Comparison between measured CaO and Al 2 O 3 normalised to CaO–Al 2 O 3 in top slag,calcium aluminate inclusions and results of theoretical calculations

Abstract

This study sets out to compare the wt-% CaO^slag and Al₂O^slag ₃ in the binary system of CaO–Al₂O₃ that were measured in the top slag before teeming, the CaO^incl and Al₂O^incl ₃ in calcium aluminates that were analysed by scanning electron microscopy of the steel samples and the CaO^calc and Al₂O^calc ₃ that were achieved from theoretical calculations using the total measured elements in the top slag and steel before teeming. The CaO^incl and Al₂O^incl ₃ in all of the calcium aluminate inclusions of this study did not have the same ratio; in the opposite, they covered a large range. The average CaO^incl and Al₂O^incl ₃ in all samples had nearly similar compositions that were close to the border of the phase Ca₁₂Al₁₄O₃₃ in the binary phase diagram of CaO–Al₂O₃. The CaO^slag and Al₂O^slag ₃ in all samples had almost similar compositions that were close to the border of the phase Ca₃Al₂O₆ in the binary phase diagram of CaO–Al₂O₃. The results of theoretical calculations were close to the CaO^slag and Al₂O^slag ₃, but they did not follow the same maximum and minimum tendencies of CaO^slag and Al₂O^slag ₃.

Introduction

One of the continuing demands at Ovako Hofors AB is to reduce sulphur and total oxygen content of bearing steels. Regarding sulphur content, it is believed that sulphides can act as crack initiators, and concerning total oxygen content, it is known that due to oxide inclusions, rolling contact fatigue life may be reduced even when the oxygen content is <10 ppm.1

The harmful effects of oxide inclusions on the fatigue properties of steel parts are highly dependent on their respective compositions, volume fractions, distributions and morphologies.2 ^– 4 Generally, big, unbreakable inclusions are the most unwanted ones, and compared with these inclusions, small, breakable inclusions or those with low melting temperatures are preferred.3, 5, 6

In most of the metallurgical secondary treatment plants of steelmaking companies, slag, steel and ladle are involved. The compositions of slag and steel are very important factors for reaching the desired final steel products, and they have a crucial effect on the type and properties of non-metallic inclusions. There have been many studies7 ^– 28 in order to measure, calculate or simulate the composition of non-metallic inclusions.

In this work, we investigate if there is any relation between the wt-% CaO and Al₂O₃ in the binary system of CaO–Al₂O₃ of calcium aluminates (CaO^incl and ) and their amount in the top slag before teeming (CaO^slag and ) and the theoretically calculated wt-% CaO and Al₂O₃ normalised to their binary system (CaO^calc and ). In case a certain relation could be established, it would be possible to modify or engineer compositions and chemical–physical properties of non-metallic inclusions based on the top slag and steel amounts and compositions during the ladle treatment process.

Steel melting plant of Ovako Hofors AB

Figure 1 presents the general scheme of the steel melting plant of Ovako Hofors AB and the respective sampling positions. Scraps are melted in the 100 t, bottom tapped electric arc furnace. The molten metal is tapped into the ASEA-SKF ladle furnace station; then, by means of the ferro silicon and ferro aluminium that were added during the tapping process, molten steel is deoxidised. Then, the slag that is the result of this deoxidation is skimmed, and the ladle filled with molten steel is sent to the ladle treatment station. In the ladle furnace, synthetic top slag that is mainly composed of CaO and Al₂O₃ is put on the molten steel, and then the melt is desulphurised, deoxidised, its alloying composition adjusted, vacuum degassed in order to remove the hydrogen of the molten steel, gas and magnetic stirred, possibly reheated and, finally, after the temperature of the melt is adjusted to the desired temperature, poured into 24 ingots each having a capacity of 4·2 t.

Figure 1.

Melting plant scheme of Ovako Hofors AB and respective sampling positions

Theoretical background

By assuming the hypothesis that in case of having a semiequilibrium between slag and steel in the ladle furnace after the vacuum degassing process and before teeming, it could be concluded that the chemical composition of non-metallic inclusions that are present in molten steel should be affected by the chemical composition of the top slag. In this study, four steel heats of grade SAE 52100 with the nominal chemical composition that is shown in Table 1, which were produced by Ovako Hofors AB, were chosen; then, we evaluated this hypothesis by comparing the CaO^slag and measured from top slag before teeming and the CaO^incl and analysed by scanning electron microscopy (SEM), as well as the CaO^calc and calculated from theoretical considerations.

