimal amount of 12%. The maximal obtained cement replacement was 20% (17.8% slag and 2.2% of silica fume).


Introduction
Coal is the energy source with the highest percentage of estimated fossil fuels (over 65%), with more evenly distributed deposits in the world than oil and gas. The USA, Russia, Australia, China, India, Indonesia, Germany, Ukraine, Poland, and Kazakhstan are some of the largest coal producers in the world [1]. Coal is a widespread fossil fuel used around the world. About 23% of the primary energy demand and 39% of electricity is obtained from coal. Due to the decline in oil and gas reserves, significant use of coal is expected to increase in the future. Today, about 40% of the world's total energy is produced from coal, and it is expected that this situation will remain in the future [2]. of limestone on the hydration of ternary slag cements. In their study, mixes with different amounts of cement-slag/quartz-limestone/quartz-anhydrite were prepared and tested. The amount of slag used was expressed as slag/quartz content, which was in the range of~28-47%. In the study of the Sum et al. [24], optimization of gypsum and slag content in blended cement containing slag was investigated. For these experiments, blast furnace slag was also used. In this study, different amounts of gypsum (0-40%), slag (60-100%) and clinker (0-40%) were mixed with Portland cement. Authors concluded that a high cement strength in the ternary slag-gypsum-clinker system may be obtained for 10% of gypsum, 20-40% of clinker and 60-75% of slag as well as for 10-25% of gypsum, less than 5% of clinker and 70-85% of slag. In the study of the Yoon et al. [25], blast furnace slag-blended Portland cement was used and tested for cobalt (Co) immobilization. In this study, authors replaced 0-70% of the cement with blast furnace slag. Their results show that regarding the retention of cobalt ions (Co 2+ ), retention increases with increasing the slag content in the mixes and the best results were obtained for cement replacement of 70%. However, the amount of slag that is obtained and already presented in the landfills is significantly higher than actual needs, and therefore there is a need for new research studies that could enable or provide guidelines for the use of slag in other industries. It is important to emphasize that the slag obtained from different sources (blast furnaces, hitting plants, etc.) possess different chemical compositions and thus potential application for all of them must be checked, investigated, and proven. From the aspect of practical application, it is very important that the use of slag has no negative effects on the environment, only positive effects, which are reflected in the protection of natural aggregates, reduction of CO 2 emissions, energy savings and, above all, reduction in waste material.
Due to its rough particles (sizes between sand and gravel), which have a glass structure and make it convenient to be used to replace natural aggregates (such as sand, gravel, crushed granite, etc.), potential practical application of slag in the cement industry was investigated in this study. The starting waste slag was characterized in detail and solutions were proposed to eliminate the defects of the slag from the aspect of application in the cement industry, and finally, the physical-mechanical characterizations of the obtained slag-Portland cement mixtures were made.

Sampling
The waste slag was sampled from landfill of a heating plant located in Valjevo, Serbia. The slag was obtained during combustion of lignite. The samples were taken from six different places at the landfill from three different depths, 30 samples in total. After sampling, the samples were dried at 105 • C for 2 h and homogenized. Homogenized slag samples were crushed, ground and sieved so the size of the grain and the particles were 100% below 63 µm before analysis.
Portland cement (CEM I 42.5R) was bought from the company CRH, Popovac, Serbia. Silica fume was bought from the company "Sika".

