Mould powder investigations for high speed

casting
J. A. Kromhout*1,3, C. Liebske1, S. Melzer1, A. A. Kamperman2 and R. Boom1,3
Mould powders significantly determine the stability of the continuous casting process of steel. The
processes leading to melting of mould powder and solidification of mould slag were studied in situ
using high temperature X-ray diffraction with additional powder X-ray diffraction and microscopic
techniques. It was shown that during heating, a powder shows a specific sequence of phase
relations before melting takes place. During cooling and solidification, one or more crystalline
phases can be formed. The findings on high temperature properties were confirmed by the
analyses of slag rims and slag films, obtained from the Corus thin slab caster. Finally, the in depth
characterisation, together with some calculations on mould slag was compared with plant data of
the thin slab caster. Results from this work give a better understanding of the mould powder
functions at the thin slab caster and are being used to guide mould powder design for the current
and even higher casting speeds.
Keywords: Mould powder, Thin slab casting, Characterisation methods, Slag films, Crystallisation

Introduction is considered as essential to understand the mould

powder behaviour and the mould powder functions.
Mould powders significantly determine the stability of This applies especially to high speed thin slab casting,
the continuous casting process of steel at all casting where the process demands are more stringent compared
speeds. The main functions of mould powder are to with conventional casting techniques.8,9
provide strand lubrication and to control mould heat Mould powder developments are still mainly based on
transfer in the horizontal direction. At higher casting trial and error. A general approach is to relate the
speeds associated with thin slab casting, the role of chemical composition of a mould powder to the
mould powder is even more important. operational behaviour during casting. Some physical
During casting, the powder melts on the steel surface, properties like the melting point and the viscosity of
forming a layer of liquid mould slag. Subsequently the mould slag are addressed as well, but it is obvious that
slag infiltrates between the steel shell and the mould this approach will not automatically result in a suitable
creating a thin slag film which solidifies into glassy and mould powder or an in depth knowledge on powder
crystalline phases. The properties of the slag film dictate design.10
the main functions of mould heat transfer and strand At Corus IJmuiden, a project started within the
lubrication. The formation of crystals is favourable for a framework of the Materials Innovation Institute (M2i),
homogeneous and controlled (horizontal) heat transfer Delft University of Technology with the aim to develop
during casting, which is required to prevent the mould powders suitable for high speed thin slab casting
formation of surface cracks.1 with a maximum casting speed up to 8 m min21. For
Several authors have reviewed the composition, this work, a fundamental understanding and quantifica-
properties and operational experiences of mould pow- tion of the melting and solidification behaviour of
ders.2,3 However, only some publications have been mould slag, as well as the mould powder functions are
issued on the design of mould powders and the choice of required. To obtain these, the chemical and mineralo-
mould powder raw materials, i.e. the chemical and gical composition of mould powders will be related to
mineralogical constituents.4,5 A few papers describe the the physical properties of mould slag and to the
mineralogical behaviour of mould powder and mould operational behaviour during casting.
slag at high temperatures.6,7 Nevertheless, knowledge on
the mineralogy and the high temperature phase relations Thin slab casting at Corus IJmuiden
The thin slab caster at Corus IJmuiden (direct sheet
1
Corus Research, Development and Technology, PO Box 10000, 1970 CA plant, DSP) started production in 2000. The liquid steel
IJmuiden, The Netherlands is produced in the adjacent BOS no. 2 in 320 t batches
2
Corus Strip Products IJmuiden, Direct Sheet Plant, PO Box 10000, 1970
CA Ijmuiden, The Netherlands and is treated in a ladle furnace. The caster has one
3
M2i, Materials Science and Engineering Department, Delft University of strand and is equipped with a funnel shaped mould, a
Technology, PO Box 5008, 2628 AL Delft, The Netherlands specially designed submerged entry nozzle (SEN) and
*Corresponding author, email [email protected] an adjustable multiple pole electromagnetic brake

ß 2009 Institute of Materials, Minerals and Mining

Published by Maney on behalf of the Institute
Received 1 October 2008; accepted 9 December 2008
DOI 10.1179/174328109X407130 Ironmaking and Steelmaking 2009 VOL 36 NO 4 291
Kromhout et al. Mould powder investigations for high speed casting

