Desulphurisation and Inclusion Behaviour of Stainless Steel Refining by Using CaO–Al₂O₃ Based Slag at Low Sulphur Levels

Abstract

The deep desulphurisation behaviour and steel cleanliness of stainless steel refining in the low sulphur content period (below 45 ppm [%S]) by using the current lime-fluorspar based slag and an optimized lime-alumina slag have been studied and compared. The desulphurisation effect was discussed based on the calculated sulphide capacity of slags and the equilibrated sulphur distribution ratio between the slag and steel. The sulphur evolution during the treatment was predicted with a previously developed kinetic model. The influence of slag chemistry on the compositional evolution of the steel and inclusions was discussed through thermodynamic considerations with respect to deoxidation and desulphurisation reactions. The sulphur content in the inclusions was predicted using the dissolved sulphur in the steel and their oxide chemistry. An equivalent steel cleanliness and desulphurisation effect was obtained with lime-alumina based slag as with lime-fluorspar based slag both in laboratory and industrial tests for low sulphur content stainless refining.

1. Introduction

Non-metallic inclusions and their characteristics, i.e. number density, size distribution, morphology and chemistry, have a complex impact on stainless steel properties. For instance, inclusions with high number density and large size can shorten the bearing and fatigue life of steel products.^1,2) Oxide inclusions, e.g. Al₂O₃, SiO₂ and TiO₂ with irregular, dendritic and clustered morphology have more detrimental effects.^3,4) These detrimental effects can be reduced by coating the inclusions with a soft, low melting temperature layer, which normally consists of MnS.^4,5) The presence of sulphur can also ensure adequate weld penetration of stainless steel and therefore improve its performance.⁶⁾ On the other hand, sulphides can induce pitting corrosion and weaken the corrosion resistance of stainless steel.^2,3,4,5) To utilize the beneficial effects of the sulphide inclusions and to depress their detrimental effects, it is essential to control the sulphur content and sulphides formation in stainless steel production.

Ladle refining plays a crucial role on impurities removal and steel cleanliness control in modern steelmaking. Nowadays, a CaO–CaF₂–SiO₂ (CSF) based top slag is being currently used to quickly remove sulphur and inclusions from liquid steel.^7,8,9) However, the valorisation of this slag poses a tough challenge because of its disintegration in the post-metallurgical process.¹⁰⁾ The CaO–Al₂O₃ (CA) based top slag, due to excellent refining and valorisation properties,¹¹⁾ has been considered as a potential substitute. In the first phase of our studies,^12,13) the desulphurisation of stainless steel with 150 ppm initial S by using the current CSF slag and various CA based slags has been performed to prove the concept of substitution. An equivalent desulphurisation efficiency, e.g. a comparable sulphur distribution ratio between slag and steel, and a similar final sulphur level (30–50 ppm) in the steel was obtained with an optimized CA based synthetic slag (with 50% CaO, 40.4% Al₂O₃, 4.3% SiO₂, 2.1% Fe₂O₃, 1.7% TiO₂ and 0.6% MgO) as compared with the current CSF slag. The use of the optimized CA based slag also resulted in an improved steel cleanliness, e.g. lower inclusion number density, smaller size and fraction area and less sulphide inclusion formation. We, therefore, concluded that the proof-of-concept of using the CA based slag for stainless steel ladle refining has been achieved.

In order to make the substitution industrially possible, further investigations, i.e. laboratory simulation and plant test with practical conditions are needed. After the AOD process, the liquid steel usually contains 30–50 ppm sulphur and is treated in a ladle furnace. Large amount of alloys, which contain considerable sulphur content are added in liquid steel during the treatment. Top slag is therefore used to remove/control the sulphur level. This so-called deep desulphurisation is essential for producing low sulphur stainless steel products, and it is highly sensitive to the secondary refining (ladle treatment) conditions, i.e. the deoxidation and reoxidation of liquid steel, the melting rate and the kinetic conditions of slag. In this context, deep desulphurisation of stainless steel was studied based on laboratory scale under industrial conditions (i.e. 30–45 ppm initial sulphur content and doloma refractory crucible) and plant scale by using the current CSF and the optimized CA synthetic slags. The application feasibility of the optimized CA synthetic slags in ladle treatment of stainless steel is discussed. Specific attention has been given to the desulphurisation behaviour and steel cleanliness.

2. Experimental

2.1. Materials

To simulate the secondary stainless steel refining process, austenitic stainless steel ingots were collected after the AOD process from a steel plant and used in the laboratory experiments. Tables 1 and 2 respectively show the chemical composition of the steel ingot and the top slags used in the tests. The CSF slag, a mixture of lime (containing 97% CaO, 1.5% SiO₂ and 1.5% MgO) and fluorspar (containing 92.3% CaF₂, 2.6% CaO and 5.1% SiO₂), is being currently used during ladle treatment in the stainless steel plant. The CA slag is the optimized CaO–Al₂O₃ based synthetic slag, which contains small amount of SiO₂, Fe₂O₃, MgO and TiO₂ as additives.

Table 1. The chemical composition of austenitic stainless steel used in the tests (wt%).

C	Mn	P	S	Si	Cr	Ni	Mo	Cu	Ti	Co
0.03	1.0	0.03	0.0030–0.0045	0.30	16.9	7.0	0.28	0.36	0.001	0.16

* C and S were measured with LECO combustion analysis, and other elements were measured with inductively coupled plasma atomic emission spectroscopy (ICP-AES).

