Non-metallic inclusions (NMI) are formed during the steelmaking process and have different origins, such as chemical reactions in the process, slag entrapments or outbreaks from refractory. The negative impact of NMIs on production and product quality leads to the necessity of understanding their behavior in the process to apply countermeasures if required. Tracing methods are used to identify the source of NMIs and track their modification. Tracing also represents an important tool in the future, considering the transformation of the iron- and steel industry to a CO2-neutral and perhaps CO2-free production and the associated impact on steel cleanness and secondary metallurgy treatments. For example, the transformation is connected to an increased use of scrap leading to higher concentrations of trace and tramp elements in the steel whose influence on the steel quality needs to be investigated.
With the state-of-the-art method, an active tracing method, NMIs are marked due to a partial reduction by rare earth elements (REEs), such as La and Ce. Using automated scanning electron microscopy with energy dispersive spectroscopy (SEM/EDS) analysis, REE-modified NMIs are clearly visible and distinguishable from common deoxidation products. One problem of REEs is their high oxygen affinity, leading to reoxidation and thereby affecting the alloying process. Another difficulty occurs in the characterization of REE-containing NMIs. Therefore, the aim of this thesis concerning this technique is to minimize the losses caused by reoxidation on a laboratory scale and develop suitable methods for the characterization of traced NMIs.
The conducted tracing trials on a laboratory scale show that the reoxidation losses of REEs can be reduced by applying specific alloying methods, such as adding them wrapped in Al-foil or as ferroalloy. REE-containing multiphase NMIs are erroneously split into single particles in the characterization with automated SEM/EDS. A self-developed tool improved the evaluation by recombining falsely split NMIs. Consequently, the occurring errors regarding the size, chemical composition, and spatial distribution of NMIs were reduced. Another important issue is the analysis of the morphology of REE-traced NMIs which was done by applying the sequential chemical extraction technique since their stability in acids was not yet known.
For a faster classification of traced and untraced NMIs with a subsequent overview of the tracing success, artificial intelligence (AI) was applied to support automated SEM/EDS analysis. The AI model distinguishing between NMIs and artifacts (binary classification) achieved an accuracy above 95 %. Also, the trained 4-class model showed promising results classifying into traced, untraced, homogeneous and heterogeneous NMIs.
In addition to the active tracing experiments with REEs on a laboratory scale, industrial trials were performed to compare the findings regarding micro-cleanness. In both setups, similar NMI types and modifications occurred. Slag analysis and investigations of the clogging layer showed that REE-traced NMIs have the tendency to separate into the slag and also contribute to the clogging layer formation.
The major disadvantage of applying REEs as tracers is their influence on the properties of NMIs. Consequently, this research deals with the feasibility of alternative tracing techniques to identify the source of NMIs. Therefore, a passive and another active tracing method were applied. The elemental fingerprint, a passive approach, is based on the natural concentration of REEs in materials. The novel active approach is the isotopic spiking, where the isotopic ratio of one possible source is changed by the addition of enriched stable isotopes. These two tracing approaches have great potential to investigate the origin and modification of NMIs during steel production without influencing the properties of NMIs.
By applying the REE-Fingerprint, it is possible to determine the contributing auxiliaries to the clogging layer or NMIs formation by measuring the natural concentration of REEs in the materials. The first results show that sliding gate sand and Al granules affect clogging of Ti-stabilized interstitial free (IF) steels. The implementation of the isotopic spiking technique allows to distinguish whether a NMI was formed or influenced by the isotopically modified source or a source with natural isotopic composition. The first experiments dealt with the investigation of the impact of a 26Mg-enriched slag on NMIs. Measuring the isotopic composition of NMIs with inductively coupled plasma–mass spectrometers confirmed a modification by the 26Mg-enriched slag. While REE-traced NMIs appear brighter in the SEM/EDS analysis, isotopically traced NMIs are optically not distinguishable from untraced ones.