MgO Based Inclusions in Aluminum Killed Steels

Neil Sutcliffe and Michael Callahan

Nucor Steel Gallatin, 4831 U.S. Hwy. 42 West, Ghent, KY 41045-9704

Email: [email protected]; [email protected]

Keywords: Inclusions, cleanliness, MgO, Al-killed

ABSTRACT
The current paper presents some aspects of MgO containing inclusions found in Al killed steels produced in a CSP based
plant. The paper details the theoretical aspects of inclusion formation, SEM analysis of inclusions, plant studies and the
resultant practice modifications. MgO containing phases are produced from the refractories, slags and steel. Determining the
source and then the method for controlling resultant MgO inclusions is also outlined.
INTRODUCTION
Aluminum killed steels inherently contain some forms of inclusions; the understanding and management of them determines
whether the steel will meet the final customers’ expectations and requirements or not. The use of aluminum leads to low
oxygen levels in the steel and variable quantities of Al 2 O 3 inclusions; these are controlled by calcium addition practices and
process controls and limits. Low oxygen and basic slags lead to spinel formation which requires modification and removal.
The holding vessel for these practices is not inert to these processes and as temperature changes with time, reactions occur
within the system.
At Nucor Steel Gallatin, we found that MgO containing inclusions were resulting in clogging at the caster and were
detrimental to quality. This investigation was conducted to understand why MgO containing inclusions formed and thus how
to control their formation. Numerous reviews of inclusion control within plant environments (1-5) and laboratory research (6-9)
have provided a significant data base of knowledge. The key to success is transforming this to understanding and application
in the plant environment, in order to produce steel which has a sufficiently low inclusion loading to exceed customer
performance expectations.
SAMPLE ANALYSIS
As part of the routine quality assessment of steel produced at Nucor Steel Gallatin (NSGAL) lollipop samples are taken from
the tundish on specific steel grades as well as all heats with numbers ending in “0” and “5”. The steel samples are then
sectioned as shown in figure 1, polished and subjected to automated SEM analysis (10). The SEM is an ASPEX PSEM with a
30 mm2 LE EDS Detector with the operating parameters shown in table 1 and table 2. A routine report shows ternary
diagrams of Al-Ca-Mg, Ca-S-Al and Ca-S-Mn with inclusion sizes (0.5-2.5, 2.5-5, 5-15 and > 15 µm) as well as inclusion
counts, internal classifications and size summaries. There are similar reports presented in literature from Nucor plants (3, 11, 12),
Vesuvius (13) (ASPEX), Arcelor Mittal (5) and U.S. Steel (2, 14) (using ASCATTM) highlighting the widespread use of the
computer controlled SEM (CCSEM) analysis techniques.
If the routine analyses indicate items of interest then the SEM can be operated manually for further investigation. The Nucor
internally developed software described by Pretorius et al(1) is used for further study and manipulation of measured data as
well as to present results in this paper.
 Table 1: ASPEX PSEM operating parameters.
Beam Emission Spot size Magnifi- Search time Search Grid Step size Optimal
Energy current cation Per pixel Resolution Step size(15)
20 kV 40µA 37.3 450X 4µs 512x512 0.88µm 1.414µm

Table 2: LE EDS Detector operating parameters

Threshold Minimum size Stub area Min counts Max counts Threshold Time
Search Measure per Pixel
0-165 2.0µm 28mm2 1000 2500 0-175 16µs

Figure 1: Sketch of lollipop sampler showing section for CCSEM. Sample holder on the right, sample between 7 and 8 has
the Al strip for reference.
Harris et al (15) show the development of settings for representative analysis techniques for lollipop type samples and the step
size vs. minimum inclusion size specified in table 1 are within their recommendations. Maddalena (13) uses similar settings but
covers a larger scan area than the NSGAL standard.
Slag samples were milled and cold pressed in a Spex machine and subsequently analyzed in a Panalytical XRF. Nitrogen
was measured by Leco and the steel analysis is performed on an OES from Thermo Fisher. The NSGAL laboratory is
certified to ISO17025, A2LA registrar. ICP chemical analysis was conducted on several steel samples for Ca and Mg.
The grades forming the basis of the present study are medium carbon, aluminum killed, silicon and manganese bearing
grades and are shown in table 3. FS 6.4 is the steel chemistry used in the FactSage 6.4 calculations.

Table 3: Steel grades studied in this work.

Grade C Mn Si Al Cr
Type 1 0.26 1.05 0.20 0.035 Residual
Type 2 0.195 0.85 0.20 0.018 Residual
Type 3 0.26 1.3 0.24 0.030 0.35
FS 6.4 0.25 1.1 0.25 0.035 Ca, Mg, O, S

PROCESS ROUTE AND BACKGROUND

The plant consists of two DC arc furnaces with eccentric bottom tapping. The heats (175 metric ton) are tapped live into
preheated mag-carbon lined ladles. The heat is then killed by means of Al addition with slag former additions for
desulphurization and then bulk alloy additions are made. The aim slag chemistry is shown in table 4. The ladle stirring is
achieved by means of a single porous plug. The heat is further processed at the 22.3 MVa AC LMF where further alloying, a
fixed Ca wire addition and metallurgical stir are applied before being shipped to the single strand thin slab caster. The final
average slag chemistry is shown in table 4.