2.2. Experimental Procedure

2.2.1. Laboratory Scale

The experiments were performed in a vacuum induction furnace (type VSG 30, 60 kW power supply and 4 kHz frequency). The details of the experimental set-up have been given elsewhere.^12,13) The experimental procedure and conditions are shown in Fig. 1 and Table 2, respectively. Fifteen kilograms of austenitic stainless steel ingot were melted under Ar atmosphere in a crucible with an inner diameter of 150 mm, outer diameter of 176 mm and the height of 275 mm. Here, the crucible was made with doloma bricks, which is being used in the industrial ladle as refractory material. The temperature of molten steel was controlled around 1650°C. Thereafter, the oxygen activity and temperature of the molten steel were measured with a Celox oxygen sensor with type B thermocouple probes (provided by Heraeus Electro-Nite). Subsequently, a steel sample in conical shape (the upper diameter is 20 mm, lower diameter is 10 mm and the height is from 16 to 25 mm) was taken by dipping the spoon sampler into the molten bath and rapidly withdrawing it, followed by quenching in air. 150 g slag was added to the molten steel through the loading chamber; meanwhile 75 g FeSi alloy was added to molten steel in Tests 1-2, while 3 g Al was added in Tests 3-4. The slags were added in 4 steps in amounts of 150 g each and in total 5 steel samples were taken according to the scheme as shown in Fig. 1. At the end of the test, the oxygen activity and temperature of the molten steel were measured, and then the melt was poured into a mould. The slag sample was collected after the mould was cooled down.

Fig. 1.

Experimental procedure of laboratory test.

Table 2. The chemical composition of top slag and laboratory experimental conditions.

Test	[S] in steel (ppm)	Slag type	Chemical composition (wt%)							Alloy addition (g)
			CaO	Al2O3	Fe2O3	SiO2	MgO	TiO2	CaF2	FeSi	Al
1	43	CSF	77.9	–	–	6.8	1.2	–	14.1	75	0
2	27
3	43	CA	50.0	40.4	2.1	4.3	0.6	1.7	–	0	3
4	28

* Measured with X-ray fluorescence spectrometer (XRF).

2.2.2. Plant Scale

After scrap smelting in the EAF and decarburization in the AOD process, liquid melts (120 ton) were tapped into a ladle where the AOD slag was removed. Top slag and deoxidizer were then added during the ladle treatment, followed by alloying and scrap addition depending on the specification of steel grade. Thereafter, the liquid steel was prepared for casting. The composition and addition of alloy elements are listed in Table 3. It should be noted that a relatively larger amount of CA slag addition in Test 6 was due to the fact that (1) the high initial sulphur content in molten steel (i.e. 43 ppm) and (2) the high sulphur content introduced by alloy addition, i.e. 0.13 kg sulphur introduced in Test 6 as compared to 0.05 kg in Test 5. Taking into account the initial steel sulphur content and sulphur pick-up during alloying, the required slag addition quantity was calculated based on the Lab experimental results to reach a necessary desulphurization level. Steel samples were taken at different stage of the ladle treatment (Fig. 2), viz. (1) after deslagging (AOD slag); (2) after top slag addition; (3) just prior to sending the ladle to the caster.

Table 3. Industrial conditions, i.e. slag amount, ferro-alloy addition and the total sulphur content of the ferro-alloys.

Test	Slag		Addition (kg)			T.S (ppm)
	Type	Amount (kg)	FeMnSi	FeTi	FeMo	FeMnSi	FeTi	FeMo
5	CSF	1600	266	15	–	164	221	614
6	CA	2200	90	11	191

* T.S measured with LECO combustion analysis.

Fig. 2.

Sampling procedure of industrial test.

2.3. Sample Analysis

The steel samples obtained during the tests were cut into three parts. The first part of the steel sample was used for total sulphur measurement with a LECO combustion analysis (type CS-444). The detailed method for total sulphur analysis in steel and that for slag have been already explained elsewhere.^12,13) The second part of the steel sample was prepared for inclusion characterization, which was conducted with a high resolution scanning electron microscope (FEI XL-40 LaB6). The SEM was equipped with an EDAX energy dispersive spectrometer (EDS) detector and NSS feature software.¹³⁾ The latter can be used to perform automatic inclusion analysis (AIA). The maximum working magnification was set to 1000 and the effective minimum detectable inclusion area was about 0.5 μm². The third part of the steel sample was used for inclusion morphology assessment by the inclusion extraction technique,¹³⁾ as well as for steel composition assessment using inductively coupled plasma atomic emission spectroscopy (ICP-AES, Varian Liberty series II instrument with an axial plasma configuration).

3. Results and Discussion

3.1. Desulphurisation Behaviour

3.1.1. Desulphurisation Results

The sulphur content in steel and slag samples, as well as the sulphur distribution ratio between the slag and steel, i.e. L_S=(%S)/[%S] are listed in Table 4. Figure 3 shows the evolution of sulphur content in molten steel as a function of time after the slag addition. It can be seen from Fig. 3(a) (lab scale) that (1) sulphur was continuously removed during the treatment, indicating the desulphurisation effect of both CSF and CA slags; (2) sulphur removal was fast in the early stage, then dropped slowly or levelled off, showing the difficulty of deep desulphurisation; (3) a relatively lower final sulphur content was obtained in the tests with low initial sulphur content. Figure 3(b) (plant scale) shows the comparison of desulphurisation effect by using the current CSF and the optimized CA slag in an industrial ladle treatment. It can be seen that (1) sulphur was also quickly removed after the top slag addition; (2) alloying and scrap addition resulted in a considerable resulphurisation of liquid steel. This is due to the high sulphur content in alloying elements (Table 3); (3) a similar level of sulphur content was achieved by using CSF and CA slag at the end of the treatment.

Table 4. The sulphur content in steel and slag samples, the oxygen activity and the temperature of the molten steel, as well as the sulphur distribution ratio between the slag and steel.

