Chemical and mineralogical characterization of silicon manganese iron slag as railway ballast

In nature, metal ores such as iron, lead, aluminum and others are found in an impure state, sometimes oxidized and mixed with silicates of other metals. During casting, when the ore is exposed to high temperatures, these impurities are separated from the molten metal and can be removed. The mass formed by these compounds is slag. Slag is the co-product of the smelting of ore to purify metals. It may be considered a mixture of metal oxides, but may also contain metal sulphites and metal atoms in their elemental form. After it is reprocessed to separate the metals contained, the co-products of this process can be used in cement, rail ballast, road paving and various other purposes. The objective of this research work is the presentation of the chemical and mineralogical characterization tests of the silicon-manganese iron slag with the purpose of reusing the coproduct as rail ballast. X-ray diffraction tests, quantitative chemical analyzes, scanning electron microscopy and free lime content were prepared for these characterizations. The results of these tests showed the technical feasibility of using slag as rail ballast.


Introduction
Ferroalloys are essential inputs in the steel industry, as they are indispensable in the manufacture of all types of steel, both in the basic processes and in the refined property and value aggregation in special steels (Andrade et al., 1999).Basically the most conventional ferroalloys are ferrosilicon, ferrochromium, ferro-siliconmanganese and ferromanganese, and are present in almost all the processes of iron and common or special steel production.
Among the types of alloys, the most prominent in their production are the manganese alloys.They are applied in the manufacture of virtually all types of steel and iron castings due to their desulfurizing and deoxidizing properties (Andrade and Cunha, 2000).
The production process of the siliconmanganese iron alloys (FeSiMn) consists of the simultaneous reduction of manganese and silicon by carbon in electric reduction furnaces (FER).The charges consist of manganese ore and quartz (or quartzite) or manganese ore, rich slag and quartz (or quartzite).These charges are complemented by fluxes and reducers, which are used to adjust the composition of the steel co-product (slag) in the process.
According to Olsen and Tangstad (2004), the distribution of Si between SiMn alloys and multicomponent MnO-SiO 2 -CaO-Al 2 O 3 -MgO slags is mainly determined by the process temperature.The silicon content from silica will increase by approximately 6 % per 50 °C in the temperature range from 1550 °C to 1700 °C.This process is only feasible if the reduction temperatures of both oxides are close to each other.An-other aspect to consider is the melting point of the slag: if it is below the silica reduction temperature, the furnace will continuously spend energy in melting, to the detriment of the reduction reaction.
Thus, two basic procedures are required in the formation of slag.The first one contemplates the reduction of the MnO activity (lower basicities of the slag) to increase the temperature of reduction of the manganese approaching the silicon reduction temperature.In the second, the formation of the slag with higher melting points compatible with the process temperatures can be obtained.This is done by acting on the charge formulation.One of the most relevant aspects of the process is the distribution of Si and Mn between the Mn-Fe-Si-Cs at alloys and the slag of the MnO-SiO 2 -CaO-MgO-Al 2 O 3 REM, Int.Eng.J., Ouro Preto, 70(4), 385-391, oct. dec. | 2017 system.The equilibrium relationships between the two systems for the Carbon Monoxide (PCO) pressure equal to 1 atm are shown in Figure 1, which shows the diagram of the system considered (Ding and Olsen, 2000).
There are several proposed equilibrium relationships for these systems.Each refers to existing conditions, whether in statistical data of one or several installations, or in specific experiments.
There are two types of slag formed by the described production processes: the rich slag, which is acidic, presents high Mn contents (above 40 %), being recyclable and reused as an input in FeSiMn production (besides having very low phosphorus content, which is an advantage for the FeSiMn production), and poor basic slag, which has low manganese content (MnO < 20 %) and about 30 % of Si is produced by upgrading the standard alloy through addition of silicon waste from the iron silicon industry, which at the end of the ferroalloy production process, is discarded (Olsen and Tangstad, 2004).Figure 2 shows the silicon iron manganese slag.