Characterization
Determination of the chemical composition and heavy metal content in slag was obtained with classical chemical analysis [26], using the atomic absorption spectrophotometer AAnalyst 300 (Perkin Elmer, Waltham, MA, USA) and the spectrophotometer Specol 1300 (Analytic Jena, Wegberg, Germany), using standard and internally developed methods.
Simultaneous non-isothermal thermogravimetric analysis (TGA) and differential thermal analysis (DTA) were used for testing the thermal properties of the slag and the Portland cement. For measuring, a Setaram Setsys Evolution 1750 instrument (Setaram, Lion, France) was used. Measuring was performed in the temperature range of 25 to 1000 • C with a heating rate of 10 min −1 in an air atmosphere, with a flow of 16 cm 3 min −1 . The mass of the sample was 15 mg.
The Sorptomatic 1990 Thermo Finnigan device (Thermo Fisher, Waltham, MA, USA) was used to measure the textural properties of the slag. The nitrogen adsorption-desorption isotherms were obtained by adsorption at −196 • C. Firstly, the slag was degassed at room temperature for 1 h and then, under the same residual pressure, for 16 h at 110 • C. For analyzing the isotherms, Software ADP Version 5.13 Thermo Electron (Thermo Fisher, Waltham, MA, USA) was used. The Brunauer-Emmett-Teller (BET) method was used for determination of the specific surface area (S BET ) of the slag [27]. The micropore volume (V mic -DR) was determined by application of the Dubinin-Radushkevich (DR) equation on the nitrogen adsorption isotherms [28], while the Dollimore and Heal (DH) method [29] was used for measuring the mesopore volume (V meso ).
The structural properties and the composition of the crystal phase of the slag and Portland cement were determined by Roentgen X-ray diffraction analysis (XRD). For XRD analysis, a PHILIPS PW-1710 X-ray diffractometer (Philips, Amsterdam, Netherlands) with a scintillation counter and a curved graphite monochromator was used. The measuring was performed at room temperature at 0.02 • /s, in the range of 4-65 • 2θ. A voltage of 40 kV and a current of 30 mA was used, while 1 • and 0.1 mm slits were used for the direction of the primary and diffracted beams.
The field emission scanning electron microscopy (FESEM) and the TESCAN Mira3 XMU instrument at 20 kV (Tescan, Brno-Kohoutovice, Czech Republic) were used for investigation of the morphological and microstructural properties of the slag. Prior to analysis, the slag samples were pre-coated with several nanometers of a thick gold layer, using the sputter coater Polaron SC503 Fision Instrument (Quorum Technologies Ltd., Lewes, UK). The chemical composition of the surfaces were determined by energy-dispersive X-ray spectroscopy (EDS) analysis in conjunction with an Oxford Instruments INCA X-sight system (Oxford Instruments, UK) attached to the Jeol JSM-6460 LV SEM (JEOL Ltd., Tokyo, Japan).
The Thermo Scientific "Nicolet iS50" instrument ((Thermo Fisher, Waltham, MA, USA) was used for infrared spectroscopic analyses (FTIR) of the slag and Portland cement. Investigations were performed in the transmission mode. In the background, the KBr spectrum was used. The wavelength range was 400-4000 cm −1 , while the resolution and number of scans were 2 cm −1 and 64, respectively. Two corrections-base line automatic correction and suppression of the atmosphere-were used after recording the spectra.

Testing of the Slag's Pozzolanic Activity
Reactions that occur between the active ingredients of pozzolan, such as lime and water, represent pozzolanic activity. The progress of the pozzolanic reaction is usually monitored by a decrease in the free lime in the system or an increase in the amounts of silica and aluminum, which are soluble in acid 105 by using the Florentin method. Additionally, pozzolanic activity can be monitored indirectly through the measurement of the activity index (AI). According to this method, a part of Portland cement is replaced with the appropriate pozzolan in a precisely defined amount, and then the properties of such mixtures are compared with the properties of a mixture that is 100% Portland cement, i.e., the activity index (AI) represents the ratio of the strength of the Portland cement mixture with and without pozzolan [30].
The pozzolanic activity includes the maximum amount of lime that pozzolan can encompass and the degree of reaction over the time or the reaction rate between a pozzolan and calcium ions (Ca 2+ ) or calcium-hydroxide (Ca(OH) 2 ) in the presence of water. The rate of the reaction is dependent on the specific surface area, chemical composition, and the active phase content in the pozzolan. Both factors depend on the nature, quality, and quantity of the active phases in the pozzolans [30].
In this study, the pozzolanic activity of slag was followed by measuring the activity index (AI). The measuring was in accordance with the International EN 450-1 standard. The AI was determined as the ratio of the compressive strength of the standard mortar prism, which was prepared with 75 wt.% of cement and 25 wt.% of slag, to the compressive strength of the standard mortar prism prepared with 100% cement, tested at the same age. The EN 450-1b standard defines that the AI of the mortar prism prepared with 75 wt.% of cement and 25 wt.% of the slag, after 28 and 90 days, must not be less than 75% and 85%, respectively, in comparison with the AI values obtained for the prism prepared with 100% cement. In addition to the activity index, the following tests were also carried out on binder material: water requirement (WR) according to EN 450-1 Annex B, setting time (ST, initial (IST) and final (FST)) according to EN 196-3 and soundness (S) according to EN 196-3.

