Feasibility of silicomanganese slag as cementitious material and as aggregate for concrete

.


Introduction
One of the main emitters of CO 2 worldwide is the cement industry. Emissions from cement plants reach 5-7 % of total global CO 2 emissions, corresponding to 900 kg CO 2 /ton cement [1]. The most effective way to reduce these emissions is not to use cement, which is impractical. The second most effective way is to reduce cement consumption, which can be achieved by replacing part of the cement with industrial wastes with pozzolanic and/or hydraulic properties, also called supplementary cementitious materials (SCM). Pozzolanic reactions (amorphous silica related) predominate in these compounds, typically being slower reactions than the hydraulic reactions of cement, so that the development of strength tends to occur over the long term [2]. Some of the most successfully used SCMs are fly ash (FA) [3,4], silica fume (SF) [5,6] and ground, granulated, blast furnace slag (GGBFS) [7,8] and the main motivation for their use it is always the use of a waste for applications with high added value. The current literature includes emerging and successfully tested SCMs such as cupola slag [9], copper slag [10] or ladle furnace slag (LFS) [11]. Fig. 1..
An emerging use contemplates silicomanganese slag (from ferroalloy), of which there are very few references in the literature, for the manufacture of cement and concrete. The production of this slag in Spain reaches 150,000 tons/year [12] and global production 8.9-10.4 million tons/year [13], of which almost all is deposited in landfills belonging to the industries themselves or to Public Administrations. Silicomanganese slag is generated in the production of ferroalloys with a high MnO content by carbothermic reduction, generally in submerged arc furnaces. Regarding its chemical composition, silicomanganese slag is richer in SiO 2 and has less CaO than GGBFS, but the composition (overall) is similar, so it can be expected to have a real application as SCM provided that the reactivity of its components is similar [14].
The few case studies found show that the replacement of GGBFS by granulated silicomanganese slag has made it possible to obtain slightly lower compressive strengths at short ages but similar at 28 days [14]. The hydration processes of these slags include as major hydrated phases Portlandite, C-S-H (CaO, SiO 2 , H 2 O) and C-A-H (CaO, Al 2 O 3 , H 2 O) along with manganese rich hydrated phases [14]. On the one hand, it has been reported that the strength gain of this compound is slow in early ages, justifying this assumption with isothermal conduction calorimetry (ICC), while it has been found that the evolution of heat is low at early ages of hydration [15]. On the other hand, it has been shown that replacement levels up to 35 % of Portland cement by silicomanganese slag hardly affect the setting times of the material [16]. There are also studies using silicomanganese slag as an alkali activated binder (AAB), showing potential, as described by several authors [17,18].
The silicomanganese slag is associated with rapid cooling that favors an amorphous structure. The valorization process continues with crushing and grading by size (sieving). After undergoing this process, the slag is transformed into siderurgical aggregates (SA), a term already used by the authors in other studies [19][20][21]. The references in the literature on this use are also scarce, however, a recent study has shown that the durability of concrete with these SAs, in terms of chloride penetration and sorptivity, is lower than that of conventional concrete at 28 days, but similar at 90 days [22]. In mechanical terms, the replacement of coarse natural aggregate by SA can reduce the compressive and tensile strength by about 9 % and 17 % respectively [22]. As previously mentioned, studies are scarce and, even more so, those using the fine fraction of these SAs.
This study attempts to cast some light on the use of silicomanganese SAs both as an alternative binder (SCM) and for the manufacture of concrete as an aggregate. The objective of this is to reduce the CO 2 emissions generated by the manufacture of cement, to reduce the costs derived from the fees for excess CO 2 emissions, as well as to reduce the consumption of natural aggregates (and everything that this entails). A waste valorization methodology has been proposed to produce SCM, as well as the design of standardized mortars applying 20 % replacement of Portland cement by this compound as well as by FAs, to make a comparison between the three materials (FA, SA and CEM) in terms of strength evolution. The design of a novel concrete has been proposed that includes not only the replacement of the coarse fraction but also of the fine fraction, comparing its physical-mechanical behavior in the fresh state with that of a conventional limestone concrete with similar mix proportions. This study will establish the technical feasibility of the use of silicomanganese slag as an alternative construction material, establishing the limit conditions for its use.