(EMBr).11 Liquid core reduction decreases the slab Referring to slag formation, the minimum liquid pool
thickness from 90 to 70 mm. The designed production depth during casting is 5 mm. The liquid pool depth is
level is 1?3 Mt y21 of coils. Since commissioning several measured by immersing a stainless steel sheet in
technological developments have been implemented12 combination with a copper strip into the mould.10 The
and current operational performance is around the operators observe rim formation and the formation of
design level. The maximum operational casting speed is powder lumps. The maximum thickness of slag rims is
5?8 m min21 and the average value is over 5?4 m min21. y10 mm and the formation of lumps should be
Steel grades produced at the slab caster are mainly low avoided. Special attention is given to control the growth
carbon and high strength low alloy. The maximum of slag rims, i.e. to prevent excessive rim formation
sequence length is 10 ladles, i.e. more than 12 h casting. during casting.
It has been decided to increase the production of the The powder consumption should not be lower than
DSP to a level of 1?8 Mt/y21 with one caster strand. To the current average value of 0?05 kg m22 (kg powder/
meet this demand, the steel in mould time has to be metre slab surface). The consumption is measured by
increased to y85% and the maximum casting speed will monitoring continuously the weight of the powder bin
be increased to 8?0 m min21. A project was started to on the tundish car. Homogeneous and controlled heat
develop and implement the essential technologies for transfer and controlled strand lubrication are desired,
this purpose. Special attention is given to the design of within the given operational window of the caster. The
mould powders for high speed casting. main aspects for mould heat transfer are to prevent slab
surface cracks (longitudinal facial cracks) and to protect
Mould powders for thin slab casting and the mould copper plates at the meniscus area, due to the
operational criteria (increased) heat load. Mould heat transfer is calculated
using mould cooling water temperatures and flow;
During the first year of operation, several mould
additional data can be obtained via the mould thermo-
powders were used with both high (.1?15) and low
(y1?0) basicity (CaO/SiO2). After approximately a year, couples. Strand friction is obtained from the hydraulic
a low basicity powder was selected as the standard. With oscillating system of the mould. Scale formation at the
this powder the casting speed and sequence length have slab surface (i.e. wustite, FeOx), especially in the tunnel
been increased successfully to the current level. furnace area, should be independent of the mould
Simultaneously, trials with alternative mould powders powder type. Scale formation is observed by inspection
have been carried out with the aim to investigate the of the rolls in the tunnel furnace.
possibilities to increase process stability or product
quality. In Table 1, an overview is given of the chemical Mould powder characterisation and
composition of the standard powder (A) and of two
alternative materials (B and C). Mould powder B is an
phase relations
alternative, especially for increased casting speeds. Characterisation
Mould powder C is used to increase mould heat transfer, It is clear that the chemical composition as given in
i.e. to increase strand solidification which will result in a Table 1, does not provide information on the miner-
thicker steel shell. alogical composition, i.e. the raw material choice of the
Note the limited decrease in basicity (CaO/SiO2) for mould powders. As mentioned before, the mineralogy is
mould powders A to C and the increase in sodium an important theme in the current mould powder
(Na2O) for these powders. In all the mould powders, one developments.
source of free carbon is used. Room temperature X-ray diffraction measurements
For mould powder evaluation, several operational (XRD) with subsequent Rietveld analyses were used to
criteria have been defined. These criteria are related to: determine quantitatively the mineralogical phase com-
(i) slag formation, which includes liquid pool position of the three mould powders. By doing this, the
depth and rim formation raw material choice of the three mould powders was
(ii) powder consumption revealed (see Table 2).
(iii) mould heat transfer
(iv) strand lubrication
(v) scale formation. Table 2 Mineralogical composition, as measured at
Corus IJmuiden (Rietveld XRD), wt-%
Table 1 Chemical composition of mould powders, as
measured at Corus IJmuiden, wt-% Mould powder type

Mould powder type Mould powder composition A B C

Component A B C Silicates
Wollastonite (CaSiO3) 50¡5 56¡6 49¡6
CaO/SiO2 1.0 0.9 0.8 Quartz (SiO2) 2¡1 3¡1 2¡1
SiO2 33.2 32.4 35.3 Diopside (CaMgSi2O6) 2¡1 7¡2 2¡2
CaO 33.6 29.3 27.0 Albite (NaAlSi3O8) 14¡3 0 0
MgO 0.7 1.3 1.3 Fluorites
Al2O3 2.9 2.7 2.7 Fluorite (CaF2) 17¡3 5¡2 7¡2
Na2O 11.9 14.2 16.6 Cryolite (Na3AlF6) 0 10¡2 8¡2
MnO 0 0 3.0 Carbonates
F 8.8 7.0 8.4 Natrite (Na2CO3) 13¡3 16¡3 14¡3
Cfree 3.7 5.7 0.6 Calcite (CaCO3) 2¡1 4¡2 2¡1
CO2 9.4 8.9 7.8 Glass 0 0 18¡4

