AISTech 2019 — Proceedings of the Iron & Steel Technology Conference

6–9 May 2019, Pittsburgh, Pa., USA

DOI 10.1000.377.216

Pre-Fused vs. Blended Fluxes: Their Impact on Cleanliness of Low-Carbon Low-Alloy Steel

Rodrigo Lencina1, Bo Zhang2, Annelies Malfliet3, Yuxin Chen1, Bruno Touzo1, Bart Blanpain3, Muxing Guo3
1
Imerys
43 Quai de Grenelle, Paris, France, 75015
[email protected]
2
School of Metallurgy, Northeastern University
Shenyang, China, 110819
[email protected]
3
Department of Materials Engineering, KU Leuven
Kasteelpark Arenberg 44, Leuven, Belgium, 3001
[email protected]

Keywords: CaO-Al2O3-based slags; calcium aluminates; ladle refining; steel cleanliness; inclusion characteristic

INTRODUCTION
The term calcium aluminate flux is generally applied in the steel industry to CaO-Al2O3 based materials with a low melting
point. Other oxides normally found in low proportion in the composition of the fluxes are SiO2, MgO and TiO2. Additionally,
certain commercially available fluxes may contain levels of C, P, S, F and V that should not be overlooked. Calcium
aluminate fluxes (also known as synthetic slags) are normally added at the beginning of the ladle refining process in a range
of 1–5 kg/t steel with the objective of forming a CaO-Al2O3 based top slag. For many decades, the traditional role of calcium
aluminate fluxes in the ladle refining operations has been the adjustment of slag viscosity and chemistry, the improvement of
desulphurization kinetics and the protection of the refractory lining [1-3]. Nowadays, with the increasing demand of high
quality clean steels, the formation of slags aimed to maximize inclusion removal is a critical part of the process [4-6].
Consequently, steelmakers require calcium aluminate fluxes that form slags adapted to meet the new cleanliness demands for
steel.
Slags with a CaO-Al2O3 composition are commonly used in the low carbon low alloy steel production because of their low
oxidation potential, low melting point (1350~1450°C), and a low viscosity (0.36~0.16 Pa.s at 1600~1700°C)[7]. One of the
main roles of CaO-Al2O3-based slags in the ladle refining process is the absorption of inclusions and elemental impurities
such as sulfur. The effect of CaO-Al2O3-based slag on the steel's cleanliness and desulfurization has been investigated by
several researchers. Yoon et al. [8] investigated the effect of the (CaO)/(SiO2) basicity ratio and (CaO)/(Al2O3) ratio on the
cleanliness and found that decreasing the melting temperature through adjustment of C/A ratio is more efficient for inclusion
removal than reaching the CaO-saturated region through adjustment of the C/S ratio. Björklund et al. [9] and Coletti et al.
[10] pointed out that the top slag is not in equilibrium with liquid steel during ladle refining. Steel cleanliness is influenced by
the interaction between slag and steel. On the other hand, the past studies [8-11] were mainly focused on the influence of the
main slag components (i.e. CaO, Al2O3, SiO2 and MgO). Industrial practice and recent studies [12], however, identified that
the minor slag components, such as S, C, F, TiO2 among others, also significantly influence the deoxidation, the
desulphurization, and the chemistry and morphology of inclusions through slag-steel interaction.
In addition to the melting temperature, the melting rate of the flux after addition into the ladle is also an important kinetic
factor for the slag-steel interaction. Actually, this melting rate is affected by the flux preparation method. Two common
industrial processes for flux production are fusion (>1400°C) and powder blending (at an ambient temperature). Pre-fused
calcium aluminate flux is formed mainly by the C12A7 mineral phase whereas a blend consists of a mechanical mixture of
different mineral phases, e.g. corundum and calcium oxide. The influence of the mineral constituents of a calcium aluminate
flux on steel cleanliness has been poorly documented. There is a need for a more fundamental understanding of the
interaction mechanisms taking place during the refining process with calcium aluminate fluxes.