Sample No.	Slag type	Time after slag addition (min)	[S] (ppm)	(S) (ppm)	a[O] (ppm)	Temp. (°C)	LS
T1-1	CSF	0	43	0	37	1655
T1-2		15	20
T1-3		30	13
T1-4		40	7
T1-5		55	5	1053	62	1656	211
T2-1	CSF	0	27	0	22	1638
T2-2		15	14
T2-3		30	5
T2-4		45	4
T2-5		55	4	552	18	1641	138
T3-1	CA	0	43	444	22	1649
T3-2		14	35
T3-3		24	29
T3-4		37	18
T3-5		49	16	951	15	1630	59
T4-1	CA	0	28	444	14	1645
T4-2		19	17
T4-3		34	11
T4-4		44	8
T4-5		54	7	949	14	1635	136
T5-1	CSF	0	33			1665
T5-2		15	28
T5-3		55	41			1570
T6-1	CA	0	43			1642
T6-2		15	29
T6-3		55	34			1570

Fig. 3.

The evolution of [S] in molten steel as a function of time after slag addition: (a) laboratory experiment and (b) plant scale experiment.

The relation between sulphur distribution ratios, viz. the mass balance ratio ( $L_{\text{S}}^{\text{mb}}{\text{=(\%S)}}^{\text{mb}}\text{/[\%S]}$ ) and the measured ratio for final samples are shown in Fig. 4. Here, (%S)^mb is the sulphur content in slag phase obtained by considering the mass balance between slag and steel. In this case, the (%S)^mb is calculated by assuming that the sulphur removed from the steel moves to the slag phase. The amount of 15 kg steel and 600 g slag are used for the calculation. Figure 4 shows that the sulphur mass balance distribution ratio agrees with the measured one, confirming the reliability of the present experimental data.

Fig. 4.

The relation between the measured sulphur distribution ratio and mass balance values.

From the above desulphurisation results (i.e. Fig. 3 and Table 4), it can be concluded that a better desulphurisation effect, i.e. lower final sulphur content and higher L_S were achieved with CSF based slag. A slightly worse desulphurisation result was obtained with CA slags on laboratory test (Tests 2 and 4 in Fig. 3(a)), while a comparable results were still achieved on plant scale test (Tests 5 and 6 in Fig. 3(b)). In order to understand this difference of deep desulphurisation effect with distinct slag chemistry better, the desulphurisation thermodynamics need to be discussed.

3.1.2. Desulphurisation Thermodynamics

Sulphide capacity (C_S), a function of slag chemistry and experimental temperature (Eq. (1)),^14,15) is used to describe the sulphur absorption capability of slags. Since the sulphide capacity of a slag is only related with the slag chemistry and temperature, it can be calculated with the aid of FactSage software by using the experimental results (Tables 3, 4, 5). The detailed calculation method has been already described elsewhere.¹²⁾ The calculated sulphide capacity values (C_S) for the initial and the final slags are listed in Table 6. It can be seen that the sulphide capacity of CSF slag is much higher than that of CA slag; additionally, a decrease in slag sulphide capacity was found after the tests. The former is resulting from the high CaO content in the CSF slag (Table 5), providing a large amount of free oxygen ions (O^2–) for the desulphurisation reaction between steel and slag (reaction (2)). While the later implies that the desulphurisation capability of the slags decreased as the reaction proceeds. This is because of the Si oxidation, resulting in an increase of SiO₂ content in slag, which combines with the free oxygen ions (O^2–) to form ${\text{SiO}}_{\text{4}}^{\text{4}-}$ tetrahedral units.   

$$C_{\text{S}}=exp\left(-\frac{\mathrm{\Delta }{G}^0}{RT}\right)\cdot \left(\frac{a_{(O^{2-})}}{f_{{\text{(S}}^{\text{2}-}\text{)}}}\right)$$(1)

  

$${\text{(O}}^{\text{2}-}\text{)}+\text{[S]}={\text{(S}}^{\text{2}-}\text{)}+\text{[O]}$$(2)

where $a_{(O^{2-})}$ is the activity of oxide ions in the slag; $f_{{\text{(S}}^{\text{2}-}\text{)}}$ the activity coefficient of sulphide ions in the slag; R the gas constant, 8.314 J/(mol·K); T the temperature in K; ΔG⁰ the standard Gibbs energy of the reaction (3),¹⁴⁾ J/mol:   

$$\begin{array}{c}\frac{1}{2}{\mathrm{S}}_2(\mathrm{g})+{\text{(O}}^{\text{2}-}\text{)}={\text{(S}}^{\text{2}-}\text{)}+\frac{1}{2}{\mathrm{O}}_2(\mathrm{g})\\ \Delta G^O=118\mathrm{\hspace{0.17em} }535-58.8157T\end{array}$$(3)

Table 5. The chemical analysis of slag after the desulphurisation treatment (wt%).

Test No.	Slag type	CaO	SiO2	MgO	Al2O3	Cr2O3	Fe2O3	MnO	CaF2
1	CSF	48.8	18.2	12.0	9.7	–	–	–	10.6
2		51.9	14.3	10.5	10.1	–	–	–	13.7
3	CA	41.7	10.8	7.1	35.9	0.8	1.3	0.4	–
4		46.1	6.7	6.8	36.5	0.8	1.4	0.1	–

Table 6. The calculated sulphide capacity for initial and final slags.