292 Ironmaking and Steelmaking 2009 VOL 36 NO 4

 Kromhout et al. Mould powder investigations for high speed casting

In general, a mould powder contains silicates, fluorites

and carbonates. Additionally, amorphous components
(glass) can be present. For each mould powder,
however, significant differences in the raw material
composition can be observed. In mould powder A, albite
(feldspar) is used. Furthermore, this mould powder has
one source of fluorine. Mould powders B and C both
have two sources of fluorine and do not contain albite.
Mould powder C also contains a significant amount of
amorphous materials (glass) which is not present in the
other two mould powders.
During the Rietveld calculations, some problems
occurred on data fitting, which is associated with the
effects of preferred orientation of wollastonite (CaSiO3).
Wollastonite occurs in three different crystallographic
structures, which have similar peak positions, such that
those structures cannot be reliably refined in multi-
phased diffraction patterns. The uncertainties on the
determined phase proportions are therefore relatively
high.
The viscosity of the mould powders and the corre- 1 Example of XRD pattern of mould powder during
sponding break point (or solidification point) was heating
measured by the supplier using a viscometer (rotating
cylinder method). The viscosity at 1300uC is considered XRD patterns were collected in 2h range from 20 to
as a characteristic value of the mould slag and is related 50u, while the temperature was raised from 500 to
to slag infiltration, i.e. mould powder consumption 1250uC and then subsequently was lowered to 600uC.
during casting.13 The break point or solidification point Heating and cooling rates were 200uC min21. X-ray
of a mould powder is related to the mould heat transfer diffraction patterns were taken in temperature incre-
between the solidifying steel shell and the copper mould ments of 50uC. A second sequence of XRD patterns was
(horizontal heat transfer) and indirectly to mould collected by heating with an offset of 25uC compared to
friction during casting. The melting trajectories of the the first run to obtain a higher resolution of phase
three mould powders were measured at Corus IJmuiden stabilities as a function of temperature. Furthermore,
using a hot stage microscope. Results of the analyses are the chemical composition of the remaining mould slag
given in Tables 3 and 4. was analysed using scanning electron microscope and
energy dispersive X-ray spectrometer (SEM/EDS) tech-
Phase relations
niques to verify that phase relations were not influenced
Initial mixtures of raw materials of the mould powders by fluorine loss during high temperature experiments.
are not in chemical equilibrium but the systems tend The data on powder melting are important to under-
towards it during heating, which results a series of stand slag formation and the formation and growth of
reactions between the phases and eventually formation slag rims during casting. Information on slag solidifica-
of secondary phases. These reactions may include solid tion, i.e. the formation and presence of crystals is
state reactions or partial melting and solid–liquid essential to control the horizontal heat transfer during
interaction. The mineralogical composition will change casting. This impacts directly the solidification of steel,
continuously during heating until the mould powder i.e. the occurrence of surface cracks at the slab surface
melts completely. The sequences of reactions are thereby (surface quality) and the lifetime of the mould copper
specific for each mould powder and controlled by the plates. A summary of the results of the three mould
mineralogical and chemical bulk composition. powders is given in Fig. 2.
Changes in mineralogy of the major phases During cooling, mould powder A shows crystal-
(.10 wt-%) for the three powders have been observed lisation of cuspidine (3CaO.2SiO2.CaF2 or
in situ using high temperature X-ray diffraction Ca4Si2O7F2) at temperatures between 1150 and
(HTXRD) with additional powder XRD and micro- 1100uC. No other major phases were detected during
scopic techniques. HTXRD. Mould powder B shows the formation of
Figure 1 shows a typical example of an XRD pattern three phases, all at the same temperature. Crystallisation
during heating. The same techniques were also applied starts around 1100uC, i.e. at a temperature range
to study the crystallisation sequence during cooling.
Table 4 Melting trajectory, as measured at Corus
Table 3 Viscosity and break point, as measured by IJmuiden
supplier
Mould powder type
Mould powder type
A B C
A B C
Softening point, uC 1066 994 1012
Viscosity at 1300uC, Pa s21 0.13 0.09 0.10 Melting point, uC 1088 1026 1045
Break point (solidification point), uC 1167 1110 1054 Fluidity point, uC 1112 1064 1070

Ironmaking and Steelmaking 2009 VOL 36 NO 4 293

Kromhout et al. Mould powder investigations for high speed casting

2 Results of HTXRD analyses and indications of phase stabilities of major phases, within three mould powders (heating
and cooling): these stabilities may be dependent on actual heating and cooling rates and are strictly valid only for
given experimental conditions