Chemical Analysis and Determination of the Content of Heavy Metals
Chemical composition is very important, and this is one of the main conditions that determines whether a material can be used and whether it finds practical application in the cement industry. For waste materials, which are the subject of research in this paper, necessary criteria prescribed by the EN 450-1:2012 standard must be satisfied in order to apply the material in the cement industry and concrete or mortar production. For this reason, chemical analyses of the waste slag have been performed. The results are shown in Table 1, where the results of the chemical composition of the Portland cement (CEM I 42.5R) and the values prescribed by the standard are also presented. The comparison of the slag properties and the requirements prescribed by the standard shows that there are two criteria that waste slag, in starting form, does not satisfy. At first, the sum of the masses of SiO 2 , Fe 2 O 3 and Al 2 O 3 (47.71%) in the slag is less than 70%. Then, the loss of ignition (LOI) of the slag (37.21%) is more than the 11% that is prescribed by the standard.
Since the loss of ignition of slag mainly comes from the loss of unburned organic constituents or physically adsorbed water during annealing, this disadvantage may be surpassed by annealing the slag at the temperature of 650 • C. This temperature was determined by TGA/DTA analyses (Section 3.2), which confirmed that 650 • C is the necessary temperature for the loss of unburned organic constituents or physically adsorbed water. After the additional thermal treatment of the slag, the percentual chemical composition changed, so that all requirements prescribed by the standard were satisfied. Results are shown in Table 2. In addition to the determination of the chemical composition, the content of heavy metals in slag was also measured and results are shown in Table 3. According to the rulebook on the categories, testing and classification of waste-Official Journal of the Republic of Serbia No 56/2010-the content of the Cr, Ni, Pb and Sb exceed the maximum values allowed for the classification of waste as non-hazardous, so slag should be categorized as hazardous waste due to its characteristics. For the use of waste materials in the cement industry in Europe, including Serbia, the rules prescribed by the "Rulebook on restrictions and prohibitions on the production, placing on the market and use of chemicals" as well as the "Rulebook on the categories, testing and classification of waste" are in use. These rulebooks do not restrict the concentration of heavy metals in the waste material. The only limitation according to the rulebooks is the restriction of releasing any heavy metals from the end products (mortars or hardened mortars) during leaching. Thus, a leaching test was performed on the mortar mixture where 17.8% of CEM I 42.5R was replaced with slag (maximal allowed amount of slag that may be used-results are shown in Section 3.7). The testing procedure consists of the leaching of crushed hardened mortar (grain size smaller than 4 mm) with water, the liquid-solid ratio was 10 dm 3 /kg, and the determination of the heavy metal concentrations was obtained from eluates. The leaching is performed by mixing the mixture for 24 h. For the test, the mortar prisms used for testing AI were crushed. The leaching test was carried out in accordance with the EN 12457-2 standard and the results are shown in Table 4:

Thermogravimetric and Differential Thermal Analysis (TGA/DTA)
The thermal properties of the slag and Portland cement were investigated by the simultaneous non-isothermal TGA/DTA analysis and the results are shown in the Supplemental Materials ( Figure S1a (for slag), S1b (for Portland cement)) and Table 5. In the DTA diagram of the slag (S1a), one endothermic and two exothermic peaks were observed. The endothermic peak of DTA curves was detected at 95 • C, originating from the dehydration of physically adsorbed water. Due to this process, weight loss of about 4% in the first temperature interval (25-214 • C) was observed. In the second temperature interval (214-600 • C), two exothermic peaks were observed at 468 and 542 • C. For many burning products, such as slags or ashes, the loss of ignition (LOI) content can be equated to the amount of unburned organic constituents. The presence of organic carbon in a sample may be manifested by the appearance of an exotherm, or exotherms in the range of 400-600 • C. If the DTA curve contains two exotherms, the lower temperature exotherm represents amorphous (isotropic) carbon, and the higher temperature is due to the crystalline (anisotropic) variety. Thus, these exotherms at 468 and 542 • C correspond to the reaction of the production of CO 2 and CO due to combustion of the organic constituents, which contain isotropic and anisotropic carbon, respectively [31]. The mass loss during these reactions, which occurs in the temperature interval 214-650 • C, may be read from the thermogravimetric curve and was about 33% and corresponds to the content of organics in the sample. From the aspect of the application of slag in the cement industry, it is very important to notice that the content of organics and unburned carbon is restricted due to: (1) the tendency to adsorb the air entrainment admixtures added to the concrete to prevent crack formation and propagation and (2) increasing water requirements and decreasing strength development in concrete [32,33].
As it may also be seen from Figure S1b, for CEM I 42.5R two characteristic peaks are visible on the DTA curves. The endothermic peak of the DTA curves was detected at 113 • C, originating from endothermic reaction dehydration i.e., removal of physically adsorbed water. This process was followed with weight loss of about 0.65% in the temperature range 25-400 • C. In the second temperature interval (400-1000 • C), an endothermic peak with high intensity was observed at 730 • C, which was accompanied with weight loss of 6.87% on the TGA curve. This peak and weight loss can be explained by decomposition of the CEM I 42.5R aluminate phase, i.e., portlandite decomposition [34][35][36], dehydroxylation of the calcium aluminate and the alumina hydrates and the decomposition of the calcite (carbonates) [34][35][36][37]. However, due to the low amount of carbonate mineral calcite in the Portland cement, this process was not visible on the DTA curve.

Textural Properties
The nitrogen adsorption-desorption isotherm of the slag sample is shown in Figure S2, while the calculated textural parameters are shown in Table 6. The position of the hysteresis loop ending is characteristic for measurements with nitrogen at −196 • C and cannot be related to the textural properties of the material, but can be related to the use of nitrogen (N 2 ) as an adsorbate (so-called tensile strength effect) [38,39]. Although there is a certain content of mesopores in the analyzed material, which is indicated by the presence of a hysteresis loop, the small volume of the mesopores (0.020 cm 3 /g), the absence of a horizontal plate at high p/p 0 and the position of the hysteresis loop end exclude the possibility of categorizing slag as an exclusively mesoporous material. Therefore, the resulting isotherm can be classified as type II, which is typical for non-porous or macroporous materials (International Union of Pure and Applied Chemistry-IUPAC nomenclature) [38,39].
The standard "EN 197-1-Cement-Part 1 Composition, specifications and conformity criteria for common cements", which is applicable in the Republic of Serbia and the EU, prescribes that the ingredients for cement should have a specific surface area (S BET ), determined in agreement with ISO 9277, of a minimum of 15 m 2 /g. The S BET of the slag sample was determined and the calculated value was S BET = 25.0 m 2 /g, which means that the slag sample meets the required criteria.

X-ray Structural Analysis (XRD)
For the determination of the crystal phase composition of the slag, XRD analysis was used and the results are shown in Figure 1. Additionally, XRD analysis of the CEM I 42.5R was performed and results are also shown in Figure 1. It was determined that the slag sample possesses low crystallinity. However, crystal phases were detected, and the following minerals/phases were present: quartz, olivine, magnetite, pyrite, tridymite, feldspars. The most abundant mineral is quartz, while olivines and feldspars and then all other minerals/phases are less present. Of the feldspars, plagioclases are dominant, relative to Kfeldspars.
The results of the XRD analysis of the CEM I 42.5R sample show its high crystallinity and the following minerals were detected: brownmillerite, calcium silicate oxide, calcite, larnite, magnesium silicate, calcium hydroxide. The most abundant phase is calcium silicate oxide, then brownmillerite, larnite and calcite, while magnesium silicate and calcium hydroxide are present as associate minerals. Of the larnite, larnite-beta (Ca2(SiO4) is dominant. The CEM I 42.5R crystallinity was high.  From the micrographs taken with different magnifications, it can be seen that the slag is mainly constituted of aggregates of irregular size, which are mostly flat or have a granular structure. Micrographs taken with higher magnifications (5000× and 10000×) show that slag is mainly It was determined that the slag sample possesses low crystallinity. However, crystal phases were detected, and the following minerals/phases were present: quartz, olivine, magnetite, pyrite, tridymite, feldspars. The most abundant mineral is quartz, while olivines and feldspars and then all other minerals/phases are less present. Of the feldspars, plagioclases are dominant, relative to K-feldspars.

