Reactivity Assessment of Modified Ferro Silicate Slag by R3 Method

Traditional methods to track the reactivity of supplementary cementitious materials (SCMs) and their contribution to the hydration mechanism mostly use Portland Cement (PC) as an activator. Alternatively, a novel method to assess the reactivity of SCMs called R3 was recently presented. This novel method uses lab grade chemicals such as portlandite (CH), K2SO4, KOH, and CaCO3 to activate the SCM by resembling the pH of the alkaline pore solution created by PC. By using this method, the reactivity of the SCM can be easily quantified from measured heat release, bound water content, and CH consumption. The primary objective of the current study is to apply the novel methodology to analyze the reactivity of Modified Ferro Silicate (MFS) Cu slag benchmarked against siliceous fly ash (FA), ground granulated blast-furnace slag (GGBFS), and inert quartz filler. GGBFS showed the highest cumulative heat release and bound water content due to its latent hydraulic behavior. Determination with XRD analysis of the major phase of the R3 model MFS slag paste showed the participation of Fe in the hydration mechanism by forming Fe-AFm. R3 paste with GGBFS showed the presence of hydrotalcite/Al-AFm, whereas FA showed the presence of ettringite (AFt) as their crystalline reaction products. The experiments also indicate that the MFS slag acts as a reactive pozzolanic material with an acceptable performance in heat release, bound water content, and CH consumption, and can be used as SCM to make concrete. With the possibility of using MFS slag as SCM to replace part of PC, sustainability and circular economy can be fairly well achieved.


Introduction
Copper (Cu) is one of the most vital metals used in a wide range of thermal applications and electrical wiring. Cu metal is usually synthesized by pyro-metallurgical refining of Cu ore such as chalcopyrite (CuFeS 2 ) or bornite (Cu 5 FeS 4 ) [1,2]. Alternatively, due to the improved technological facilities and state of the art on smelter and refining processes, industries also focus on the use of metal scraps as recyclable materials to synthesize Cu metal. During the smelting and refining of Cu ore or Cu rich metal scraps into Cu metal, a slag is obtained as a by-product [2]. This by-product is produced in high volume, as the production of 1 ton of Cu metal usually generates approximately 2.2-3.0 tons of Cu slag. Usually the slag is dumped, stockpiled, or to a limited extent used in low grade applications. Despite its limited usage, Cu slag (the term Cu slag in the literature refers to its origin, rather than its mineralogy, which is mainly a siliceous amorphous phase) is synthesized with PC-GCS showed increased heat release after seven days and compressive strength at 28 and 91 days, and (2) TGA and SEM analyses showed more gel phases and less pores in PC-GCS paste. In the work of Hallet et al. [28], the impact of slag fineness on the reactivity of blended cements with non-ferrous slag by R 3 was examined. It was stated that R 3 isothermal calorimetry showed that the slag was similar in reactivity to siliceous fly ash and increasing the fineness increases the reactivity of the slag.
Moreover, Pontes et al. [24] studied the pozzolanic reactivity of artificial pozzolanic materials by the Frattini method, Chapelle method, and SAI test. The reactivity was assessed for eight different materials of siliceous and aluminosilicates composition, and the authors stated that further investigation is needed to study the reaction mechanism of SCM due to the influence of clinker phases during the reactivity assessment. In the work of Edwin et al. [25], isothermal calorimetry, Chapelle test, Frattini test, and SAI test were used to assess the pozzolanic reactivity of Cu slag. The authors concluded that the pozzolanic activity of the Cu slag mainly depends on temperature, the curing age, and the specific surface area. In another recent work of Edwin et al. [29], porosity of reactive powder concrete (made with PC, copper slag, and additives) based on automated analysis of back-scattered-electron images was analyzed. However, the proposed analysis did not show similar results to the MIP data due to the different sizes and characteristics of the pores examined with MIP and back scattered electron images. The studies [21][22][23][24][25] used PC as an activator to study the reactivity of the SCM. Due to the influence of clinker phases on the reactivity assessment, only speculative conclusions can be made on the reactivity of the SCM. To overcome this drawback, a recent report [30] published by RILEM technical committee TC 267-TRM (which partly turned into the ASTM standard C1897-20 (2020)) proposed a novel method to evaluate the reactivity of SCMs. The method is termed as "R 3 ", which stands for rapid, reproducible, and relevant. The method uses mixes of SCM, CH, and H 2 O, supplemented with alkali sulfate (SO 4 2− ) and carbonate (CO 3 2− ) resembling the pH of the alkaline pore solution created by PC. By using this method, the reactivity of the SCM can be easily quantified from measured heat release, bound water, and CH consumption. In this article, reactivity of Modified Ferro Silicate (MFS) Cu slag was assessed according to the novel R 3 model system reported in RILEM TC 267-TRM. Reactivity of the R 3 model MFS slag paste was compared with quartz filler and the widely used pozzolanic material siliceous FA and latent hydraulic SCM GGBFS. The R 3 model system was selected to eliminate the influence of PC while assessing the reactivity of the SCMs. The objective was to compare the reactive behavior of MFS slag with FA and GGBFS on the one hand and with inert filler on the other hand to understand its reaction mechanism, which should substantiate the use of MFS slag as a sustainable binder material for concrete.