Slag type	CAF		CA
Test No.	1	2	3	4
Initial slag	1.5×10–2	1.5×10–2	2.8×10–3	2.7×10–3
Final slag	1.0×10–2	1.4×10–2	1.2×10–3	2.7×10–3

Figure 5 shows the relation between the measured sulphur distribution (L_S) and the calculated C_S. Although scatter is found in Fig. 5, the data points follow approximately a straight line with slope of 1, implying the sulphur distribution ratio was mainly influenced by the sulphide capacity of the slag. This can be understood by the equilibrated sulphur distribution ratio between slag and steel (Eq. (4)), which is obtained by relating the slag sulphide capacity (Eq. (1)) with the desulphurisation reactions (2) and (5).¹⁵⁾   

$$log{L}_S=log{K}_5+log{C}_S+log{f}_{[\mathrm{S}]}-log{a}_{[\mathrm{O}]}$$(4)

  

$$[\mathrm{S}]+\frac{1}{2}{\mathrm{O}}_2(\mathrm{g})=[\mathrm{O}]+\frac{1}{2}{\mathrm{S}}_2(\mathrm{g}),\mathrm{\hspace{0.17em} }log{K}_5=-\frac{2\mathrm{\hspace{0.17em} }154}{T}+3.166$$(5)

in which   

$$log{f}_{[\text{S}]}=\sum e_{\text{S}}^i[\%i]$$(6)

where K₅ is the equilibrium constant of the reaction (5); T is the temperature in K; f_[S] is the activity coefficient of sulphur in molten steel¹⁶⁾ and a_[O] is the oxygen activity in molten steel; i represents dissolved element i in molten steel; and $e_{\text{S}}^i$ is the interaction parameter of i for sulphur (see Table 7).

Fig. 5.

Relation between the measured sulphur distribution ratio L_S and the calculated sulphide capacity C_S.

Table 7. Interaction parameters used in the calculation.^17,18,19)

j	i
	O	Si	Cr	Mn	S	Al
S	0.27	0.065	0.01	–0.025	–0.028	–
Al	–34740/T+11.95	0.056	0.0096	0.035	–	0.044
Si	–0.119	0.103	–	0.033	–	0.058
Mn	–0.083	–	0.0042	–	–	–

It can be seen from Eq. (4) that under thermodynamic equilibrium conditions the calculated log L_S and log C_S have a linear relation with the slope of 1, which was verified in the present work as shown in Fig. 5. In addition to the slag C_S, which depends on slag chemistry and operation temperature, other parameters, such as steel chemistry also have a complex influence on sulphur distribution ratio due to their effect on the sulphur and oxygen activities in molten steel (Eq. (6)). By using the measured slag chemistry, steel chemistry and experimental temperature, the equilibrated L_S can be calculated through Eq. (4).

Figure 6 shows comparison of the measured L_S and the calculated equilibrated L_S, in which the solid squares are obtained by using the chemistry of the initial slags and the solid circle by that of the final slags. The data points are more close to the one by one relation (dash line) when using the chemistry of the final slags than that of initial slags, implying that the slag-steel interaction with respect to desulphurisation was approaching the equilibration during the treatment. In the end of the experiment, however, the calculated equilibrated L_S values are still larger than the measured values, meaning the desulphurisation equilibrium between slag and steel has not been reached. Therefore, the desulphurisation capacity of the slag had not been fully utilized during the treatment. In other words, the desulphurisation effect can be improved by using the present slags if kinetics allows.

Fig. 6.

Comparison between the measured L_S and calculated equilibrated L_S.

3.1.3. Desulphurisation Kinetics

A kinetic model for the stainless steel desulphurisation has been developed in our previous work¹²⁾ based on the two film theory.²⁰⁾ In the kinetic model, the dissolved sulphur was assumed to be removed by the slag through ion exchange, viz. O^2– and S^2– at the slag/steel interface and the sulphur transfer from the interface towards slag phase was considered to be the rate-limiting step. The desulphurisation kinetics can be expressed as follow:   

$$\begin{array}{l}ln\left\{\left(L_S+\frac{W_m}{W_s^{}}\right){[\%S]}^t-\frac{W_m}{W_s^{}}[\%S]°-(\%\text{S})°\right\}\\ =-\frac{A\cdot {\rho }_s\cdot k_s}{W_m}\left(L_S+\frac{W_m}{W_s^{}}\right)t+ln\left\{L_S\cdot [\%S]°-(\%\text{S})°\right\}\end{array}$$(7)

where A is slag/steel contact area, m²; W_m and W_s are, respectively, the weight of molten steel and slag at time t, kg; [%S]° and [%S]^t are, respectively, the sulphur content in steel at time = 0 and t, ppm; (%S)° is the sulphur content in slag at time = 0; k_S is the total mass transfer coefficient of sulphur in slag phase, m/s; ρ_s is the density of slag at experimental temperature, kg/m³; and L_S is the calculated equilibrated sulphur distribution ratio. During the calculation, the mass transfer coefficient of 1.2×10^–6 m/s¹²⁾ was used in this work. Since the slag only covered 1/4 of the molten steel surface during the treatment,¹²⁾ the contact area between the slag and steel is considered to be 1/4 of the cross section area of the crucible. The detailed parameters used in the calculation are listed in Table 8.

Table 8. The parameters and experimental conditions used in the calculation of Eq. (7).

Test	slag	LS	ρ (g/cm3)	A (m2)	k (m/s)	[S]° (ppm)
1	CSF	721	2.80	4.4×10–3	1.2×10–6	45
2	CSF	800				27
3	CA	300				43
4	CA	450				28

Figure 7 shows the comparison between the calculated and experimental data. The good agreement between calculated and measured sulphur evolutions shows a promising application of this model in practice. It can be seen from Table 7 and Fig. 7 that: (1) for the same slag, a lower initial sulphur content in the steel and a high equilibrated sulphur distribution ratio between slag and steel resulted in a better desulphurisation effect (see the comparison between Tests 1 and 2 and that between Tests 3 and 4). This is apparently due to the influence of the thermodynamic driving force on the desulphurisation reaction; (2) for a given test, the desulphurisation rate (i.e. slope of the calculated curve in Fig. 7) decreases with decreasing sulphur content in the steel. This is more obvious for the tests with CSF slag (Tests 1 and 2); (3) with a similar initial sulphur content in the steel and mass transfer coefficient of sulphur in the slag phase, the CSF (Tests 1 and 2) slag shows a relatively better desulphurisation effect than the CA slag. However, comparable final sulphur content and sulphur removal speed (i.e. Test 4) were still achieved with the synthetic CA slag as with the CSF slag in case of a low initial sulphur content condition (see results of Tests 2 and 4 in Fig. 7). Based on the results of both the industrial and laboratory tests, it can be concluded that the present synthetic CA slag is feasible for the low sulphur stainless steel refining process. (4) the fitted mass transfer coefficients of sulphur in slag phase is 1.2 × 10^–6 m/s, which is in the same order of magnitude obtained by other researchers,^8,12,21) suggesting the reliability of the present kinetic model.