© 2019 by the Association for Iron & Steel Technology. 2101

In this paper, two calcium aluminate fluxes were investigated with respect to their refining capabilities, such as the ability to
absorb non-metallic inclusions and to remove impurities (i.e. S) from low carbon low alloy steel at ladle refining
temperatures. These fluxes had similar chemical compositions but different mineral phases, where one was a pre-fused
homogeneous material and the other a blend of reagents. The distinct slag-steel interaction mechanisms for each refining flux
have been identified by using experimental data. The steel cleanliness with the two flux materials have been evaluated
through chemical analysis and inclusion characterization of the steel and slag samples.

EXPERIMENTAL METHOD

Materials
For the steel-slag (flux) interaction experiments, one steel grade and two different fluxes were selected. The steel used was an
industrial sample provided by a European producer. Table 1 shows the chemical composition of the low carbon low alloy
steel measured with inductively coupled plasma-atomic emission spectroscopy (ICP-AES, Varian Liberty series II instrument
with an axial plasma configuration). Table 2 shows the chemical composition of the two fluxes measured with X-ray
fluorescence spectroscopy (XRF, Philips PW 2400). The pre-fused flux is a commercial product with a 0 – 15 mm particle
size made in a fusion kiln located in France. The flux named blend in Table 2 is a mixture of chemical reagent grade oxides,
i.e. lime, corundum, silica, iron oxide and titanium oxide. The blended flux’s composition was designed to match that of the
pre-fused flux.

Table 1. Chemical Composition of the Low Carbon Low Alloy Steel (wt%)
C Si Mn S Cr Ti Al
0.0036 0.040 0.11 0.0076 0.016 0.044 0.045

Table 2. Chemical Composition of the Flux (wt%)

Flux CaO Al2O3 SiO2 Fe2O3 MgO S C MnO TiO2 CaF2
Pre-fused 47.72 50.39 0.56 0.60 0.41 0.03 0.13 0.03 0.07 -
Blend 48.00 49.00 0.70 0.50 1.50 0.01 0.03 - 0.10 -

Experimental Procedure
A schematic drawing of the experimental set-up is shown in Figure 1. The interaction experiments were performed in a
vertical tube furnace (Gero HTRV, MoSi2 heating elements). The experimental procedure is illustrated in Figure 2. The
crucible containing 60 g low carbon low alloy steel was melted under purified argon atmosphere at 1600°C. An oxygen
probe, suitable for measurement of the oxygen concentration at refining temperatures, was used to measure “in-situ” the pO2
during the test. After the furnace temperature reached 1600°C, the interaction cell was held at that temperature for 30 min to
homogenize the temperature and chemistry of the liquid steel. Afterwards, 25 g flux was introduced to the liquid steel in three
times (to avoid a large temperature drop) within 5 min using a SiO2 tube fixed on the top flange of the furnace. To facilitate
the addition into the furnace, the samples of fluxes were processed using the following methods. The lumps of pre-fused flux
were crushed and sieved with a 3 mm mesh, the fine fraction was used for the experiments. The chemical reagents used to
prepare the blend were already in a fine particle size. Pellets of 5 mm x 10 mm were prepared with the blend using an
isostatic press.
The steel-slag interaction time was counted after completion of the flux addition. After the interaction time (10 or 30
minutes), the crucible was lifted into the cold area of the furnace for a relatively rapid cooling. Steel and slag samples were
separated and subjected to analysis.

2102 © 2019 by the Association for Iron & Steel Technology.