Table 4: Aim start slag chemistry and final LMF chemistry

Al 2 O 3 CaO FeO MgO MnO SiO 2
Aim <30 54.5 <1 <6 <1 4
Start
Final 28.5 53.9 0.87 5.8 0.19 3.6
Standard lollipop samples are taken from the 22 metric ton tundish at 100 metric tons cast. CCSEM analysis is conducted on
the lollipop sample. The tundish has an impact pad, continuous temperature measurement and has a stopper rod for flow
control to the mould. Cast widths vary from 990 to 1626 mm (39-64 inch) and the strand exit thickness is fixed at 65mm.
Casting speed is width and grade dependent and can vary from 4.57 to 5.33 m/min (180-210 imch/min)for medium carbon
grades. Figure 2 shows the process route for NSGAL.

Figure 2: Process route at Nucor Steel Gallatin. Kill after tap, desulphurization and bulk alloy additions on the car. Alloy
trimming, Ca addition and stir at LMF. Single strand thin slab caster.
The process route has an influence on the inclusions found and generated during the steelmaking operation (4, 5, 14, 16-20). Story
et al (14) showed that lance stirring can influence the proportion of solid vs. liquid calcium aluminates with calcium treatment
and Mendez et al (4) showed the impact of bottom stirring energy on MgO content in inclusions; both papers support the
practice of gentle rinse / stir after Ca additions. Adaba et al (16) highlights the difference in inclusion levels for kill on tap
versus kill after tap in the same plant. The kill on tap practice generates spinel versus alumina in the kill after tap technique.
Careño-Galindo et al (19) also showed the kill on tap practice, modeled plant results and then predicted the impact of varying
the order and type of additions during tap. The result was the formation of manganese silicate and alumina in the steel with
varying quantities generated based on order of addition. Pretorius et al (18) and Story et al (14) highlighted the influence of high
and low levels of Ca additions and the impact on the type of inclusions generated. High levels can lead to CaS formation and
stopper erosion or clogging, whereas low levels can lead to incomplete modification of Al 2 O 3 . It was also presented that the
total level of Ca does not reflect how effective the treatment of the steel had been (14). Mendez et al (4) and Graham and
Irons(17) both discuss the influence of FeO and MnO contents in the ladle slag and the resultant level of MgO in inclusions,
with low levels of FeO and MnO (sum < 1%), giving the highest levels of MgO in the inclusions. Alatarvas et al (20) used
CCSEM type analysis and electrolytic extraction techniques to study the inclusions generated in a vacuum and calcium
treated Al and Si killed steel. They observed higher levels of MgO rich inclusions in higher Ca containing steel. Kaushik et al
(5)
studied the impact of different processing routes and steel grades on the resultant cleanliness levels. They found that high
Al steels with a low total oxygen content did have Ca-free magnesium spinels, however with increasing calcium levels, some
calcium containing spinels formed and these heats had a tendency to experience clogging(5).
The inclusion evolution through the process from tap to cast was assessed on several heats at NSGAL; the results of one heat
are shown in figure 3(a) through 3(f). The inclusions in the E01 (after tap and kill) sample are predominantly Al 2 O 3 based
with some Calcium aluminate inclusions from the ladle glaze (see below). After 10 minutes stir and the addition of high
carbon ferro-manganese the L01 (L0* defines LMF) sample indicated the presence of spinel inclusions. The L02 sample was
taken after an addition of carbon, Al cone and FeSi for trimming. L02 shows an enlargement of the spinel inclusions and an
increase in the Al 2 O 3 levels again. The L03 sample is taken shortly after the Calcium wire addition. There is a significant
change in the average inclusion chemistry as indicated in figure 3(d) and 3(f). Also note the formation of MgO rich
inclusions in figure 3(d). The C02 (tundish) sample is taken from the tundish after approximately 100 metric tons has been
cast. Note that the average inclusion size has decreased and the spread in analysis has decreased. Table 5 presents the average
inclusion chemistry and table 6 the steel chemistry of the samples from figure 3.

a) b)

c) d)
 e) f)Average analyses per sample
Figure 3: Samples through the process

Table 5: Average inclusion chemistry for the samples in figure 3.

Sample %Mg %Al %Si %S %Ca %Ti %Mn Ca/Al
E01 0.56 66.71 1.37 2.93 6.55 1.56 11.19 0.1
L01 14.3 45.44 3.3 2.11 17.25 0.65 11.71 0.38
L02 15.05 40.68 5.43 1.91 20.09 0.95 9.69 0.49
L03 6.55 16.79 6.19 2.26 42.24 2.09 15.27 2.52
C02 4.85 29.72 4.15 2.86 44.23 2.86 8.48 1.49

It will be noticed in table 5 that the Mn content of the inclusions is quite high. The Mn is normally associated with S and is
precipitated on to the existing inclusions during solidification of the sample.

Table 6: Chemical analysis of samples in figure 3(a) to 3(e).