Fig. 7.

Experimental and modelling results for the [S] evolution during treatment.

3.2. Steel Cleanliness

3.2.1. Inclusion Population, Morphology and Chemistry

The steel cleanliness with respect to inclusion population and chemistry are measured with SEM-AIA technique. The upper part of Fig. 8 shows the evolution of the inclusion size after the slag addition. For a given test, there is no considerable change in the average inclusion size, and no significant effect of the slag chemistry on inclusion size evolution was observed, while the maximum size of inclusions are found to be decreased during the treatment. The latter is attributed to the flotation of large inclusions. The evolution of inclusion area fraction, i.e. F.A., during the treatment is shown in the lower part of Fig. 8, where F.A. represents the inclusion area fraction, which is the ratio of the observed area of inclusions to the measured area of steel matrix. It decreased in all the laboratory tests (Fig. 8(a)), while it slightly increased in the industrial treatment. The former shows that the inclusion flotation and absorption occurred with all types of slag, and the latter is the consequence of slag entrapment and reoxidation due to strong stirring and alloying operation. Similar inclusion area fractions were obtained in the final steel samples by using CSF and CA slags both in laboratory and industrial scale.

Fig. 8.

The evolution of inclusion size and area fraction (F.A.) as a function of treatment time.

Figure 9 shows the population density functions (PDF) of inclusions, which is obtained by dividing the frequency of inclusions in a given bin (inclusion number per area) with the bin width.²²⁾ No considerable difference can be found between PDFs of final samples, indicating that a similar inclusion size distribution is obtained with CA and CSF slags. The PDF curves calculated with experimental data (based on probability density function) in Fig. 9 are approximately quadratic, suggesting that the size distribution can be represented with log-normal PDFs. On the basis of the analysed results of inclusion size, area fraction and size distribution, it can be concluded that an equivalent steel cleanliness was obtained with the optimized CA as with CSF based slag.

Fig. 9.

Inclusion size distribution determined with the SEM-AIA technique for the final steel samples.

Figure 10 shows a typical evolution of the inclusions during the treatment of Test 4. The SiO₂–MnO–Cr₂O₃ complex inclusions in spherical and angular shape are already present in the steel before the treatment (Fig. 10(a)). Apparently, they are the result of a reduction operation at the end of AOD process, where FeSi is added to recover Cr from high chromium oxides containing slag. Small holes can be found on these original inclusions, which are probably due to the MnS dissolution during the inclusions extraction.¹³⁾ Al₂O₃ based inclusions with plate-like shape are formed after the slag and alloy addition, while the spherical inclusions containing Al₂O₃–SiO₂–MnO can still be observed in the samples (Fig. 10(b)). Pure Al₂O₃ inclusions in octahedral shape and a small amount of Al₂O₃–MgO spinel were observed in the final samples (Fig. 10(c)). This evolution of inclusions during the experiment is believed to be the result of the slag-steel-refractory-inclusions interaction.

Fig. 10.

The inclusion morphology evolution in Test 4: (a) before the slag and alloy addition, (b) during the treatment and (c) at end of the treatment.

For each steel sample, at least 100 inclusions were measured with the SEM-AIA and the average composition was then obtained. Since the oxygen content in the oxysulphide inclusions is not directly measured with EDS, it is assumed that all measured Si, Al and Cr are in the form of SiO₂, Al₂O₃ and Cr₂O₃, respectively. All S in the oxysulphide inclusions is assumed to form MnS first. The remaining Mn content after the formation of MnS is considered to be in the form of MnO. Table 9 shows the analysed composition of inclusions and that of steel samples (with ICP-AES), in which the symbol of [ ] denotes the dissolved element in molten steel. The compositional evolution of inclusions during the treatment is illustrated in Fig. 11. In order to fix the inclusion evolution into a ternary diagram (Al₂O₃, SiO₂ and MnO), the Cr₂O₃ is not taken into account, small amount compounds like MgO and MnS are also not presented.

Table 9. Chemical composition of the inclusions and steel samples (wt%).

