Rob Dekkers, Ph.D.

Thesis, Katholieke Universiteit Leuven, Leuven, Belgium (2002) 133

Chapter 10

IMPROVEMENT OF STEEL CLEANLINESS


Possibilities to improve steel cleanliness are discussed. Two major issues
are recognised, i.e. removal of inclusions and prevention of steel
reoxidation. Parameters affecting both removal and reoxidation are
discussed and illustrated with industrial data.


10.1. Introduction

Theoretically, aluminium killed steels contain total oxygen contents (O
tot
) of only a few
part per million (ppm) dissolved oxygen. In industrial melts the presence of oxide
inclusions usually rises O
tot
a factor ten compared to the equilibrium values. Saxena [1]
studied the kinetics of aluminium deoxidation and concluded that deoxidation rate is
controlled mainly by the rate of separation of aluminium oxide inclusions from the bath.
Inclusion removal takes place by flotation, stirring and filtration. Competitive to inclusion
removal is reoxidation by air, slag, refractory and ladle glazes (slag left on the walls),
which reoxidation decreases the effective deoxidation rate. In general, reoxidation occurs if
the oxygen activity in the steel bath is less than the oxygen activity in any phase in contact
with the steel. In this chapter, sources of reoxidation are discussed and illustrated with
industrial data. Finally steel flow in ladle, tundish and mould is considered. In particular in
the tundish, control of steel flow can be a tool for inclusion removal, but also a source of
reoxidation, e.g. through slag entrainment.


10.2. Flotation of inclusions

The density of aluminium oxide inclusions is about 60% of that of liquid steel. Therefore
inclusions tend to float out of the liquid steel. According to Stokes law their emerging
velocity is given by:

9
) ( 2
2
.
.
gr
v
incl steel
incl

= (10.1)

Due to the small dimensions of the inclusions and the high viscosity ( ) of the liquid steel
the emerging velocity of the inclusions is limited to only a few micrometers per second, e.g.
30 ms
-1
for a 10 m large aluminium oxide particle. Therefore dragging of inclusions by
steel convection is expected to overrule vertical rising as predicted by Stokes law. Plotting
O
tot
of a large number of heats of different steel grades versus their holding times reveals a
Chapter 10 134
small tendency of increasing O
tot
with increasing holding time in the ladle. Though the
relationship is rather weak, it is opposite to what is expected if inclusion removal would be
exclusively controlled by Stokes law. However, excluding degassed steel grades, which
have a shorter holding time after aluminium addition and lower O
tot
than the grades
submitted to traditional ladle metallurgy, a weak negative relationship is observed (fig.
10.1).

time between deoxidation and end of ladle metallurgy (min.)
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
10
20
30
40
50
60
70
80
0 10 20 30 40 50 60 70 80

Fig. 10.1 O
tot
at the end of the ladle metallurgy versus holding time (no degassed steel grades
included).

In order to homogenise the steel bath and to enhance collisions between inclusions, the
steel is stirred electro-magnetically (induction), mechanically or by inert gas bubbling [2].
For induction stirring a current in the induction coils forces the steel to move along the
magnetic fields. The higher the current or power input, the stronger the stirring. Figure
10.2. shows that stirring enhances oxygen removal in aluminium killed steels [3]. The
oxygen removal decreases exponentially with stirring time. After about 10 minutes
prolonged stirring gives barely rise to improved cleanliness.

Mechanical stirring results from rotating, tilting or shaking the steel ladles. In laboratory
experiments stirring rods or propellers are used. Degassing of steel in a Ruhrstahl Heraeus
(RH) vessel, which is common for ultra low carbon grades, is an efficient practice for
stirring and inclusion removal [4]. Due to the vacuum in the vessel, the steel is sucked-up
through a first nozzle and streams back into the ladle through a second nozzle, creating a
turbulent steel flow.

For non-degassed steel grades, stirring by an inert gas, either argon or nitrogen, is common
practice. The non-wetting properties of the inclusions help them to separate from the liquid
steel [5] and attach to the gas bubbles (and the refractory walls and the slag-steel interface).
Improvement of steel cleanliness 135
Since calcium aluminates separate from the steel less easily than aluminium oxide particles,
calcium addition can best be postponed till the rate of aluminium oxide removal has
become insignificant. Calculations on the efficiency of inclusion removal by gas bubbling
have been carried out by several authors [6-9]. Wang et al. [6,7] consider the probability
that an inclusion encounters a gas bubble as well as the probability that an inclusion
becomes attached to the bubble (fig. 10.3). The latter depends on the sliding rate of an
inclusion along the bubble and the rate of steel film drainage and rupture between inclusion
and bubble. A large number of small bubbles increases the chance that inclusions and
bubbles meet, but the capacity of inclusion removal decreases. Furthermore if the bubbles
are smaller than 0.5 mm they become entrapped by the steel convection and are not
removed from the steel bath. In a recent paper Choo and Lee [9] refined the model. They
confirm that probabilities of collision and attachment are important, but add that for the
removal of fine inclusions also the interfacial energy is important. A large interfacial
energy would prevent detachment of inclusions from the bubbles.



Fig. 10.2 Decrease of total oxygen contents versus stirring times at different electro-magnetic
intensities. The higher the current or power input, the stronger the stirring [3].

