Fundamental Research on Fluorine-Free Ladle Furnace Slag for Axle Steel of Electric Multiple Unit Vehicles

Abstract

Fluorine-bearing refining slag (FBS) is used to produce axle steel for electric multiple unit vehicles. To avoid environmental pollution caused by fluorine, a fluorine-free ladle furnace slag (FFS) was designed based on an industrial FBS. The effects of main components on the physical and metallurgical properties of slag were investigated via theoretical analysis and laboratory tests. The composition range of components of the designed FFS are w(CaO) = 40–55 wt.%, w(SiO₂) = 2–6 wt.%, w(Al₂O₃) = 30–40 wt.%, w(MgO) = 6–8 wt.%, and w(CaO)/w(Al₂O₃) = 1.25–1.50. Industrial-scale test results indicate that the FFS has similar deoxidation and desulfurization capabilities to industrial FBS.

1. Introduction

China has the world’s largest high-speed rail network. Axle steel is a special steel used to manufacture the axles of electric multiple unit (EMU) vehicles, which guarantees safe transportation. The oxygen content and level of inclusions are of vital importance to the properties of axle steel, and therefore refining plays a significant role in the melting process [1]. It was shown that an appropriate refining slag plays a significant role in improving the quality of steel products [2,3]. Some components of the refining slag were found to affect the slag viscosity and melting characteristics [4,5]. Therefore, the refining slag composition must be strictly controlled within limits to meet the melting requirements.

In China, the adopted production of axle steel is generally the electric arc furnace (EAF) → ladle furnace (LF) → vacuum degassing or Ruhrstahl-Heraeus process (VD or RH) → continuous casting (CC) route. The LF slag of axle steel is mainly designed based on a high-basicity CaO-Al₂O₃ slag system [6], and the mass ratio of CaO to Al₂O₃ is usually in the range w(CaO)/w(Al₂O₃) = 1.0–2.0 [7]. In addition, calcium fluoride is widely used as fluxes to satisfy the slag high-temperature physicochemical properties during the LF process. Studies have shown that the introduction of CaF₂ to the CaO-A1₂O₃-SiO₂-MgO system lowers the melting point and viscosity, thus improving the fluidity of the slag [8,9,10,11,12]. Recently, the mechanisms by which slag components affect the behavior of the slag melt have been studied by molecular dynamics and Raman spectroscopy [13,14]. Zhang and co-workers proposed that CaF₂ can depolymerize the network structure of the melt in the CaO-Al₂O₃-CaF₂ system and studied means of improving the fluidity of the slag based on molecular dynamics simulations [14].

However, calcium fluoride will react with SiO₂ to generate the low melting point SiF₂ that tends to evaporate in the atmosphere during the melting process [15]. Zhao et al. investigated the volatilization characteristics of three kinds of typical fluorine-containing slag systems for steelmaking process, being the remelting electroslag, the continuous casting mold flux, and the traditional refining slag [16]. Results show that the CaF₂-CaO refining slag system generates mounts of SiF₄, and the volatility can be reduced by adding a small amount of SiO₂. Using thermodynamic calculation, Mao and co-workers proposed that the volatiles are NaF, KF, SiF₄, AlF₃, NaAlF₄, and HF during the continuous casting process [17]. Apparently, industrial releases from the steelmaking process are contributing to increased levels of fluoride in the environment. Therefore, it is urgent to develop environment-friendly fluorine-free refining slag for the refining process.

At present, extensive research has attempted to design fluorine-free mold flux, since more than 95% of the crude steel has been produced through the process of continuous casting technology [18]. The substitutes for fluorides in mold flux are primarily focused on TiO₂, B₂O₃, and Na₂O [19,20,21]. Although many industrial trials in fluorine-free mold flux were carried out and some claimed to successfully replace the conventional fluorine-bearing mold flux, while the majority of study is only limited to crack-insensitive steels [18]. Very few studies about fluorine-free refining slag have been conducted for the axle steel, which is a special railway axle steel. Therefore, the development of fluorine-free refining slag is of great significance to the sustainable and cleaner production of axle steel smelting.