Scanning Electron Microscopy (SEM) Analysis
The results of the XRD analysis of the CEM I 42.5R sample show its high crystallinity and the following minerals were detected: brownmillerite, calcium silicate oxide, calcite, larnite, magnesium silicate, calcium hydroxide. The most abundant phase is calcium silicate oxide, then brownmillerite, larnite and calcite, while magnesium silicate and calcium hydroxide are present as associate minerals. Of the larnite, larnite-beta (Ca 2 (SiO 4 ) is dominant. The CEM I 42.5R crystallinity was high. It was determined that the slag sample possesses low crystallinity. However, crystal phases were detected, and the following minerals/phases were present: quartz, olivine, magnetite, pyrite, tridymite, feldspars. The most abundant mineral is quartz, while olivines and feldspars and then all other minerals/phases are less present. Of the feldspars, plagioclases are dominant, relative to Kfeldspars.

Scanning Electron Microscopy (SEM) Analysis
The results of the XRD analysis of the CEM I 42.5R sample show its high crystallinity and the following minerals were detected: brownmillerite, calcium silicate oxide, calcite, larnite, magnesium silicate, calcium hydroxide. The most abundant phase is calcium silicate oxide, then brownmillerite, larnite and calcite, while magnesium silicate and calcium hydroxide are present as associate minerals. Of the larnite, larnite-beta (Ca2(SiO4) is dominant. The CEM I 42.5R crystallinity was high.  From the micrographs taken with different magnifications, it can be seen that the slag is mainly constituted of aggregates of irregular size, which are mostly flat or have a granular structure. Micrographs taken with higher magnifications (5000× and 10000×) show that slag is mainly constituted of irregularly shaped aggregates of different sizes in the range of 2 to 10 µm. From the morphology of the investigated sample, it is also evident that the slag sample possesses cavities, which most likely originate from the loss of the bound aggregates on the surface during the From the micrographs taken with different magnifications, it can be seen that the slag is mainly constituted of aggregates of irregular size, which are mostly flat or have a granular structure.

Scanning Electron Microscopy (SEM) Analysis
Micrographs taken with higher magnifications (5000× and 10000×) show that slag is mainly constituted of irregularly shaped aggregates of different sizes in the range of 2 to 10 µm. From the morphology of the investigated sample, it is also evident that the slag sample possesses cavities, which most likely originate from the loss of the bound aggregates on the surface during the combustion process of lignite or the drying of the sample, such as water molecules from the surface, organic molecules or carbonates.
The chemical composition of the surfaces of the slag is presented in Figure 3 and Table 7.
Minerals 2020, 10, x FOR PEER REVIEW 9 of 15 The chemical composition of the surfaces of the slag is presented in Figure 3 and Table 7.  Five different positions and two measurement mods (spot and area) were used in order to get precise information about the chemical composition of the slag surface. Any chosen position for the EDS analysis of the surface composition of slag represents an aggregate of particles. This fact, pointed out previously in the analysis of the higher magnification micrographs, is confirmed by the analysis of the EDS spectrum results. The composition of the two spectra (one and five), of which spectrum one contains the results of the average composition of the surface at almost 0.1 mm 2 , and spectrum five contains the results of the composition at a point in the immediate vicinity of the edge of said surface, differ significantly. Indeed, the carbon content at position five is three times smaller, while the contents of Aluminum and Silicon are over two times higher in relation to the average composition of the chosen surface (spectrum one). Taking into account the processes, which simultaneously occurred during the burning of lignite (dehydration, partially or complete oxidation) and slag cooling and settling (possible carbonation and rehydration) as well as the different phases identified by the XRD analysis, complex highly heterogeneous surface composition is expected. Nevertheless, a reasonable degree of agreement is obtained by comparing the average values of the major analyte of the EDS analysis with the results obtained by classical chemical analysis (Section 3.1).