	Time (min)	Element content				Inclusion
		[Al]	[Si]	[Mn]	[Cr]	MgO	Al2O3	SiO2	MnO	Cr2O3	MnS
T1-1	0	–	0.30	0.99	14.0	0.1	7.2	24.5	42.8	16.1	6.7
T1-2	15	–	0.24	0.97	14.0	0.0	3.5	20.1	51.9	12.6	6.2
T1-3	30	–	0.2	1.00	13.9	0.9	8.9	12.0	42.2	23.2	2.9
T1-4	40	–	0.17	0.98	13.9	6.0	10.5	6.6	47.2	22.9	0.0
T1-5	55	–	0.14	0.99	13.9	1.0	7.9	7.6	56.7	22.4	1.1
T2-1	0	–	0.29	0.72	18.3	0.0	4.6	27.2	46.6	19.1	2.3
T2-2	15	0.002	0.25	0.72	18.2	0.0	11.5	26.8	37.0	22.6	1.4
T2-3	30	0.002	0.21	0.71	18.2	6.9	16.1	18.2	27.5	29.3	0.9
T2-4	45	0.001	0.20	0.69	17.9	1.7	17.2	8.7	36.7	34.9	0.3
T2-5	55	–	0.17	0.71	18.2	0.6	9.2	15.6	41.0	30.4	0.6
T3-1	0	–	0.33	1.17	16.5	0.2	4.3	24.1	41.3	12.6	12.4
T3-2	14	0.002	0.31	1.16	16.4	0.9	22.9	11.9	36.2	16.2	4.9
T3-3	24	0.003	0.28	1.13	16.5	1.3	31.7	11.9	27.5	18.0	2.5
T3-4	37	0.003	0.28	1.17	16.5	0.9	34.7	4.8	22.3	22.3	4.2
T3-5	49	0.003	0.26	1.16	16.6	1.9	47.2	5.8	21.6	12.3	3.3
T4-1	0	0.002	0.31	0.45	18.3	0.0	20.2	24.5	30.6	18.3	2.4
T4-2	19	0.005	0.28	0.42	18.1	3.1	52.7	3.9	14.3	22.8	0.0
T4-3	34	0.005	0.28	0.41	18.2	4.3	69.9	0.1	6.3	16.2	0.0
T4-4	44	0.006	0.26	0.41	18.2	8.7	67.2	2.0	1.4	19.3	0.0
T4-5	54	0.006	0.25	0.40	18.2	8.2	69.1	0.9	0.9	20.5	0.0

Fig. 11.

The compositional evolution of inclusions during the treatment at 1650°C (wt%).

It can be seen from Table 9 and Fig. 11 that: (1) the original inclusions were mainly in the form of MnO–SiO₂–Cr₂O₃–MnS, apparently this is the result of the reduction operation at the end of AOD process; (2) after the slag and alloy additions (FeSi and/or Al), Al₂O₃ content in the inclusions rapidly increases with an accompanying decrease in SiO₂ and MnO content in the tests with the CA slag and 3 g Al additions (Tests 3-4), while in the tests with the CSF slag and no Al additions (Tests 1-2) a small rise in Al₂O₃ content is observed at the early stage and then the Al₂O₃ content levels off followed by a small decrease. This compositional evolution of inclusions suggests that the original inclusions are reduced by aluminium through the reaction (8), leading to a considerable increase in Al₂O₃ content. The less obvious increase of Al₂O₃ content in the tests with the CSF slag (Tests 1 and 2 in Fig. 11) is a result of a small input of Al (from FeSi alloy); (3) In the end of the treatment, the inclusions contain mainly Al₂O₃–SiO₂–MnO–Cr₂O₃ with low MnS content (Table 9). According to the thermodynamic calculation, these inclusions would be saturated with corundum phase (mainly Cr₂O₃) at the experimental temperature; (4) minor levels of MgO are detected in the inclusions in all the tests, probably due to the slag-steel-refractory-inclusions interaction.   

$$\text{[Al]}+{\text{SiO}}_{\text{2}}{\text{(or MnO, Cr}}_{\text{2}}{\text{O}}_{\text{3}}\text{)}\to {\text{Al}}_{\text{2}}{\text{O}}_{\text{3}}+\text{[Si}\hspace{0.17em}\text{(or Cr, Mn)]}$$(8)

Since the top slag has considerable impact on the steel chemistry through the desulphurisation and deoxidation reaction, and the molten steel directly reacts with inclusions, an improved understanding of the thermodynamics and kinetics of the slag-steel-inclusions interaction is of significant importance for inclusion chemistry control during ladle treatment.

3.2.2. Thermodynamics of the Slag-steel-inclusions Interaction

3.2.2.1. Slag-steel Interaction

The FeSi and Al were added during the treatment for alloying and deoxidation purposes, thus the deoxidation reaction with respect to Al and Si are mainly considered for the slag-steel reactions. These reactions can be expressed as follows:^17,23)   

$$\text{2[Al]}+\text{3[O]}={\text{(Al}}_{\text{2}}{\text{O}}_{\text{3}}\text{),}\mathrm{\hspace{0.17em} }log{K}_9=\frac{45\mathrm{\hspace{0.17em} }300}{T}-11.62$$(9)

  

$$[\mathrm{S}\mathrm{i}]+2[\mathrm{O}]=(\mathrm{S}\mathrm{i}{\mathrm{O}}_2)\text{,}\mathrm{\hspace{0.17em} }log{K}_{10}=\frac{24\mathrm{\hspace{0.17em} }600}{T}-8.4$$(10)

where K_i is the equilibrium constant of Eq. (i). (Al₂O₃) and (SiO₂) are the Al₂O₃ and SiO₂ in slag, respectively. Through the Eqs. (9) and (10) the activity of the [Al] and [Si] after slag/steel equilibration can be predicted, and is denoted as ‘equilibrated value’. During the calculation, the measured a_[O] and temperature at the end of the tests are used (Table 4); the activities of the Al₂O₃ and SiO₂ in slag are estimated following two approaches. One is based on the chemistry of the final slag (Table 5) with the aid of FactSage and hereinafter denoted as ‘equilibrated with top slag’. The database of FactPS and FToxide are used during estimation. The other approach assumes the activity of Al₂O₃ and SiO₂ is unity, i.e. the slag has been saturated with Al₂O₃ and SiO₂. Figure 12 shows the comparison of the equilibrated and measured activity of [Al] and [Si]. It should be mentioned that the ‘measured activity’ of [Al] and [Si] is obtained by using Wagner equation (Eq. (11))¹⁶⁾ with the measured steel composition (Table 9) and interaction parameters (Table 7):   

$$log{a}_{[j]}=\sum e_j^i[\%i]+log[\%j]$$(11)

Fig. 12.

The comparison of calculated and measured activity of [Al] and [Si], where FS and Unity represent equilibration with top slag and pure oxides, respectively.