The fluidity characteristic of slag, which contains the melting temperature and viscosity, is an important physicochemical property that determines the stability and productivity in steelmaking [22]. The choice of the initial composition of synthetic slag could have important implications for the secondary refining process [23]. Results showed that viscosity decreased with increasing MgO content (1–8 wt.%) and the mass ratio of CaO/Al₂O₃ (0.7~0.9) in the CaO-Al₂O₃-MgO slag system [24]. Moreover, the melting temperature increases first and then decreases with the increase of CaO/Al₂O₃ from 1.17 to 1.52 in the CaO-Al₂O₃-10 wt.% SiO₂-based slags [22]. Therefore, a better understand the relationship between the properties of refining slag and components and the design of a fluorine-free LF slag (FFS) for cleaner production of EMU axle steel is the motivation for carrying out this work [25,26].

To achieve the goal of sustainable development, development at the expense of the environment can no longer be allowed. A fluorine-bearing LF slag (FBS) from a steel company in China was studied and redesigned in this work. The production route of the axle steel is EAF → LF → VD → CC. The effects of MgO, SiO₂ and w(CaO)/w(Al₂O₃) of refining slag on the physical and metallurgical properties were theoretically and experimentally investigated. Furthermore, an FFS was designed and used in industrial-scale tests to verify its performance.

2. Research Methods

2.1. Theoretical Analysis

To explore the effects of MgO content (5–10 wt.%), SiO₂ content (2–14 wt.%), and w(CaO)/w(Al₂O₃) on the basic physical and metallurgical properties of the CaO-Al₂O₃-SiO₂-MgO-CaF₂ system, several multicomponent phase diagrams, isothermal cross-section diagrams, liquid phase area diagrams, and iso-activity diagrams were studied with the help of FactSage 8.0 software (Thermfact/CRCT, Montreal, Canada and GTT-Technologies, Aachen, Germany). The thermodynamic data of the liquid phase and pure solid phase in the slag were selected from the FactPS and FToxid databases, respectively.

2.2. Laboratory Tests

An FFS system was proposed and studied based on theoretical analysis. Moreover, the chemical composition of the FFS was further optimized based on the results of an investigation of the viscosity and melting point of the experimental slags. The chemical compositions of the experimental slags are listed in

Table 1. Chemical compositions of experimental slags (wt.%).

No.	CaO	Al2O3	SiO2	MgO	CaF2	w(CaO)/w(Al2O3)
1	46.11	36.89	10.00	7.00	-	1.25
2	48.33	38.67	6.00	7.00	-	1.25
3	50.56	40.44	2.00	7.00	-	1.25
4	43.50	43.50	6.00	7.00	-	1.00
5	52.20	34.80	6.00	7.00	-	1.50
6	55.07	30.91	1.84	7.18	5.00	1.78

, corresponding to different SiO₂ contents (No. 1, 2, and 3) and w(CaO)/w(Al₂O₃) ratios (No. 2, 4, and 5). In addition, the sample of FBS (No. 6) used by the axle steel plant (Stainless steel Co., Ltd., Taiyuan, China) was also tested under the same conditions, with the purpose of comparing the properties of the FFS and FBS slags.

High-grade pure calcium oxide (CaO), magnesium oxide (MgO), silica (SiO₂), aluminum oxide (Al₂O₃), and calcium fluoride (CaF₂) were employed to prepare experimental slags. CaO was calcinated at 1273 K in air atmosphere for 10 h to remove H₂O and CO₂. After that, the chemical reagents were uniformly mixed for the laboratory tests.

The rotating cylinder method was used to test the viscosity of the experimental slags. About 140 g of the pre-melted slag sample was put into a graphite crucible (φ 40 mm × 80 mm) lined with Mo sheet and heated to 1823 K under Ar atmosphere, which was kept for 30 min for a complete melting. A Mo spindle (φ 10 mm × 25 mm) was placed in the center of the graphite crucible and slowly lowered into the molten slag to a level 10 mm from the crucible bottom. The spindle was rotated with a speed of 200 rpm during the continuous viscosity measurement period. To ensure that the measured value is close to the equilibrium viscosity, the cooling rate of the rotating viscometer furnace (RTW-16, China) was set as 3 K·min⁻¹. The test was stopped when the viscosity of the slag was 2 Pa·s, and the viscosity and temperature were recorded continuously at intervals of 0.1 K.