Aggregate properties
This research deals with the use of a by-product as a potential construction material, so its characterization must be as complete as possible. This characterization consists of a battery of standardized tests that determine the physical-chemical and mechanical properties of the material that justify its use in concrete. Table 1 shows the oxide composition obtained in a vacuum atmosphere, using a sequential wavelength dispersive X-ray fluorescence spectrometer (WDXRF) of the PANalytical brand, model AXIOS, equipped with a Rh tube and three detectors (gas flow, scintillation and Xe sealing). The main elements present in SA are, in this order, silicon and calcium, which together account for more than 62 % of the weight of SA. Next, elements that are also significant due to their abundance are manganese, aluminum, magnesium and, to a lesser extent, iron. This chemical composition highlights that this material has a high content of silicon and calcium oxides, similar to traditional pozzolanic materials and even to Portland cements. This SA has a high content of manganese (11.9 %) and a significant presence of sulfides (1.7 %).

Chemical properties
Defining hydraulicity as the ability of a material to harden due to hydration processes to obtain calcium aluminate hydrates and calcium silicates, the pozzolanicity of a material can be approximated through the hydraulicity index (HI) [23]. This index is calculated using the Eq. (1): The hydraulicity index obtained is 1.12, which is greater than 1 and therefore exceeds the requirement for good performance to establish that the material has hydraulic behavior.
The crystallographic properties of the aggregates were obtained using X-ray diffraction (XRD) techniques. The XRD spectrum (diffractogram) of the SA sample was obtained on a Bruker D8 Advance commercial instrument, the equipment operating at a power of 40 kV/ 30 mA, provided with a Cu anticathode tube and a wavelength of λ = 1.5418⋅10 − 10 m. The spectrum was produced at Bragg angles (2θ) between 10 and 70 • (Δ2θ = 0.02 • ).
As can be seen, although it is possible to identify some crystalline phases in the sample, mainly the calcium-magnesium silicate called akermanite (Ca 2 MgSi 2 O 7 ), it stands out that the amorphous character of the material predominates, preventing the sharp formation of diffraction peaks and contributing to a very pronounced and irregular radiation background.
The fact that these SAs have an eminently glassy character and are not fully crystallized implies that the phases present are in a metastable state and therefore the SA will not be completely inert in the long term, which may limit its potential use as an aggregate in manufacturing of concretes. This metastability, a priori, does not make this aggregate ideal for use in concrete that is going to be exposed to aggressive external agents. Table 2 shows the results of the leaching test and the limits established by the decrees 106/2006 y 100/2018 of slag recovery in the Autonomous Community of Cantabria.
In addition, some physical properties of the analyzed sample are reported, such as its humidity (H = 7 %), the pH of the leachate (9.08 at 21.6 • C) and the conductivity of the leachate (118.6 µS/cm at 20 • C). As can be seen, the values reported for all the elements are well below the limits required by current regulations (European and local), so the SA does not present leaching problems. In addition, it should be remembered that when SA is part of a concrete, it will be encapsulated within the cementitious matrix itself, so the risk of leaching will be even lower than that of the free SA analyzed here. Table 3 shows the results of the pozzolanic activity test carried out on the SA supplied according to  and using a pure sample of powdered SA. As can be seen, the result of the pozzolanicity test is positive (calcium ion concentration <4 mmol/l), so the SA will present pozzolanic activity predictably in the medium or long term. The result obtained in this test, therefore, provides great added value to the SAs, since they can a priori be used as an addition or partial replacement of the binder in cementitious matrices, providing strength at late ages. Table 4.
For the estimation of the content of light particles, the test method described in standard EN 1744-1 [25] was used. The aim of this test was to determine the existence of substances that can cause stains or blisters on the surface of the concrete (for example, carbon or wood). The test was carried out on samples of the SA fractions with a size greater than 3 mm, immersing a quantity of SA in a liquid with a density slightly lower than 2.0 g/cm 3 . In this case, a supersaturated solution of zinc chloride was used, with a density of 1.98 ± 0.02 g/cm 3 . The lower density particles float on the liquid, all the liquid is filtered through a 0.3 mm size sieve, and the retained particles are washed, dried and weighed.