A comparison of the measured activities (a_[Al] and a_[Si]) with the equilibrated values is given in Fig. 12. The equilibrated activities are assumed to be equilibrated with the top slag and pure oxide, i.e. Al₂O₃ and SiO₂, respectively. It is clear that the measured values (a_[Al] and a_[Si]) are larger than the equilibrated values when steel is assumed to be equilibrated with the top slag, while the measured values are lower than the calculations with the oxides saturated slag. This is more obvious for the Si deoxidation reaction, i.e. Si activity in molten steel is far removed from the equilibrium both with top slag and pure SiO₂. Figure 12 shows: (1) the slag-steel interaction did not reach equilibrium during the treatment; (2) the reactions can proceed to increase the oxides (Al₂O₃ and SiO₂) content in the slag and cease before the slag has been saturated with them if a longer slag-steel interaction is allowed. The later agrees with the compositional analysis results for steel samples that the [Si] content decreased during the treatment (Table 9). In general, it is believed that the reaction has reached equilibrium at the slag/steel interface at high temperature and this is corroborated by a number of laboratory studies and industrial trials.^8,9,12,23) Therefore, there must be Al₂O₃ and SiO₂ concentration gradients from the slag/steel interface towards the slag bulk. Actually, these concentration gradients are considered to be the driving force for the reactions.

3.2.2.2. Steel-inclusions Interaction

The deoxidation reactions to form complex oxide inclusions can be described with Eqs. (9), (10), (12) and (13).^9,17,23) Similar with the case of the slag-steel reactions, the activity of Al₂O₃, SiO₂, MnO and Cr₂O₃ in the inclusion complex can be calculated based on the element activities in molten steel. Here, the activity of [Al], [Si], [Mn] and [Cr] can be obtained through Eq. (11)¹⁸⁾ based on the measured steel composition (Table 9), and then the activity of Al₂O₃, SiO₂ MnO and Cr₂O₃ in the inclusions complex can be calculated and is denoted as ‘Equilibrated’ value. The oxygen activity and temperature measured at the end of treatment is used during the calculation. On the other hand, the activity of Al₂O₃, SiO₂ and MnO can be directly estimated based on the average inclusion chemistry (Table 9 measured with SEM-AIA) with the aid of the FactSage software and is denoted as ‘FactSage’ value.¹³⁾ The database of FToxide and FactPS is used during the estimation. The Cr₂O₃ exists both in the ‘slag’ and ‘corundum’ phases, and to simplify the estimation, its activity is set to one.   

$$\text{[Mn]}+\text{[O]}={\text{(MnO)}}_{\text{incl}}\text{,}\mathrm{\hspace{0.17em} }log{K}_{12}=\frac{11\mathrm{\hspace{0.17em} }070}{T}-4.53$$(12)

  

$$\text{2[Cr]}+\text{3[O]}={{\text{(Cr}}_{\text{2}}{\text{O}}_{\text{3}}\text{)}}_{\text{incl}}\text{,}\mathrm{\hspace{0.17em} }log{K}_{13}=\frac{36\mathrm{\hspace{0.17em} }200}{T}-16.00$$(13)

Figure 13 shows the comparison of the Equilibrated and FactSage estimated activities of Al₂O₃, SiO₂, MnO and Cr₂O₃ in inclusions. Although the data points scatter to a considerable degree, the equilibrated activities for Al₂O₃, SiO₂ and MnO are approximately in agreement with the FactSage estimated values, this implies that the steel-inclusions reaction is fairly close to the equilibrium with respect to the deoxidation of Al, Si and Mn during the treatment. Comparing to the non-equilibrium of the slag-steel reaction (Fig. 12), this is probably attributed to the large specific surface area of the inclusions, which allows enough steel/inclusion interface for the reactions. It should be noted that the estimated activity of Cr₂O₃ at its saturated state (corundum phase) is far away from the equilibrated value. This is probably due to (1) the imprecision of Cr₂O₃ activity estimation (simply to be 1); and (2) limited steel-solid corundum interaction with respect to Cr deoxidation.

Fig. 13.

Comparison of the equilibrated and FactSage estimated activity of Al₂O₃, SiO₂, MnO and Cr₂O₃ in inclusions.

Based on the above thermodynamic analysis of the slag-steel and steel-inclusions reactions (Figs. 12 and 13) and the experimental results (Table 9 and Fig. 11), it can be concluded that the slag-steel equilibrium has not been attained during the treatment, and the reaction between slag and steel is much slower than that between steel and inclusions. Therefore in the slag-steel-inclusions reaction system, it is the top slag-steel reaction, which affects the molten steel chemistry, and consequently influences the compositional evolution of inclusions during the ladle treatment.

3.2.3. Sulphur Content of the Inclusions

As seen in Table 9 the inclusions in the steel sample are mainly in the form of oxides with a minor amount of sulphides. As been understood that excepting the influence of solidification process, the sulphur content in inclusions is mainly dependent on the sulphide capacity (C'_S) of the inclusions and the sulphur content ([%S]) in the steel.¹³⁾ Here, the sulphide capacity of the inclusions could be calculated based on the inclusion oxides chemistry and the experimental temperatures. Furthermore, the equilibrated sulphur distribution ratio, i.e. L'_S (= (%S)^incl/[%S], between the steel and inclusions can be calculated. The calculation method of C'_S and L'_S for inclusions is similar with that for slag. The detailed calculation approach has been given in our previous paper.¹²⁾ It should be noted that the measured oxides chemistry (Table 9) and sulphur content in the steel samples (Table 4) were used in the calculation. The upper part of Fig. 14 shows the relation of the calculated C'_S and the measured L'_S between inclusions and steel, while the lower part shows the relation between the measured L'_S and the calculated equilibrated L'_S.

Fig. 14.

The relation between the measured L'_S and calculated C'_S, as well as between the measured L'_S and calculated equilibrated L'_S.