Table 1.

Nominal chemical composition of steel grades SAE 52100/wt-%

	C	Si	Mn	P	S	Cr	Ni	Mo
Min.	0·95	0·20	0·20	…	…	1·35	…	…
Max.	1·00	0·35	0·40	0·025	0·015	1·60	0·25	0·08

For calculating CaO^incl and in the measured and calculated results of calcium aluminates, four elements were considered: Ca, Al, Mg and S (or any compounds that only had some of these elements). Then, it was assumed that all available Mg would, together with Al, form the spinel phase (MgAl₂O₄), and the rest of the Al would, together with the available Ca, form calcium aluminates (the available Ca was assumed to be the rest of the Ca that had already formed the CaS phase with all available S). For calculating CaO^slag and in the result of the top slag, because Mg in the top slag can also come from ladle bricks and is not reliable, only three elements were considered: Ca, Al and S. Regarding CaO^calc and , the measured amounts of the elements that were used for theoretical calculations are shown in Table 2. Finally, all of these calculated and measured results were compared with each other.

Table 2.

Total amount of elements in top slag and steel measured in four different heats used for theoretical calculations

Sample designation	Total weight of element in top slag and steel/kg
Sample designation	Fe	S	Ca	Al	Mg
S1	99 436	21	610	267	32
S2	99 432	20	585	275	44
S3	99 464	21	561	277	43
S4	99 422	22	538	272	59

Experimental

In this study, four steel heats of grade SAE 52100 with the nominal chemical composition that is shown in Table 1 were chosen, and then CaO^incl and that were achieved by means of analysing the result of SEM were compared with CaO^slag and before teeming, as well as CaO^calc and .

In order to perform SEM analysis for calculating CaO^incl and , steel bars with the cross-section of 147×147 mm were chosen, and then longitudinal samples with a rough area of 40×90 mm from the part of the bars that used to be the bottom part of the teemed ingots were cut, surface machined and polished. Then, by means of the INCA feature software,29 which was specifically developed for analysis and classification of steel inclusions using energy dispersive X-ray microanalysis, the SEM results were analysed, and then calcium aluminates were chosen as the inclusion types of interest for this study. All scanned inclusions that could fulfil the chemical compositions as well as the optical properties that are shown in Tables 3 and 4 were considered as calcium aluminates. The main reason for these cutoff criteria was to achieve the most reliable results by considering limitations of SEM analysis and sampling preparation.

Table 3.

Cutoff criteria for chemical composition of selected calcium aluminates/wt-%

	O	Al	Ca	Mg	S
Max.	100	70	70	10	14
Min.	15	10	10	0	0

Table 4.

Cutoff criteria for optical properties of selected calcium aluminates

	Area/μm²	Width/μm	Aspect ratio
Max.	No limit	No limit	6
Min.	25	5	No limit

The reason for choosing bar samples instead of lollipop samples was because it was one of the objectives of this study to find out if it was possible to predict the morphology and types of inclusions in the products by knowing the top slag and molten steel amount and compositions during the ladle treatment process. Thus, oxides in the bar samples were supposed to represent the same inclusion types of the final product without considering the effect of reduction ratio on them.

After separating the calcium aluminates according to the cutoff criteria mentioned in Tables 3 and 4, the composition of these non-metallic inclusions was normalised to Ca, Al, Mg and S. It was then assumed that all the available Mg in these calcium aluminates form the spinel phase (MgAl₂O₄) and that the rest of the Al would, together with the available Ca, form calcium aluminates with different Ca/Al ratios (the available Ca was assumed to be the rest of the Ca that had already formed the CaS phase with all available S).

It should be mentioned that in some calcium aluminates, Mg was found as pure MgO; these inclusions were here considered as exceptions and were excluded from the results. The total number of observed and excluded calcium aluminates and the total scanned area for each sample are shown in Table 5.

Table 5.

