© B. Sammer et al., Published by EDP Sciences, 2026

1 Introduction

Due to the shift toward electric vehicles and the growing emphasis on enhancing the energy efficiency of industrial motors, the demand for non-oriented electrical steel has seen a rise in recent years, making this steel grade the key functional soft magnetic material in the process of electromagnetic energy conversion [1]. Non-oriented electrical steel is known for its excellent magnetic properties, such as high permeability, high saturation magnetization, and low losses. Therefore, it is widely used as core material in a variety of electrical devices and plays an important role in electronics production, power construction and aerospace applications [2,3]. Si is the primary alloying element in non-oriented electrical steel. However, Mn and Al are also commonly added. The most frequently used alloys contain approximately 3.2% Si [4]. Si significantly increases the electrical resistivity of steel, which leads to a reduction in induced eddy currents and ultimately a decrease of core losses [5]. Nevertheless, the magnetic properties are influenced by a variety of different factors including elemental composition, grain size and orientation, precipitations and non-metallic inclusions (NMIs) [6–8]. While small amounts of (Al)N and (Mn)S can have a beneficial effect on grain growth, NMIs generally lead to higher iron losses by hindering domain wall motion [9]. However, the formation of NMIs is tightly connected to the utilized raw materials along with the smelting, casting, rolling, and other processes [10]. In particular, alloying, deoxidation, and desulfurization as well as argon blowing directly influence the number density, size, morphology and composition of NMIs [11]. Therefore, a comprehensive understanding of the evolution of NMIs throughout the different stages of the industrial manufacturing process is important in order to implement countermeasures aiming to control or reduce NMIs.

As many studies show, one option of modifying the inclusion landscape of non-oriented electrical steel is the addition of rare earth elements (REEs). Due to their strong deoxidation and desulfurization ability, they can change the NMIs regarding their number, size and morphology, which leads to a mitigation of the harmful effects of NMIs in the steel and ultimately to an improvement of the properties of the steel product [12–16]. One major issue associated with NMIs is their pinning effect on both grain boundaries and domain walls, which impedes the magnetization process. This effect becomes stronger as the NMIs become smaller [17]. The addition of REEs can increase the NMI size and thereby weaken the pinning of both grain boundaries and domain walls, resulting in easier magnetization, reduced hysteresis loss and overall improved magnetic properties of the non-oriented electrical steel [18–20]. As an example, Wang et al., [21] achieved a reduction in number density by REE treatment from 54458.2 /mm² to 24230.2 /mm² while simultaneously the average NMI size increased from 0.388 μm to 0.427 μm. This resulted in a decrease in iron loss by 0.068 W/kg and an increase in magnetic induction by 0.007 T.

This work provides a comprehensive characterization of the inclusion landscape of non-oriented electrical steel, focusing on its evolvution throughout the various stages of the industrial production process. For this evaluation, samples from different manufacturing steps are examined with respect to NMI number per mm², mean NMI size, and occurring inclusion classes. Additionally, the influence of REEs, particularly Ce and La, on NMIs in non-oriented electrical steel is investigated. Therefore, two laboratory trials are carried out in which Ce and La are individually added to industrially produced non-oriented electrical steel. The findings aim to provide valuable insights into the modification of the inclusion landscape by REEs and their potential impacts on the magnetic properties of non-oriented electrical steel.

2 Materials and methods

2.1 Cleanness evaluation on industrial scale

A non-oriented electrical steel, manufactured at the voestalpine Stahl GmbH steel plant in Linz, Austria, is investigated with respect to its inclusion landscape and how it changes during production. Non-oriented electrical steel is produced via the blast furnace (BF) – basic oxygen furnace (BOF) route. After crude steel production in the BOF, the molten steel is tapped into a ladle. In the ladle furnace (LF), no alloying elements are added; its primary purpose is temperature regulation. All alloying processes are carried out in the Ruhrstahl–Heraeus (RH) degasser. The first Al addition is made after a decarbonization phase, followed by successive additions of ferrosilicon. The final ferrosilicon addition is accompanied by the introduction of other alloying elements as well as a second Al addition. Finally, the RH treatment concludes with flushing,to remove impurities and homogenize the composition and temperature of the molten steel. Casting of the steel is performed using a continuous single-strand slab caster. The next step after casting is hot rolling, followed by batch annealing and subsequent cold rolling. To complete the production process, the cold-rolled non-oriented electrical steel is continuously annealed. Figure 1 depicts the steel production process, including continuous casting.