Materials and Characterisation
Patented MFS slag was used as the starting material. MFS slag is a fumed secondary Cu slag synthesized during the recycling of Cu scraps into Cu metal. Initially, the MFS slag was dried in a laboratory oven at 105 • C for 24 h. Next, the dried MFS slag was milled in a attritor mill (Wiener 1S) to achieve a similar particle size distribution as PC. Quartz flour with a particle size distribution equivalent to MFS slag was also used as a starting material to synthesize the reference R 3 model sample. The particle size distribution was determined using a laser diffraction analyzer (Mastersizer 2000, Malvern Instrument Ltd.). Portland cement supplied by Holcim (CEM 52.5 N) and fly ash (class C) were used in this study. GGBFS, a by-product from the manufacture of iron, was supplied by Ecocem Benelux. Lab grade KOH, K 2 SO 4 , and CaCO 3 , along with Ca(OH) 2 (CH), which possesses less than 5 wt.% of CaCO 3 , were used as the starting material to prepare the R 3 model paste. Initially, the chemical composition of the materials was determined by wavelength-dispersive X-ray fluorescence spectrometry (XRF). Mineralogical composition of all starting materials was obtained using X-ray diffraction (XRD, Thermo Scientific, ARL X'TRA). Quantitative XRD analysis was carried out by using 10 wt.% of analytical grade crystalline ZnO (95% purity) as external standard. The material was blended with ZnO manually with the help of a mortar and pestle. An X-ray diffraction scan was recorded on the homogenized material using Cu Kα radiation with acceleration voltage 30 kV, a current of 30 mA, a step size of 0.020 • , and a counting time of 2.5 s per step. The diffraction pattern obtained using the XRD device was analyzed and quantified with Topas Academic V.5 software, using the Rietveld technique.

Synthesis of R 3 Model Paste
The mix design used to synthesize R 3 model paste as recommended by TC 267-TRM [1] is given in Table 1. Initially, the SCM was weighed and mixed with the CH and CaCO 3 gently using the spatula until a homogeneous color was achieved. KOH and K 2 SO4 were weighed and dissolved in the deionized H 2 O. The prepared mix and solution were stored in an oven at 40 ± 2 • C for 24 h. After 24 h, the solution (pH around 13.6) was added to the dry powder and mixed at 1600 rpm for 2 min to achieve a homogeneous paste. The prepared paste was cast into an appropriate container, tapped gently to remove air bubbles. The container was covered with Parafilm M (plastic paraffin) to achieve a sealed layer and stored at 40 ± 2 • C for a period of 7 days to accelerate the reaction.