comparable to mould powder A. Together with cuspi- Comparing to previous work, however, the obtained
dine, the phases combeite (2Na2O.4CaO.6SiO2 or results do not completely fit.6 Especially, the formation
Na4Ca4Si6O18) and villiaumite (NaF) are formed. For of villiaumite and the presence of only two phases in
powders A and B the temperature range of complete mould powder C differ from the previous work. Several
melting (stability during heating) is comparable to the authors have described the formation of cuspidine
temperature where the first crystallisation is detected. in more detail.14,15 The presence of more than one
Mould powder C shows crystallisation of cuspidine at crystalline phase, as in mould powder B, is not described
a temperature around 950uC, followed by crystallisation and possible effects on mould heat transfer are not
of villiaumite at a temperature around 700uC. Note that known.
the crystallisation of cuspidine in powder C starts y100
to 150uC lower than the observed stability during
heating and also 100 to 150uC lower than the crystal-
Some calculations
lisation of cuspidine in mould powder A or B. For the To obtain an indication of the behaviour of the mould
given experimental cooling rates, crystallisation in powders during casting, some calculations were made.
mould powder C is suppressed (undercooling) compared The calculations are based on the chemical composition
with A and B; which is an effect of the chemical and are focussed on the crystallinity and on the break
composition (e.g. increase in Na and/or decrease in point or solidification point.
basicity compared to A and B) and is controlled by The crystallinity of mould slag, i.e. the crystallisation
nucleation rates of such glass forming liquids. This effect tendency is calculated using the NBO/T ratio: the
is not yet fully understood. number of non-bridging oxygen per tetrahedrally
After HTXRD measurements, the remaining mould coordinated atom.16 This approach is based on the
slag was analysed using XRD measurements at room
temperature. Mould powder A showed cuspidine as
major phase with some minor amounts of combeite and Table 5 Estimation of relative amounts of crystalline
a sodium–aluminium silicate (e.g. nepheline). Mould phases in mould slag after HTXRD
powder B showed the three phases cuspidine, combeite measurements
and villiaumite with some minor amounts of nepheline. Mould powder type
Mould powder C showed the two detected phases
cuspidine and villiaumite and no additional (minor) A B C
crystalline components. Based on the intensities of the
XRD patterns obtained after high temperature runs, the Main phases
Cuspidine zzz* zz{ zz
abundance of crystalline phases can be estimated
Combeite zz
qualitatively. Note that the samples contain significant Villiaumite zz z{
proportions of glass, which is not considered in these Minor phases
analyses. A summary of the crystalline phases in mould Combeite z
powders A, B and C after HTXRD measurements is Nepheline z z
given in Table 5. *zzz: major/dominating phase.
In general, cuspidine is predominantly formed as a {
zz: phase present.
{
crystalline phase during solidification of mould slag. z: minor phase.

294 Ironmaking and Steelmaking 2009 VOL 36 NO 4

 Kromhout et al. Mould powder investigations for high speed casting

Table 6 Slag calculations Table 7 Comparison of several slag properties

Mould powder type Mould powder type

A B C A B C

Crystallinity (NBO/T), mol.-% 81 79 71 Melting point (measured), uC 1088 1026 1045

Break point (Sridhar), uC 1182 1135 1072 Break point (measured), uC 1167 1110 1054
Break point (calculated, Sridhar), uC 1182 1135 1072
Start of crystallisation (HTXRD), uC* y1150 y1100 y950
relation between the structure or the depolymerisation Number of crystalline phases (HTXRD)* 1 3 2
of silicate slags and the physical properties. Crystallinity (NBO/T), mol.% 81 79 71
The NBO/T ratio has been adopted to represent the Basicity (CaO/SiO2) 1.0 0.9 0.8
degree of depolymerisation.17 The index is calculated as *Note that these values are only strictly valid for the given
follows experimental heating/cooling rates.
NBO=T~ (1)
(1)
2xCaO z2xBaO z2xCaF2 z2xNa2 O {2xAl2 O3 z6xFe2 O3 z(2xMgO z2xMnO )
C show a similar trend. This trend can be recognised via
xSiO2 z2xAl2 O3 zxTiO2 z2xB2 O3 z(xMgO zxMnO ) the start of crystallisation, as measured during HTXRD.
There is a similarity between the break point and the
where x is mole fraction of the component in the mould start of crystallisation, as measured with HTXRD.
flux. The bracket in the denominator/numerator means However, crystallisation of mould slag needs to be
it will be included into the denominator if MgO is larger investigated in more detail. Note that mould
than 7% and/or MnO is larger than 4%, otherwise it will powder C shows less crystallisation starting at a lower
be included in the numerator. The recommended
temperature.
equation to calculate the %crystallinity is
%crystallinity~141:1(NBO=T){284:0 (2) Operational experiences
where %crystallinity is given in mol.-%. A general overview of the operational performance of
A relation between the chemical composition and the the three mould powders at a casting speed of
break point is given by Sridhar.18 In this study, the 5?4 m min21 is given in Table 8. Subsequently, the slag
relationship for dynamic conditions is used formation during casting (liquid pool depth and rim
Tbr ~1120{8:43%Al2 O3 {3:30%SiO2 z8:65%CaO{ formation) is described in more detail.