The quenched slag sample obtained after the viscosity measurement was dried and crushed as a specimen for the following melting-point test. The melting property of the slag was studied by a standard method [27]. The melting point, considered to be the hemisphere point temperature, was measured by melting point and melting rate detectors. A small cylindrical 3 mm high specimen, with a diameter of 3 mm, was used for each test. The sample was placed on an alumina plate on the sample-holder, which was heated at a rate of 10 K·min⁻¹. The changes in the shape of the samples during heating were observed with a microscope. The temperature at which the sample showed a shrinkage of 50% was considered as the melting temperature. Each experiment was repeated three times and the average value of the results was taken and reported.

2.3. Industrial Scale Tests

Three sets of industrial scale tests were carried out to evaluate the smelting effects of the designed FFS.

Table 2. Chemical compositions of the slag for industrial scale tests (wt.%).

No.	CaO	Al2O3	SiO2	MgO	CaF2	w(CaO)/w(Al2O3)
#1	53.54	37.31	3.72	5.43	-	1.44
#2	57.50	31.18	7.10	4.22	-	1.84
#3	55.15	30.01	1.84	7.18	5.82	1.84

lists the chemical compositions of the slag for the industrial scale tests. The chemical composition of No. #1 is within the optimized range obtained by the laboratory study, while #2 is not but has the same w(CaO)/w(Al₂O₃) ratio as FBS (No. #3). The final (No. #3) is a slag from the axle steel plant, which has a similar chemical composition as No. 6 studied in the laboratory tests.

The industrial scale tests of axle steel were carried out in a steel plant in China. The production route is 80 t EAF → LF → VD → CC.

Table 3. Composition requirement of axle steel (wt.%).

C	Si	Mn	P	S	Cr	Mo	Ni	V	Al	Cu
0.24–0.32	0.20–0.40	0.60–0.90	≤0.01	≤0.01	0.90–1.20	0.20–0.30	0.50–1.50	≤0.06	0.01–0.04	≤0.20

shows the composition requirement of axle steel. The endpoint carbon and phosphorus contents of EAF were controlled on average 0.20 wt.% and 0.002 wt.%, respectively, for three industrial tests. For industry test No. #3, the refining slag (400 kg), fluorite (100 kg), and lime (400 kg) were added for slagging when tapping. Aluminum shot (250 kg), silicon-manganese alloy (500 kg), micro carbon ferrochrome (VCr3, 1100 kg), and ferromolybdenum (250 kg) were used for deoxidation and alloying. The tapping weight per furnace of 8.2 t. in the earlier stage of LF refining, the deoxidization process of feeding aluminum wire was applied according to the deoxidation condition. Lime and fluorite were added to make reductive slag during the LF refining. After deoxidation and slagging, silicon-manganese alloy, manganese, ferromolybdenum, and ferrovanadium were added to adjust the composition of molten steel according to the target composition of axle steel. Additionally, the industry tests of No. #1 and #2 have the same LF refining conditions, except that fluorite was not added.

The smelting results of the designed FFS was evaluated by measuring the total oxygen and sulfur content in the steel samples. Steel was sampled from two positions, including early LF and after LF. Cylindrical samples (Φ 4 mm × 5 mm) were machined from the steel sample after LF and used to analyze the total oxygen and sulfur content. The contents of sulfur in the steel sample were determined by the infrared carbon and sulfur detector (CS-2300, America). The total oxygen content was measured by oxygen and nitrogen analyzer (ON736, America). Furthermore, the cylindrical samples were embedded in epoxy resin, polished by SiC paper, and diamond suspensions for further analysis. The inclusions in the steel were analyzed by scanning electron microscopy (Phenom ProX, China) with energy-dispersive X-ray spectroscopy (SEM-EDS).