The most important alloying elements for the production of non-oriented electrical steels are Si, Mn and Al. In addition to the range of these alloying elements, the limits of C and S of the investigated non-oriented electrical steel are listed in Table 1.

To characterize the evolution of NMIs during the industrial production of non-oriented electrical steel, samples were systematically taken at defined stages along the steelmaking and subsequent processing route. Two samples were obtained before and after treatment in the ladle furnace (LF1 and LF2). Additional samples were taken before and after the first Al and ferrosilicon additions in the RH degasser (RH1 and RH2), as well as at the end of the degassing process (RH3). During the casting process, three samples were taken from the tundish (T1–T3). Further samples were acquired from the slab (S), after hot rolling (HR), following batch annealing (BA), and finally from the cold-rolled and continuously annealed finished product (FP). The sampling procedure is schematically shown in Figure 2.

Fig. 1 Steel production process including continuous casting of non-oriented electrical steel.

Fig. 2 Sampling positions throughout the manufacturing process.

2.2 NMI modification on laboratory scale

Ce and La exhibit a strong affinity for O and S, making them highly reactive elements in metallurgical processes. Due to this strong chemical reactivity, REEs tend to partial or even a full reduction of pre-existing NMIs, leading to the formation of new REE-containing NMIs. Through this reduction process, Ce and La can significantly alter the physical and chemical properties of NMIs. Specifically, they can modify NMI morphology and change their shape, size, and distribution within the steel. Additionally, these elements influence the chemical composition of the NMIs, often leading to the formation of new REE-based compounds or phases. This study considers all NMIs containing Ce or La to be modified. The Richardson – Ellingham diagrams shown in Figures 3a and 3b illustrate the stability of Ce and La oxides and sulfides, respectively. Other elements that are relevant to the formation of NMIs in non-oriented electrical steel are also included in these diagrams. For the thermodynamic calculations underlying these diagrams, the software FactSage (version 8.3), with the database FactPS 2023 was used [22].

To investigate the impact and modification potential of REEs on NMIs in non-oriented electrical steel, two laboratory trials were conducted. These experiments were carried out using a resistance-heated Tammann-type furnace (Ruhstrat HRTK 32 Sond., Ruhstrat Power Techology GmbH, Bovenden, Germany), as it is possible to alloy and do sampling under an inert atmosphere over the process duration. A detailed description of the alloying and sampling process, along with a schematic illustration of the Tammann-type furnace, can be found in the work of Dorrer et al., [23], providing additional context for the experimental setup. The selection of REE amounts for these trials is based on prior research. Specifically, 20 ppm of Ce and La were selected for the NMI modification experiments, as suggested by the findings of Wang et al., [24]. To minimize losses during the alloying process, the metallic REEs were wrapped in Al foil. The chosen alloying concept has been proven to be effective technique, according to a previous work performed by Thiele et al., [25]. For both trials, 300 g of hot-rolled electrical steel,with the elemental composition given in Table 1 were used as raw material. The raw material was placed in a zirconia crucible, which is resistant to high temperatures, thermal shock, and chemical reactions, ensuring integrity of the material during the experiment. The crucible was then positioned in the Tammann-type furnace and heated to 1600 °C. After holding at this temperature for five minutes to ensure thermal equilibrium, the first sample was taken, and Ce or La was added, respectively. Sampling was conducted systematically throughout the experiment to monitor the evolution of NMIs. Samples were taken every five min for a total duration of 40 min following the initial addition of the REEs, resulting in nine samples per trial. Each sample was quenched in water. Following a total holding period of 45 min at 1600° C, the remaining melt was cooled to room temperature inside the furnace. Elemental compositions of the raw material (M0) and the remaining melts (RM Ce & RM La) are listed in Table 2. For measuring the concentrations of Ce and La in the remaining melts, an inductively coupled plasma quadruple mass spectrometer (ICP-QMS, Agilent 7500ce; Agilent Technologies, Tokyo, Japan) was used. For each sample, a small piece weighing approximately 500 mg was digested and subsequently analyzed. A detailed description of the ICP-QMS measurement procedure is given in the work of Bandoniene et al., [26]. However, the sample digestion technique was adapted for solid steel samples. The dissolution of the samples was performed using a mixture consisting of 30 mL concentrated HCl, 10 mL concentrated HNO₃, as well as 2 mL HBF₄ with a concentration of 48 wt.%.