13:86%MgO{18:40%Fe2 O3 {3:21%MnO{ Liquid pool depth

The most important parameters for powder melting and
9:22%TiO2 z22:86%K2 O{3:20%Na2 O{
the formation of the liquid pool are the vertical heat
6:46%F (3) transfer during casting and the amount of free carbon
Cfree. Trials with powders A, B and C, with different
where Tbr is given in uC and the chemical composition of
amounts of free carbon, showed the same values for the
the mould slag given in wt-%.
liquid pool depth, i.e. between 2 and 4 mm. There is no
Based on the chemical composition of the three mould
significant effect of the amount of free carbon. This
powders, the crystallisation tendency and the break
indicates that the vertical heat transfer (or the lack of
point were calculated. Results are given in Table 6.
vertical heat transfer) dominates powder melting and
Compared with mould powders A and B, a lower
slag formation. Most likely, this result can be related to
crystallinity was calculated for mould powder C. The
calculated break point decreased significantly from the EMBr practice at the DSP caster; more specifically
mould powder A via B to C. This can possibly be the EMBr settings during casting. Electromagnetic
related to the increase in Na and the decreased value of breaking will affect the (vertical) heat transfer in the
basicity for powder C. mould due to the reduction of turbulent metal flow
velocities. Previous trials showed that a decrease in
EMBr settings will result in an increase in liquid pool
Summary of data depth during casting.19 The EMBr practice at the DSP
An overview of the data obtained on slag properties is caster, i.e. the vertical heat transfer mainly determines
given in Table 7. It is clear that the data on measured powder melting and slag formation. New trials with
and calculated break points for mould powders A, B and decreased EMBr settings are being considered.
Table 8 Mould powder performance at DSP caster: vc55?4 m min21

Mould powder type

A B C

Liquid pool depth, mm* 2–4 2–4 2–4

Rim/lump formation* Some rims, ,10 mm Some rims, ,10 mm No rims
Powder consumption, kg m22 0.055 0.065 0.080
Average mould heat transfer, MW m22 Wide face 2.9 3.0 3.4
Narrow face 2.4 2.4 2.1
Scale formation No No No
*Liquid pool depth and rim formation behaviour as measured during stable casting conditions (stable mould meniscus).

Ironmaking and Steelmaking 2009 VOL 36 NO 4 295

Kromhout et al. Mould powder investigations for high speed casting

The effects of the inorganic mould powder constitu-

ents and the mould slag were studied. Rims, obtained
from mould powder A were sampled and investigated
using optical microscopy and SEM/EDS techniques.
These rims showed a layered structure of mould slag,
intermediate phases/sintered products and mould pow-
der. Also small Fe droplets and particles of carbon black
(free carbon source) were detected. Furthermore, it was
shown that thin crystals of a condensed Na compound
coat and cement the various layers. As a result, the rims
are dense, strong and will survive relatively long during
the process, i.e. the rims will not melt easily. The
mechanism leading to the formation and growth of slag
rims can be considered as a ‘painting’ mechanism.2,9
Analyses of the slag rims also showed the melting
sequence and phase relations, as detected by HTXRD
3 Slag rim analysis showing melting sequence: mould (powder melting). An illustration of this is given in Fig. 3.
powder A An excess of sodium, especially at lower temperatures,
possibly enhances the formation of dense and strong rims.
Another effect on rim formation is the melting point
Rim formation
(melting trajectory) of the mould powder and, in the
During casting, slag rims are formed which adhere to the authors’ experience, the solidification behaviour of the
mould walls close to the meniscus. Under stable casting mould slag. A mould powder with a low melting point will
conditions, slag rims are small but play a role during the be less sensitive to rim formation during casting.23 This
infiltration of mould slag. However, rims can grow, was confirmed by the behaviour of mould powder C
disturb and even interrupt the casting process.20 Based during casting. Mould powder C clearly showed no
on the experience at the DSP caster, excessive rim formation of rims during casting, where mould powders A
formation should be avoided. The infiltration process and B give some restricted formation of rims.
(mould slag) can be hindered, parts of the rim can break
and fall into the mould and bridges (very big rims) can Influence of process conditions
be formed, especially at the narrow faces or between the An effect on the formation and growth of rims is the
SEN and the mould wall. Owing to the mould mould heat transfer near the meniscus area. Microscopic
dimensions (mould thickness) of a thin slab caster and analyses of mould powder A, focussing on the miner-
the high casting speed, with increasing demands to alogical composition, showed that rims from the wide
meniscus stability, the control of rim formation and face differ from rims obtained from the narrow face, due
especially rim growth is very important. The formation to different heat loads.
and growth of rims can be affected by the properties of Above all, the rims that have been investigated
the mould powder and by the conditions at the mould from mould powder A, especially the bigger ones,
meniscus. clearly show the effects of mould meniscus fluctuations,
i.e. turbulence at the meniscus area during casting.
Influence of mould powder The so called painting mechanism will considerably
Some literature sources suggest effects of free carbon be enhanced by these fluctuations. Stable casting
and powder melting on rim formation.21,22 However, no conditions, i.e. a stable mould meniscus will not result
clear mechanism is given until now. Previous work at the in excessive rim formation. There is no direct need to
Corus thin slab caster concentrated on the role of free adapt the mould powder composition and slag proper-
carbon and the corresponding melting behaviour on the ties to control rim formation and rim growth.
formation and growth of rims.9,10 It was shown that
rims and lumps can be formed due to insufficient melting Slag films
of mould powder (slag formation) which is caused by an After infiltration, a thin slag film is formed between the
excess of free carbon, a specific choice of free carbon steel shell and the mould. During the casting process, the
sources or an inhomogeneous distribution of free carbon slag film moves downwards and solidifies into a glassy
particles within the mould powder granules. An phase and one or more crystalline phases. The slag film
improper use of free carbon will be a source of rim realises the main powder functions mould heat transfer
formation during casting. and strand lubrication and for this reason the slag film