Geometric and dimensional properties
The granulometric classification test was carried out in accordance with EN 933-1 [26] and EN 933-2 [27]. Fig. 2 shows all the fractions used in the design of concrete with SA, labeled as siderurgical aggregate (SA), limestone (LS) and siliceous aggregate (SS), as well as their minimum and maximum size (mm/mm). The coarse fractions of SA do not contribute fines, while the SA sand has a 6 % content of fines (<0.063 mm). The natural aggregate fractions (LS and SS) will be used to make the appropriate corrections in the concrete mix to improve its compactness.
The shape of the particles decisively determines the compactness of skeleton in the hardened state of the concrete as well as its workability and docility in the fresh state. This parameter was analyzed in two different ways. Through standard EN 933-3 [28], the flakiness index was determined, which measures the flakiness of particles using bar sieves with openings half that of mesh sieves. The shape index was also determined in accordance with EN 933-4 [29], establishing as noncubic particles those whose length/thickness ratio is greater than 3. Table 5 shows the results obtained.
The SAs in this study comply with the Spanish Structural Code [30] in terms of flakiness index, being <35 %, so their use is a priori suitable for making concrete.

Physical properties
Density determination was carried out using the procedures described in standard EN 1097-6 [31] for the determination of particle density and water absorption, and in EN 1936 [32] for the determination of real density, apparent and total open porosity. Table 6 shows the physical properties of all the aggregates used, both natural and siderurgical. The density of SA is higher than that of natural aggregate, although this difference is only 6 %, mainly due to the presence of metals. On the other hand, the absorption of SA is higher, but in any case lower than the 5 % limit established by the Structural Code [30]. This superior absorption will require greater use of kneading water, or put another way, silico manganese aggregate is water intensive.

Mechanical and tribological properties
These properties are determined by the chemical composition of the   SA and by the process of generating and cooling them. The mechanical performance of the aggregate will influence that of the concrete, in as much as the concrete collapses due to the aggregate, when it does not achieve a performance similar to that of the paste. The resistance to fragmentation was established by determining the Los Angeles index, carried out in accordance with EN 1097-2 [33], using an abrasive load made up of 12 steel balls and a speed of 31 rpm up to 500 revolutions. The material tested is screened through a 1.60 mm sieve and the passing material is expressed as a percentage. The results show a resistance to fragmentation of 37 %, slightly lower than the limit of 40 % proposed by current regulations [30].
On the other hand, the aggregates crushing value (ACV) was determined according to ISO 20290-3 [34]. The aggregate of the 10/14 fraction is compressed, subjecting it to a loading ramp at a speed of 40 kN/min until it reaches 400 kN. The tested material is sieved through a 2.50 mm sieve and the passing material is expressed as a percentage.
Finally, the content of soft particles was determined. The hardness of the aggregates (and hence the softness) is related to its compressive strength [35] and thus to the compressive strength of the concrete. This test was carried out in accordance with the guidelines contained in the UNE 7134-58 standard [36]. A particle is considered to be soft when sliding it under the cylindrical Cuzin probe produces a groove due to the effect of scratching, without metal being detached on it, or when the detachment of fragments from the surface of the particle is observed. The soft particles are weighed and expressed as a percentage. The results obtained are shown in Table 7; not finding soft particles present in these aggregates.

Durability
The durability of concrete is its ability to behave satisfactorily under aggressive physical or chemical action during the service life of the structure. In this capacity, the quality of the aggregates used plays a predominant role, since they make up most of the volume of the concrete. The freezing resistance of SA was determined following the guidelines of standard EN 1367-1 [37]. The purpose of this standard is to determine the behavior of SA under volumetric variations produced by cycles of freezing and thawing. To do so, the 8/16 mm fraction was subjected to cycles of exposure at − 20 • C for 8 h and exposure to water at 20 • C for 16 h. 10 cycles were carried out and the percentage of detached material was determined.
The behavior of SA under volumetric variations produced by repeated cycles of water saturation and drying was also determined. This test was carried out in accordance with UNE 146,510 [38], exposing the samples to successive cycles of saturation in water at 20 • C for 16 h and subsequent drying in an oven at 110 ± 5 • C for 8 h. A total of 20 cycles were completed and the percentage of material removed was determined. The results obtained are shown in Table 8, where hardly any mass variation was observed with the passing of cycles.