Fig. 10.3 Collision and cohesion probabilities versus the bubble diameter for different inclusion
sizes [6].
Chapter 10 136
At Sidmar, no relation between the O
tot
at the end of the ladle metallurgy and the stirring
time, the amount of stirring gas and the flux of the gas flow have been found (fig. 10.4-6).
Apparently prolonged gas stirring has no significant effect. Intensive stirring in the ladle
without vacuum treatment may cause atmospheric reoxidation during splashing of the
bubbles at the surface of the bath [10,11]. Nitrogen pickup is used as an indirect measure of
reoxidation. In order to investigate whether reoxidation occurs, we have looked for a
relationship between nitrogen and O
tot
at the end of the ladle metallurgy. No relationship


gas stirring time (sec)
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
20
40
60
80
0 500 1000 1500 2000 2500 3000

Fig. 10.4 O
tot
at the end of the ladle metallurgy versus gas stirring time.


volume of stirring gas (l)
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
25
50
75
100
125
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000

Fig. 10.5 O
tot
at the end of the ladle metallurgy versus volume of stirring gas.
Improvement of steel cleanliness 137
between nitrogen content and O
tot
was found (fig. 10.7). Moreover, O
tot
and nitrogen
change between the end of ladle metallurgy and the mould is not related. This may be very
dependent on the steel plant. Brinkmeyer and Melvill [12] state, shrouding of the steel
stream from the ladle to the tundish has proven to be the single most important practice
affecting continuously cast steel cleanliness. Sidmar shrouds the steel with submerged
ladle nozzles and submerged entry nozzles (SENs). Argon shielding (overpressure) is
carried out only between tundish and mould. Submerged nozzles are not only


average stirring flux (l/sec.)
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
25
50
75
100
5 10 15 20 25

Fig. 10.6 O
tot
in the ladle versus average flux of gas stirring.


N (ppm) at end of ladle treatment
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
20
40
60
80
100
10 20 30 40 50 60 70

Fig. 10.7 Relationship between O
tot
and the nitrogen content of steel at the end of ladle metallurgy.
Chapter 10 138
designed to prevent atmospheric reoxidation, but also turbulence of the steel-slag interface.
After tundish change ladle nozzles are not submerged, because of the empty tundish
heats show only limited loss of cleanliness. Although atmospheric oxidation during ladle
metallurgy can not be proven, it can neither be excluded, because aluminium loss is
significant during ladle metallurgy as discussed in chapter 4 (fig.4.8). The required amount
of oxygen related to the aluminium loss comes from the slag or refractory and the oxygen
flux may be maintained through convection of the slag the slag acts as a transport
medium or through the pores of the refractory material.


10.3. Calcium addition

Clogging of submerged entry nozzles (SEN) significantly decreases in case of calcium
treated steels. Higher casting capacities and less replacement of SENs result in increased
quantities of cast steel. Moreover, there are less casting irregularities, which results in less
surface defects. Calcium treatment has been introduced in the seventies and is nowadays
commonly used to control the properties of inclusions [13-28]. Calcium is added by wire
injection of, e.g. ferro-calcium or manganese-calcium alloys. Appropriate additions of
calcium to the liquid steel react with aluminium oxide particles and result in the
transformation of solid aluminium oxide inclusions into liquid calcium aluminates. The
amounts of calcium reacting with aluminium oxides must be controlled within narrow
ranges, because otherwise solid calcium aluminates are formed (fig. 10.8). This is less
straightforward than it seems: calcium has a very low solubility in liquid steel and therefore
the efficiency of calcium additions is low. Furthermore calcium must react homogeneously
with all aluminium oxides in the steel melt, because otherwise calcium poor and calcium
rich aluminates are formed. Castability decreases by presence of the high melting calcium
aluminate inclusions, which may be even worse than the aluminium oxides [15,21,22]. The
cost, low efficiency and necessary high process control make calcium treatments difficult.

Calcium treatment does not result in lower total oxygen contents or smaller inclusions. In
fact, inclusions are less easily removed from the melt. Unlike aluminium oxides, which
easily form aggregates if particles encounter, liquid calcium aluminates do not coalesce. In
general this is due to the preferential surface occupation by certain atoms, which
subsequently results in surface charging of liquid inclusions. Despite of that, costumers
demand for calcium treated steels is growing, because calcium treatment is also applied for
desulphurisation. Lower residual levels of sulphur in the liquid steel, prevent the formation
of manganese sulphide that may cause centre-line segregation and heterogeneous properties
of the steel after rolling. Formation of solid calcium sulphide occurs after the
transformation of aluminium oxide into calcium aluminate, which is energetically
preferred. Therefore the calcium treatment is more efficient if the amount of aluminium
oxide inclusions is minimal. Similar results are obtained by additions of rare earth
elements, which favour the formation of globular non-deformable sulphides or oxy-
sulphides if added after calcium treatment [29].
Improvement of steel cleanliness 139

A: Al
2
O
3

C: CaO
L
C
A
6

C
A
2

C
3
A

C
A

Al
2
O
3
CaO

Fig. 10.8 The Al
2
O
3
-CaO phase diagram. There is only a narrow range (at about 1600C) for the
formation of liquid calcium aluminates (FactSage).

Calcium treatment is an example of morphology control by changing the composition of
the inclusions. The number and size of the calcium aluminates are closely related to the
steel cleanliness prior to calcium addition, because the aluminium oxide inclusions created
during the deoxiation practice are transformed. Since calcium aluminates exhibit minor
tendency to separate from the steel, control of the number and the size of aluminium oxide
inclusions is of paramount importance for obtaining clean steels.

From laboratory experiments, the shape of inclusions has been found to depend on the
dissolved oxygen content, on the concentration gradient of oxygen and aluminium in the
liquid steel, on the holding time and on the stirring method (chapter 2). Clusters are
removed within about 15 minutes after aluminium addition. Enhancement of cluster
formation during the deoxidation process, will therefore probably result in a high inclusion
removal rate and a low residual O
tot
in the steel bath.