For sampling during the laboratory trials, E235 steel tubes with a carbon content of 0.17 wt.% were used. These tubes were immersed in the melt, and a small volume of melt was extracted using a three-valve pipette filler. The observed increase in carbon content within the remaining melts (RM Ce and RM La) can be attributed to dissolution of the steel tubes during the sampling process. Additionally, opening the viewport during sampling operations introduces a small amount of air into the otherwise inert furnace atmosphere. The oxygen present in the introduced air reacts with the melt, leading to a reduction in the Si and Al content.

Fig. 3 Richardson-Ellingham diagrams of a) oxides and b) sulfides.

2.3 Micro-cleanness characterization

Both the samples from industry as well as the laboratory trials were examined regarding their NMI types, number of NMIs per mm², and size expressed as the mean equivalent circular diameter (ECD). The method used for those characterizations of NMIs was the automated scanning electron microscopy with energy-dispersive spectroscopy (SEM/EDS) analysis, carried out using a field-emitting SEM (JEOL 7200 F; JEOL Germany GmbH, Freising, Germany) equipped with a 100 mm² SDD-EDS detector (Oxford Instruments Ultim Max 100; Oxford Instruments GmbH NanoAnalysis, Wiesbaden, Germany). Automated particle analysis was performed using the SEM software AZtec feature (AZtec 6.0, Oxford Instruments GmbH NanoAnalysis, Wiesbaden, Germany).

All samples were measured for a total of eight hours, with a working distance of 10 mm, resolution of 1024 x 768 pixels, 400x magnification, and a process parameter of 5. Additionally, only features larger than 1 μm were considered by the software. However, the parameters for measuring the samples of industrial and laboratory trials differ regarding acceleration voltage, probe current and measuring time per particle. Those parameters are summarized in Table 3.

In the industrial samples, the majority of NMIs are relatively small and contain N. To optimize the analysis of these NMIs, an acceleration voltage of 10 kV was selected, minimizing the excitation volume and thereby reducing the influence of the surrounding steel matrix while simultaneously achieving a higher yield of the N signal. This approach facilitated precise quantification of small N-containing NMIs. For the Ce- or La- containing NMIs observed in laboratory trials, an acceleration voltage of 15 kV was determined to be more effective. The higher voltage improved the excitation of REEs, enabling more accurate characterization. The remaining parameters were adjusted accordingly to ensure an adequate count rate.

During post-processing, the measurement results from the automated SEM/EDS analysis were cleaned up to eliminate artifacts, including grinding or polishing residues and scratches. For the laboratory trials, an additional data correction step was required. In backscattered electron images, REE-bearing phases appear brighter than common NMIs, causing heterogeneous multiphase NMIs containing Ce or La to be erroneously split into smaller, separate NMIs. To address this issue, these NMIs were recombined using a specifically programmed tool to correct the resulting errors in the inclusion population, elemental composition, and size [27]. The recombination method used in this tool is based on the initial approach developed in the work of Meng et al. [28].

3 Results

3.1 NMI-Evolution during industrial production

Figure 4 provides a comprehensive overview of both the number of NMIs per mm² as relative values as well as the proportional difference of the mean ECD, illustrating how these parameters evolve throughout the various stages of the manufacturing process.

The data reveal a significant reduction in the number of NMIs per mm² during the ladle furnace treatment and the RH degassing process. In the samples taken from the tundish, no notable changes in the number of NMIs per mm² are observed. However, the number density in the slab sample shows a distinct reduction. The number of NMIs per mm² remains at this low level throughout the subsequent stages of hot rolling and batch annealing. The lowest number of NMIs per mm² is present in the sample taken after the batch annealing process. In the final sample taken from the finished product, a marginal increase in the number density is observed.