Ca* Ca# Mg# C Mn P S Si Al N
E01 4 30.4 4.4 0.0543 0.1222 0.0063 0.0391 0.0011 0.2446 0.0121
L01 1 10.6 1.8 0.2118 1.0474 0.0099 0.0078 0.1201 0.0186 0.0072
L02 6 16.4 3.7 0.2463 1.0244 0.0090 0.0030 0.2126 0.0396 0.0072
L03 27 26 6 0.2543 1.0155 0.0100 0.0031 0.2105 0.0338 0.0073
C02 17 20.3 3.8 0.2434 1.0466 0.0102 0.0029 0.2104 0.0347 0.0076
Ca and Mg are ppm, balance of analysis in wt. %; OES analysis = *; ICP analysis = #;
Figure 3a shows large quantities of alumina inclusions due to the primary deoxidation of the bath. With continued stirring
and the addition of slag formers, alloys and CaC 2 to kill the slag, the alumina inclusions are modified to MgO·Al 2 O 3 spinels
(figure 3(b) and 3(c)). Pretorius et al (3) proposed that the primary mechanism of spinel formation is via the slag/metal
interaction generating Mg in the steel. Kim et al (21) model the formation of spinel due to the reduction of slag and refractory
by Al in the steel with the slag contribution being the largest. Mu et al (6) measured the reduction of MgO and CaO by Al in
the steel and again show the large contribution of slag reduction as the source of Mg in the steel. Harada et al (7) show Mg
levels increase in the steel with and without slag in the system indicating again that both slag and refractory reduction can
occur. Yang et al (9) present an extensive review of the mechanisms and theory of spinel formation and several basic reactions
are presented in table 7; where {} represents refractory species, ( ) slag species and [ ] steel species. Key aspect from all of
the equations is the availability of Mg in the steel.
 Table 7: Reactions involved in the spinel formation.
3{MgO} + 2[Al] = 3[Mg] + Al 2 O 3 (1) Reduction of refractory by Al in the steel.(7)
{MgO} +{C} = Mg (g) + CO (g) (2) Reduction of oxide by carbon in the brick.(22,23)
Mg (g) = [Mg] (3) Condensation of Mg gas to Mg in the steel.(22)
3(MgO) + 2[Al] = (Al 2 O 3 ) + 3[Mg] (4) Reduction of MgO in the slag by Al in the steel.(7)
[Mg] + 2[Al] + 4[O] = MgO·Al 2 O 3 (5) Reaction between Al and Mg in the steel.(7)
(MgO) + [Ca] = (CaO) + [Mg] (6) Reduction of MgO in the slag by Ca from CaC 2 addition.(3)
3[Mg] + 4(Al 2 O 3 ) incl = 3Mg·Al 2 O 4 + 2[Al] (7) Reaction between inclusion and Mg in the steel.(3,24)
MgO·Al 2 O 3 + 3[Mg] = 4MgO + 2[Al] (8) Al-Ca-Mg-O equilibrium(25)

Using FactSage 6.4, equilibrium calculations for the slag and steel were conducted at 1580oC (temperature and chemistry for
steel and slag prior to Ca addition) and the resultant Mg in solution is 9.7 ppm for a 4% MgO slag and as high as 18 ppm for
a 7.5% MgO containing slag. The corresponding Ca levels were 9.8 and 10.5 ppm; samples from the plant, tested with ICP,
showed lower Mg levels, but within the equivalent ranges. Ca levels in the plant were higher just after tap, but dropped at
L01 and L02 (prior to Ca wire addition).
After the Ca wire addition (figure 3(d)) the inclusion distribution changes again as the spinel and alumina inclusions have
been modified. Pistorius et al (8) highlighted that Ca replaced both Al and Mg in MgO.Al 2 O 3 spinels resulting in liquefaction
of the inclusion. Pretorius et al (3) adopted Ca reduction of the MgO as the predominant mechanism (reaction 9 in table 8).
Yang et al (9) shows a three route process for the modification of an MgO.Al 2 O 3 spinel in figure 4 and Deng and Zhu (24)
concentrate on the path 1 and path 2 of reference 9 and utilize the un-reacted core model to describe the modification process.

Figure 4: Inclusion modification paths after Yang et al (9). MA = MgO·Al 2 O 3 ; CA = CaO-Al 2 O 3 ; MAC = MgO-CaO-Al 2 O 3 .

Table 8: Proposed reactions involved in the modification of spinel (3, 9, 24, and 25)
y(MgO·Al 2 O 3 ) + x[Ca] = (xCaO.yAl 2 O 3 ) + y[Mg] (9)
[Ca] + [O] = (CaO) (10)
2[Al] + 3[O] = (Al 2 O 3 ) (11)
x[Ca] + 2y[Al] +(x+3y)[O] = (xCaO.yAl 2 O 3 ) (12)
x(CaO) + y(Al 2 O 3 ) = (xCaO.yAl 2 O 3 ) (13)

Reaction 9 in table 8 is the main reaction for the spinel modification (3, 9, 24, and 25) and reactions 10 through 13 and 4 (table 7)
can occur at the same time. As the liquid layer around the inclusion core increases in thickness, the rate of Mg diffusion into
the steel is reduced, thus path 1 (partial diffusion) and path 2 (full diffusion) type final inclusions (9).
Path 1 in figure 4 has a spinel reacting with Ca to form a mix of MgO-CaO-Al 2 O 3 and a plant sample is shown in figure 5.
Plant samples have shown this as the main type of inclusion (figure 3 (d) and (e)).

Oxygen Magnesium

Aluminum Calcium Sulphur

Figure 5: Example of path 1 modification of the spinel. CaS formation is also evident.

Oxygen Magnesium
No Sulphur in this
Inclusion.

Aluminum Calcium

Figure 6: An example of path 2 spinel modification.