Infrared Spectroscopy with Fourier Transform (FTIR)
The infrared spectra of the starting slag sample and CEM I 42.5R are shown in Figure 4. In the infrared spectrum of the slag, spectral bands at 3697, 3619, 3340, 1618, 1585, 1509, 1454, 1264, 1025,  1005, 911, 796, 777, 692, 522 and 460 cm −1 are presented. The spectral bands at 3697 and 3619 cm −1 are assigned to the stretching vibrations of structural hydroxyl groups in Al-OH and Si-OH [40,41]. The spectral bands at 3340 and 1618 cm −1 may be assigned to the presence of the physical bounded and surface adsorbed water and stretching and banding vibrations of OH − groups, respectively [40,41].  Five different positions and two measurement mods (spot and area) were used in order to get precise information about the chemical composition of the slag surface. Any chosen position for the EDS analysis of the surface composition of slag represents an aggregate of particles. This fact, pointed out previously in the analysis of the higher magnification micrographs, is confirmed by the analysis of the EDS spectrum results. The composition of the two spectra (one and five), of which spectrum one contains the results of the average composition of the surface at almost 0.1 mm 2 , and spectrum five contains the results of the composition at a point in the immediate vicinity of the edge of said surface, differ significantly. Indeed, the carbon content at position five is three times smaller, while the contents of Aluminum and Silicon are over two times higher in relation to the average composition of the chosen surface (spectrum one). Taking into account the processes, which simultaneously occurred during the burning of lignite (dehydration, partially or complete oxidation) and slag cooling and settling (possible carbonation and rehydration) as well as the different phases identified by the XRD analysis, complex highly heterogeneous surface composition is expected. Nevertheless, a reasonable degree of agreement is obtained by comparing the average values of the major analyte of the EDS analysis with the results obtained by classical chemical analysis (Section 3.1).