The size of the NMIs, as indicated by the mean ECD, also exhibits notable variations throughout the process. After the first addition of alloying elements (between samples RH1 and RH2), the NMIs increase slightly in size. Similar to the number of NMIs per mm², the mean ECD remains constant in the samples taken from the tundish. The largest NMIs occur in the slab sample, before decreasing in size during the hot rolling process. Batch annealing, on the other hand, causes a further slight increase in the mean ECD. Finally, after the cold rolling and continuous annealing process, a reduction in NMI size is observed.

Depicted in Figure 5 is the relative distribution of the various NMI classes observed throughout the manufacturing process. Samples LF1, LF2 and RH1 contain mainly oxidic NMIs, with only a small number of NMIs additionally bearing S. These three samples are not included in the diagram, since they have a significantly higher number density compared to the rest, and, consequently, a detailed analysis of the other samples would be difficult. This exclusion highlights the pronounced reduction in NMIs during the early stages of processing, particularly during ladle furnace treatment and the early phase of RH degassing. A detailed comparison of the samples RH2 and RH3 shows a notable transition in the dominant NMI class over the alloying steps and during the RH treatment. The primary inclusion class changes from nitrides to multiphase NMIs containing both N and S as non-metallic bonding partners. Furthermore, no more oxide-nitrides (ON) are observed at the end of the degassing process. In the tundish samples, the distribution of NMI classes remains relatively unchanged. However, once the non-oriented electrical steel is cast, a transformation in the inclusion population occurs. Sulfides emerge as the dominant inclusion class, while the prevalence of other inclusion classes – primarily oxide – nitride – sulfides (ONS), nitrides, oxides and particularly nitride – sulfides (NS) – decreases. The inclusion landscape of the samples HR and BA exhibits a similar pattern, with oxide-sulfides (OS) being the most common class, followed by sulfides. In the sample taken from the finished product, OS remains the most prevalent NMI class. However, there is an observed increase in the amounts of nitrides and NS. Additionally, noticeable quantities of ONS are detected.

Within the class of nitridic NMIs, AlN consistently stands out as the most frequently occurring type throughout the entire production process. Similarly, the most common type in the NS class remains unchanged across all stages of production. These NMIs are characterized by a heterogeneous morphology, consisting of small MgS particles attached to larger AlN. In contrast, the classes of sulfides and OS exhibit a transition in their most frequent NMI type after casting. Before casting, (Mg,Mn,Zr)S is the most prevalent type of S-containing NMIs, while the most common O and S-bearing NMIs are (Al,Ca,Mg)OS. However, after casting, these dominant types are changed to (Ca,Mg,Mn)S and (Al,Ca,Mg,Mn)OS, respectively. Figures 6a to 6d provide a detailed visualization of the main types of NMIs observed after casting. These figures illustrate the morphological and compositional diversity of the NMIs found in non-oriented electrical steel. For the other classes of NMIs, no type of NMI is clearly dominating throughout the different manufacturing steps.

Fig. 4 Number of NMIs per mm2 and mean ECD of investigated industrial samples.

Fig. 5 Relative number per mm2 of different NMI classes.

Fig. 6 Main NMI types after casting, a) AlN, b) (Al,Mg)NS, c) (Ca,Mg,Mn)S, d) (Al,Ca,Mg,Mn)OS.

3.2 Modification of NMIs by REEs

To investigate the influence of the REEs Ce and La on the pre-existing NMI population, the raw material, samples taken during the experiment and the remaining melt from the laboratory trials were analyzed using automated SEM/EDS measurements. This comprehensive analysis provides detailed insights into the modification of NMIs by REE additions. Figures 7a and 7b summarize the changes in the relative number of NMIs per mm² over time for both laboratory trials. In these figures, the observed NMIs are categorized into different classes. Additionally, Figures 7c and 7d illustrate the impact of Ce and La on the size of the NMIs, with the mean ECD plotted as a relative value over the duration of the experiment. These figures collectively highlight changes in NMIs during the laboratory trials and the effects of Ce and La on their population and size.