Scanned area of four steel samples

Sample designation	Scanned area/cm²	Calcium aluminates
Sample designation	Scanned area/cm²	Total found	Total removed as exceptions
S1	24·2	29	3
S2	21·9	28	2
S3	20·6	21	1
S4	16·4	24	1

The achieved SEM results attained as explained above were normalised to the binary system of CaO–Al₂O₃, and then CaO^incl and were calculated. These results were compared with CaO^slag and in the top slag before teeming, which were measured by the laboratory instruments that are named and briefly explained in the section on ‘Sample analysing methods’. Finally, these results were compared with CaO^calc and , which were theoretically calculated based on the total measured elements in top slag and steel. It was also here assumed that all the available Mg in these calcium aluminates form the spinel phase (MgAl₂O₄), and that the rest of the Al would, together with the available Ca, form calcium aluminates with different Ca/Al ratios (the available Ca was assumed to be the rest of the Ca that had already formed the CaS phase with all available S).

Sample analysing methods

Figure 1 shows the sampling position for the tested heats. For all the tested heats, one steel sample (lollipop) and one slag sample were taken. All lollipop samples were tested by two testing instruments as follows: OES ARL 4460 Metal analyser from Thermo Fisher Scientific Inc. was used for metal analyses, and the instrument called Leco CS-444 was used in order to analyse the sulphur and carbon contents of the samples; this instrument works based on the carbon–sulphur combustion method (ASTM E1019). All slag samples were analysed by an X-ray fluorescence instrument called PANalytical Axios^mAX- Minerals.

The Celox instrument30 that was equipped with consumable MgO stabilised ZrO₂ electrodes was applied in order to measure the oxygen activity of molten steel.

For scanning the steel samples regarding the non-metallic inclusions, the analytical SEM brand Leo Supra 35 was used, and the INCA feature software29 was employed for the analysis and classification of non-metallic inclusions, using energy dispersive X-ray microanalysis.

Results

Figures 2 and 3, which were taken using the element mapping feature of the INCA feature software29 by applying energy dispersive X-ray microanalysis, show the SEM–energy dispersive spectroscopy (EDS) mapping analysis of two of the observed calcium aluminates considered in this study.

Figure 2.

Scanning electron microscopy–EDS mapping analysis of one of observed calcium aluminates

Figure 3.

Scanning electron microscopy–EDS mapping analysis of one of observed calcium aluminates

Table 6 shows the statistical summaries for all calcium aluminates achieved from the SEM analysis of four samples of this study.

Table 6.

Statistical summaries for all calcium aluminates observed in this study

Sample designation		Area/μm²	Aspect ratio	Width/μm	ECD/μm	Length/μm
S1	Average	223	2·3	11·7	14·8	26·3
S1	Standard deviation	259	1·3	5·8	8·0	20·9
S2	Average	209	3·0	9·7	14·3	30·6
S2	Standard deviation	237	1·6	4·2	7·8	23·6
S3	Average	213	3·1	10·2	14·7	34·4
S3	Standard deviation	243	2·0	3·7	7·5	32·5
S4	Average	177	2·8	9·4	13·5	27·5
S4	Standard deviation	193	1·6	3·5	6·6	20·8

Figure 4 presents CaO^incl, CaO^slag and CaO^calc in all four samples of this study. In this figure, there are also two other lines indicating one standard deviation above and one standard deviation below the average CaO^incl. Figure 5 presents , and in all four samples of this study; in this figure, there are also two other lines indicating one standard deviation above and one standard deviation below the average .

Figure 4.

Measured and calculated CaO^{incl, slag, calc} in all four samples of this study

Figure 5.

Measured and calculated Al₂O₃ ^{incl, slag, calc} in all four samples of this study

Figure 6 shows the calculated and measured positions of CaO^incl, CaO^slag and CaO^calc in all four samples of this study in the binary phase diagram of CaO–Al₂O₃.31 In this figure, there are also two other lines indicating one standard deviation above and one standard deviation below the average CaO^incl in four different samples of this study.

Figure 6.