 Figure 1. Schematic representation of the experimental set-up

Figure 2. Overview of the heat treatment procedure

Composition and inclusion characteristics analysis

The steel samples were cut into two parts along the vertical axis, and each part was further sliced horizontally in three pieces.
All analyses were performed on the top, middle and bottom samples to evaluate differences in inclusion characteristics with
height. Figure 3 shows the steel sample after the treatment with the pre-fused flux, which was embedded in low viscosity
resin (Epofix) by vacuum impregnation, and polished to perform EPMA-AIA (automatic inclusion analysis) analysis to
characterize the inclusions in the steel sample, such as composition, size distribution, spatial distribution, etc. A picture with
at least 1.91 mm2 of steel matrix was obtained by stitching 200 pictures at 1000 times magnification (Figure 4). Within this
area, the number and size of the inclusions were recorded and for the inclusions with a surface area of at least 0.5 µm², the
chemical composition was measured with EDS. In order to eliminate the misjudgment caused by the dust and cavities on the
steel surface, all of the inclusions recognized by EPMA-AIA were manually re-checked one by one. The EPMA was
equipped with an energy-dispersive spectrometer (EDS) detector and a particle analysis system, which can be used to conduct
automatic inclusion analysis (AIA). The second part of the sample was subjected to combustion analysis using LECO
instruments to determine the total oxygen, carbon, and sulfur. For each position, the measurement was repeated thrice and the
average values are reported in Table 3. The estimated measurement error was ±0.0005 wt%.

© 2019 by the Association for Iron & Steel Technology. 2103

 Figure 3. Picture of the steel sample after treatment with the pre-fused flux

Figure 4. Stitched picture using 200 pictures (x1000) of the steel sample

RESULTS AND DISCUSSION

Variation of total oxygen, carbon and sulfur in the steel

Table 3. Results of LECO analyses

Interaction [T.O] [C] [S]
Flux Position
time (min) (wt%) (wt%) (wt%)
Initial steel - 0 0,0057 0,0036 0,0076
Pre-fused Top 10 0,0060 0,0026 0,0044
Pre-fused Middle 10 0,0024 0,0023 0,0040
Pre-fused Bottom 10 0,0030 0,0033 0,0040
Pre-fused Top 30 0,0037 n/a n/a
Pre-fused Middle 30 0,0021 0,0055 0,0028
Pre-fused Bottom 30 0,0015 0,0034 0,0020
Blend Top 10 0,0147 0,0027 0,0043
Blend Middle 10 0,0105 0,0020 0,0047
Blend Bottom 10 0,0072 0,0019 0,0041
Blend Top 30 0,0030 0,0048 0,0007
Blend Middle 30 0,0027 0,0048 0,0026
Blend Bottom 30 0,0024 0,0035 0,0028

The total oxygen consists of the oxygen dissolved in the steel and the oxygen bonded in the oxide inclusions. Since the
dissolved oxygen is less than 0.001 wt% at the level of [Al] content (0.045 wt%) in the industrial steel according to the [Al]-
[O] equilibrium [13], the T.O content can be used to measure steel cleanliness. Figures 5 - 7 show the comparison of the
LECO analysis results between the initial steel and the samples with pre-fused flux and blended flux treated for 10 and 30
minutes.

2104 © 2019 by the Association for Iron & Steel Technology.

 Figure 5. Concentration of total oxygen in the steel before and after the treatment using different fluxes for 10 and 30 min
Figure 5 shows that with the treatment of pre-fused flux, T.O has decreased after 10 and 30 minutes of slag-steel interaction,
whereas with the treatment of the blended flux, T.O content in steel increased remarkably at 10 minutes, but decreased to a
low level at 30 min. The variation tendency of T.O is consistent with that of the inclusion number density and area fraction.
This further confirms that the pre-fused flux has a higher ability to improve the steel cleanliness than the blended flux for a
shorter treatment time (10 min). For the longer treatment time (30 min), the advantage of pre-fusing is not obviously
observed.

Figure 6. Concentration of sulfur in the steel before and after the treatment using different fluxes for 10 and 30 min
According to the results in Figure 6, both pre-fused flux and blended flux are able to remove sulfur from the steel by
approximately half after 10 minutes flux treatment. A further decrease in S content of the steel is observed with longer
interaction time. On the other hand, Figure 7 shows that there is almost no change in the C content of the steel after the
treatment with both fluxes. As evidenced by their chemical composition, the fluxes are not designed for steel decarburization.

Figure 7. Concentration of carbon in the steel before and after the treatment using different fluxes for 10 and 30 min

© 2019 by the Association for Iron & Steel Technology. 2105

Inclusion number density, size and area fraction
The number density, size and area fraction of the inclusions in the steel samples before and after the flux treatment with the
two different fluxes for 10 and 30 minutes are given in Table 4.