Compared to the raw material (M0), the first two samples taken during the trials exhibit a significantly lower number of NMIs per mm². However, ten minutes after the REE addition, the number of NMIs per mm² significantly increases again. This trend is observed in a similar way for both Ce and La. Samples with La addition generally show a higher number of NMIs per mm² throughout the trial, in contrast to those with Ce. However, the opposite is true for the remaining melts, where La leads to fewer NMIs.

An evaluation of the inclusion classes of the samples taken between 0 and 40 min, shows that mainly oxides, sulfides and OS NMIs are present throughout this holding period. With the addition of Ce, an increase in the proportion of oxides over time is observed.

Comparision of the remaining melts to the raw material shows that both Ce and La are effective in reducing the overall number of NMIs per mm², with sulfides showing a particularly pronounced decrease. In total, the addition of Ce results in a 51.9% reduction in the number of NMIs per mm², while with La, an even greater reduction of 58.0% is achieved. Nonetheless, the classes of the occurring NMIs remain unchanged. The mean ECD of the NMIs also exhibits notable changes over the duration of the trials. Quantitatively, the addition of Ce and La leads to an increase between the raw material and the remaining melts in the mean ECD of 16.7% and 37.5% respectively.

Both laboratory trials achieve an overall modification rate exceeding 80%. Among the various inclusion classes, the most modified are OS and ONS, both of which reach modification rates of over 90% with the addition of either Ce or La. No modifications are observed in the ON class. When comparing the effects of Ce and La on other inclusion classes, subtle differences are observed. While Ce exhibits a slightly higher modification rate for nitrides, La, on the other hand, has a more pronounced effect on oxides, sulfides and NS NMIs, achieving modification rates approximately 20% higher than those observed with Ce. Table 4 provides a summary of the modification success rates for each inclusion class in the laboratory trials, offering a clear comparison of the effects of Ce and La on NMIs.

The laboratory trials with the addition of Ce and La resulted in the formation of various types of heterogeneous as well as homogeneous REE-modified NMIs. Figures 8a to 8d provide a visual representation of common NMI types observed during these trials, offering a closer look at both heterogeneous as well as homogeneous types. The specific NMIs depicted in these figures were selected because they represent highly modified NMI classes and are among the most frequently observed types within these classes.

Fig. 7 Evolution of number per mm2 and mean ECD during the laboratory trials.

Fig. 8 Typical NMIs observed in the laboratory trials, a) heterogenous (Al,La)NS, b) heterogenous (Al,Ca,La)OS, c) homogenous (Al,Ca,Ce)OS and d) heterogenous (Al,Ca,Ce)ONS.

4 Discussion

4.1 Changes in inclusion landscape

The examination of the number density of NMIs per mm² and their ECD, together with the relative distribution of NMI classes at different stages, has yielded several key trends and transformations in the evolution of the inclusion landscape throughout the manufacturing process of non-oriented electrical steel. Moreover, the findings highlight the persistence of specific NMI types as well as the shifts in their predominant types after casting.

The significant reduction in the number of NMIs per mm² during ladle furnace treatment and RH degassing process highlights the importance of these early refining steps. These processes are designed to remove NMIs through mechanisms such as agglomeration and flotation. The absence of notable changes in the number density of the tundish samples indicates stabilization of the inclusion landscape throughout this stage, with minimal further removal or transformation occurring. The distinct reduction in the number of NMIs per mm² observed in the slab sample may be induced by agglomeration of NMIs during the solidification of the steel during the casting process. Subsequent hot rolling and batch annealing maintain the low NMI number density achieved after casting, suggesting that these process steps have no significant influence on the number of NMIs per mm². In the sample of the finished product, an increase in the number density of NMIs is measured. This observed increase (in the number of NMIs per mm²) can be attributed to the fragmentation of existing NMIs occurring due to the mechanical deformation during the cold rolling process.