Reactivity Assessment of the Synthesized R 3 Model Paste
A primary objective to assess the reactivity by the R 3 test was to avoid the influence of cement hydration on the SCMs and to assess the degree of reaction of SCMs individually. Here, the reactivity of the synthesized R 3 paste was assessed by measuring: (1) Consumption of CH by ThermoGravimetric Analysis (TGA), (2) bound water by the oven method, (3) heat production rate by isothermal calorimetry, (4) pore solution analysis, and (5) presence of binder phase by XRD analysis.

CH Consumption Analysis for R 3 Model Paste
Approximately 3 g of R 3 hydrated model paste, cured for 7 days, was used to prepare the sample for TGA. The initial sample was crushed (1-3 mm) by mortar and pestle. The crushed sample was placed in 50 mL of isopropanol for 15 min for hydration stoppage through solvent exchange. The suspension (sample + isopropanol) was stirred constantly for a period of 2 min and vacuum filtration was carried out by a 2.7 µm pore size filter to separate the crushed sample from the isopropanol. The sample was placed on a watch glass and dried at 40 • C for 8 min in an oven. Next, the sample was stored in a low vacuum desiccator till the measurement was carried out. Approximately 50 mg of the stored sample was used for the TGA measurement. TGA was carried out in an Al 2 O 3 crucible, with a heating regime of 30 to 950 • C at 10 • C/min in a N 2 atmosphere. Proteus thermal analysis software was used to measure the CH content vs. sample mass at 950 • C (m 950 ), using the tangential approach, with the mass loss attributed to CH (∆m H2O,CH ) situated between 400 and 480 • C [31]. Three repetitions for a single measurement were carried out to calculate the mean value and standard deviation, and to check the reproducibility. Finally, residual CH (m CH , ignited ) and CH consumption (m CH,consumption ) (g/100 g SCM) of the SCM paste synthesized as per R 3  A sample prepared and cured for 7 days as per the R 3 procedure was used to measure the bound water content through heating using a laboratory oven. A sample, crushed (1-3 mm) by hand, was placed in an oven at 105 • C for 24 h and was used as a starting material. Prior to the experiment, the precise weight of the Al 2 O 3 crucible, thermally treated at 350 • C for 1 h, termed as w c , and the precise weight of the crucible with starting materials, termed as w o , were determined.
The sample, along with the crucible, was calcined at 350 • C for 2 h in the box furnace. After 2 h, the oven was turned off and the sample was cooled till 150 • C. Final weight of the sample including the crucible was measured and termed as w t . Finally the H 2 O bound, dried was calculated (Equation (3)) and rescaled to g/100 g anhydrous (H 2 O bound,anhydrous ) (Equation (4)).

Isothermal Calorimetry Protocol for R 3 Model
The isothermal calorimetry technique (TAM Air, TA instruments) was used to investigate the reaction kinetics and to measure the heat production rate for the R 3 model paste until the age of 7 days. The calorimeter was set to 40 • C and calibration was carried out after temperature was stabilized. All necessary equipment along with the materials used for the preparation of the samples was also placed in the oven at 40 • C. Exactly 9.405 g of deionized water was poured in the reference ampoule. After calibration, a sealed reference container was inserted in the reference channel of the calorimetric device. The baseline of the calorimetry was measured for 3 h. Fifteen g (termed as m p ) of R 3 model paste was introduced into the ampoule, immediately after mixing, and inserted in the measuring channel of the calorimeter. The heat profile was recorded at 40 • C for 7 days. The cumulative heat of the sample was calculated by neglecting the initial 1.2 h due to the mixing at 20 • C, which causes initial negative flow. The cumulative heat release is obtained in J/g SCM as given by Equation (5).
cumulative heat release = Heat released m p × 0.0997 (5) 0.0997 in the denominator corresponds to the ratio of SCM present in the R 3 model paste.