10.4. Method of aluminium addition

Tiekink and co-authors [30,31] studied the morphology of aluminium oxide inclusions in
steel at an argon stirring station and at a RH degasser. Sampling was done both with
standard production samples (lollypop samples) and with bomb samples in 315-metric-ton
ladles. At the argon stirring station three experiments were carried out with respect to
Chapter 10 140
aluminium addition: (1) aluminium was added by wire in one batch, (2) aluminium was
added by wire in two batches and (3) aluminium was added during tapping (and trimming
at the stirring station). In all cases low carbon aluminium killed (LCAK) steel was
produced. Deoxidation during tapping resulted in large (10-20 m) single particles. Best
results were obtained when aluminium was added in one batch. In this case clusters were
formed that were largely removed within a few minutes. At Sidmar aluminium lumps are
usually added in one batch. The inclusions formed are very similar to these formed by wire
injection during the above mentioned experiments. Unknown is whether the relative
amount of inclusion types is also similar. No relationship has been found between the
oxygen content at tapping and O
tot
at the end of ladle metallurgy (fig. 10.9). The total
amounts of aluminium added is only slightly related to O
tot
at the end of ladle treatment
(fig. 10.10). On average silicon-aluminium killed steels have lower O
tot
at end of ladle
metallurgy than aluminium killed steels. Partially, this may be explained by the lower
amount of aluminium added to silicon-aluminium steels, but also at higher aluminium
additions O
tot
of silicon-aluminium steels are below average. It has been suggested (chapter
4) that manganese silicate inclusions, formed during tapping, affect the deoxidation
kinetics, but the mechanism by which it results in cleaner steel is still not understood.


10.5. Filtration

Ceramic filters have been successfully used to remove inclusions from non-ferrous metals,
especially in aluminium industries. Laboratory experiments show that in particular large
inclusions can be eliminated successfully [32-40]. Yet only one application of filtration in
steel industry has been reported [37]. Filtration of inclusions concerns in depth filtration,
which means that on the average the pores are significantly larger than the inclusions.
Attachment of particles at the pore walls occurs by diffusion, direct interception and surface
forces. Surface forces are important with respect to inclusions smaller than 30 m, while
larger inclusions are entrapped mechanically. After attachment inclusions may sinter to the
refractory material of the filter.

No significant difference of removal efficiency between different filter materials has been
found, because the non-wetting properties of most inclusions ensure good attachment to the
pore walls. The efficiency of inclusion removal increases with decreasing melt velocity and
increasing filter length.

The application of filter techniques in the steel industry is difficult, because of the high
steelmaking temperatures, serious flow resistance (low casting rate), erosion of filter
material and reoxidation introducing new particles, and high amounts of small inclusions
that are less efficiently removed. [35,38-40]. Liu [37] reports the successful application of
corundum and calcium oxide filters in continuous casting tundishes.

Improvement of steel cleanliness 141
O (ppm) at tapping
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
20
40
60
80
200 300 400 500 600 700 800

Fig. 10.9 O
tot
at the end of ladle metallurgy versus the oxygen content (Celox) at tapping.

Al added in total (kg/ton)
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
20
40
60
80
0,5 1,5 2,5 3,5

Fig. 10.10 O
tot
at the end of ladle metallurgy versus mass of aluminium added.


10.6. Slag control

Steelmaking slags have three main goals: (1) preventing atmospheric reoxidation, (2)
thermal insulation and (3) take-up of non-metallic inclusions. These conditions are best
satisfied if a sufficiently thick and (continuous) liquid slag layer is present on top of the
steel bath. In order to have a liquid slag that is able to take up inclusions it must not be
saturated with respect to the components of the inclusions.

Chapter 10 142
Care should be taken to avoid that the slag does not give rise to refractory wear. An
undersaturated slag in contact with refractory material dissolves the bricks till it is saturated
with respect to all of the components in the slag-refractory system. If for example, slag with
the composition denoted with a black spot in fig. 10.11 comes in contact with a magnesium
oxide (or calcium oxide) based refractory, the latter starts to dissolve into the slag.
Dissolution continues till the slag is saturated with magnesium oxide, i.e. at the point where
the line between the original composition and MgO enters the triangle liquid (L)-CaO-
MgO. At this stage the slag is already oversaturated with respect to calcium oxide and thus
the liquid slag contains solid calcium oxide particles. Such an oversaturated slag may cause
abrasion of the refractory bricks at the slag zone. Slag penetration and refractory wear is
subject of several papers [41-45], which indicate that the degree of dissolution of the
refractory depends also on the composition and the orientation of the refractory grains in
contact with the slag. An elaborate research on slag-refractory interactions in VOD ladles
has been carried out by Jones [46].


L
L + MgO
CaO
+
L
L + CaO + MgO
Fig. 10.11 Isothermal section of the Al
2
O
3
-CaO-MgO phase diagram at 1600C. The dark area
corresponds to the liquid one-phase stability field (FactSage).

At Sidmar three different slags may affect the steel cleanliness, i.e. (1) the synthetic ladle
slag, (2) the tundish slag and (3) the mould slag. These slags are chemically very distinctive
(table 10.1). In case of silicon-aluminium killed steels the synthetic ladle slag contains
about 10% silicon oxide, while in case of aluminium killed steel the percentage is only a
few per cent. Slag used for production of ultra low carbon steel has not been considered.
Improvement of steel cleanliness 143

After tapping the steel is covered by a slag, which may contain high amounts of iron-,
manganese-, and silicon oxide. This slag consists of entrained convertor slag, steel
oxidation products, deoxidation products from oxidisers added during tapping, and ladle
glazes. In particular the high amounts of iron-manganese oxides in the slag are considered
as detrimental with respect to steel cleanliness. A number of studies describe a relationship
between total oxygen content in the steel and the iron-manganese oxide content in the ladle
slag [3,10,47-50]. Ahlborg found spinel and calcium hexaluminate inclusions if the iron-
manganese content of the ladle slag was below two per cent. Despite that spinel and
calcium hexaluminate badly affect the castability, low iron-manganese contents in the ladle
slag are preferred, because more accurate control of aluminium and other alloys in the steel
is obtained. At Sidmar a synthetic slag replaces the slag formed during tapping. In this way
one can immediately add a slag which is optimal for ladle treatment. Without using a
synthetic slag, the ladle slag has to be reduced by adding aluminium or another slag
deoxidiser.