Recovery of SA as an addition
This process is based on obtaining a very fine material, with a large specific surface that favors the start of chemical hydration reactions as there is a greater surface in contact with water and aggregates.
The valorization process of the silicomanganese siderurgical aggregates, used to obtain a pozzolanic addition, was carried out following the methodology proposed by Sosa et al. [9]. In order to exploit the benefits that could be derived from its pozzolanic activity, a specific preparation is previously required to obtain SA as a powdery material with fineness similar to that of cement itself. To this end, a wet grinding process was implemented in a ball mill (Fig. 3) that enables finely divided SA to be obtained within a slurry. The material is then dried in     an oven at 105 ± 5 • C until it reaches constant weight, thus obtaining a greenish-brown color. It should be noted that the grinding process should enable a particle size of about 20 µm to be obtained, in this case 82 % of the material passes through the 20 µm sieve, which was determined on a dry sample.

Mix proportions
A reference mortar (NM) was manufactured with the proportions described in EN 196-1 [39] for standardized mortars (40x40x160 mm). For all mixes, cement (CEM) I 52.5R [40] and tap water were used, as well as the same amount of silica sand CEN (EN 196-1) and water. Two mixes were also developed in which 20 % of the CEM is replaced by the addition of silicomanganese SA powder (SAP) called SAM, and by fly ash (FA) called FAM. The design of these concrete mixes aims to compare the performance of the SAP with those of the CEM and with an additive with contrasted pozzolanic character (FA). Table 9 shows the mix proportions of the manufactured mortars.
After verifying the pozzolanic capacity of SA as an additive, in the second phase of this research, the suitability of SA as an aggregate for the manufacture of conventional concrete was verified. This verification was carried out by direct comparison with a conventional limestone reference concrete.
The distribution of the granulometric skeleton of both mixtures were designed using the Fuller method [41] and both mixes have been designed using the same amount of CEM I 52.5 R, 1 % superplasticizer additive (SP), the same volume of aggregates and the same amount of effective water (Table 10), to facilitate comparison between aggregates. The type of superplasticizer used is a high-activity superplasticizer/ water reducer for the production of low-viscosity concrete with improved rheology. The effective water (water available in the mix) was calculated as the difference between the total water and the water absorbed by the aggregates, which is greater in the case of SA ( Table 6).
The two mixes designed were as follows: • A conventional reference concrete mix with fine and coarse limestone fractions (REF).
• A concrete mix with fine and coarse siderurgical aggregate fractions (SAC), which also incorporates natural sands to increase the compactness of the concrete, in this case limestone sand (NA 0/4) and siliceous sand (NA 0/2).
The concrete was manufactured in a 100 L capacity vertical axis Fig. 3. Diagram of the production process of silico-manganese SA powder.  (planetary) mixer, using the mixing times recommended by ASTM C 192 [42] for structural concrete, consisting of 3 min of mixing, 3 min of rest and 2 min of kneading. The concrete was compacted by means of a vibrating needle and was demolded 24 h after manufacturing, curing in a humidity chamber at 21 ± 5 • C and 95 ± 5 % relative humidity.

Mortar properties
The mechanical properties of the mortars were determined in terms of flexural strength and compressive strength. The tests were carried out in accordance with EN 196-1 [39], using a universal servo-hydraulic press with a load capacity of 150 kN. The load application speeds for the flexural strength and compressive strength tests were 50 N/s and 2500 N/s respectively. Three mortar specimens were tested per mix and age (7, 28 and 91 days of curing submerged in water at 20 ± 2 • C), making a total of 18 specimens.