XRD and Pore Solution Analysis
X-ray diffraction scans with a step size 0.020 • and counting time of 2.5 s per step in a 2θ range from 5 • to 15 • were recorded on the 7 days old hydration stopped R 3 samples to investigate the presence of crystalline binder phases. Pore fluid from the hydration stopped MFS slag-R 3 samples was also extracted with a pressure up to 200 MPa. After extraction, the pore solution was filtered and diluted with acid. The diluted solutions were analyzed using ICP-OES.

Synthesis of Mortar
Compressive strength was assessed on PC, 15 M, 30 M, 50 M, and 70 M mortars, prepared as per EN 196-1 and cured at 20 ± 1 • C and 95% relative humidity for 24 h ( Table 2). For the mortar with 70 wt.-% MFS slag in the binder, additional powder activation by Na 2 SO 4 was foreseen, resulting in a hybrid mortar (HM). After 24 h, the mortars were demolded and stored in a curing chamber at 20 ± 1 • C and 95% relative humidity until the compressive strength tests were performed.

Characterization of the Starting Materials
Chemically, the starting materials are mainly composed of four oxides of Si, Fe, Al, and Ca (Table 3) Table 4. Since MFS and GGBFS are industrial by-products obtained by a quenching process, they are mainly composed of an amorphous or glassy phase. FA is an amorphous rich by-product synthesized during the coal combustion that is mainly composed of fine particulates which are collected by the electrostatic precipitator. In this study, siliceous rich FA of Class F was used to see the relative comparison with MFS slag. Particle size distribution of the starting materials is shown in Table 5. The calculated CH content is shown in Table 6. From this, it can be seen that all SCMs and inert quartz filler consumed CH at a different degree compared to its initial presence.
Appl. Sci. 2021, 11, 366 7 of 14 R 3 paste synthesized from MFS slag has consumed less CH than the well-established pozzolanic source FA. MFS slag has an amorphous phase that is rich in SiO 2 (around 30 wt%). However, although fly ash has lower amorphous content than MFS slag, it contains more SiO 2 (around 50 wt%, assuming all SiO 2 is present in glass phase). Initially, Ca(OH) 2 is dissolved in presence of H 2 O to release Ca 2+ ions which consume the Si 2+ ions (or silicates with different degrees of H dissociation) from MFS slag to produce C-S-H as a result of the pozzolanic mechanism. Furthermore, due to the presence of a minor content of Al 2 O 3 in the MFS slag, C-S-H can take-up aluminate ions to form C-A-S-H. GGBFS shows less CH consumption compared to the MFS slag and FA due to the latent hydraulic behavior. Due to the presence of a reactive amorphous phase rich in CaO and SiO 2 , GGBFS has the potential to self react with H 2 O to produce C-S-H or C-A-S-H binder phase [32]. Bound water calculated by the oven method also shows that GGBFS paste possessed a high amount of hydrated compounds, which decomposed between 105 and 350 • C, compared to MFS slag and FA (Table 6). However, by considering this temperature range, certain hydrates such as ettringite cannot be taken into account. Ettringite dehydrates below 100 • C as mentioned elsewhere [33,34]. Note that the mechanism governing GGBFS reactivity explained through CH consumption and bound water cannot be considered as decisive, and the results should be interpreted with caution by accounting the results shown in Sections 3.2.2 and 3.2.3. A slight decrease in the CH content compared to the initial presence was observed with the quartz filler material. This indicates that the amorphous phase (around 3 wt.%) of quartz can consume a small amount of CH to produce C-S-H, and this mechanism might be due to the high temperature curing (40 • C). For instance, different authors have described the mechanism of decrease in the CH content with increased ages of the inert filler (quartz) blended system [35][36][37]. Additionally, the study published by Li et al. [30] to assess the reactivity using the R 3 model system showed slight CH consumption for quartz inert filler. Figure 1 shows the calorimetric data of the R 3 model paste containing MFS slag, GG-BFS, FA, and quartz filler expressed per g of SCM. GGBFS shows a higher heat production rate than MFS, FA, and quartz. The following observations can be made on their reactivity:

Isothermal Calorimetry
(1) GGBFS paste is prone to dissolve by itself in H 2 O and can synthesize more binder such as C-S-H or C-A-S-H, AFm, and AFt, which results in a strong exothermic heat release. Bound water analysis of GGBFS explained in Section 3.2.1 also showed that GGBFS possessed a high amount of hydrates. (2) MFS slag paste shows increased heat release in comparison to FA. This behavior is attributed to the presence of a higher content of reactive amorphous phase (as shown in Table 4), possibly suggesting faster dissolution of the reactive phase and formation of binder. (3) As the MFS slag has released substantially more heat compared to inert filler, this indicates that extra heat release is probably due to the pozzolanic reaction (Figure 1b). (4) The inert quartz filler proved to be a comprehensive inert material to investigate the filler effect due to its limited heat production attributed to its CH consumption observed in Figure 1a). (3) As the MFS slag has released substantially more heat compared to inert filler, this indicates that extra heat release is probably due to the pozzolanic reaction ( Figure 1b). (4) The inert quartz filler proved to be a comprehensive inert material to investigate the filler effect due to its limited heat production attributed to its CH consumption observed in Figure 1a). From Figure 2, the heat production rate of the R 3 model paste can be categorized into two regions: (1) Initial heat release up to 2 h, likely due to the dissolution of reactive ions such as SO4 2− , Al 3+ , and CO3 2− , (2) later heat release until 10 h, likely attributed to the gradual dissolution of the amorphous phase and the formation of binder phases such as C-S-H or C-A-S-H, AFt, and AFm. This heat release intensity tends to differ for FA due to the increased presence of crystalline phases (25 wt.%) which act as a barrier for the dissolution of the amorphous phase. For instance, Chancey et al. [38] characterized the amorphous and crystalline phases of the FA class F and showed that the amorphous phase was partially surrounded by crystalline phases such as quartz, mullite, and maghemite, thus limiting the dissolution and reaction towards formation of binder. Heat release till 10 h is only seen with MFS slag, GGBFS, FA, and not with inert filler, suggesting that this relates to the dissolution of amorphous (glassy) phases to form hydrates. Moreover, R 3 model paste incorporating MFS slag showed an increased heat release in the period between 4 to 6 h. This heat release can possibly be attributed to the dissolution of Zn 2+ ions present in the MFS slag. For instance, ZnO present in the glassy structure of the MFS slag could potentially tend to dissolve and influence the hydration mechanism. (3) As the MFS slag has released substantially more heat compared to inert filler, this indicates that extra heat release is probably due to the pozzolanic reaction ( Figure 1b). (4) The inert quartz filler proved to be a comprehensive inert material to investigate the filler effect due to its limited heat production attributed to its CH consumption observed in Figure 1a). From Figure 2, the heat production rate of the R 3 model paste can be categorized into two regions: (1) Initial heat release up to 2 h, likely due to the dissolution of reactive ions such as SO4 2− , Al 3+ , and CO3 2− , (2) later heat release until 10 h, likely attributed to the gradual dissolution of the amorphous phase and the formation of binder phases such as C-S-H or C-A-S-H, AFt, and AFm. This heat release intensity tends to differ for FA due to the increased presence of crystalline phases (25 wt.%) which act as a barrier for the dissolution of the amorphous phase. For instance, Chancey et al. [38] characterized the amorphous and crystalline phases of the FA class F and showed that the amorphous phase was partially surrounded by crystalline phases such as quartz, mullite, and maghemite, thus limiting the dissolution and reaction towards formation of binder. Heat release till 10 h is only seen with MFS slag, GGBFS, FA, and not with inert filler, suggesting that this relates to the dissolution of amorphous (glassy) phases to form hydrates. Moreover, R 3 model paste incorporating MFS slag showed an increased heat release in the period between 4 to 6 h. This heat release can possibly be attributed to the dissolution of Zn 2+ ions present in the MFS slag. For instance, ZnO present in the glassy structure of the MFS slag could potentially tend to dissolve and influence the hydration mechanism.