Path 2 type reaction is the reduction of MgO from the spinel and the diffusion of the Mg into the steel. This is evident in
figure 6 that shows a small MA core surrounded by CA slag. The spherical nature of the inclusion indicates a liquid
inclusion. A detailed study of the different concentrations of elements from the middle to the edge of individual inclusions, as
performed by Deng and Zhu (24) was beyond the scope of the current work.
Path 3 inclusions are very similar to path 2, however they are not normally seen in tundish (as in figure 3 e) samples or in coil
samples, suggesting that they are really path 2 inclusions that haven’t reacted completely.
Verma et al (26, 27, and 28) present the mechanism of spinel modification via a transient CaS formation, as shown in reaction 16 in
table 9. They (28) indicated that the rate of modification is influenced by the level of S in the steel, with lower S levels leading
to faster modification due to Ca reacting primarily with the MgO in the spinel, rather than the S in the steel. Local levels of S
will however influence the formation of CaS on the modified inclusion.

Table 9: Reactions involved in the transient CaS mechanism of Verma (26,27,28)

[Ca] + [S] = CaS (14) Initial wire addition
[Ca] + (MgO) = (CaO) + [Mg] (15) Reduction of MgO from the spinel
3CaS + (Al 2 O 3 ) = (CaO) + (Al 2 O 3 ) + 2[Al] + 3[S] (16) CaS reacting with the oxides
On the heat presented in figure 3a through 3f, the MgO content of the inclusions is low and this heat did not show any
indication of clogging at the caster. When the steel has high levels of MgO based inclusions, the caster experiences some
clogging, as indicated in figure 7.

Figure 7: Caster parameters and indication of SEN clogging – increases in stopper rod opening with constant casting speed.
The SEM results of the final LMF sample, tundish sample and coil samples from a heat that showed the type of clogging
illustrated in figure 7, are shown in figure 8(a) through 8(c). The average inclusion chemistry is shown in table 10.
 Figure 8:
a) L03 denotes ladle sample
b) C02 denotes tundish sample
c) CL denotes coil sample.

Figure 8: CCSEM analysis of samples from a heat that exhibited clogging at the caster.

Table 10: Average Inclusion chemical composition of samples in figure 8.

Sample %Mg %Al %Si %S %Ca %Ti %Mn Ca/Al
L03(ladle) 45.25 21.67 1.45 3.13 20.03 0.8 3.9 0.92
C02 RR(caster) 20.28 22.91 3.39 10.97 37.14 0.45 2.56 1.62
CL (Coil) 16.09 20.73 1.08 18.25 37.55 0.5 3.4 1.81
The Mg content of these inclusions is significantly higher than the inclusions shown in table 5 and figure 3. The S levels in
the CL sample (coil) are impacted by the precipitation of CaS and to a lesser extent MnS on the existing inclusions on
solidification. There is some indication of reoxidation from ladle to tundish as there is an increase in < 2µm inclusions (16).
However, the surface area measured in the three samples is different and thus the inclusion count is not an ideal indicator of
steel deterioration or chemical reactions. The L03 sample was taken a few minutes after the Ca wire addition at the LMF.
The CaO-Al 2 O 3 inclusion levels are lower than the tundish (C02) sample. In the tundish sample there is a noted shift in the
inclusion chemistry from the Al 2 O 3 corner and the MgO.Al 2 O 3 forms of inclusions, i.e. Ca has had time to modify some of
the inclusions. There has been an increase in the size of the MgO rich inclusions. The CL sample had the same inclusion
types as the tundish sample indicating that there was no interaction with the casting powder or refractories.
Figure 9 shows an example of the MgO rich inclusion normally associated with clogging at the caster as well as historically
with some weld rejects at pipe customers.

Oxygen Magnesium

Aluminum Calcium Sulphur

Figure 9: MgO rich inclusion from a heat that showed SEN clogging at the caster.
The central core of this type of inclusion is MgO and it is surrounded by CA; note the presence of S on the edges of the
inclusion. Other inclusions do not have the S present; the MgO core surrounded by CA of varying thicknesses is the common
trend. When the CCSEM results indicate this type of inclusion and there is clogging at the caster, the heat in question will be
downgraded or diverted from the original customer.
The balance of this paper will describe how these MgO particles originate and what modifications were made to plant
practices to control them.