Concrete properties
The consistency in the fresh state was determined by means of the slump test, in accordance with EN 12350-2 [43], for the REF and SAC mixes, to check the effect of the shape of the aggregates on the workability of the concrete.
The physical properties of hardened concrete at 28 days of age were determined on six 150x100 mm cylindrical specimens (thirds of normalized specimens) per mix. These properties include bulk density, apparent density and saturated surface dry (SSD) density, obtained through EN 12390-7 [44]. The open porosity and water absorption were also determined in accordance with UNE 83,980 [45].
Regarding the mechanical properties of concrete, the uniaxial compressive strength at 28 days of age was determined on 4-5 cubic specimens of 100 mm in accordance with EN 12390-3 [46]. A universal servo-hydraulic press with a capacity of 2500 kN and a load application speed of 5 kN/s (0.5 MPa/s) was used. The flexural strength at 28 days of age was also determined on 4 prismatic specimens of 100x100x400 mm in accordance with EN 12390-5 [47] using the three-point support variant. A universal servo-hydraulic press with a capacity of 250 kN and a load application speed of 110 N/s (0.5 MPa/s) was used. Fig. 4 shows the mechanical properties of mortars with 20 % CEM replacement by SCM, as well as those of the reference mortar. Both flexural and compressive strength increase with age, and their behavior can be adjusted to a logarithmic model with a high correlation (R 2 ~0.9). Fig. 4 (left) shows that the reference mortar exhibits greater strength at any curing age, although the relative differences are greater at earlier ages, 13 % and 20 % higher than SAM and FAM, respectively. At 90 days this difference is practically halved, which indicates a greater development of strength at advanced ages when these SCM are used. In the case of FA, this delayed gain is due to the reaction with the cement by joining the Ca(OH) 2 with free silica with a pozzolanic reaction that forms an insoluble CSH structure [48]. In the comparison between the use of both types of additions, the addition of SA has shown to have a higher activity than the FA (traditional addition), reaching average compressive strength values 5 MPa higher after 28 days of curing. Fig. 5 shows the appearance of the mortars tested, highlighting the greenish color of the SAM and the presence of more pores in the mixes that incorporate SCM.

Mortar properties
The flexural strength of the mortars (Fig. 4 right) shows that the differences between mixes are small. The SAM mortars show values of 7 % to the FAM at 90 days of age. The main difference compared to the evolution of the compressive strength is the lack of evolution of the NM mortars at 90 days of age, this means that the average strength at both ages is approximately 50 MPa.
On the other hand, ASTM C311 [49] defines the pozzolanicity index (strength activity index) of supplementary cementitious materials as the ratio between the compressive strength of mortars that incorporate 20 % replacement and mortars without replacement, for ages of 7 and 28 days. The SA mortars in this study have a strength activity index of 85 and 90 % at 7 and 28 days, respectively, those that reach 75 % being considered active additions.
This new material, based on the recovery of silicomanganese slag, has great potential as an SCM or as a partial replacement for part of the cement after analyzing its binding capacity in standardized mortars. Given that the reactions of SCM are slow, a potential application would be in the manufacture of cements with low hydration heat, which would also have the advantage of being a binder with a reduced environmental impact, derived from reducing the CO 2 generated during calcination processes in clinker production.