It can be seen from the upper part of Fig. 14 that the data points approximately follow the line with a slope of 1, indicating that the sulphur distribution ratio between the inclusions and steel was mainly influenced with C'_S of the inclusions. A good agreement between the equilibrated and measured L'_S is obtained in the lower part of Fig. 14, suggesting the equilibrium between inclusions and steel with respect to sulphur was reached. Therefore, the sulphur content of inclusions could be predicted through the calculation of inclusion sulphide capacity and sulphur content in steel during the treatment. The former can be obtained by using the inclusion oxides chemistry and the treatment temperature, and the latter can be predicted through the developed desulphurisation kinetic model.

4. Conclusions

The deep desulphurisation and steel cleanliness of stainless steel refining by using an optimized CA based synthetic slag was studied under the industrial conditions on a laboratory scale. The sulphide capacity, sulphur distribution ratio and desulphurisation kinetics were investigated with respect to the desulphurisation effect of the current CSF and the optimized CA slags. The steel cleanliness was compared and the slag-steel-inclusions interaction was discussed based on thermodynamic considerations. The main results can be summarized as follows:

(1) Deep desulphurisation to the level of several ppm sulphur was achieved with an optimized CA based synthetic slag under industrial conditions in laboratory tests (i.e. 30–45 ppm sulphur content and doloma refractory crucible). A similar desulphurisation effect with current CSF and the optimized CA slags was obtained in the industrial experiment. This confirms the application feasibility of the optimized CA based synthetic slag for deep desulphurization purpose of stainless steel ladle refining.

(2) The deep desulphurisation was mostly influenced by the slag sulphide capacity, the steel chemistry and the deoxidation level of stainless steel. The sulphur evolution in steel during the ladle refining can be predicted with a developed kinetic model.

(3) An equivalent steel cleanliness was achieved with two current CSF and an optimized CA based synthetic slag, although a small amount of Al₂O₃–MgO spinel was observed in the final samples with CA based synthetic slag. This is believed to be the result of the slag-steel-refractory-inclusions interaction.

(4) The equilibrium was more easily reached for steel-inclusions interaction with respect to deoxidation than that for slag-steel interaction. Therefore in the slag-steel-inclusions reaction system, it is the top slag-steel reaction, which affects the molten steel chemistry, and consequently influences the compositional evolution of inclusions during the ladle treatment.

(5) The sulphur content in inclusions was predicted through the calculation of inclusion sulphide capacity and sulphur content in steel.

References

• 1)   R.  Kiessling: Met. Sci., 14 (1980), 161.
• 2)   M.  Wakoh,  T.  Sawai and  S.  Mizoguchi: ISIJ Int., 36 (1996), 1014.
• 3)   W. A.  Spitzig: Metall. Trans. A, 14 (1983), 271.
• 4)   M.  Wakoh,  T.  Sawai and  S.  Mizoguchi: Tetsu-to-Hagané, 78 (1992), 1697.
• 5)   H.  Kim,  H.  Lee and  K.  Oh: Metall. Mater. Trans. A, 32 (2001), 1519.
• 6)   D. K.  Aidun and  S. K.  Martin: J. Mater. Eng. Perform., 4 (1996), 496.
• 7)   T.  Miki,  J.  Björklund,  M.  Andersson and  P.  Jönsson: Proc. of 3rd Int. Conf. on Process Development in Iron and Steelmaking, MEFOS, Lulea, 2 (2008), 581.
• 8)   J.  Lehmann and  M.  Nadif: Rev. Miner. Geochem., 73 (2011), 493.
• 9)   B.  Yoon,  K.  Heo,  J.  Kim and  H.  Sohn: Ironmaking Steelmaking, 29 (2002), 215.
• 10)   D.  Durinck,  F.  Engström,  S.  Arnout,  J.  Heulens,  P. T.  Jones,  B.  Björkman,  B.  Blanpain and  P.  Wollants: Resour. Conserv. Recy., 52 (2008), 1121.
• 11)   M.  Hino and  S.  Ban-Ya: Ultra High Purity Base Metal, The Japan Institute of Metals, Sendai, (1995), 396.
• 12)   P.  Yan,  M.  Guo,  X.  Guo,  S.  Huang,  J. V.  Dyck and  B.  Blanpain: ISIJ Int., 53 (2013), 459.
• 13)   P.  Yan,  S.  Huang,  L.  Pandelaers,  J. V.  Dyck,  M.  Guo and  B.  Blanpain: Metall. Mater. Trans. B, 44 (2013), 1105.
• 14)   C.  Fincham and  F.  Richardson: Proc. of the Royal Society of London on Math. Phys. Sci., London, 223 (1954), 40.
• 15)   A. H.  Chan and  R.  Fruehan: Metall. Mater. Trans. B, 17 (1986), 491.
• 16)   C.  Wagner,  S.  Mellgren and  J. H.  Westbrook: Thermodynamics of Alloys, Vol. 4. Addison-Wesley Press, Cambridge, MA, (1952), 51.
• 17)   K.  Suzuki,  S.  Ban-Ya and  M.  Hino: ISIJ Int., 42 (2002), 146.
• 18)   B.  Deo and  R.  Boom: Fundame ntals of Steelmaking Metallurgy, Prentice Hall International, New York, (1993), 130.
• 19)   T.  Miki and  M.  Hino: ISIJ Int., 45 (2005), 1848.
• 20)   W.  Nernst: Z. Phys. Chem, 47 (1904), 52.
• 21)   G.  Roy,  P.  Chaudhary,  P.  Miji and  R.  Goel: Metall. Mater. Trans. B, 32 (2001), 558.
• 22)   B. D.  Marsh: Contrib. Mineral. Petrl., 99 (1988), 277.
• 23)   J.  Park: Metall. Mater. Trans. B, 38 (2007), 657.