Table 4. Inclusion characteristics of the steel samples before and after flux treatment
Interaction time Inclusion number Inclusion Size (μm) Inclusion area
Position
(min) density (/mm2) Ave. Max. Min. Fraction (%)
Initial 0 - 26.2 2.95 5.76 0.81 0.0158
Top 24.6 4.18 8.31 0.84 0.0299
10 Middle 14.1 3.31 6.81 1.02 0.0107
Bottom 13.6 2.44 7.27 0.87 0.0056
Pre-fused
Top 11.5 4.11 8.11 0.84 0.0135
30 Middle 5.2 4.71 9.17 0.96 0.0081
Bottom 6.3 3.59 7.84 0.85 0.0056
Top 42.4 3.61 11.28 0.84 0.0384
10 Middle 36.6 3.85 16.33 0.89 0.0377
Bottom 33 2.35 6.14 0.82 0.0126
Blend
Top 16.2 3.10 6.26 0.85 0.0109
30 Middle 3.1 4.75 6.85 0.94 0.0049
Bottom 7.9 3.49 5.36 1.06 0.0066

Comparison of the inclusion number density, average size and area fraction of the steel samples before and after the treatment
with both fluxes is shown in Figure 8 to 10 respectively.

Figure 8. Comparison of the inclusion number density of the steel samples before and after the treatment with both fluxes
Figure 8 shows the variation tendency of the inclusion number density. Compared to the initial steel, the inclusion number
density decreased after the 10 min treatment with the pre-fused flux, and further decreased with the 30 min treatment. For the
blended flux, the inclusion number density increased by 42.5% after the 10 min treatment, but then decreased remarkably
with the 30 min treatment. The pre-fused flux is more effective with a shorter treatment time (10 min) than the blended flux
to absorb and remove inclusions from the steel. For the longer treatment time (30 min), the advantage of the pre-fusing flux is
not clearly observed. Figure 9 shows that there was no clear variation of the average inclusion size with time for both fluxes.

2106 © 2019 by the Association for Iron & Steel Technology.

 Figure 9. Comparison of the inclusion average size of the steel samples before and after the treatment with both fluxes

Figure 10. Comparison of the inclusion area fraction of the steel samples before and after the treatment with both fluxes
The inclusion area fraction shows the similar variation tendency as the inclusion number density (see Figure 10).
Additionally, the inclusion area fraction reduced from top to bottom for all the samples due to inclusion flotation. Based on
the data in Figures 8 and 10, it can be concluded that the pre-fused flux is more effective to remove inclusions when the flux
treatment is less than 10 minutes.
The morphology of the typical oxide inclusions in the steel sample after the treatment with the pre-fused flux for 30 min was
presented in a previous work [12]. There was no obvious difference in the inclusion morphology between the steel samples
after the treatment with pre-fused and blended fluxes for 30 min treatment time.

TiO2 content in the inclusions and number density of the Ti containing inclusions
Table 5 lists the average chemistry of the inclusions in the different steel samples. The TiO2 content in the inclusions
increased significantly from 7.9 wt% to 15.7 wt% after the treatment with the blended flux for 10 min, and then decreased to
0.8 wt% after 30 min treatment with this flux. In contrast, the TiO2 content in the inclusions continuously decreased with time
in case of the pre-fused flux treatment, i.e. from the initial 7.9 wt% to 5.1 wt% after 10 min and to 0.6 wt% after 30 min.

Table 5 Average chemical composition of the inclusions in the different samples (wt%)
Interaction Inclusion composition
Sample
time (min) Al2O3 SiO2 MnO Cr2O3 TiO2 MnS
Initial Steel 0 90.4 0.1 1.1 0.1 7.9 0.4
10 87.6 0.9 5.6 0.2 5.1 0.4
Pre-fused
30 67.0 7.4 9.6 5.0 0.6 8.7
10 78.7 0.4 3.6 0.1 15.7 0.8
Blend
30 84.6 1.0 12.0 0.5 0.8 0.6

© 2019 by the Association for Iron & Steel Technology. 2107

The variation of the number density of the TiO2-containing inclusions is shown in Figure 11. The number density of the
TiO2-containing inclusions decreased after the treatment with the pre-fused flux, whereas with the blended flux it increased
after the treatment for 10 min, and then decreased after the 30 min treatment.