Influence of Zn 2+ in the Hydration Mechanism of MFS Slag
Several authors have reported the influence of Zn 2+ in PC and alkali activated materials [39][40][41][42][43][44]. For instance, Gawlicki et al. [44] studied the hydration of PC in the presence of ZnO and found that amorphous Zn(OH)2 could be formed in the reaction with H2O, which could potentially transform into Ca(Zn(OH)3)2.2H2O, provoking the heat release. In From Figure 2, the heat production rate of the R 3 model paste can be categorized into two regions: (1) Initial heat release up to 2 h, likely due to the dissolution of reactive ions such as SO 4 2− , Al 3+ , and CO 3 2− , (2) later heat release until 10 h, likely attributed to the gradual dissolution of the amorphous phase and the formation of binder phases such as C-S-H or C-A-S-H, AFt, and AFm. This heat release intensity tends to differ for FA due to the increased presence of crystalline phases (25 wt.%) which act as a barrier for the dissolution of the amorphous phase. For instance, Chancey et al. [38] characterized the amorphous and crystalline phases of the FA class F and showed that the amorphous phase was partially surrounded by crystalline phases such as quartz, mullite, and maghemite, thus limiting the dissolution and reaction towards formation of binder. Heat release till 10 h is only seen with MFS slag, GGBFS, FA, and not with inert filler, suggesting that this relates to the dissolution of amorphous (glassy) phases to form hydrates. Moreover, R 3 model paste incorporating MFS slag showed an increased heat release in the period between 4 to 6 h. This heat release can possibly be attributed to the dissolution of Zn 2+ ions present in the MFS slag. For instance, ZnO present in the glassy structure of the MFS slag could potentially tend to dissolve and influence the hydration mechanism.

Influence of Zn 2+ in the Hydration Mechanism of MFS Slag
Several authors have reported the influence of Zn 2+ in PC and alkali activated materials [39][40][41][42][43][44]. For instance, Gawlicki et al. [44] studied the hydration of PC in the presence of ZnO and found that amorphous Zn(OH) 2 [43] showed the mechanism of ZnO in GGBFS based alkali activated materials through X-ray pair distribution function. The proposed mechanism by the authors states that ZnO dissolves in the highly alkaline solution, reacting with the dissolved Ca 2+ ions from the slag to form Ca(Zn(OH) 3 ) 2 .2H 2 O. This mechanism continues as long as there is availability of free Zn in the pore solution. This proposed mechanism can be applicable to the MFS slag due to the presence of Zn 2+ ions in the glassy phase. Upon addition of H 2 O in the R 3 mix containing MFS slag as SCM, Zn 2+ (from MFS slag) and Ca 2+ (from Ca(OH) 2 , CaCO 3, MFS) slowly tend to dissolve into the solution and eventually form Ca(Zn(OH) 3 ) 2 .2H 2 O, which results in an exothermic reaction and heat release. This mechanism is expected to continue until all free ZnO present in the glass is consumed.

Pore Solution vs. Calorimetric Analysis
The diluted pore solution from the R 3 model MFS slag paste has been extracted at a curing age of 7 days and analyzed to determine the concentration of Si and Zn ( Table  7). The ICP-OES analysis showed the presence of Zn that is possibly dissolved from the MFS slags. The Si concentration in the pore solution also implies the pozzolanic reactivity of the slag. Moreover, in another study [45], we have analyzed the reactivity of MFS slag along with thermally and chemically adapted MFS slag through R 3 calorimetric analysis. In this work, a distinct increased heat release between 4 to 6 h in the calorimetric analysis is only observed for the reference MFS slag; the thermally adapted MFS slag, which has less ZnO, showed no increased heat release in the specified time window. This supports the conclusion that this part of the heat release in the MFS slag is related to the presence of Zn 2+ ions in the glassy phase.