The mean ECD of NMIs exhibits dynamic behavior throughout the manufacturing process, reflecting the interplay of thermal, chemical, and mechanical factors. The slight growth of NMIs after the first addition of alloying elements (between RH1 and RH2) suggests that alloying may promote the agglomeration or coarsening of NMIs, likely due to chemical changes. The constant NMI size observed in the tundish samples further supports the assumption of a stable inclusion landscape at this stage. The largest NMIs appear in the slab sample, likely due to the segregation and precipitation of elements during the solidification, leading to the formation of larger NMIs. However, a reduction in mean ECD is observed in the subsequent hot rolling process, which is likely due to mechanical deformation and fragmentation. Batch annealing causes a slight increase in NMI size, which may be induced by thermal effects that promote growth or coarsening of NMIs. Finally, the cold rolling and continuous annealing process results in a reduction in NMI size, presumably caused by further mechanical fragmentation.

The distribution of NMI classes throughout the manufacturing process shows significant transformations. While the early stages (LF1, LF2, and RH1) are dominated by oxidic NMIs, nitrides are the main inclusion class present in the sample RH2. This shift may be explained by the fact that the first alloy addition takes place between the sampling of RH1 and RH2. The subsequent change in dominant inclusion class between RH2 and RH3, where nitrides are being replaced by NS NMIs, is likely driven by the further alloying steps and the ongoing RH treatment. The complete disappearance of ON by the end of the degassing suggests that these NMIs are either removed or transformed into other NMI types. After casting, sulfides emerge as the dominant class, which is accompanied by a decrease in oxides, nitrides, NS and ONS. These shifts may be due to changes in the solubility as well as the segregation of elements during solidification. OS being the most common inclusion class in the samples HR, as well as the BA, suggests that the thermal and mechanical conditions during hot rolling and batch annealing promote the formation or stabilization of these NMIs. The reappearance of some inclusion classes in the final sample suggests that cold rolling and continuous annealing processes may create conditions favorable for the formation of these NMIs.

AlN consistently emerging as the most frequently occurring type of nitride throughout the entire production process indicates that AlN is thermodynamically stable under the conditions present during the manufacturing of non-oriented electrical steel. This stability is likely due to the high affinity of Al for N, which promotes the formation of AlN NMIs even in the presence of other competing elements. Similarly, (Al,Mg)NS remains the dominant type of NS NMIs across all production stages, suggesting a high stability of this type of NMI under the different process conditions. The transition from (Mg,Mn,Zr)S to (Ca,Mg,Mn)S and from (Al,Ca,Mg)OS to (Al,Ca,Mg,Mn)OS during the solidification process may be attributed to segregation mechanisms.

The distribution of NMIs in electrical steel throughout the production route reflects a complex interplay of heating rates, annealing atmosphere, elemental composition, precipitation kinetics and mechanical processing. These factors must be carefully controlled to achieve the desired microstructure and magnetic properties in the final product.

4.2 Effects of REEs on NMIs

The laboratory trials conducted in this study provide valuable insight into the effects of REE additions, specifically Ce and La, on the NMI landscape of non-oriented electrical steel by demonstrating their influence on NMI size and number density, as well as their ability to modify different inclusion classes.

The initial decrease in the number of NMIs per mm² observed after REE addition was followed by a subsequent increase after 10 minutes for both Ce and La. This higher number density suggests a dynamic interaction between REEs and the melt, likely influenced by the kinetics of NMI formation and modification. Compared to the raw material and remaining melts, samples taken during the process show differences in present NMI classes along with an overall smaller mean ECD. This can be attributed to rapid cooling during sampling. Rapid solidification limits the time available for NMI growth, agglomeration and chemical reactions, resulting in finer NMIs more complex classes.

In the remaining melt, the addition of Ce and La led to a reduction of over 50% in the number of NMIs per mm² compared to the raw material, demonstrating the effectiveness of both REEs in refining the melt by reducing the overall inclusion population. Both Ce and La led to an increase in mean ECD, indicating that REE additions not only reduce the number of NMIs but also promote growth of larger NMIs, possibly through coalescence or agglomeration mechanisms. However, the addition of La yielded a stronger reduction in the number of NMIs per mm² as well as a more pronounced reduction in size compared to Ce, resulting of the La content being higher than that of Ce.