MgO RICH INCLUSIONS

Ladle Glaze
Plant based investigations conducted by researchers at the Royal Institute of Technology (KTH) Sweden (29-31) have
highlighted the interactions between ladle glaze and steel. Tripathi (29) presents a mechanism for glaze formation whereby the
liquid top slag adheres to the ladle walls on draining. On subsequent heats the glaze has the potential to react with the steel
thus producing inclusions. Beskow et al (30) analyzed glazed ladle samples where they observed a four layer structure – un-
reacted MgO-carbon system, a carbon depleted MgO layer, and a slag infiltrated layer and an outer glaze layer of slag.
Refractory samples were taken from a used ladle below the slag line area. Samples from below the slag line showed minimal
slag coating in the order of 1 to 2 mm. Figure 10 shows the elemental mapping of some used refractory. Samples of the
“glaze” were scratched off the used brick and sent for XRD analysis. The sample indicated Periclase (MgO), Graphite and
Calcium aluminate in the form Ca 12 Al 14 O 33 (Mayenite or C 12 A 7 ), shown in figure 11. The presence of graphite was found in
some of the samples subjected to SEM analysis, but it was within the un-reacted brick. Lehmann et al (22) showed the
formation of a dense layer of MgO on the brick surface in an Al killed steel, however, the plant based samples indicated a
coating of slag and the penetration of slag into the refractory brick as presented by Beskow et al(30).
Beskow et al (31) also presented the impact of ladle life on inclusion generation. It was found that with age the ladle glaze
became thicker and more porous leading to greater penetration by slag and increased slag refractory reactions leading to
greater inclusion formation. In contrast, plant results at NSGAL have not shown an increase in inclusion generation with
ladle life. This may be as a result of less penetration of the refractory by the slags. Also note that the work in Sweden (29-
31)
was conducted in a plant with a vacuum degasser, which would cause more turbulence around the slag line in particular in
comparison to an LMF process. Du Sichen et al (32) did show high MgO contents in inclusions with new ladles and at
NSGAL this was also observed and is discussed below.
Figure 10: Slag line sample of refractory / slag interface. Bottom right corner of each image is the un-reacted refractory brick.
FactSage 6.4 was utilized to study the influence of slag cooling and solidification under equilibrium conditions as it coats the
ladle during draining and the results are presented in figure 12. It will be noted that the slag being utilized in the ladle is lime
saturated and has been developed for improved stirring and alloy addition practices at the ladle furnace. The ladle slag
precipitates out CaO on cooling and it is only at much lower temperatures that other species start to form. In contrast to the
XRD analysis, FactSage modeling indicated CaO, Ca 3 Al 2 O 6 and MgO at steelmaking temperatures. FactSage models
equilibrium conditions whereas the plant sample would have experienced a different cooling rate; the slag chemistry used in
the FactSage calculations may not have been identical to the last slag in the ladle. Both results indicate a liquid calcium
aluminate coated the ladle during draining. Thus when steel is tapped into a ladle with a slag glaze the glaze will melt and be
stirred into the heat, MgO grains from the slag infiltrated area of the brick, may be washed into the steel.

Figure 11: XRD analysis of ladle glaze.

 Figure 12: FactSage study of ladle slag solidification.
Mu et al (6) in laboratory studies found the reduction of MgO from both slag and refractory by Al in Fe-Al alloys down to the
0.04% Al level in the samples. Inclusions were spinel based in MgO crucibles at 0.04wt% Al whereas at higher Al levels,
there were indications of Mg rich inclusions at equilibrium. In the plant, during killing and desulphurization, there will be
high localized [Al] that can lead to MgO reduction from the ladle glaze and the refractory, thus MgO loading in the steel will
increase.

Ladle Slag
Historically it had been noted at NSGAL that with >6.5% MgO in the LMF final slag, there was an increased potential for
MgO based inclusion formation as well as clogging at the caster. FactSage 6.4 was used to produce the multi-component
phase diagram for the LMF slag shown in figure 13.

Figure 13: SiO 2 -Al 2 O 3 -CaO-MgO quaternary phase diagram at 5 (red) and 7% MgO (blue) generated with FactSage 6.4 at
1575oC.
The ladle slag is composed primarily of carryover slag from the furnace on tapping; Al 2 O 3 from the Al killing practice;
desulphurization mix; products from the calcium carbide addition and lime added to control temperature and slag thickness
during final processing at the LMF.
In the plant environment, there are other components within the slag system – F, Na, FeO, MnO etc. These components will
modify the liquid phase field of the slag in contact with the steel in the ladle. However what is important is that as MgO
increases, Periclase forms in the system – i.e. MgO will precipitate as a stable phase.
To avoid the MgO rich phase of the slag system we are working with, it is theoretically possible to modify the slag system.
However, the requirements of the process still need to be achieved, i.e. desulphurization, insulation, inclusion capture
refractory protection and minimal re-entrainment. If the ladle slag had an addition of SiO 2 , it would be possible to form di-
calcium silicate (Ca 2 SiO 4 ) in figure 13 or to form a liquid inclusion CAS2 (Anorthite) at around 50% SiO 2 . However if more
SiO 2 was added to the slag, then the desulphurization ability of the slag would be reduced due to a dilution of CaO.
Reduction of SiO 2 by Al in the steel would also occur and the required steel chemistry would not be achieved. If more CaO
was added, the slag would crust over more quickly – the liquid field below 1600oC only allows 54 to 55% CaO in solution
before crusting would occur. On the Al 2 O 3 side of the phase diagram there is more room for slag modification. The primary
inclusion formation from the slag would be spinel. The spinel is modified by the Ca addition to form liquid inclusions, as
outlined above, which are more easily removed by flotation. However the slag is thin and can lead to greater temperature
loss and some refractory attack. The normal practice at the LMF for a thin slag is the addition of lime during the process.
Figure 14 shows the average %MgO in the LMF final slag from January 2012 to December 2014. Over the years there have
been several modifications to the ladle slags to reduce costs, improve desulphurization, improve product quality and improve
tundish temperature control.