Materials and methods
As mentioned earlier, there are two types of slag formed by the described production processes: rich and poor slag.The rich slag is acidic, because it contains high levels of Mn, undergoes vitrification that decreases its resistance and makes it dangerous for handling.This high content of Mn contained in the rich slag is reused as an input in FeSiMn production.Thus, because it is a basic co-product with low manganese content, has a resistance greater than that of the rich slag and is not being reused in the production of FeSiMn, the silicon-manganese iron slag used for this study was the basic (poor) slag.For that, slag samples provided by a silicon-manganese iron producer in Minas Gerais, at its Ouro Preto-MG unit, were used, which has a production capacity of 5000 t / month, with 3000 t / month of material processed for use as a rail ballast.The sample of the material was initially subjected to X-ray diffraction, quantitative chemical analyses, scanning electron microscopy and free lime content assay.
For the accomplishment of these tests the material was submitted to the granulometric reduction process.It was then sieved, where approximately 20 kg of the sample passed through a sieve of either 200 or 75 μm.All the tests have as reference the standards established by ABNT (Brazilian Association of Technical Standards), AREMA (American Railway Engineering and Maintenance Association).

X-Ray Diffraction Analysis
A Shimadsu diffractometer, model D / MAXB, with iron tube and scanning range of 7° to 70° and radiation of 20 kV and 5 μA was used in the analysis.The speed adopted in the test was 2 degrees / minute lasting ± 35 minutes.
The analysis of the results is processed through specific software (MDJ Jades), which compares the peak of the incidence of electrons with a database available for the characterization of the constituent minerals.
A monochromatic X-ray radiation was imposed on the samples under various angles of incidence.The dispersion of the different wavelengths obeys Bragg's law which is represented by the Equation below:

nl = 2dxsenq
Being: n -integer known as the order of the diffracted beam; l -wavelength of the apparatus tube; q -incidence angle and d -interplanar distance (spacing).
The records of this optical process are collected in a detector in the form of an amplified electrical signal, which is recorded in digital or analog form, cor-responding to the diffractograms.The diffractograms present characteristic peaks that are the results of X-ray diffraction in crystallographic planes of the sample, related to the position, intensity and shape of the crystalline structure, thus characterizing the mineralogical composition of the analyzed materials.

Quantitative chemical analysis
It is by measuring the intensity of the characteristic X-rays emitted by the sample's elements that the quantitative analysis by X-ray fluorescence is based on.A radioactive source emits X-rays, which excite the constituent elements and, in turn, emit spectral lines with characteristic energies of the element and whose intensities are related to the concentration of the element in the sample.X-ray fluorescence analysis consists of three phases: excitation of the sample elements, dispersion of the characteristic X-rays emitted by the sample and detection of these X-rays emitted.The X-ray fluorescence spec-trometer is an instrument that enables the quantitative determination of the elements present in a given sample by applying X-rays to the surface of the sample and the subsequent analysis of the fluorescent rays emitted.
In the analysis, a Shimadsu Energy Dispersion X-ray spectrometer, model EDX-700 HS / 800 HS / 900 HS was used.The samples used in this experiment went through two types of preparation: The first preparation was carried out with 20 g of pulverized sample passed through the 75 μm sieve.Then the sample was heated between 100 ° C and 120 °C in the furnace, after which 0.6 g of estuary acid -C 18 H 36 O 2 (acts as a 'glue' to fix the sample) were added.Next, the sample was further sprayed for 4 minutes to mix the sample material with the estuary acid, and then the mixture was pressed into the pallet form.In the second preparation, 10 g of pulverized sample was passed through the 75 μm sieve, and then the sample was placed in a plastic ring sealed by a plastic filter and inserted into another plastic bag with the tips cut to form a vacuum in the sample for the test.Samples prepared for the X-ray fluorescence assay are shown in Figures 3 and  4 for the first and second assemblies.

Chemical analysis by scanning electron microscopy
For the observation and distinction of different types of minerals stemming from the emission and interaction of electron beams on a sample, the Scanning Electron Microscopy technique was used, which allows the characterization from the point of view of its morphology, its organization and its chemical composition.What is observed in the scanning electron microscope is due to the contrast variation, which occurs when a primary electron beam sweeps the surface of the sample in point-to-point analysis.The qualitative distinction of the particle morphology of the samples of the study materials was made using an electronic scanning electron microscope of the JEOLJSM-5510 type.
In the test, the samples need to be conductive in the presence of carbon and, for this, they undergo the metallization process.The samples are thus arranged in the form of monolayers in metal cylindrical pieces (pallets), coated with graphite by a double-sided tape, then deposited in a vacuum evaporator for a period of about 1 hour.The graphite electrodes are then vaporized for about 10 minutes.After this process, the sample becomes conductive and the pallet is introduced into the chamber of the microscope, making a vacuum of 3 minutes.The samples are subjected to electron bombardment processes, and then photomicrographs are made.In this study, the adopted increases were 50x, 100x, 150x to 270x in the samples, using different accessories such as X-ray spectrometers, retro-scattered electron detector, absorbed electron detector, and transmitted electron detector.