3. Results and Discussion

3.1. Theoretical Analysis

3.1.1. MgO

Figure 1 shows isothermal cross-section diagrams of CaO-Al₂O₃-SiO₂-xMgO (x = 5–10 wt.%) at 1873 K. It can be seen that MgO phase can exist in the slag when its MgO content is 6–10 wt.%. Moreover, the stable existence area of MgO expands with the increase in the MgO content, showing the same trend as reported the literature [28,29]. Magnesia-carbon (MgO-C) refractories are widely used in the ladle slag line. Therefore, the MgO content in the studied system should be controlled in the range of 6–10 wt.%, aiming to reduce the damage of the refining slag to the furnace lining during the refining process [30]. Furthermore, the influence of the MgO content on the meltability of the slag should be considered [31,32]. The effects of MgO content of the CaO-Al₂O₃-SiO₂-MgO system on the liquid phase area at different temperatures is shown in Figure 2. The liquid phase area of slag expands at 1873 K when the MgO content increases from 5 wt.% to 7 wt.%. This result indicated that the increase of MgO content (5–7 wt.%) improves the fluidity of slag. Likewise, Kim and co-workers [33] found that the addition of MgO decreased the viscosity of CaO-SiO₂-20% Al₂O₃-MgO slag system. The main reason for this is a modification of the slag network structure with the addition of MgO. By contrast, the liquid phase area shrinks as the MgO content increases at 1873 K, if the content is in the range of 7–10 wt.%. The reason for this may be that a large amount of MgO precipitates in the slag and forms a high-melting-point spinel phase, resulting in an increase in the melting point of the slag [34]. Therefore, the MgO content in FFS is suggested to be controlled between 6 wt.% and 8 wt.%, which could protect the furnace lining and give the molten phase a workable melting performance.

3.1.2. SiO₂

To investigate the effects of SiO₂ content on the melting property of the FFS, liquid-phase area diagrams of the CaO-Al₂O₃-MgO-xSiO₂ system with different SiO₂ content (2–14 wt.%) are shown in Figure 3. It can be observed that the liquid phase area expands with SiO₂ in the studied range at 1873 K. Therefore, it is beneficial to appropriately increase the SiO₂ content of the slag to improve the melting performance of the slag.

However, it has been reported the SiO₂ component in the refining slag affects the deoxidation reaction of the dissolved aluminum in the melt. As shown in Equation (1).

$$3\left({\mathrm{SiO}}_2\right)+4[\mathrm{Al}]=2\left({\mathrm{Al}}_2{\mathrm{O}}_3\right)+3\left[\mathrm{Si}\right]$$

(1)

The SiO₂ in the slag can oxidize the dissolved elemental aluminum in molten steel, subsequently results in lowering the deoxidization degree of the melt. Therefore, the activity of SiO₂ in the slag should be reduced. Figure 4 shows iso-activities diagram of SiO₂ in the CaO-Al₂O₃-SiO₂-10 wt.% MgO system at 1873 K. The data was selected from the Slag Atlas [28] (black solid lines) and the FactSage 8.0 software database (red solid lines), respectively. There is a difference between the activity of SiO₂ calculated by the two databases. It can be seen in Figure 4 that the activity of SiO₂ is low when its content in the slag is less than 10 wt.%, which could prevent the oxidation of dissolved aluminum in the melt. Since some SiO₂ could be brought from the raw materials and tapping of the EAF, the content of SiO₂ of FFS is suggested to be controlled within the range of 2–10 wt.%.

3.1.3. Mass Ratio of w(CaO)/w(Al₂O₃)

It is known from the binary phase diagram of CaO-Al₂O₃ and earlier experiences that the two components can react under LF conditions and form low-melting-point compounds, such as 12CaO·7Al₂O₃, affecting the metallurgical and melting properties of the slag. Figure 5 shows the pseudo ternary phase diagram of CaO-Al₂O₃-SiO₂-10 wt.% MgO, which shows that when the SiO₂ content is 2–10 wt.%, the liquid phase area at 1873 K is within the w(CaO)/w(Al₂O₃) range of 0.66–1.50. In the LF process, the refining slag absorbs the Al₂O₃ inclusions in the melt, causing a decrease in w(CaO)/w(Al₂O₃), and the absorption capacity of the slag has a stronger relationship with the Al₂O₃ and CaO activities in the molten phase. Therefore, the iso-activity curves of Al₂O₃ and CaO were plotted to find a means to further optimize the wt. ratio w(CaO)/w(Al₂O₃).