Table 10.1 Typical slag compositions.
Al
2
O
3
CaO SiO
2
MnO FeO MgO Na
2
O F C
ladle 38-43 46 3-10 1-4 1-4 2 - - -
tundish - - 90-94 - - - - - < 5
mould 3-9 26-40 28-40 - 0-3 0-6 4-14 5-9 1-7

In particular iron oxide and manganese oxide in the slag are detrimental for the steel
cleanliness, because due to the lower oxygen activity of the liquid steel they can be
reduced. At Sidmar a positive relationship between the iron oxide content of the basic ladle
slag and O
tot
of the steel has been found (fig. 10.12), but is lost if O
tot
in the tundish is
considered. Intensive gas stirring or mass transfer reactions, e.g. due to the Marangoni
effect [51,52], may give rise to violent turbulence at the steel-slag interface and even
emulsification of its interface. Emulsification has been observed in case of bulk aluminium
addition to steel that was covered by an iron rich slag [53]. The Marangoni effect refers to
the interfacial tension gradient term, which in combination with the thermal gradient and
activity gradient define the shear stress at the liquid metal-slag interface. Intensive
interaction between steel and basic slag (provided that the slag is unsaturated) promotes
cleanliness, but in case of iron oxide containing slag reoxidation occurs. Steel droplets in
the slag, as a result of emulsification of intensive stirring, may act as a transport medium
for (atmospheric) oxygen into the steel bath. At the slag surface they oxidise, while at the
steel-slag interface they are reduced by dissolved aluminium or silicon in the steel.

The tundish slag consist of silicon oxide (rice hull ashes), because of its good insulation
properties. Most of the rice hull slag is solid. Impurity components at the slag-steel
interface give rise to a liquid slag layer. Since the silicon oxide slag will be a significant
source of oxygen to the steel, experiments with basic tundish slag have been carried out at
Sidmar. The basic slag contained mainly aluminium oxide and calcium oxide. In case of
Chapter 10 144
basic tundish slag O
tot
was significantly lower in tundish and mould than in case of rice hull
slag, but no difference in O
tot
was found in the solidified slab. Furthermore, oxidation of
aluminium and titanium dissolved in the steel was decreased and less silicon pick-up
occurred in case of the basic slag than in case of the rice hull ashes. Therefore it is
concluded that rice shall slag causes reoxidation of the steel in the tundish, but
simultaneously oxide inclusions are removed to the tundish slag. Similar results have been
obtained by other authors [38,54]. Goto and Miyazawa [54] compared the degree of
reoxidation of an aluminium-killed steel and a non-killed steel in the tundish. Although the
tundish slag contains less silicon oxide, reoxidation occurred. In the aluminium killed steel
O
tot
remained constant, because aluminium oxide inclusions float out of the steel. In the
unkilled steel O
tot
increased, because no inclusions were present and subsequently no
flotation occurred. Since the thermal conductivity of the basic tundish slag is high,
commonly rice hull ashes cover the basic tundish slag. The rice hull ashes dilute the basic
slag, i.e. becomes enriched in silicon oxide, and as a result the tundish slag has to be
replaced by a new one after a number of heats.


% FeO in ladle slag
O
t
o
t

(
p
p
m
)

a
t

e
n
d

o
f

l
a
d
l
e

m
e
t
a
l
l
u
r
g
y
0
25
50
75
100
125
0 5 10 15 20

Fig. 10.12 O
tot
in the steel at the end of ladle treatment versus the iron oxide content in the ladle
slag.

The mould slag mainly consists of calcium oxide and silicon oxide and in less extent of
aluminium oxide, sodium oxide and calcium fluoride. During industrial practice the mould
slag becomes enriched in aluminium oxide due to flotation of aluminium oxide inclusions
and reactions between aluminium dissolved in the steel with reducible slag components
[55]. As a result the viscosity of the mould slag increases. Yet it is not expected to affect its
ability to lubricate the strand during casting [55]. Mould powders with high interfacial
tension, high viscosity and low solidification temperature appear advantageous in order to
Improvement of steel cleanliness 145
prevent slivers [56] or oscillation marks [57,58]. Much research has been carried out on
thermal conductivity, viscosity and powder consumption of mould fluxes [59-60].

Studies on modelling of inclusion removal by the slag [61-64] divide the process in stages.
First the inclusion approaches the interface, followed by steel film thinning and finally
rupture of the steel film resulting in an inclusion-gas or an inclusionslag contact. The
interfacial energy plays a major role with respect to the steel film rupture. In case of
removal of inclusions into the slag, the process continues with complete separation of the
inclusions from the steel-slag interface or dissolution of inclusions into the slag. Both
separation and dissolution are usually no problem, because of the high steel-inclusion
interfacial energy. In order to enhance dissolution of inclusions into the slag, the initial
ladle slag composition is chosen dependent on the expected amounts of inclusions formed
during killing of the steel, such that the slag does not become saturated in, for instance,
aluminium. For steelmaking practice, other properties of the slag such as thermal and
atmospheric insulation and viscosity of slag are important as well.