XRD Analysis
The XRD analysis in Figure 3 shows the presence of hydrated phases in the R 3 paste after 7 days curing. The main hydration products of the R 3 hydrated paste may include C-(A)-S-H, AFt, and AFm phases depending on the chemical composition of the starting SCM. General expected AFm phases are hydroxy-AFm, Al-hemicarbonate (Al-Hc), and Al-monocarbonate (Al-Mc). It has been reported elsewhere [46,47] that Al-Hc and hydroxy-AFm are unstable with respect to Al-Mc in the presence of CaCO 3 , and [48] showed that with rising CO 3 2− ion concentration, hemicarbonate (Hc) is transformed into monocarbonate (Mc).
In the FA containing model R 3 system, a sharp and distinct peak of ettringite (2θ = 9.1 • ) was observed. Ettringite can be synthesised in a system by reacting stoichiometric amounts of Ca 2+ , SO 4 2− , and Al 3+ ions with H 2 O [49,50] and formation of ettringite in FA is mainly attributed to the high content of Al 2 O 3 . However, no presence of AFm phase can be observed in FA due to the fact that FA just reacts so slowly that the SO 4 2− ions were not completely consumed, thus maintaining the saturation index which can stabilise the AFt (ettringite) phase [51]. In the GGBFS system, monocarboaluminate (Al-Mc) along with hydrotalcite (HT) can be observed, where the added calcite is the carbonate source to form these AFm phases. Some literature has observed the transformation of hemicarbonate (Hc) to monocarbonate (Mc) happening in the period between 14 to 28 days [46,52]. However, R 3 model paste of GGBFS showed the presence of monocarboaluminate even after 7 days due to the presence of excess CO 3 2− ions with an accelerating curing condition at 40 ± 2 • C.  [54][55][56]. An interesting study [57] proposed two possible hypotheses by modelling and calculating the role of CaCO3 in cement hydration. They proposed that the role of CaCO3 has two functions: (1) Active participant, and (2) inert participant. With respect to their findings, XRD results of the presented work can be interperted as follows. As an active participant, the CaCO3 present in the R 3 mix tends to dissolve and releases CO3 2− into pore solution forming AFm phases, as is the case for GGBFS and MFS slag. As an inert participant, the CaCO3 tends to just act as an inert filler or reacts slowly after 7 days without participating in the short term reaction mechanisms as in the case for FA. MFS slag contains around 40 to 41 wt% Fe 2 O 3 . During hydration, Fe containing AFm and/or Fe-AFt phases may form [53]. As seen in the XRD analysis around 2θ = 10.3 • , it is known from literature that Al containing AFm phases (generally Al-Hc) are formed first and later converted to Al-Mc (2θ = 11.8 • ) with an increase in CO 3 2− content as seen in GGBFS [47]. However, this peak of 2θ = 11.8 • has undergone a shift in the MFS slag and it is seen at 2θ = 12. Al 3+ in presence of CO 3 2− and Ca 2+ forms Al-Mc and/or Hc. However, the role of CaCO 3 in the hydration mechanism is still an open discussion. Different authors have already published their hypotheses, but there is no general agreement on the role of CaCO 3 during cement hydration [54][55][56]. An interesting study [57] proposed two possible hypotheses by modelling and calculating the role of CaCO 3 in cement hydration. They proposed that the role of CaCO 3 has two functions: (1) Active participant, and (2) inert participant. With respect to their findings, XRD results of the presented work can be interperted as follows.
As an active participant, the CaCO 3 present in the R 3 mix tends to dissolve and releases CO 3 2− into pore solution forming AFm phases, as is the case for GGBFS and MFS slag. As an inert participant, the CaCO 3 tends to just act as an inert filler or reacts slowly after 7 days without participating in the short term reaction mechanisms as in the case for FA.