Figure 6a,b show the iso-activity diagrams of Al₂O₃ and CaO in the CaO-Al₂O₃-SiO₂-10 wt.% MgO system, respectively, using data selected from the Slag Atlas (black solid line) and the FactSage 8.0 software database (blue solid line). The diagrams show that when the content of SiO₂ in the slag is constant, the activity of Al₂O₃ in the melt decreases, and the activity of CaO increases with the w(CaO)/w(Al₂O₃) ratio [35]. Therefore, a high w(CaO)/w(Al₂O₃) ratio in a workable range is beneficial for the absorption of Al₂O₃ inclusions.

3.2. Laboratory Tests

Based on the discussion above, the preferable theoretical ranges for the chemical composition of the FFS for axle steel is proposed as w(CaO) = 40–55 wt.%, w(SiO₂) = 2–10 wt.%, w(Al₂O₃) = 30–50 wt.%, w(MgO) = 6–8 wt.%, and w(CaO)/w(Al₂O₃) = 1.00–1.50. To further optimize the chemical composition of the FFS, the viscosity and melting point were determined. The chemical compositions studied are listed in Table 1.

3.2.1. SiO₂

Viscosity-temperature curves of the samples are shown in Figure 7. All samples are seen to present a similar trend, where the viscosity of the slag slowly increases when the temperature decreases, but is above the break temperature. When the temperature decreases beyond the break temperature, the viscosity increases strongly.

Figure 8 shows the viscosity-temperature bar graph of the experimental slags with different SiO₂ contents. The viscosity of the samples increases as the SiO₂ content increases from 2 wt.% to 10 wt.% at a constant temperature, which indicates that the increase in SiO₂ content becomes rate-limiting for the mass transfer at the slag-melt interface region. The reason for this phenomenon is that SiO₂ increases the tendency to network formation and the degree of polymerization of the refining slag [36,37]. Furthermore, comparing the viscosity values of the designed FFS and industrial FBS at the same temperature, it is seen that when the SiO₂ content is 2 wt.%, the viscosity is less than that of the industrial FBS, and increases with the SiO₂ content. As the SiO₂ content reaches 6 wt.%, the viscosities of FFS and FBS show similar values. A higher SiO₂ content, e.g., 10 wt.%, may seriously affect the fluidity of the slag.

Figure 9 shows the melting point of slags of various contents of SiO₂ (2–10 wt.%). When the SiO₂ content increases from 2 wt.% to 10 wt.%, melting point of slag decreases slightly (from 1619 K to 1608 K), implying that the SiO₂ has little effect on the melting point of the slag. The melting point of the designed FFS was lower than for the industrial FBS and much lower than the refining temperature (1893–1913 K). Based on these results, one may conclude that the preferable SiO₂ content of FFS is 2–6 wt.%.

3.2.2. Mass Ratio w(CaO)/w(Al₂O₃)

The viscosity of experimental slags with various ratios of w(CaO)/w(Al₂O₃) was summarized in Figure 10. As shown in Figure 10, the viscosity shows an increased trend with the decrease of temperature when the ratio of w(CaO)/w(Al₂O₃) is constant. Results also indicated that the viscosity decreased with the ratio of w(CaO)/w(Al₂O₃) of slag at the constant temperature. When this ratio increased from 1.00 to 1.50, the viscosity decreased from 0.29 Pa·s to 0.10 Pa·s. The explanation of this phenomenon is attributed to that more free oxygen was introduced into the molten slag with increasing w(CaO)/w(Al₂O₃) [38]. It has been reported that the (AlO₄)⁵⁻ tetrahedron is a main network former, and free oxygen can disrupt the network structure of molten slag by cutting off the bridging oxygen bond, and thereby decreases the slag viscosity [7]. In addition, it can be found that the viscosity of FFS was less than or proximately equal to the industrial FBS when the ratio of w(CaO)/w(Al₂O₃) was 1.25 or 1.50 and gave the opposite result when the ratio was 1.00 in the same temperature conditions.

Figure 11 shows the melting point of slags with various ratios of w(CaO)/w(Al₂O₃). It can be observed that when the ratios of w(CaO)/w(Al₂O₃) of slag increases from 1.25 to 1.50, the melting point increases less than 2 K. Moreover, the melting point of FFS was much lower than that of industrial FBS and the actual refining temperature (1893–1913 K). It can be speculated that the melting property of the designed FFS in the LF process may be better than the currently used FBS. Therefore, the w(CaO)/w(Al₂O₃) in FFS was further optimized to 1.25–1.50.