When an inclusion becomes attached to an interface it will stay there for a moment, i.e. the
residence time. Yin et al. [65,66] observed that solid particles agglomerate at the interface
due to capillary attraction, which on turn depends on interfacial energy. The capillary
attraction extended up to 100 m, but was not observed for liquid particles. After
agglomeration, densification (sintering) of the particles took place. For detailed
investigations on the concentration of inclusions at interfaces and the dissolution of
inclusions in a slag phase the reader is referred to [65-71]. So far it is unknown whether
differences in interfacial energy of different inclusions significantly affect their rate of
removal into slag.


10.7. Role of refractory

Ideally refractory materials do not interact with the steel bath or the slag. Decreased service
life of refractory can be due to chemical attack, abrasion, thermal shock or transformations
at high temperatures. With respect to reoxidation of steel both chemical and mechanical
wear can play a role [72-75]. During the present research only rarely refractory particles
have been observed in steel samples. Their presence is expected to be insignificant
compared to the inclusion content.

The oxygen potential of the refractory material is closely related to the achievable O
tot
of
the steel. Bannenberg [75] illustrates this by the relationship between oxygen activity and
the aluminium content of the liquid steel for acid refractories and for basic refractories (fig.
10.13). The solid line represents the contents of dissolved oxygen and aluminium at
thermodynamic equilibrium. In case of basic refractory, the dissolved oxygen contents are
close to the thermodynamic equilibrium. In contrary, steel in contact with acid refractory
results in significant larger dissolved oxygen contents. Since the oxygen activity in the steel
Chapter 10 146
can not become smaller than the oxygen activity of the refractory in contact with the liquid
steel, the acid refractory material dissolves, which causes in a flux of oxygen into the steel.


acid ladle refractory
basic ladle refractory
O

a
c
t
i
v
i
t
y

o
f

t
h
e

s
t
e
e
l

Al content (10
-3
wt%)
Fig. 10.13 Achieved oxygen activities in relationship to aluminium in case of acid and basic
refractory materials [75].


In the slag zone the ladle lining contains magnesium oxide-carbon (magnesia-C) bricks. A
layer of about 1.8 till 2 mm of the magnesia-C bricks dissolves per heat. In order to
decrease dissolution during tapping magnesium oxide particles are added. The synthetic
slag may also cause dissolution of the magnesia-C refractory, because it is not saturated
with respect to magnesium oxide (fig. 10.11). Below the slag zone the refractory consists of
mainly aluminium oxide with small amounts of silicon oxide, iron oxide or titanium oxide.
Besides corundum the mineralogy of the bricks may contain small amounts of mullite
(3Al
2
O
3
2SiO
2
) or hercynite (FeAl
2
O
4
). Both mullite and hercynite decompose if they are in
contact with liquid aluminium killed steel, i.e. dissolved oxygen content less than 5 ppm
(fig. 10.14). This means that some wear of aluminium oxide refractory and subsequent
reoxidation of the steel melt may occur.

Significant oxidation of dissolved aluminium after killing occurs in the ladle (fig. 4.10), but
does not necessarily result in a higher O
tot
of the steel. The aluminium may react with
oxygen at the steel-slag or steel-refractory boundaries and subsequently the inclusions are
removed from the steel bath. Aluminium oxide particles observed in steel samples that may
derive from the refractory show commonly a shell-shaped fracture surface, which indicates
Improvement of steel cleanliness 147
abrasion or thermal shock of the refractory material. Therefore, refractory wear in the ladle
is expected to occur mainly mechanically rather than chemically.



Fig. 10.14 Thermodynamic equilibrium lines for metal oxides show the thermodynamic achievable
oxygen content in relationship to the concentration of the (most strong) deoxidising metal metal
[73].

Despite of the low oxygen potential (high stability) of aluminium oxide refractory, wear
can occur in case of calcium treatment [20,69]. Not only the oxygen activity of the liquid
steel may then be less than that of the refractory, but also the calcium dissolved in the steel
reacts with the refractory material.

The refractory material of the tundish consists of a 10-15 cm thick layer of forsterite
(Mg
2
SiO
4
) with some aluminium oxide. Forsterite decomposes in contact with steel with
low oxygen contents [73] and therefore may be a source of reoxidation of the steel.
Furthermore decomposition of forsterite may give rise to spinel formation, which is
considered as very detrimental in particular in calcium-treated steels. The tundish refractory
is renewed every 15-18 heats.

The last refractory material in contact with the steel before casting is that of the SENs. In
general two types of refractories are used, i.e. silicon oxide (white nozzles) and aluminium
oxide carbon (black nozzles). The latter is preferentially used, although clogging is
enhanced. The silicon oxide nozzles are severely attacked by the steel and subsequently
give rise to reoxidation of the steel [77].
Chapter 10 148
10.8. Steel flow control

Steel flow plays a very important role with respect to steel cleanliness. It transports
inclusions to the slag, but simultaneously it may entrain inclusions and oxygen (or other
unfavourable elements) from the steel-slag interface into the steel bath.

Research on steel flow concerns both water modelling and mathematical modelling
[31,32,78-101]. Especially mathematical modelling has been developing continuously
during the past decade. In order to model inclusion removal phenomena, flow pattern,
thermal buoyancy, turbulent motion, collisions, flotation, and reoxidation must be
incorporated. Modelling of the ladle convection focuses on the mixing of addition elements
for deoxidation, desulphurisation and alloying and the subsequent steel-slag interactions.
Most work on modelling concentrates on tundishes, because the dimensions of the tundish
in combination with dams and weirs enable control of the steel flow from the ladle to the
mould. Removal of inclusions may be enhanced by high collision probabilities in the first
part of the tundish followed by optimal flotation conditions in the second part of the tundish
(fig. 10.15). Brinkmeyer and Melville [12] found that the efficiency of inclusion removing
by basic tundish fluxes could be improved by optimising the steel flow through the tundish
and therefore carried out water model experiments, which resulted in less reduced
cleanliness of the cast steel during ladle changes.