Fig. 11 Evolution of the TiO2-containing inclusion number density during the treatment with both fluxes
According to the equilibrium distribution ratio of Ti between slag and steel at the present experimental condition, Ti should
move from the steel to the slag during the treatment. The consequent decrease in Ti content in the steel would cause a
decomposition of the TiO2-containing inclusions in the steel during the treatment, resulting in a decrease of the TiO2 content
in the inclusions. Hence, a decrease in number density of the TiO2-containing inclusions would be expected for both fluxes.
However, a large amount of TiO2-containing inclusions was observed in the test with blended flux for 10 minutes. The
increased number of TiO2-containing inclusions is likely caused by the introduction of TiO2 particles into the steel coming
from the chemical reagent grade TiO2 powder used in the blended flux, which was added to the blend recipe in order to match
the content of the perfused flux (see Table 2). After the 30 min treatment, however, these TiO2 inclusions have floated to and
dissolved into the top liquid slag, resulting in the expected decrease in TiO2-containing inclusions. Actually, the
contamination extent of steel by using the blended and pre-fused fluxes may be traced/indicated through the evolution of
TiO2 content in the inclusions (see Table 5) and that of number density of the Ti containing inclusions (Figure 11) during the
experiments. In the case of the pre-fused, since the TiO2 is dissolved in the main constituting mineral, this contamination was
eliminated for the 10 min treatment time. The results in Table 5 and Figure 11 confirm that with a shorter time (10 min)
treatment the blended flux brings contamination to the steel, whereas the pre-fused flux not only can protect the steel from
that contamination, but also further removes the existing inclusions and other impurity elements (i.e. sulfur).

Effect of the flux mineralogy on the steel cleanliness

The mineral phases, which were analyzed using X-ray diffraction (XRD), in the two flux types are listed in Table 6. The
mineralogical quantification was carried out using the Rietveld method by the Topas software. Pre-fused flux is mainly
composed of 12CaOꞏ7Al2O3, which has a low melting temperature as shown in Figure 12. Because this melting temperature
is much lower than the experimental temperature of 1600 °C, the main part of pre-fused flux could melt rapidly and has
therefore a longer time to interact with liquid steel (e.g. absorb S and inclusions from the steel). In the blended flux, the main
phases are CaO and corundum, with a small fraction of Ca(OH)2 which will decompose into CaO and H2O at high
temperature. CaO, and corundum are all high-melting-point phases (Figure 12). Thus, after the addition of the flux into the
molten steel, the phases in these fluxes need a relatively long reaction time to form low-melting-point phases, leading to a
slow melting of the flux. In other words, the low-melting-point phases in pre-fused flux increase the melting rate of the flux
and facilitates rapid removal of inclusions. The rapid melting of the pre-fused flux provided a longer time to absorb
inclusions, so the inclusion number density was already reduced within 10 minutes as shown in Fig. 11.

2108 © 2019 by the Association for Iron & Steel Technology.

 Table 6. Mineral phases in the fluxes (wt%)
Mineral Phase Prefused Mixed Melting Point
12CaOꞏ7Al2O3 98.5 0 1455 °C*
CaOꞏAl2O3 0 1.1 1605 °C*
2CaOꞏSiO2 1.5 2 2130 °C**
CaO 0 40.8 2572 °C**
Ca(OH)2 0 4.2 -
Corundum 0 49 2054 °C**
Quartz 0 0.7 1723 °C**
Periclase 0 1.5 2825 °C**
Hematite 0 0.05 -
* Data were obtained from ACerS-NIST Phase Equilibria Diagrams 3.1.0 database.
** Data were obtained from FactSage 7.1 database (FactPS and FToxid).