Calculated and measured positions of CaO^{incl, slag, calc} in all four different samples of this study in binary phase diagram of CaO–Al₂O₃ (Ref. 31)

Figure 7 presents the ternary diagram of the calcium aluminates in one of the four samples of this study. This diagram shows the ratio of CaO/Al₂O₃/MgO of the inclusions and their respective equivalent circle diameters (ECDs) in micrometres. In this figure, liquidus surfaces with isotherm lines every 100°C were calculated by means of Thermo-Calc software.32

Figure 7.

Equivalent circle diameters (μm) of calcium aluminates in one of samples

In Fig. 7, a clear relationship between the chemistry and size of inclusions could be observed, i.e. the larger the inclusion, the more Ca it contains. This fact could be more clarified by Fig. 8; this figure presents wt-%Ca versus wt-%Al in all of the oxides that were observed in all four samples of this study, and the bubble sizes represent the ECDs of the inclusions.

Figure 8.

wt-%Ca versus wt-%Al observed in all samples: bubble sizes represent ECDs of oxides

Discussion

In this study, it was assumed that in the calcium aluminates that were considered, most of the times the available Mg bounded with Al and appeared as a spinel phase, and then the rest of the Al that was not bounded to the Mg would form phases that were composed of Ca and Al with different ratios of these two elements (the available Ca was assumed to be the rest of the Ca that had already formed the CaS phase with all available S). However, sometimes, Mg could be found as pure MgO that did not form the spinel phase. These inclusions were treated as exceptions and were not considered in this study.

Regarding these calcium aluminates, it should be mentioned that some amounts of Ca were bounded with Si; however, in this study, this amount was neglected. This neglection of Si was because of the fact that the Si contents of the calcium aluminates of this study were always <4 wt-%, and the steel matrix of steel grade SAE 52100 consisted of about 0·2–0·35 wt-%Si; the measured Si amount of the inclusions, especially the small ones, could not be relevance (due to the Si that is present in the steel matrix, Si contents of the inclusions were measured higher than the true amount of Si). It was therefore decided to neglect Si and concentrate on the amounts of Mg, Ca, Al and S in the inclusions.

By regarding Figs. 4 and 5, it was clear that the CaO^incl and in all samples of this study did not have the same ratio. On the contrary, they covered a large range.

By considering Figs. 4 and 5, it could also be concluded that the results that were theoretically calculated were close to CaO^slag and ; however, they did not follow the same trend of CaO^slag and .

By considering Fig. 6, it could be seen that the average CaO^incl and in all samples had nearly similar compositions that were close to the border of the phase Ca₁₂Al₁₄O₃₃ (C₁₂A₇) in the binary phase diagram of CaO–Al₂O₃,31 and the CaO^slag and in all samples had compositions close to the phase Ca₃Al₂O₆ in the binary phase diagram of CaO–Al₂O₃.31

These results clearly indicate that the top slag, steel and inclusions of this study did not reach equilibrium; however, there were clear indications that they had obvious trends to reach some equilibria. For instance, the chemical composition of calcium aluminates tended to reach equilibrium at the eutectic point in the binary phase diagram of CaO and Al₂O₃. Moreover, it could be mentioned that the assumption that the top slag–steel systems were closed systems that did not react with the factors outside was not necessarily true, and high probably contents of Mg, O and N could change; however, these changes were not big enough to affect the calculated results of this study.

It should also be considered that the results that were achieved from the SEM analysis might have some errors that could be due to sampling, machining, polishing, analysing techniques and interpreting the results.

Conclusions

The following statements briefly point out the achieved results in this study.

In the calcium aluminates, Mg was often seen as spinel, and Al that did not form spinel formed calcium aluminates with different amounts of CaO^incl and (the available Ca was assumed to be the rest of he Ca that had already formed the CaS phase with all available S).

The CaO^incl and in all samples did not have the same ratio; in the opposite, they covered a large range.

The average CaO^incl and in all samples had nearly similar compositions that were close to the border of the phase Ca₁₂Al₁₄O₃₃ (C₁₂A₇) in the binary phase diagram of CaO–Al₂O₃.31

The CaO^slag and had almost similar compositions that were close to the border of the phase Ca₃Al₂O₆ in the binary phase diagram of CaO–Al₂O₃.31

The results of theoretical calculations were close to CaO^slag and , but they did not follow the same maximum and minimum tendencies of CaO^slag and .

References

Lange

K.-W

: Int. Mater. Rev., 1988, 33, (2), 53–89.