The use of a so-called turbostop may prevent splashing and subsequent reoxidation of the
steel during filling of the tundish. Shielding the steel with argon from the atmosphere can
prevent reoxidation after tundish change as well. Likewise an argon atmosphere is created
around the contact between tundish and exchangeable SENs. During continuous casting, an
underpressure is created in the SEN, which causes entrapment of air into the SEN.




Fig. 10.15 A schematic drawing of a tundish with turbostop, dam and weir to control the steel flow.
In the first part (A) turbulence causes collisions between inclusions and subsequent agglomeration,
while in the second part (B) laminar flow along the slag-steel interface gives rise to inclusion
separation.
Improvement of steel cleanliness 149



Fig. 10.16 Four modes of mould flux entrainment at the flux-steel boundary: by turbulence caused
by melt flow (A), by vortexing (B), by turbulence caused by argon bubbles (C) and by sedimenting
crust caused by excessive heat loss (D) (adapted from [102]).


Entrainment of slag in ladle, tundish and mould must be prevented. Especially in the mould
entrapment of mould powder may results in defects. Entrainment of slag can be caused by
turbulence due to steel flow or argon bubbles, by vortexing [60] or by sedimentation of
(solidified steel) crust due to excessive heat loss (fig 10.16). Vortexing may also occur
when the ladle or tundish bath level reaches a critical height during drainage. The lighter
slag is then drawn into the vortex. Slag detectors, e.g. infra-red or electro-magnetic, detect
when slag starts to emerge with the steel flow. The slag detectors can be combined with
stoppers that prevent continued slag drainage [103].


10.9. Conclusions

In order to obtain high cleanliness both removal of inclusions as well as reoxidation have to
be controlled. After addition of aluminium the O
tot
of the steel drops rapidly and after about
15 minutes it stabilises at a low level. The obtained O
tot
, however, is much larger than the
thermodynamic achievable value.

Based on our observations the initial decrease of O
tot
is correlated to the removal of
aluminium oxide clusters from the bath. Argon stirring enhances the separation of
inclusions, but the present stirring procedure is insufficient to remove small particles.
Accretion of small inclusions during ladle treatment leads to the formation of agglomerates
up to 20 m on average. The agglomerates are not removed, because they are likely to be
dragged with the steel convection.

Chapter 10 150
Lower total oxygen contents can be achieved by a more efficient stirring procedure and by
enhancement of cluster formation during killing of the steel. O
tot
before killing or the
amount of aluminium added do not affect the cleanliness.

No evolution of inclusions during ladle metallurgy has been observed in size and
composition. This indicates that no significant oxide formation in the steel occurs, though
the dissolved aluminium content decreases. It is supposed that aluminium oxidation takes
place at the steel-slag or steel-refractory interfaces and that the inclusions do not dilute the
steel bath.

It has been shown that the use of a basic tundish slag will decrease the total oxygen content,
the silicon pickup and oxidation of aluminium and titanium. The advantages of a basic
tundish slag are lost in the slab, because probably reoxidation occurs during casting or by
the mould powder.