2100

2000

1900 Liquid phase

1800

CaO(s) + Liquid
Temperature (°C)

1700
LDSF Process Temperature
1600

1500

1400
C12A7(s)

Al2O3(s)
1300
C3A(s)

CA2(s)

CA6(s)
CA(s)

1200

1100
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Al2O3 /(CaO+Al2 O3) (g/g)

Figure 12. CaO – Al2O3 phase diagram. The arrow indicates the composition of a pre-fused flux also known as LDSF®

CONCLUSIONS
The effect of the preparation method of the flux on the steel cleanliness (in terms of the inclusion characteristics, including
number density, size, composition and impurity content) was investigated. The pre-fused flux shows a better performance of
removing inclusions and sulfur from liquid steel due to the predominant presence of the low-melting point phase
12CaOꞏ7Al2O3, resulting in a fast melting and a consequent ability to remove the inclusions rapidly. In the case of the
blended flux, the steel cleanliness was deteriorated after 10 min treatment due to the introduction of oxide particles with high
melting point from the flux. In summary, the present additional tests confirm that with a shorter treatment time (10 min) the
blended flux (mechanical mix of reagents) brings contamination to the steel, whereas the pre-fused flux (homogeneous
mineralogy) treatment not only can protect the steel from that contamination, but also further removes the existing inclusions
and other impurity elements (e.g. sulfur). With a longer treatment time (30 min) the advantage of the pre-fused flux is not any
longer clearly observed.

ACKNOWLEDGEMENT
The authors would like to thank ArcelorMittal Ghent for the steel sample and Aperam Isbergues for the analysis of T.O, C
and S in the steel samples.

REFERENCES
1. P. Yan, X. Guo, S. Huang, J. Van Dyck, M. Guo, B. Blanpain, “Desulphurization of Stainless Steel by Using CaO–
Al2O3 Based Slags during Secondary Metallurgy”, ISIJ International, Vol. 53 (2013), No. 3, pp. 459-467.
2. C. Wöhrmeyer, E. Elorza-Ricart, R. Jolly, S. Guichard, C. Brüggmann, A. Sax. “The impact of synthetic slags on steel
ladle refractory lifetime”. 51st International Colloquium on Refractories, Aachen, 2008.

© 2019 by the Association for Iron & Steel Technology. 2109

3. E. Pretorius, “Ladle refining and slag conditioning”, Electric Furnace Steelmaking Seminar AISE, 1997.
4. H. Suito and R. Inoue: ISIJ Int., 1996, vol. 36, pp. 528-536.
5. P.K. Iwamasa and R.J. Fruehan: Metall. Mater. Trans. B, 1997, vol. 28B, pp. 47-57.
6. S. Sridhar and A.W. Cramb: Metall. Mater. Trans. B, 2000, vol. 31B, pp. 406-410.
7. P.C. Yan, S.G. Huang, L. Pandelaers, J. Van Dyck, M.X. Guo, and B. Blanpain: Mater. Trans. B, 2000, vol. 31B, pp.
406-410.
8. B. Yoon, K. Heo, J. Kim, and H. Sohn: Ironmak. Steelmak., 2002,vol. 29, pp. 214–17.
9. J. Björklund, M. Andersson, and P. Jönsson: Ironmak. Steelmak., 2007, vol. 34, pp. 312-324.
10. B. Coletti, S. Smets, B. Blanpain, P. Wollants, J. Plessers, C. Vercruyssen, and B. Gommers: Ironmak. Steelmak., 2003,
vol. 30, pp. 217-222.
11. M. A. T. Andersson, P. G. Jönsson and M. Hallberg: Ironmak. Steelmak., 2000, vol. 27, pp. 286.
12. B. Zhang, A. Malfliet, R. Lencina, B. Touzo, M. Guo, “Effect of ladle refining flux on the cleanliness of low carbon
low alloy steel”, 10th International Conference on Clean Steel, Budapest, 2018.
13. M.K. Paek, J.M. Jang, Y.B. Kang, and J.J. Pak: Mater. Trans. B, 2015, vol. 46B, pp. 1826-1836.

2110 © 2019 by the Association for Iron & Steel Technology.