Liquefaction and crystallization of mold powders and mold slags

This work summarizes 11 publications of the applicant as a necessary condition for obtaining
a postdoctoral lecture qualification. Its focus is on the key processes of mold powder
liquefication and solidification of their slags during continuous casting of steels and on the
related experimental methods. First, published devices used to investigate the melting
behavior of mold powders are described and evaluated with respect to their ability to represent
the heating conditions of the mold powder in the continuous casting process. For this purpose,
the methods are divided into two groups with respect to their heating rates and characteristic
values, e.g., Biot number or characteristic length, are used for evaluation. The experimental
set-ups of one group enable a sufficiently high heating rate and sample geometry does not
exceed the respective characteristic length and may be used to simulate liquefaction in course
of heating in the mold. Contrary, for the other group slow heating rates are used and results
close to the steady state are obtained. Thus, data regarding reaction kinetics cannot be
concluded form their results. Nevertheless, they provide essential information for
understanding equilibrium phase compositions in dependence of temperature. This has been
especially necessary, when thermodynamic data for calculations of mold powder compositions
have not been available as they are now.
For casting ultra-low carbon steels recarburization of the liquid steel has a detrimental effect
on the product quality. Thus, mold powders without free carbon are required. Several nonoxides
are suggested as substitutes, e.g., BN or SiC, to control the liquefaction of mold
powders and some industrial trials pose promising results. To understand the role of carbon in
mold powders during heating, graphite is added to a commercially available mold powder in
different amounts. Mineralogical investigations of temperature treated samples reveal that
carbon delays solid-solid reactions and shifts them to higher temperatures. As graphite was
not incorporated into the granules but simply added to the granules, the direct contact to raw
material particles is reduced and its effect is lower than when added to the powdery mixture.
Contrary, at higher temperatures the granules liquefy independently from each other and delay
the formation of a continuous liquid phase, which is not the case for powdery samples. Based
on these results, SiC is selected to be investigated in more detail. Also, the addition of
antioxidants to increase its stability is tested. Results of thermodynamic calculations of
mixtures containing different raw materials which form a slag of the same chemical composition
after oxidation and liquefaction, and those obtained by mineralogical investigations of
tempered specimens are compered. With respect to liquefaction SiC shows similar behavior
as carbon. At low temperatures it delays the solid-solid reactions. Even though higher amounts
of liquid phases are formed at lower temperatures, with increasing temperature their amount
is lower compared to the mold powder as delivered, because the necessary amount of SiO2 is
not available due to the stability of SiC up to high temperatures. Following, its content can be
considerably reduced related with the reduction of CO2 emission. The addition of antioxidants
has not the desired effect. Contrary, it even contributed to the formation of liquid phase
especially at lower temperatures. Another approach to reduce or even remove melt controlling
additives is the increase of the distance between such raw material particles which are
expected to react with each other. This is realized by separating the raw materials into “basic”
and “acidic” granules. The mineralogical investigations reveal that proper allocation of the raw
materials to these two granule types enables the adjustment of a similar melting behavior as
shown by the standard mold powder without any liquefaction controlling additive. Even though
the application of these products may need special measures regarding homogenous mold
powder feed to the mold, these carbon free products may contribute to increase steel quality
and considerable reduce CO2 emission.
To characterize the crystallization behavior of mold slags under near service conditions, the
Double Hot Thermocouple Technique is used, but there are still some deficiencies. Its results
are mainly represented by visually comparing pictures taken during the experiment in
dependence on time. Furthermore, a continuous temperature gradient from the hot to the cold
side is desired, but the minimum temperature is observed in the center. Thus, the set-up of
this method to prevent a minimum temperature in the center is improved and the representation
of the results in order to reliably compare the crystallization behavior of different mold slags is
developed. Furthermore, the crystallization behavior if slags during experiments using the
Single Hot Thermocouple Technique is compared to that of slag films in the continuous casting
process and discussed.
For mold slags, the viscosity is represented in dependence on the reciprocal absolute
temperature to determine the Break Temperature manually by intersecting the tangents
approaching the linearized curve progression. Consequently, the obtained results are often
dependent on the measuring person. To overcome this disadvantage and to automize its
determination, a proper procedure is developed, tested and verified.
Due to its significant contribution to liquefaction and crystallization, fluorine is added to mold
powders. Otherwise, it is regarded as health hazardous and corrodes the continuous casting
machine. Thus, the development of new slag compositions without fluorine is required. Several
attempts have already been made to use TiO2 or B2O3 in mold slags to achieve similar
properties as shown by the standard fluorine containing slags. The main focus lies on the
formation of a proper crystalline fraction within the slag film to control the heat transfer. For
casting of soft steels, primarily glassy solidification is desired. Different chemical compositions
based on selected oxides are calculated with FactSage and optimized. Following, quenching
tests are carried out to prove the formation of a glassy phase. Based on these results, samples
are selected to be tested in the laboratory regarding viscosity and crystallization behavior. One
of these composition reveals similar properties as the reference fluorine containing slag and is
selected for tests in a pilot caster in comparison to the standard one. During operation no
differences in the measured values are observed, but the structures of the slag films differ and
the surface of the slab shows lower oscillation marks than the one casted with the fluorine
containing one. This reveals the applicability of the used laboratory methods to develop new
mold flux compositions.
Redox reactions between the steel containing Mn, Ti and Al and the SiO2 of the mold slag take
place, changing their chemical compositions and following their properties. Especially, slag
viscosity, crystallization tendency and liquidus temperature are affected. Also, ZrO2 from the
submerged entry nozzle is dissolved into the slag. To quantify the effect of each oxide on the
slag properties, different amounts of these oxides are added to selected slags. Whereas TiO2
and Al2O3 resulted in a decrease of crystallization temperature, it increases for ZrO2.
Furthermore, the shapes and sizes of cuspidine crystals are affected and new crystal phases
are formed in dependence on the added oxide and its amount. These investigations permit the
estimation of the maximum amount of the respective oxide still enabling safe operation during
continuous casting.
Additionally, the corrosion of the slag band material is investigated in contact to fluorine
containing fluxes. In literature a cyclic mechanism is proposed, which is not observed here. It
is shown that the slag wets the graphite of the refractory material. Therefore, graphite does not
act as a barrier for infiltration as it is the case for other refractory applications. Furthermore,
the reaction of CaO stabilizing the ZrO2 with the mold slag and the following transformation of
zirconia into smaller crystals is observed contributing to its dissolution and erosion in the slag.
Contrary, for fluorine free slags corrosion is reduced due to its higher basicity.
The research done in the last years contribute to understand of the impact of mold powder
liquefaction and mold slag crystallization on the continuous casting process, but there are still
questions which have not been answered yet. Digital methods applying both classical
simulation concepts and AI methods may be needed, but the necessary parameters are partly
not available now. Many of them cannot be measured on site and experimental procedures
are missing. Thus, new designed experimental set-ups are required, too.