Infrared Spectroscopy with Fourier Transform (FTIR)
The infrared spectra of the starting slag sample and CEM I 42.5R are shown in Figure 4. In the infrared spectrum of the slag, spectral bands at 3697, 3619, 3340, 1618, 1585, 1509, 1454, 1264, 1025,  1005, 911, 796, 777, 692, 522 and 460 cm −1 are presented. The spectral bands at 3697 and 3619 cm −1 are assigned to the stretching vibrations of structural hydroxyl groups in Al-OH and Si-OH [40,41]. The spectral bands at 3340 and 1618 cm −1 may be assigned to the presence of the physical bounded and surface adsorbed water and stretching and banding vibrations of OH − groups, respectively [40,41]. The position of the spectral band is related to the banding vibrations of OH − and groups of water molecules may be an indication that water molecules interact through oxygen atoms with surface cations and act as an H-bond acceptor. Namely, according to the literature, a spectral band is usually obtained at 1630 cm −1 and shifting in its position may give information about the acidity/alkalinity of the surface. When a water molecule acts as an H-bond acceptor, the δ (H 2 O) is shifted to lower wave numbers; conversely, when it acts as H-bond donor, the δ (H 2 O) is shifted to higher wave numbers [42]. Spectral bands at 1025, 1005, 796, 777, 692, 520 and 460 cm −1 can be assigned to the Si-O vibrations. According to their positions and shape, they are clear confirmation of the presence of quartz, plagioclases and tridymite in the waste slag, which were also confirmed by XRD analysis [43][44][45]. Additional confirmation of the presence of the plagioclases (albite (NaAlSi 3 O 8 ) and anorthite (CaAl 2 Si 2 O 8 )) is the spectral band at 911 cm −1 , which originated from Al-Al-OH banding vibrations [40]. Bands originating from other minerals presented in the crystal phase of the slag sample were not visible due to them overlapping with much more pronounced bands in the spectrum. Spectral bands at 1585, 1509, 1454 and 1264 cm −1 originated from slag components that are in the amorphous phase and were not visible by XRD analysis. These spectral bands may be assigned to the organic components of the slag. Lignite, whose incomplete combustion produces slag, which is the subject of this research, is humus coal from deposits in Serbia, Kolubara basin and belongs to type III kerogen, which is formed from terrestrial plants, rich in cellulose and lignin, with low lipid content [46,47]. Such coals are mainly aromatic in nature, so the vibrations visible in the spectrum of the slag sample are mainly related to the vibrations in the aromatic rings. Thus, spectral bands at 1509 and 1454 cm −1 are assigned to the ν 19a and ν 19b in-plane, aromatic ring vibrations. Vibrations ν 8a and ν 8b (in-plane, aromatic ring stretch) are assigned in the FTIR spectrum with spectral bands at 1603 and 1585 cm −1 . However, due to the overlapping of the band with lower intensity at 1603 cm −1 with those that originated from the vibration of the −OH group (at 1618 cm −1 ), only the line at 1585 cm −1 is visible in the spectrum. Stronger intensity of the band at 1585 cm −1 in comparison to the band at 1603 cm −1 indicates that the substituent on the ring is conjugated to the ring [48]. Finally, the spectral band at 1264 cm −1 may be assigned to enolic stretching C-O vibrations [49]. cations and act as an H-bond acceptor. Namely, according to the literature, a spectral band is usually obtained at 1630 cm −1 and shifting in its position may give information about the acidity/alkalinity of the surface. When a water molecule acts as an H-bond acceptor, the δ (H2O) is shifted to lower wave numbers; conversely, when it acts as H-bond donor, the δ (H2O) is shifted to higher wave numbers [42]. Spectral bands at 1025, 1005, 796, 777, 692, 520 and 460 cm −1 can be assigned to the Si-O vibrations. According to their positions and shape, they are clear confirmation of the presence of quartz, plagioclases and tridymite in the waste slag, which were also confirmed by XRD analysis [43][44][45]. Additional confirmation of the presence of the plagioclases (albite (NaAlSi3O8) and anorthite (CaAl2Si2O8)) is the spectral band at 911 cm -1 , which originated from Al-Al-OH banding vibrations [40]. Bands originating from other minerals presented in the crystal phase of the slag sample were not visible due to them overlapping with much more pronounced bands in the spectrum. Spectral bands at 1585, 1509, 1454 and 1264 cm −1 originated from slag components that are in the amorphous phase and were not visible by XRD analysis. These spectral bands may be assigned to the organic components of the slag. Lignite, whose incomplete combustion produces slag, which is the subject of this research, is humus coal from deposits in Serbia, Kolubara basin and belongs to type III kerogen, which is formed from terrestrial plants, rich in cellulose and lignin, with low lipid content [46,47]. Such coals are mainly aromatic in nature, so the vibrations visible in the spectrum of the slag sample are mainly related to the vibrations in the aromatic rings. Thus, spectral bands at 1509 and 1454 cm −1 are assigned to the ν19a and ν19b in-plane, aromatic ring vibrations. Vibrations ν8a and ν8b (in-plane, aromatic ring stretch) are assigned in the FTIR spectrum with spectral bands at 1603 and 1585 cm −1 . However, due to the overlapping of the band with lower intensity at 1603 cm −1 with those that originated from the vibration of the −OH group (at 1618 cm −1 ), only the line at 1585 cm −1 is visible in the spectrum. Stronger intensity of the band at 1585 cm −1 in comparison to the band at 1603 cm −1 indicates that the substituent on the ring is conjugated to the ring [48]. Finally, the spectral band at 1264 cm −1 may be assigned to enolic stretching C-O vibrations [49]. In the spectrum of the CEM I 42.5R, the band of 1433cm −1 is attributed to stretching vibrations of the CO3 -2 , and originates from the presence of calcium-carbonate (CaCO3) in cement [8,41] and also, this spectral band is characteristic for brownmillerite [50], which was also detected by the XRD analysis. The spectral band at 914 cm −1 originated from Al-Al-OH bending vibrations, and the band at 876 cm −1 may be assigned to Fe-Al-OH deformations, which also confirmed the presence of brown millerite (  In the spectrum of the CEM I 42.5R, the band of 1433cm −1 is attributed to stretching vibrations of the CO 3 −2 , and originates from the presence of calcium-carbonate (CaCO 3 ) in cement [8,41] and also, this spectral band is characteristic for brownmillerite [50], which was also detected by the XRD analysis. The spectral band at 914 cm −1 originated from Al-Al-OH bending vibrations, and the band at 876 cm −1 may be assigned to Fe-Al-OH deformations, which also confirmed the presence of brown millerite (Ca 2 (AlFe) 2 O 5 ) while the band at 516 cm −1 originated from Al-O-Si bending vibrations. The band at 657 cm −1 is due to coupled Al-O and Si-O out of plane vibrations [8,39], while the band at 602 cm −1 is due to vibrations of internal modes of FeO 6 in brown millerite. The bands at 420 and 465 cm −1 originated from O-Mg-O and O-Ca-O bending modes in minerals, which contain calcium (brown millerite, larnite, calcium silicate oxide and calcite) and magnesium silicate, respectively [51]. Spectral bands in the range 1200-1000 cm −1 are due to stretching (in plane and out of plane) vibrations of Si-O groups [8,40,41].