4 Slag droplets, sampled from inside mould: y50 cm depth

296 Ironmaking and Steelmaking 2009 VOL 36 NO 4

 Kromhout et al. Mould powder investigations for high speed casting

The thickness of the slag films is between 0?2 and

0?3 mm. Microscopic analyses showed slow growth
cuspidine crystals in a glass matrix; these predominant
crystals can be observed in the middle of the slag films
and at the mould (Cu) side. No other crystals have been
detected. The strand (steel) side of the films shows glass
with some rapid growth cuspidine crystals and, occa-
sionally, mould powder particles, adhering to this side.
These crystals and powder particles are probably formed
directly after emptying the mould, i.e. after casting.
Furthermore, gas bubbles and small steel droplets can be
found in the slag films. Incidentally, small ZrO2
particles, most likely coming from the SEN can be
found within the slag films. These particles act as
nucleation site for cuspidine crystals. Composition and
phases of the slag films appeared similar to those
5 Slag films, sampled from inside mould (y50 cm depth), measured slag droplets.
showing glass and cuspidine crystals and to lesser Based on the slag consumption during casting and the
degree bubbles and small steel droplets density of the mould slag, the average thickness of the
liquid film can be calculated1
can be considered as a key for a further understanding of dl ~Qs =r&Qs =2600 (4)
the mould powder functions and to guide mould powder
design.24,25 Sampling and characterisation of slag film where dl is average thickness of the liquid film, Qs slag
structure and properties is strongly required, but for consumption in kg m22 and r density of the liquid flux
reasons of process stability and safety, is very difficult to in kg m23. Given a slag consumption of 0?05 kg m22
do. At the thin slab caster, it was tried to sample slag and a slag density of 2600 kg m23, the liquid film
films at three locations: thickness is y0?0192 mm. As a rule of thumb, the
(i) at the meniscus area average liquid film thickness is at least a tenth of the
(ii) from inside the mould and total film thickness. This indicates an average film
(iii) under the mould. thickness of y0?2 mm or more.
Work described in this paper concentrates on location Special attention has been given to investigate the
(ii) using mould powder A. Sampling slag films from surface roughness of the slag films at the mould side. It is
inside the mould was carried out directly after casting generally accepted that this surface roughness, more
a sequence and proved to be extremely difficult. Surpri- specifically the interfacial thermal resistance Rint, plays a
singly, in addition to slag films, small slag fragments and major role in controlling the mould heat transfer
droplets of mould slag were detected and obtained. (horizontal heat transfer) during casting.27,28 This
roughness is a result of the crystallisation (and
Slag droplets corresponding shrinkage) of the glassy solid slag film;
Slag droplets adhered to the wide face of the mould wall. values between 20 and 50 mm and even more are
The droplets were between 6 and 20 mm long with a mentioned. In this view, the surface roughness acts as
thickness between 0?6 and 2?3 mm (Fig. 4). The an important barrier for mould heat transfer which can
mechanism of droplet formation is not fully understood. be described by
Contraction of the very thin slag film, directly after 
dfilm