Figure 14: MgO levels in LMF slag samples

(33)
As shown by Yin et al line pipe material is tested with ultrasonic equipment and a series of rejection criteria are utilized by
manufacturers. Figure 15 presents rejection rates for “stringers” within or close to the weld metal. On-site analysis showed
that some of the stringers were associated with MgO rich inclusions, Wang et al (34) classes them as B type inclusions and
Story and Asfahani(2) use tears and laps to describe similar but larger defects. In figure 15 it can be seen that in 2010, the
levels of rejections were high and by mid-2011, the shipment levels had dropped significantly. Due to the quality
deterioration, a number of investigations were conducted into tundish designs, shrouds, slide-gate systems, ladle slag
detection as well as steelmaking practices. The high MgO content inclusions were a contributor to the quality problems but
not necessarily the only problem. The ladle slags were modified in 2011 and again in 2012 with a resultant decrease in
rejections at the customer. With further modifications in 2013, the levels of defects at the customer stayed low and the
quantity of material shipped to them increased.
 Figure 15: Customer rejections and shipped material
In May and June 2014, we had an increase in MgO content in the ladle slag this also resulted in an increase in MgO rich
inclusions in the CCSEM results and incidences of clogging. The problem was eventually isolated to one furnace and a
difference in the rapid tilt back speed after tap between it and the other furnace. The slower tilt back speed resulted in a
higher volume of furnace slag carryover in the ladle and thus higher MgO levels in the ladle slag. With adjustment of the tilt
back speed as well as an indicator on the system, the MgO levels within the ladle slag were reduced.
As highlighted in table 7 and the discussion above, the primary driving force for spinel formation is the presence of Mg in the
steel. FactSage 6.4 was used to study the influence of slag chemistry on the slag/steel equilibrium at various Mg levels in the
steel and the results are shown in figure 16. Even at low levels of Mg in the steel, a 7.5% MgO slag will have a driving force
for MgO particles to be present in the steel. As can be seen in figure 3 e, the NSGAL additions and slags result in a Ca rich
inclusion system as indicated by the CaO particles with both slag chemistries.

Figure 16: Results of slag / steel equilibrium calculations.

Table 11 shows the average ladle slag chemistries for the three types of steel covered by this investigation. The restricted type
1 is the average chemistry of heats that showed clogging and had high levels of MgO based inclusions as represented by
those heats in figures 7 and 8 respectively. Table 12 and figure 17 show a trial slag and resultant caster parameters showing
no indications of clogging.
Table 11: Ladle slag chemistries.
Grade Al 2 O 3 MgO SiO 2 CaO FeO MnO P2O5
Type 1 28.50 5.81 3.62 53.93 0.87 0.19 0.03
Type 2 27.07 5.68 4.94 55.19 0.59 0.13 0.03
Type 3 31.16 5.39 3.21 54.29 0.49 0.14 0.03
Restricted 27.97 6.69 3.95 53.90 0.35 0.10 0.02
Type 1

Table 12: Trial slags

Grade Al 2 O 3 MgO SiO 2 CaO FeO MnO P2O5

Type 1 29.81 5.61 4.22 54.40 0.36 0.13 0.02
Type 3 29.15 5.46 4.38 53.10 0.84 0.13 0.02

Figure 17: Caster parameters showing no evidence of clogging.

Ladle life
It was noted that the MgO level in slags for new ladles ran 1 to 1.5% higher than older ladles. Figure 18 shows the mean and
variation around that mean, MgO levels in ladle slags vs. the age of the ladle.
Figure 18: Distribution of MgO in ladle slags vs. ladle life.
Samples were taken at the tundish on the first 4 heats for 5 new ladles. The MgO rich inclusions were seen on the first and to
a partial extent on second heat as indicated in figure 19; however there were indications for clear first heats and poor second
heats; clear first, second and poor third heats on the other ladles.

First heat on the ladle Second heat on the ladle

Third heat on the ladle Fourth heat on the ladle

Figure 19: CCSEM plots for heats used on the first four lives of a ladle.
To compensate for the increased MgO levels in the slag and the potential for MgO rich inclusion formation, the
desulphurization mix and hence final LMF slag was modified by increasing the calcium aluminate addition to increase the
Alumina level and thus dilute the MgO levels without altering the slag properties. Extra lime has to be added towards the end
of the processing at the LMF as the slag is thin and tundish temperature control may become problematic. For these Si
bearing grades, only ladles with more than 3 lives are used to limit the possible formation of MgO rich inclusions.
The market conditions of 2015 led to cyclic downtime in the plant on a weekly basis (24 to 48 hours). This led to extended
periods for ladles under preheat conditions. When the ladles were brought into service, they occasionally behaved like a new
ladle and had indications of clogging at the caster. Figure 20 shows the effect of preheat time on the MgO content of the ladle
slag. Scheduling was modified as much as possible to produce less sensitive grades when the plant restarted and slag former
additions were monitored carefully. It was also found that Ti containing grades were not sensitive to this clogging problem
with high MgO ladle slags. Park et al (35) indicated that the spinel with Ti was the predominant inclusion and any non-
equilibrium MgO was tied up in complex MgO+(Mg-Ti-Al-O)+MnS inclusions.
For new ladles and for ladles with extended preheat time, it is believed that the decarburization of the surface layer of the
refractory brick and the thermal shock of moving from a preheat temperature of 1100oC to tap temperatures of 1600oC, leads
to spalling of the surface layer and thus refractory reporting to the steel and then slag. Once the ladle has been in service for
a few heats, the thermal shock is reduced, spalling is thus reduced, decarburization extent is minimized, the ladle glaze is in
position and thus the slags become more consistent.