Concrete properties
SA concrete show less workability in the fresh state, showing greater interlocking when poured and placed. Fig. 6 shows the slump of the manufactured concrete, obtaining a slump of 60 mm for the REF and 45 mm for the SAC, which correspond to the "plastic" and "soft" consistency classes, respectively, according to the Structural Code [30]. Having used the same effective w/c ratio and amount of superplasticizer additive, the variation in slump is solely caused by the shape of the aggregates, which as shown previously ( Fig. 3 and Table 5) is loose or planar. This shape, in addition to increasing the friction with the paste Fig. 4. Mechanical properties of designed mortars. and with other aggregates (interlocking effect), requires a greater volume of paste to perfectly cover the aggregates because the specific surface of the aggregates is increased, with respect to the reference calcareous aggregate. On the other hand, Fig. 6 shows how the paste coating on the lateral surface of the cone seems adequate, with no cavities being observed. A priori, this demonstrates the effectiveness of the concrete vibrating process. Table 11 shows the physical properties of the concrete mixes designed. The bulk density of both mixes is very similar, which means that their own weight is not excessive so as not to contemplate their use in structural applications, as is the case with other types of siderurgical aggregates. This is due to the low iron content of this aggregates (1.22 %) compared to 40 % in other types of siderurgical aggregates [19]. On the other hand, the water absorption of SAC is 40 % higher, exceeding 11 % wt. This is due to the fact that siderurgical aggregate is on average 45 % more porous than natural aggregate and due to the previously mentioned effect of the use of more planar particles. These planar particles require a greater volume of paste than natural aggregates (more rounded), which can cause a lack of paste and therefore the appearance of voids within the concrete.
Regarding the evolution of the compressive strength of concrete with age (Fig. 7), a great difference can be found in the mechanical performance of the two mixes. At 28 days, the average compressive strength of the REF mix is 62 MPa while that of the SAC mix is 49 MPa, which translates into a difference of 20 %. This difference is explained by the difference in mechanical behavior of the two aggregates, whose resistance to fragmentation and its crushing value is approximately 20 % higher in the case of limestone aggregate ( Table 7). The rapid cooling undergone by these slags causes the formation of an amorphous and fragile structure, a structure that forms the granular skeleton of the concrete and determines its response to external stresses. Using electric arc furnace slag (EAFS) siderurgical aggregates and analogous mix proportions (same cement and w/c ratio) the authors have obtained resistances 81-124 % higher, showing the importance of the cooling process and the nature of the slag [21].
The activity of this siderurgical aggregate as SCM does not seem to provide concrete resistance capacity at late ages (90 days). This is because the effectiveness of SCM depends fundamentally on its specific surface, which is increased by grinding and defines the material's intrinsic reactivity [50]. Fig. 8 shows the appearance of the test specimens at 28 days of age. All the cracks were satisfactory, with their vertical faces detaching due to shear stress. As demonstrated by the results of the absorption in water, the structure of the mortar paste shows a greater volume of voids in the case of the SAC mix, while all the aggregates on the fracture surface are fragmented. The greenish color is striking, which could open up the application of this concrete to ornamental or decorative uses.    Fig. 9(right), the appearance of one of the fracture surfaces of the prismatic specimens can be observed. Breakage has always occurred through the aggregates, which shows that these are the weakest link in the chain, an effect produced not only by their fragility but also by the planar shape of these particles.
It can be concluded that, in view of the results obtained, the use of siderurgical aggregates from silicomanganese slag (amorphous structure) produces a reduction of around 25-30 % in the mechanical performance of concrete when almost total replacements of the natural  aggregate are used. For this reason, the authors believe that its use is recommendable in more conservative replacements so that its impact is not so noticeable or for concrete for non-structural or ornamental use.

Conclusions
In this research, the performance of mortars and concretes manufactured using silicomanganese siderurgical aggregates was characterized, starting with a very exhaustive characterization of the physicalchemical and mechanical properties of this material. The studies found in the literature are very scarce, so this study represents an advance in the knowledge about this new construction material.
On the one hand, the aggregate was valorized to obtain a reactive filler that was incorporated, replacing part of the cement for the manufacture of mortars. On the other hand, the siderurgical aggregates were used as an almost integral part of the granulometric skeleton of the concrete. After analyzing the results, the following conclusions can be drawn: • These siderurgical aggregates have a silica content close to 40 %, an amorphous structure and when ground, a hydraulicity index of 1.7 and a positive response to the pozzolanic activity test, which demonstrates the high reactivity of this material. • The aggregate complies with each of the leaching limits permitted by current regulations and also has an acceptable shape and geometry for the manufacture of concrete, being 6 % denser than limestone aggregate. • The aggregate complies with the regulatory restrictions for water absorption (1.5<5 %) and wear resistance (37<40 %), showing excellent durability under freezing-thawing and drying-wetting cycles. • The recovery of the aggregate was proposed in order to obtain a reactive filler, obtaining a strength activity index of 90 % at 28 days. The compressive strength of mortars with the SA filler is barely 10 % lower than the control mortar at 28 days, a very similar strength for both mortars at 90 days. This material exhibits great potential as a supplementary cementitious material or as a binder. • The shape and mechanical properties of the SA reduce the strength of the concrete by 25-30 % when used as an almost total replacement. This supports the assertion that it should be used in small replacements, in concrete for non-structural use or for ornamental use due to its attractive aesthetical appearance.

Declaration of Competing Interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Pablo Tamayo reports financial support was provided by COBASA Grupo Logístico S.L.

Data availability
No data was used for the research described in the article.