References

[1] S. K. Saxena, Ironmaking Steelmaking, 7 (1980) 6, 281-287.
[2] K.W. Lange, Int. Materials. Review, 33 (1988) 2, 53-89.
[3] K. Schwerdtfeger, Arch. Eissenhttenwes. 54 (1983) 3, 87-98.
[4] Y. Miki et al., Iron Steelmaker, 24 (1997) 8, 31-38.
[5] A. Nicholson and T. Gladman, Ironmaking Steelmaking, 13 (1986) 2, 53-69.
[6] L. Wang, H-G. Lee and P. Hayes, ISIJ Int., 36 (1996) 1, 7-16.
[7] L. Wang, H-G. Lee and P. Hayes, ISIJ Int., 36 (1996) 1, 17-24.
[8] L. Zhang and S. Taniguchi, Int. Mat. Rev., 45 (2000) 2, 59-82.
[9] J-S. Choo and H-G. Lee, ISIJ Int., 41 (2001) 2, 151-157.
[10] D. Mu and L. Holappa, in Proc. 6
th
Int. Conf. on refining processes, Part II, Ed. MEFOS, Lule,
Sweden (1992) 265-273.
[11] H. Suito and I. Inoue, ISIJ Int., 36 (1996) 5, 528-536.
[12] L. Brinkmeyer and S.D. Melville, Ironmaking Steelmaking, 22 (1995) 6, 502-507.
[13] D.C. Hilty and J.W. Farrell, Iron Steelmaker, 2 (1975) 5, 17-22.
[14] D.C. Hilty and J.W. Farrell, Iron Steelmaker, 2 (1975) 6, 20-27.
[15] G.M. Faulring, J. W. Farrell and D.C. Hilty, Iron Steelmaker, 7 (1980) 2, 14-40.
[16] H.K. Tnshoff, W. Kaestner and R. Schnadt, Stahl u. Eisen, 109 (1989) 16, 743-750.
[17] D-Z. Lu, G.A. Irons and W-K. Lu, in Proc.6
th
Int. Conf. On Refining Processes, Part II, Ed. MEFOS,
Lule, Sweden (1992) 239-263.
[18] B. Bergman, N. Bannenberg and R. Piepenbroch, in Proc. 1
st
Eur. Conf. on Continuous Casting,
Florence, Italy (1991) 1501-1508.
[19] K. Larsen and R. J. Frueham, ISS Transactions, 12 (1991) 125-132.
[20] J. Krj, H. Nevala, U. Hintzen and M. Wiesel, Stahl u. Eisen, 113 (1993) 10, 73-75.
[21] N. Bannenberg, in Proc. Steelmaking Conf., Nashville, USA (1995) 457-463.
[22] E. Martinez, Iron Steelmaker, 22 (1995) 6, 19-23.
Improvement of steel cleanliness 151
[23] L. Ferro et al., Iron Steelmaker, 23 (1996) 11, 45-48.
[24] J. Madias, C. Cicutti and J.C. Gonzalez, Iron Steelmaker, 24 (1997) 2, 155-159.
[25] A.R. McKague et al., Iron Steelmaker, 25 (1998) 11, 35-41.
[26] A. Hamoen and W. Tiekink, in Proc. ISS/AIME, Toronto (1998).
[27] D. Janke et al., ISIJ Int., 40 (2000) 1, 31-39.
[28] J.M.A. Geldenhuis and P.C. Pistorius, Ironmaking Steelmaking, 27 (2000) 6, 442.
[29] W.G. Wilson, L.J. Heaslip and I. D. Sommerville, J. Metals, 37 (1985) 9, 36-41.
[30] W.K. Tiekink, A. Pieters and J. Hekkema, Iron Steelmaker, 21 (1994) 7, 39-41.
[31] W.K. Tiekink, J. Brockhoff and J. van der Stel, in Proc. 4
th
Int. Conf. on Clean Stee , Balatonszeplak,
Hungary (1992) 704-717.
[32] D. Apelian and R. Mutharasan, J. Metals, 32 (1980) 9, 14-19.
[33] S. Ali, R. Mutharasan and D. Apelian, Met. Trans. B, 16B (1985) 725-742.
[34] K. Uemura, M. Takahashi, S. Koyama, in Proc. 6
th
Int. Iron and Steel Congress, Vol. 3 Steelmaking
I, ISIJ, Japan (1990) 535-542.
[35] K. Uemura et al., ISIJ Int., 32 (1992) 1, 150-156.
[36] K. Raiber, P. Hammerschmid and D. Janke, ISIJ Int., 35 (1995) 4, 380-388.
[37] X.T. Liu et al., Ironmaking and Steelmaking, 19 (1992) 3, 221-225.
[38] J. van der Stel, M. Cornelissen, B. Deo and R. Boom, in Proc. 29
th
Annual Conf. of Metallurgists,
Hamilton, August (1990).
[39] J. Pindor, K. Michalek, M. Cieslar and V. Bednarz, in Proc. 5
th
Int. Conf. on Clean Steel, Vol. 2,
Hungary (1997) 11-17.
[40] A. Reynaud, Rev. de Mtallurgie-CIT, (2000) 1, 73-82.
[41] B. Jackson, W.F. Ford and J. White, Trans. Int. Br. Ceram. Soc., 62 (1963) 577-601.
[42] I.M. Stephenson and J. White, Trans. Int. Br. Ceram. Soc., 65 (1965) 443-483.
[43] J. White and C. Richmond, Refractories J., 46 (1970) 6-18.
[44] J.D. Dewendra, C.M. Wilson and n.H. Brett, Trans. Int. Br. Ceram. Soc., 82 (1983) 87-90.
[45] W.E. Lee and S. Zhang, Int. Mat. Review, 44 (1999) 3, 77-104.
[46] P.T. Jones, Degradation mechanisms of basic refractory materials during the secondary refining of
stainless steel in VOD ladles, Ph.D. thesis (K.U. Leuven, May 2001).
[47] N. Bessho et al., ISIJ Int., 32 (1992) 1, 157-163.
[48] T. Isono, K. Ohnuki and K. Umezawa, in Proc. 4
th
Int. Conf. on Molten Slags and Fluxes, Ed. ISIJ,
Sendai (1992) 493-498.
[49] N. Bannenberg, 37
th
Int. Refractory Colloquium, Aachen, Germany (1994) 17-34.
[50] K.C. Ahlborg, in Proc. 5
th
Int. Conf. on Clean Steel, Hungary (1997) 151-156.
[51] S.C. van Wijk et al., in Proc. Steelmaking Conf. (2000) 231-236.
[52] K. Mukai, ISIJ Int., 32 (1992) 1, 19-25.
[53] M. Olette, ISIJ Int., 33 (1993) 10, 1113-1124.
[54] H. Goto and K. Miyazawa, ISIJ Int., 38 (1998) 3, 256-259.