casting (during emptying the mould) seems to be most Rtotal ~Rfilm zRint ~ zRint (5)
keff
reasonable. The occurrence of slag droplets and slag film
fragments was earlier reported after experiments at a where Rtotal is total thermal resistance, Rfilm thermal
billet caster.26 resistance of the slag film (including radiation and
Microscopic analyses (SEM/EDS) of the droplets conduction), Rint interfacial thermal resistance, dfilm
showed cuspidine crystals and a glass phase. However, thickness of the slag film and keff the heat transfer
the local amounts of crystalline and glass differ within coefficient of the slag film.
the several droplets. In some samples, the side in contact Recently, it has been found that the effect of the
with the mould (Cu) showed significantly more crystals interfacial thermal resistance and the corresponding
than the upper part of the droplets. At the mould side, surface roughness is not as large as widely assumed. This
slow growth cuspidine crystals embedded in a glass work is based on slag films, sampled from the mould of a
matrix can be detected, suggesting formation during the pilot caster and on corresponding one-dimensional heat
casting process. On this layer, a second glassy layer can transfer calculations. As a result, a surface roughness up
be found, together with some cuspidine crystals which to 10 mm is suggested.29
show a rapid growth. These crystals are possibly formed Up to now, the surface roughness of slag films
after the casting process. As before, the films also show originating from the mould of a thin slab caster has
the presence of gas bubbles, metal droplets and mould never been reported. First results of the obtained slag
powder components.9 films showed a very smooth surface with a surface
roughness at the mould side of y2 mm (Fig. 6).
Slag films The other side of the slag film, the strand side with a
Thin pieces of slag films were obtained directly after cast- surface of glass, showed a very smooth surface with no
ing. An illustration of some slag films is given in Fig. 5. surface roughness. These findings indicate that the

Ironmaking and Steelmaking 2009 VOL 36 NO 4 297

Kromhout et al. Mould powder investigations for high speed casting

a b
6 a surface of slag film (mould side), sampled from inside mould, b detail showing surface of slag film and cuspidine
crystals at fractured surface

control of mould heat transfer during thin slab casting is like metal droplets and mould powder compo-
mainly realised by the slag film and by the slag film nents. The thickness of the slag films is between
properties. The interface at the mould side plays a less 0?2 and 0?3 mm.
dominant role. (v) The surface roughness of slag films is very low.
It is clear that this work will be continued. Special This indicates that the control of mould heat
attention will be given in measuring and calculating the transfer during casting is mainly realised by the
slag film properties.30 Furthermore, effects of the high slag film properties; the interfacial thermal
cooling rate, as practiced during thin slab casting, are resistance playing a less dominant role.
another point for future research.31
Acknowledgements
Progress The authors would like to thank C. J. de Moel for the
Future work will concentrate on the solidification and excellent work on sampling slag films from inside the
crystallisation of mould slag. In particular, the influence mould. Furthermore, thanks are due to M. Koolwijk for
of cooling rate will be given special attention. Slag films the support during the characterisation of slag films.
from inside the mould are needed to increase knowledge This research was carried out under the project number
on slag film properties and surface roughness, with the MC5?05225 in the framework of the Research
final aim to define mechanisms for mould heat transfer Programme of the Materials Innovation Institute M2i
and strand lubrication during high speed thin slab (www.m2i.nl), the former Netherlands Institute for
casting. Metals Research. This paper is based on a presentation
at the 6th European Conference on Continuous Casting,
organised by AIM in Riccione, Italy on 3–6 June 2008.
Conclusions
(i) The equilibrium phase relations of a mould References
powder at high temperatures can be studied in
1. K. C. Mills, A. B. Fox and A. S. Normanton: Proc. ISSTech 2003,
situ using HTXRD. This provides knowledge Indianapolis, IN, USA, April 2003, Iron and Steel Society, 693–
on the melting mechanisms of mould powder 701.
and solidification and crystallisation of mould 2. P. Riboud: in ‘Metallurgie des Stranggießens, Gießen und
slag. Each mould powder shows a specific Erstarren von Stahl’, (Herausgeber K. Schwerdtfeger), 233–255;
1992, Düsseldorf, Verlag Stahleisen mbH.
sequence of melting and solidification.
3. S. Ogibayashi, K. Yamaguchi, T. Mukai, T. Takahashi,
(ii) Rim formation and growth can be related to the Y. Mimura, K. Koyama, Y. Nagano and T. Nakano: ‘Mold
composition of a mould powder, the heat load powder technology for continuous casting of low-carbon alumi-
around the meniscus area and in particular num-killed steel’, Nippon steel technical report, Nippon Steel
Corporation, Chiba, Japan, 1987, 1–10.
the mould level fluctuations during casting. The 4. M. Kawamoto, K. Nakajima, T. Kanazawa and K. Nakai: ISIJ
mechanism for rim formation and growth of the Int., 1994, 34, 593–598.
standard mould powder can be considered as a 5. R. Scheel and W. Korte: Stahl u. Eisen, 1987, 107, 37–43.
‘painting’ mechanism. A casting process, with 6. P. Grieveson, S. Bagha, N. Machingawuta, K. Lidell and K. C.
the emphasis on controlled meniscus behaviour Mills: Ironmaking Steelmaking, 1988, 15, 181–186
7. T. Hiromoto, T. Shima and R. Sato: Proc. 62nd National Open
is crucial to prevent the formation of excessive Hearth and Basic Oxygen Steel Conf., Detroit, MI, USA, March
slag rims. 1979, Iron and Steel Society, 40–47.
(iii) Inside the mould, the slag film can form 8. D. Blevins, M. Ingold, A, Schaefer, J. Neal, F. Neumann and
droplets of mould slag. These are found directly C. Sowa: Ironmaking Steelmaking, 2000, 27, 85–88.
after casting. 9. J. A. Kromhout, S. Melzer, E. W. Zinngrebe, A. A. Kamperman
and R. Boom: Steel Res. Int., 2008, 79, 143–148.
(iv) Slag films from inside the mould showed 10. J. A. Kromhout, A. A. Kamperman, M. Kick and J. Trouw:
cuspidine crystals in a glass matrix, together Ironmaking Steelmaking, 2005, 32, 127–132.
with gas bubbles and some minor components 11. M. C. M. Cornelissen and R. Boom: Steel Res., 2003, 74, 716–723.