Suito

Inoue

: ISIJ Int., 1996, 36, (5), 528–536.

Bertran

Molinero

Landa

Elvira

Wild

Barthold

Valentin

Schifferl

: Ironmaking Steelmaking, 2003, 30, (2), 165–169.

Hagiwara

Kawami

Ueno

Kido

Urbancic

M.-J

Friedel

R.-D

: Wire J. Int., 1991, 24, 29–34.

Chapetti

M.-D

Tagawa

Miyata

: Mater. Sci. Eng. A, 2003, A356, (1–2), 236–244.

Abe

Furuya

Matsuoka

: Eng. Mater. Struct., 2004, 27, (2), 159–167.

Zhang

Thomas

B.-G

: ISIJ Int., 2003, 43, 271–291.

Choudhary

S.-K

Ghosh

: ISIJ Int., 2008, 48, 1552–1559.

Suito

Inoue

: ISIJ Int., 199, 131, 1381–1388.

10.

Taguchi

Nakazato

H.-O

Usui

Marukawa

Katogi

Kosaka

: ISIJ Int., 2005, 45, 1572–1576.

11.

Cho

S.-W

Suito

: ISIJ Int., 1994, 34, 265–269.

12.

Fujiwara

Tano

Yamamoto

Ichise

: ISIJ Int., 1995, 35, 1063–1071.

13.

Suito

Inoue

: Trans. ISIJ, 1984, 24, 301–307.

14.

Kobayashi

: ISIJ Int., 1999, 39, 664–670.

15.

Goto

Miyazawa

K.-I

: ISIJ Int., 1998, 38, 256–259.

16.

Jung

I.-H

Decterov

S.-A

Pelton

A.-D

: Metall. Mater. Trans. B, 2004, 35B, 493–507.

17.

Ohta

Suito

: ISIJ Int., 1996, 36, 983–990.

18.

Kang

Y.-B

Jung

I.-H

Decterov

S.-A

Pelton

A.-D

Lee

H.-G

: ISIJ Int., 2004, 44, 975–983.

19.

Kang

Y.-B

Jung

I.-H

Decterov

S.-A

Pelton

A.-D

Lee

H.-G

: ISIJ Int., 2004, 44, 965–974.

20.

Hayashi

Sano

Fredriksson

: ISIJ Int., 2004, 44, 1783–1786.

21.

Björkvall

Sichen

Seetharaman

: Calphad, 2000, 24, 353–376.

22.

Eriksson

Pelton

A.-D

: Calphad, 1993, 17, 189–205.

23.

Oertel

L.-C

Silva

A.-C

: Calphad, 1999, 23, 379–391.

24.

Decterov

S.-A

Jung

I.-H

Jak

Kang

Y.-B

Hayes

Pelton

A.-D

: Proc. VII Int. Conf. on ‘Molten slags fluxes and salts’, Cape Town, South Africa, January 2004, The South African Institute of Mining and Metallurgy, 839–850.

25.

Kang

Y.-B

Lee

H.-G

: ISIJ Int., 2004, 44, 1006–1015.

26.

Kang

Y.-B

Kim

H.-S

Zhang

Lee

H.-G

: J. Phys. Chem. Solids, 2005, 66, 219–225.

27.

Jung

I.-H

Decterov

S.-A

Pelton

A.-D

: ISIJ Int., 2004, 44, 527–537.

28.

Shannon

White

Sridhar

: Mater. Sci. Eng. A, 2008, A459, 310–315.

29.

INCA feature software (computer programme), Oxford Instruments, Oxford, UK, 2010, available from : http://www.x-raymicroanalysis.com/x-ray-microanalysis-explained/pages/main/inca.htm. (Accessed 24 January 2011)

30.

Celox Instrument, Heraeus Electro Nite, Hanau, Germany, 2011, available from : http://www.electro-nite.com/Products/IronSteel_OxygenActivityControl.aspx. (Accessed 24 January 2011)

31.

Chatterjee

A.-K

Zhmoidin

G.-I

: Mater. Sci., 1972, 7, (2), 93–97.

32.

ThermoCalc Software AB, 2011, available from : http://www.thermocalc.com (accessed 24 January 2011).