According to the laboratory results, the optimal composition range of components of FFS is w(CaO) = 40–55 wt.%, w(SiO₂) = 2–6 wt.%, w(Al₂O₃) = 30–40 wt.%, w(MgO) = 6–8 wt.%, and w(CaO)/w(Al₂O₃) = 1.25–1.50.

3.3. Industrial Scale Tests

Steel samples were obtained using three kinds of slag. The three types of slags are the chemical composition of slag that is within (No. #1) and not (No. #2) in the optimized range and fluorine-containing slag (No. #3) obtained from axle steel plant (Stainless Steel Co., Ltd., Taiyuan, China), respectively. The compositions of slags correspond to Table 2. The total oxygen and sulphur contents in different steel samples after the LF process are shown in Figure 12. It can be concluded that the FFS (#1 and #2) have a similar desulphurizing capacity as FBS (#3). According to the product standard of the plant, the S content of axle steel should not be higher than 0.001 wt.%. Thus, the desulphurizing capacities of all three refining slags meet the requirement. Furthermore, the total O content in the steel sample controlled by optimized FFS (#1) was slightly higher than FBS (#3), while for steel sample #2 was much higher than for #1 and #3. The main reason for this is that the CaO content and w(CaO)/w(Al₂O₃) of #2 slag are higher than for #1, which may negatively affect the melting performance and fluidity [39]. Therefore, the results show that optimized FFS can replace FBS, which reduces the risk of environmental pollution while ensuring the smelting performance.

Morphology and chemical composition of inclusions in the steel samples before and after the implementation of the LF process were investigated by SEM-EDS, and the results are shown in Figure 13 and

Table 4. Inclusion composition in SEM images (atomic%).

Point	Al	O	Mg	Ca
+1	27.96	72.04	-	-
+2	25.95	64.03	10.02	-
+3	8.29	46.12	44.45	1.15
+4	3.86	45.86	50.30	-
+5	25.27	55.85	4.40	14.77

. The shape of the inclusions in the original LF is, in Figure 13a,b, seen to be angular. The EDS results indicate that the inclusions were mainly Al₂O₃ as a result of melt deoxidation by Al, and a small amount of magnesium aluminum spinel. The Al₂O₃ inclusions were generally <3 μm, while the size of magnesium aluminum spinel inclusions was generally 2–6 μm. The inclusions in the steel samples after the LF process changed to MgO-Al₂O₃ and CaO-MgO-Al₂O₃ system phases (cf. Figure 13c,d), which were formed by chemical reactions between MgO, CaO, and Al₂O₃ during the refining process [40,41]. In addition, inclusions will decrease the steel quality from multiple aspects. The most evident factor that affects fatigue property is the inclusion size. The average size of inclusions in the steel sample is ~2–6 μm, which is much lower than the size observed near the fatigue crack initiation site (12.5–33.2 μm) [42].

4. Conclusions

To avoid the use of CaF₂ in LF slag in the production of axle steel in the future, a fluorine-free refining slag (FFS) was designed through theoretical analysis and laboratory test. The effects of MgO, SiO₂, and w(CaO)/w(Al₂O₃) on the physical and metallurgical properties of the slag were also studied. Based on the theoretical analysis, it is beneficial to protect the furnace lining and prevent the aluminum dissolved in the melt from oxidizing on the condition of w(SiO₂) = 2–10 wt.% and w(MgO) = 6–8 wt.%, respectively. By integrating theoretical analysis and laboratory test, the results showed that for designing an appropriate slag, the range of components of FFS should be w(CaO) = 40–55 wt.%, w(SiO₂) = 2–6 wt.%, w(Al₂O₃) = 30–40 wt.%, w(MgO) = 6–8 wt.%, and w(CaO)/w(Al₂O₃) = 1.25–1.50. Industrial scale test results demonstrated that the designed FFS has similar deoxidation and desulfurization capabilities as industrial fluorine-bearing refining slag (FBS) and that it therefore meets the requirements for treating axel steel. Use of the designed FFS could protect the environment from fluorine pollution in the LF process of axle steel making.