Testing of the Slag's Pozzolanic Activity
The pozzolanic activity is one of the most important and basic parameters in order to determine if the investigated material can be used as a replacement for cement. For this reason, the AI of the slag was determined, and these investigations were performed in accordance with the EN 450-1 standard and the results are shown in Table 8. Based on the results presented in Table 8, according to the parameters IST and S, the waste slag satisfied criteria, while according to parameters AI and WR, the waste slag does not satisfy parameters prescribed by the EN 450-1 standard, and thus, it may be concluded that the investigated slag does not possess pozzolanic activity in the presented form, and therefore it cannot be used as a replacement for cement without additional modifications and actions. In this study, this problem was solved with the addition of silica fume to the slag. Namely, it is well known that silicate fume possesses good pozzolanic properties and because of this, it was tested as an addition to the slag in order to increase slag's pozzolanic activity and improve its properties. At first, determination of the chemical composition of the silica fume was performed and the results shown in Table 9 confirm its very good quality (≈94% of SiO 2 ). After the determination of the chemical composition, the silica fume was used as an addition to slag during the production of cement mixtures. For this aim, five samples of the cement mixtures were prepared with different amounts of CEM I 42.5 R (the content was in range 100-75%), slag (the content was in range 8.9-22.2%) and silica fume (the content was in range 1.1-2.8%) as described in Table 10.
After the mixtures were prepared, the AI, as the most important parameter of all prepared mixtures, were determined, and the results are presented in Table 11.  From the presented results, it may be seen that slag with the addition of the silica fume possesses pozzolanic activity, which satisfies requirements prescribed by the EN 450-1 standard. Based on the measuring of the AI, the maximal possible cement replacement is 20% with 17.8% of slag and 2.2% of silica fume.

Conclusions
Results presented in this study show that waste slag-obtained from a heating plant (due to combustion of lignite) and taken from the landfill in Valjevo, Serbia-should be categorized as a hazardous waste due to its chemical composition and heavy metals content. From this point, it represents a serious environmental problem that must be solved.
Regarding slag utilization in the cement industry, results have shown that tested waste slag in the original form does not completely satisfy necessary criteria prescribed by the rulebooks from the aspect of the content of SiO 2 , Al 2 O 3 and Fe 2 O 3 as well as from the aspect of the content of the unburned carbon and loss of ignition. The proposed solution and the easiest way for solving and overcoming these problems is thermal treatment of the slag prior to use. The results of the thermal analysis show that, after thermal treatment up to 650 • C, all criteria related to heavy metals content and chemical composition can be met. Additional physical-chemical characterizations (XRD, FTIR, N 2 physisorption, SEM, determination of the density) of the slag showed good correlation and that, regarding all additional tested physical-chemical properties, the slag completely satisfied the criteria prescribed by the rulebooks for utilization in cement industry.
Based on results of testing the physical-mechanical properties of the slag, it may be concluded that the investigated slag does not possess pozzolanic activity in its raw form, but with additional modifications and actions, it can be used as a replacement for cement. This may be overcome by mixing the slag with silica fume. The results show that mixing the slag with silica fume in minimal amounts (the slag/silica fume ration ≈0.88%/≈0.12%) improve the properties of the slag so that it possesses satisfactory pozzolanic activity. Maximal possible cement replacement was 20% with 17.8% of slag and 2.2% of silica fume. The results of leaching tests also prove that it is possible to use slag as mineral binders in cement composites, as there was no significant release of heavy metals.