Free lime content
In determining the free lime content, the test method prescribed by NBR NM 13/2004 -Portland Cement -Determination of free calcium oxide by ethylene glycol (ABNT) was used as reference.This standard specifies the determination of free calcium oxide by dissolution in ethylene glycol and subsequent titration in standard solution of hydrochloric acid (HCl) using a mixture as a pH indicator (mixed indicator solution).
For the assay, three samples containing 1.00 g each were weighed and mixed in three capped Erlenmeyer flasks, each containing 30 ml of corrected (green) pH ethylene glycol.The flasks were then heated in a water bath at 65 °C to 70 °C for 30 min, shaking the flask every 5 minutes.
After this process, the filtration was done with a slow filtration filter and a vacuum pump.Thereafter, the filtrate is transferred to a clean flask and 3 to 5 drops of mixed indicator solution are added.Then the above solution is titrated with 0.1 mol / l HCl until it changes from green to pink.Thus, the formula below is used to calculate the free lime content: Where, F -HCl Factor; M -Mass of crushed sample employed in the test; V -Volume of HCl used in the titration.
This method does not differentiate oxide [Ca (OH)] from hydroxide [Ca (OH) 2 ] and does not determine the calcium that is chemically combined in solid solution with FeO and MnO or in the form of silicates.However, in the scope of the analyses intended in this study, the results of this test are more than satisfactory.They serve as a basis for estimating the percentage of free lime, which causes the expansion that is responsible for causing the spontaneous breaking of the grains, which in turn causes the excess of fines that is harmful to the ballast.

Results of tests and discussion
Since there is no specific technical standard for silicon-manganese iron slag both in Brazil and abroad, the values found in the analyzes were compared

X-Ray Diffraction Analysis
The X-ray diffraction patterns of the silicon-manganese iron slag show the mineralogical phases of the material.Figure 5 shows the X-ray diffrac-tion patterns of the silicon-manganese iron slag.

Quantitative Chemical Analysis
For this first preparation, the pure chemical species were analyzed.Table 1 shows the pure chemical species in percentage.For the second preparation, the composite chemical species were analyzed.

Chemical analysis by scanning electron microscopy
Semi-quantitative chemical analysis was performed on the pulverized material.Table 3 systematizes the com-posite chemical species found and their respective percentages for each point of analysis.Figure 6 shows the morpho-logical distribution obtained in the slag powder analyses.At the analyzed points, the carbon oxide, silicon oxide, calcium oxide and manganese oxide contents presented higher percentages, whereas carbon oxide presented the highest percentage among all elements.The analysis of the slag showed that point 5 presented the highest values of calcium oxide (31.74 %), and sum of aluminum and iron oxides (15.88 %), proving that the results found were below the values recommended by AREMA (2001).

Free lime content
In determining the free lime content, the test method prescribed by NBR NM 13/2004 -Portland Cement -Determination of free calcium oxide by ethylene glycol was used as a reference.Figure 7 shows the percent free lime content graph.

Figure 3
Figure 3Samples prepared in the first assembly for the assay.

Figure 4
Figure 4Samples prepared in the second assembly for the assay.
with the indexes recommended by AREMA (2001) for LD steel mill slag.In order to use the LD steel mill slag ballast, it must have a percentage of cal-cium oxide (CaO) of less than 45 % and a sum of aluminum oxides (Al 2 O 3 ) and iron (Fe 2 O 3 and FeO) of less than 30 % (AREMA, 2001).
Figure 5 X-ray diffraction patterns of slag.
Figure 6 Typical morphology found in pulverized slag.

Table 3
Composite chemical species and their percentage by mass.