Implementation of MFS Slag as Sustainable Binder in Mortar and Concrete-Strength and Ecological Benefits
Mortar with replacement levels of 15, 30, and 50 wt% of PC by MFS-slag were prepared to evaluate the compressive strength in function of time (Table 8). SCM mortar synthesized from MFS slags showed a slow increase in compressive strength from 2 days to 7 days, and a high increase in the period from 7 to 91 days compared to the reference PC value. Especially from the age of 28 days, the MFS slag clearly contributes to the strength, since the mixes with 15%, 30%, and 50% MFS slag reach higher compressive strength than the nominal reference values corresponding to 85%, 70%, and 50% of the PC compressive strength. This foreseen strength development is due to the possible mechanism of slow dissolution of amorphous silica from MFS slag in the alkaline pore solution and the formation of strength providing phases such as C-S-H and C-A-S-H as mentioned in Sections 3.2.1 and 3.2.4. Moreover, mortar with MFS slag with a replacement level of 70 wt% of MFS slag in PC with presence of solid activators (Na 2 SO 4 ) were prepared to study their compressive strength. Strength development of hybrid mortar (HM) also showed similar trends to the SCM mortars. However, HM mortars showed improved strength in comparison with 50 wt% MFS-slag SCM mortars after 28 days. This is due to the combined effect of pozzolanic and (geo)polymeric reaction. Nevertheless, HM mortars showed lower strength after 91 days with respect to 50 wt% MFS-slag SCM mortars. With respect to the increase in the global annual production of PC, global CO 2 emissions are also expected to increase to approximately 6 billion tons by the end of 2030. These MFS slags, if used in the concrete, can make the material much 'greener'. MFS slag can be incorporated in concrete as SCM or with additional activation, depending on the application. Partial replacement of PC up to certain replacement levels can be achieved without comprising the mechanical properties of the concrete (or at least with no larger effect than expected for traditional SCMs). However, a larger replacement level of around 70 wt% of MFS slag in PC can also be achieved with activation. At the same time, a by-product from Cu industry finds a high-level application. With less cement used to make concrete and the possibility of using MFS slag as SCM, sustainability and circular economy can be fairly well achieved.

Conclusions
The presented work gives a deep insight into the reactivity of a novel SCM, namely a fumed secondary copper slag called MFS slag. Reactivity of the MFS slag was successfully assessed by the recently developed R 3 model technique. The R 3 model paste designed by RILEM for assessing the reactivity of the SCMs allows us to overcome the limitations of the traditional reactivity test methods with PC-based pastes. From the R 3 model analysis, it was found that MFS slag can be regarded as a pozzolanic SCM due to its reactive behavior and can be used as SCM to make concrete.
The following can be stated as important findings from this work: (1) Bound water analysis of the R 3 model paste showed that GGBFS and MFS slag possess a high amount of binder hydrates due to their latent hydraulic and pozzolanic behavior, respectively. These hydrates decomposed between 105 and 350 • C during the thermal treatment. (2) CH consumption studied by TGA analysis showed that MFS slag had similar pozzolanic properties as other traditional SCMs. (3) R 3 calorimetric analysis indicated that MFS slag provoked a cumulative heat release in the range between FA and GGBFS (although some of the heat release could be due to the presence of Zn). Inert filler of quartz showed almost no heat release with R 3 calorimetric analysis, corresponding with consumption of only a slight amount of CH found by TGA analysis. (4) XRD analysis of the R 3 model MFS slag indicated the participation of "Fe" in the hydration mechanism and the existence of Fe-AFm, next to other hydrated binder phases.
One of the main research priorities in the future is to better understand the dissolution and role of ZnO and Fe 2 O 3 in the hydration mechanism of MFS slags, and the role of CaCO 3 in the formation of AFt/AFm phases for the R 3 model paste. In this respect, future experimental research such as Mössbauer and Raman spectroscopy analysis are needed to verify the participation of Fe 2 O 3 and ZnO in the hydration mechanism. Future research is also needed to verify the performance over long time scales, including complete durability assessment such as resistance to chloride and alkali silica reaction.