298 Ironmaking and Steelmaking 2009 VOL 36 NO 4

 Kromhout et al. Mould powder investigations for high speed casting

12. M. C. M. Cornelissen, J. A. Kromhout, A. A. Kamperman, 21. C. Lefèbvre, J. P. Radot, J. N. Pontoire and Y. Roux: La Rev.
M. Kick and F. Mensonides: Ironmaking Steelmaking, 2006, 33, Métall., 1997, 94, 489–496.
362–366. 22. J. Sardemann and H. Schrewe: Proc. 74th Steelmaking Conf.,
13. M. Wolf: Proc. METEC Cong. ’94, 2nd Eur. Continuous Casting Washington, DC, USA, April 1991, Iron and Steel Society, 719–
Conf., Düsseldorf, Germany, June 1994, Verlag Stahleisen mbH, 728.
Vol. 1, 78–85. 23. W. H. Emling: Ironmaking Steelmaking, 1994, 21, 43–44.
14. T. Watanabe, H. Fukuyama and K. Nagata: ISIJ Int., 2002, 42, 24. B. Tarrant and G. Brooks: Ironmaking Steelmaking, 2003, 30, 52–
489–497. 60.
15. M. Hayashi, T. Watanabe, H. Nakada and K. Nagata: ISIJ Int., 25. P. O. Hooli: Ironmaking Steelmaking, 2002, 29, 293–296.
2006, 46, 1805–1809. 26. E. Lainez and J. C. Busturia: ‘Summary of the research performed
16. K. C. Mills: ISIJ Int., 1993, 33, 148–155. under ECSC 7210-CA/932 and 7210-CA/935 projects’, Sidenor
17. Z. Li, R. Thackray and K. C. Mills: Proc. 7th Int. Conf. on ‘Molten IzD, Basauri, 1994.
slags, fluxes and salts’, Cape Town, South Africa, January 2004, 27. A. Yamauchi: ‘Heat transfer phenomena and mold flux lubrication
The South African Institute of Mining and Metallurgy, 813– in continuous casting of steel’, Doctoral thesis, Royal Institute of
819. Technology, Stockholm, Sweden, 2001.
18. S. Sridhar, K. C. Mills, O. D. C. Afrange, H. P. Lörz and R. Carli: 28. J.-F. Holzhauser, K.-H. Spitzer and K. Schwerdtfeger: Steel Res.,
Ironmaking Steelmaking, 2000, 27, 238–242. 1999, 70, 252–258.
19. K. C. Mills, J. A. Kromhout, A. Hamoen and R. Boom: Proc. 29. M. Hanao and M. Kawamoto: ISIJ Int., 2008, 48, 180–185.
Admet 2007, (ed. G. Stovpchenko), Vol. 2, 174–181; 30. S. Ozawa, M. Susa, T. Goto, R. Endo and K. C. Mills: ISIJ Int.,
Dnipropetrovsk, National Metallurgical Academy of Ukraine. 2006, 46, 413–419.
20. M. Emi: Proc. 74th Steelmaking Conf., Washington, DC, USA, 31. K. Tsutsumi, T. Nagasaka and M. Hino: ISIJ Int., 1999, 39, 1150–
April 1991, Iron and Steel Society, 623–630. 1159.

Ironmaking and Steelmaking 2009 VOL 36 NO 4 299