Figure 20: Slag MgO vs. preheat time for the ladle.
CONCLUSIONS
This paper presented the types of MgO based inclusions present in Aluminum killed steels of the Nucor Steel Gallatin plant.
MgO was found as part of MgO·Al 2 O 3 spinels and as an MgO core of a calcium aluminate with or without calcium sulphide.
The MgO cored inclusions were not modifiable and caused clogging and potential rejections at customers.
The investigation showed how MgO cores could be produced from several different sources and reactions: from the ladle
glaze being washed into the steel on tap; higher levels of EAF carryover slag due to equipment conditions; solubility products
in the steel and from the ladle slag as calculated from FactSage 6.4, as well as the result of the refractory changing the slag
chemistry.
Steelmaking practices and scheduling were adjusted to compensate for MgO core inclusions.
- Slag former additions were modified to reduce the MgO content of the slag.
- Tap return speed became a new indicator.
- Critical grades were produced later in production runs to avoid ladle spalling problems from extended preheating.
- Critical grades were not produced on ladles with less than 3 heats.
- Evidence of clogging and MgO core inclusions results in a diverted heat
Transforming the database of knowledge to understanding and application in the plant ensures the customer performance
expectations are exceeded.
 ACKNOWLEDGEMENTS
The authors wish to thank Nucor for allowing the publication of this work and the Meltshop and lab team mates that applied
and developed the practices to ensure quality steel production. Thanks also to Eugene Pretorius for training with FactSage as
well as helpful discussions. Thanks to Roger Maddalena for refractory sample preparation and discussions.