[55] G.A. Bezuidenhout and P.C. Pistorius, Ironmaking and Steelmaking, 27 (2000) 387-391.
[56] R. Bommaraju et al., Ironmaker Steelmaker, 19 (1992) 4, 21-27.
[57] E. Takeuchi and J.K. Brimacombe, Met. Trans. B, 15B (1984) 9, 493-506.
[58] T. El Gammal and U. Schoneberg, Stahl und Eisen, 112 (1992) 1, 45-49.
Chapter 10 152
[59] C.A. Pinheiro, I.V. Samarasekera and J.K. Brimacombe, Mold flux for continuous casting of steel, in
Ironmaking Steelmaking (Oct. 1994-June 1996).
[60] S.R. Sankaranarayanan and R.I.L. Guthrie, CD-rom of the 6
th
Int. Conf. on Molten Slags, Fluxes and
Salts ,Ed. S. Seethamaman and Du Sichen, Sweden-Finland (2000) 134.pdf.
[61] A.W. Cramb and I. Jimbo, ISIJ Int., 16 (1989) 6, 43-45.
[62] K. Nakajima and K. Okamura, in Proc. 4
th
Int. Conf. Molten Slags and Fluxes, Ed. ISIJ, Sendai (1992)
505-510.
[63] L. Jianguo, Rare Metals, 13 (1994) 2, 81-91.
[64] W. Liu et al., Ironmaker Steelmaker, 19 (1992) 6, 51-55.
[65] H. Yin, H. Shibata, T. Emi and M. Suzuki, ISIJ Int., 37 (1997) 10, 936-945.
[66] H. Yin, H. Shibata, T. Emi and M. Suzuki, ISIJ Int., 37 (1997) 10, 946-955.
[67] A. Kusano et al., Steelmaking Conf. Proc. (1991) 147-151.
[68] S. Sridhar et al., CD-rom Proc. 6
th
Int. Conf. On Molten Slags, Fluxes and Salts, Ed. S. Seethamaman
and Du Sichen, Sweden-Finland (2000) 128.pdf.
[69] S. Kimura, Y. Nabeshima, K. Nakajima and S. Mizoguchi, Met. Trans B, 31B (2000) 11, 1013-1021.
[70] P. Rocabois, H. Saint Raymond and C. Gatellier, Pre-prints 2
nd
Int. Congress on the Science and
Technology of Steelmaking, Vol. 1, Swansea (2001) 255-264.
[71] K. Nakajima and S. Mizoguchi, Met. Trans B, 32B (2001) 8, 629-641.
[72] J. Hrkki et al., Scand. J. Met, 19 (1990) 116-126.
[73] J. Hrkki and M. Palander, Interceram, 40 (1991) 5, 284-289.
[74] P. Artelt, Stahl u. Eisen, 115 (1995) 2, 83-89.
[75] N. Bannenberg, Stahl u. Eisen, 115 (1995) 9, 79-86.
[76] J. Krj et al., Stahl u. Eisen, 113 (1993) 10, 73-75.
[77] Y. Vermeulen, Chemische interacties tussen vloeibare fasen en refractaire materialen bij het continu
gieten van staal, Ph.D. thesis, K.U. Leuven (2000).
[78] K-H. Tacke and J.C. Ludwig, Steel research, 58 (1987) 6, 262-270.
[79] F. Oeters et al, Steel Research, 59 (1988) 5, 192-201.
[80] O.J. Ilegbesi and J. Szekely, I&SM, 18 (1991) 12, 23-30.
[81] B. Kaufmann et al., Steel Research, 64 (1993) 4, 203-209.
[82] B. Grimm et al., Stahl u. Eisen, 115 (1995) 2, 71-79.
[83] M-Y. Zhu, T. Inomoto, I. Sawada and T-C. Hsiao, ISIJ Int., 35 (1995) 5, 472-479.
[84] M. Reza Aboutalebi, M. Hasan and R.I.L. Guthrie, Met. Trans B, 26B (1995) 8, 731-744.
[85] F.M. Najjar, B.G. Thomas and D.E. Hershey, Met. Trans B, 26B (1995) 8, 749-765.
[86] T. Ishii, S.S. Sazhin and M. Makhlouf, Ironmaking Steelmaking, 23 (1996) 3, 267-272.
[87] Y. Sahai and T. Emi, ISIJ Int., 36 (1996) 6, 667-672.
[88] R.W. Crowley, G.D. Lawson, B.R. Jardine and J.C. Grosjean, Rev. de Mtallurgie-CIT (1996) 7/8,
967-972.
[89] M. Neifer, N. Bannenberg and H. Lachmund, Stahl u. Eisen, 117 (1997) 5, 55-63.
[90] S. Wang and S. Louhenkilpi, in Proc. 5
th
Int. Conf. on Clean Steel, Vol. 1, Balatonfred, Hungary
(1997) 92-101.
[91] L. Jonsson, D. Sichen and P. Jnsson, ISIJ Int., 38 (1998) 3, 260-267.
[92] G.J. Hassall et al., Ironmaking and Steelmaking, 25 (1998) 4, 273-278.
[93] L.J. Heaslip and J. Schade, Ironmaker Steelmaker, 26 (1999) 1, 33-41.
Improvement of steel cleanliness 153
[94] H. Berg et al., Ironmaking and Steelmaking, 26 (1999) 2, 127-139.
[95] H. Tozawa, Y. Kato, K. Sorimachi and T. Nakonishi, ISIJ Int. 39 (1999) 5, 426-434.
[96] Y. Miki and B.G. Thomas, Met. Mat. Trans. B, 30B (1999) 8, 639-654.
[97] L. Zhang, S. Taniguchi and K. Cai, Met. Mat. Trans B., 31B (2000) 253-266.
[98] D-Y. Sheng, M. Sder, P. Jnsson and L. Jonsson, CD-rom Proc. 6
th
Int. Conf. On Molten Slags,
Fluxes and Salts, Ed. S. Seethamaman and Du Sichen, Sweden-Finland (2000) 127.pdf.
[99] D. Morales, S. Lpez-Ramrez, J. Palafox-Ramos and D. Zacharias, Ironmaking Steelmaking, 28
(2001) 1, 3-.
[100] K. Beskow, N.N. Viswanathan, L. Jonsson and D. Sichen, Met. Trans. B, 32B (2001) 4, 319-328.
[101] P.G. Jnsson and T.I. Jonsson, ISIJ Int., 41 (2001) 11, 1289-1302.
[102] T. Emi, Refining and casting of quality products, San Nicolas, Argentina (2000).
[103] G. DeWinter, Iron and Steel Engineer (1996) 11, 29-32.
Rob Dekkers, Ph.D. Thesis, Katholieke Universiteit Leuven, Leuven, Belgium (2002) 154