REFERENCES
1. E.B. Pretorius, H.G. Oltmann, and B.T. Schart. “An Overview of Steel Cleanliness From an Industry Perspective.”
AISTECH 2013, pp. 993-1026.
2. S.R. Story and R.I. Asfahani. “Control of Ca-Containing Inclusions in Al-Killed Steel Grades.” AISTECH 2013,
pp.1201-1213.
3. E.B. Pretorius, H.G. Oltmann, and T. Cash. “The effective modification of spinel inclusions by Ca-treatment in
LCAK steel”. Iron and Steel Technology, Vol. 7, No. 7, July 2010, pp. 31-44.
4. J. Mendez, A. Gomez, C. Capurro, R. Donayo, and C. Cicutti. “Effect of Process Conditions on the Evolution of
MgO Content of Inclusions during the Production of Calcium Treated, Aluminum Killed Steels”. 8th International
Conference on Clean Steel, Budapest, Hungary, 2012.
5. P. Kaushik, H. Yin, H. Pielet, and M. Lowry. “How to Evaluate a Process for Clean Steelmaking and Quality
Control” AISTech 2011 Proceedings, Volume II, pp. 493-506.
6. H. Mu, B.A. Webler, and R.J. Fruehan. “Reduction of Slags and Refractories by Al in Steel and Inclusion
Modification”. AISTech 2015, pp. 3272-3278.
7. A. Harada, G. Miyano, N. Maruoka, H. Shibata, and S-y. Kitamura. “Dissolution behavior of Mg from MgO into
Molten Steel Deoxidized by Al”. ISIJ International, Vol. 54 (2014), No. 10, pp. 2230-2238.
8. P.C. Pistorius, P. Presoly, and K.G. Tshilombo. “Magnesium: Origin and role in Calcium-treated Inclusions.” Sohn
International Symposium Advanced Processing of Metals and Materials Volume 2 – Thermo and physicochemical
principles: Iron and Steelmaking. Edited by F. Kongoli, and R.G. Reddy. TMS (The Minerals, Metals and Materials
Society), 2006, pp. 373-378.
9. S. Yang, Q. Wang, L. Zhang, J. Li, and K. Peaslee. “Formation and Modification of MgO·Al 2 O 3 -Based Inclusions
in Alloy steels”. Metallurgical and Materials Transaction B, Volume 43B, August 2012, pp. 731-750.
10. A. Ahlborg, R.J. Fruehan, M.S. Potter, W.R. Badger, and G.S. Casuccio. “Inclusions in Aluminum-Killed Steel with
Varying Calcium Additions.” ISSTech 2003 Proceedings Vol. I, pp. 177-194.
11. H.G. Oltmann, E.B. Pretorius, and B.T. Schart. “Analytical Techniques to Assess Cleanliness in Ca Treated Steel”.
9th International Conference on Clean Steel, Budapest, Hungary, 2015.
12. A. Pitts, B. Tate, and L. Nastac. “FeSi Residuals and Their Effects on Steel Cleanliness.” AISTech 2015
Proceedings, 2015, pp. 3480-3490.
13. R. Maddalena. “Advanced SEM Methodology for Tracking Reoxidation in the Tundish.” AISTech 2013
Proceedings, pp. 2113-2121.
14. S.R. Story, Q.E. Liu, C.R. Cathcart, and M. Molnar. “Effect of Process Route on Inclusion Modification by Calcium
Treatment”. 9th International Conference on Clean Steel, Budapest, Hungary, 2015.
15. M. Harris, O. Adaba, W.N. Lekakh, R.J. O’Malley, and V.L. Richards. “Improved Methodology for Automated
SEM/EDS Non-Metallic Inclusion Analysis of Mini-Mill and Foundry Steels.” AISTech 2015 Proceedings, pp.
2315-2325.
16. O. Adaba, M. Harris, R.J. O’Malley, S.N. Lekakh, V.L. Richards, and N. Sutcliffe. “An SEM/EDS Statistical Study
of the Effect of Mini-Mill Practices on the Inclusion Population in Liquid Steel”. 9th International Conference on
Clean Steel, Budapest, Hungary, 2015.
17. K.J. Graham and G.A. Irons. “The behavior of Non-metallic Inclusions during Ladle Refining”. AISTech 2009
Proceedings, Volume 1, pp. 1003-1014.
18. E.B. Pretorius, H.G. Oltmann, and J.M.A Geldenhuis. “SEM Inclusion Analyses as a Process Control Tool”.
Proceedings of the Richard J. Fruehan Symposium, Pittsburgh, PA, 2011.
19. V. Carreño-Galindo, R.D. Morales, J.A. Romero, J.F. Chavez, and M. Valazques Toledo. “Thermodynamic analysis
of steel deoxidization process during operations”. Steel Research 71 (2000), No. 4, pp. 107-114.
20. T. Alatarvas, L. Määttä, T. Antola, M. Leinonen, S. Ollila, E.P. Heikkinen, and T. Fabritius. “Evolution of Oxide
Inclusion Characteristics During Continuous Casting: From Mould to Product.” 9th International Conference on
Clean Steel, Budapest, Hungary, 2015.
21. S.J. Kim, A. Harada, and S. Kitamura. “Condition to Suppress Spinel Formation in Ladle Treatment Predicted by
the Kinetic Simulation Model”. AISTech 2015 Proceedings, 2015, pp. 3261-3271.
22. J. Lehmann, M. Boher, H. Soulard, and C. Gatellier. “Metal/Refractory Interactions: A Thermodynamic Approach.”
7th Biennial Worldwide Congress, UNITECR’01, Cancun, Mexico. November 2001.
23. G. Du, J. Li, Z-b. Wang and C-b. Shi. “Effect of Mg-Al-Fe Alloy Deoxidation on Inclusions in Hot Work Die Steel
H13”. AISTech 2015 Proceedings, 2015, pp. 2215-2225.
24. Z. Deng and M. Zhu. “Evolution of Non-metallic Inclusions in Al-Killed Alloyed Steel during Secondary Refining
Process”. ISIJ International, Vol. 53 (2013), No. 3, pp. 450-458.
25. Y-J. Kang, F. Li, K. Morita, and Du Sichen. “Mechanism Study on the Formation of Liquid Calcium Aluminate
Inclusion from MgO·Al 2 O 3 Spinel”. Steel Research International 77 (2006) No. 11, pp. 785-792.
26. N. Verma, P.C. Pistorius, R.J. Fruehan, M. Potter, H.G. Oltmann, and E.B. Pretorius. “Calcium Modification of
Spinel Inclusions in Aluminum-Killed Steel: Reaction Steps.” Metallurgical and Materials Transactions B, Volume
43B, August 2012, pp. 830-840.
27. N. Verma, P.C. Pistorius, R.J. Fruehan, M. Potter, M. Lind, and S.R. Story. “Transient Inclusion Evolution during
Modification of Alumina Inclusions by Calcium in Liquid Steel: Part II. Results and Discussion.” Metallurgical and
Materials Transactions B, Volume 42B, August 2011, pp. 720-729.
28. N. Verma, P.C. Pistorius, R.J. Fruehan, M. Potter, M. Lind, and S.R. Story. “Transient Inclusion Evolution during
Modification of Alumina Inclusions by Calcium in Liquid Steel: Part I. Background, Experimental Techniques and
Analysis Methods.” Metallurgical and Materials Transactions B, Volume 42B, August 2011, pp. 711-719.
29. N.N. Tripathi, M. Nzotta, A. Sandberg, and Du Sichen. “Effect of ladle age on formation of non-metallic inclusions
in ladle treatment”. Ironmaking and Steelmaking, 2004, Vol. 31, No. 3, pp. 235-240.
30. K. Beskow, N.N. Tripathi, M. Nzotta, A. Sandberg, and Du Sichen. “Impact of slag-refractory lining reactions on
the formation of inclusions in steel”. Ironmaking and Steelmaking, 2004, Vol. 31, No. 6, pp. 514-518.
31. K. Beskow and Du Sichen. “Ladle glaze: major source of oxide inclusions during ladle treatment of steel.”
Ironmaking and Steelmaking, 2004, Vol. 31, No. 5, pp. 393-400.
32. Du Sichen, A. Sandberg and N.N. Tripathi. “A Study on the Population and Chemical Characterization of Non-
Metallic Inclusions in the Tool Steel Making Process”. ICS Proceedings of 2005, pp. 649-660.
33. H. Yin, R. Beischneil, F. Demmon, E. Shale, R. Kittridge, H.T. Tsai, B. Forman and M.M. Yavuz. “Reduction of
oxide inclusion induced hook cracks in tubular steel at Arcelor Mittal USA Indiana Harbor N. 2 Steel Producing”,
6th European Continuous Casting Conference, Riccione, Italy, June 2008.
34. X. Wang, Q. Li and F. Huang. “Control of B type Non-metallic Inclusions in Line pipe Steel Plates”. 8th
International Conference on Clean Steel, Budapest, Hungary, 2012.
35. S-C. Park, I-H. Jung, K-S. Oh, and H-G. Lee. “Effect of al on the Evolution of Non-metallic Inclusions in the Mn-
Si-Ti-Mg Deoxidized Steel during Solidification: Experiments and Thermodynamic Calculations”. ISIJ
International, Vol. 44 (2004), No. 6, pp. 1016-1023.