The high overall modification rates achieved in both trials highlight the efficiency of REE additions in modifying NMIs. Among the various inclusion classes, OS and ONS exhibited the highest modification rates, exceeding 90% for both Ce and La. These high rates indicate that these inclusion types are particularly susceptible to REE-induced modification. In contrast, no modifications were observed in the ON class, suggesting that this inclusion type is resistant to REE modification under the experimental conditions. The higher sulfur affinity of La explains why La addition resulted in modification rates approximately 20% higher than those observed with Ce in sulfides and NS. On the other hand, Ce has a higher affinity to N, which explains the slightly higher modification rate of nitrides by Ce. While the oxygen affinities of Ce and La are comparable, the higher modification rate of oxides by La may be a result of the La content being almost double that of Ce.

The modification trials on laboratory-scale demonstrated the ability of Ce and La to significantly reduce the number of NMIs per mm², increase their mean ECD, and modify their composition, improving the cleanness of the final product. These factors suggest that the magnetic properties of non-oriented electrical steel may be improved by adding REE, since larger NMIs are beneficial. However, the exact mechanisms of how the properties of the new NMIs are influenced by REEs still need to be further investigated.

5 Conclusion

Since steel cleanness has a crucial impact on the magnetic properties of non-oriented electrical steel, the findings of this work provide valuable and detailed insight into the evolution of the NMI landscape throughout the industrial manufacturing process. Furthermore, the results of the laboratory trials conducted in this study show that the addition of REEs, specifically Ce and La, proves to be an effective strategy for modifying NMIs, thereby contributing to a deeper understanding of REE-induced NMI modification. The main findings can be summarized as follows.

• During the industrial manufacturing process of non-oriented electrical steel, the number of NMIs per mm² notably decreased, whereas the mean ECD rose over the process. The main reduction in number density was observed during LF and RH treatment, followed by another noticeable reduction during solidification. Also, the size of the NMIs significantly increased during solidification.

• The most common NMI class changed during the process from oxides (LF1, LF2, and RH1) to nitride (RH2), NS (RH3 – T3), sulfides (S) and finally OS (HR, A, and FP).

• AlN remained the most frequently occurring nitride throughout the entire manufacturing process. Furthermore, the most prevalent type of NS NMI, (Al,Mg)NS, also remained unchanged during production. Sulfides and OS, on the other hand, showed a shift in their dominant NMI types after casting. While (Mg,Mn,Zr)S and (Al,Ca,Mg)OS are the most frequent before casting, they are subsequently replaced by (Ca,Mg,Mn)S and (Al,Ca,Mg,Mn)OS.

• In the laboratory trials, Ce and La each led to total NMI modification rates of over 80%. Over 90% of OS and ONS were modified due to the addition of either Ce or La. The ON class showed no modifications in either case. La demonstrates a stronger effect on oxides, sulfides, and NS inclusions, while Ce exhibits a slightly higher modification rate for nitrides.

• The same NMI classes occur both in the raw material and the remaining melts of the modification trials. However, the latter contains significantly fewer sulfides.

• The addition of REEs led to a reduction in NMIs’ number per mm² and to an increase in NMI size. Based on previous results reported in the literature, this reduction in number and simultaneous growth in size might improve the magnetic properties of non-oriented electrical steel.

To further understand the influence of REEs on the inclusion landscape of non-oriented electrical steel, additional laboratory trials testing different Ce and La contents as well as different cooling conditions are required. For a validation of the laboratorial findings, an industrial trial applying REEs as modifier for NMIs is required. Industrial trials would also enable investigations regarding the influence of REEs on steel cleanness as well as the magnetic properties of non-oriented electrical steel. Additionally, the effect of adding REEs on the mechanical properties of non-oriented electrical steel should be considered in future studies.

Acknowledgments

The authors would like to gratefully acknowledge the financial support provided by the Austrian Federal Ministry of Economy, Energy and Tourism, the National Foundation for Research, Technology and Development, the Christian Doppler Research Association and voestalpine Stahl GmbH. Expert support by Dr.techn. Christoph Walkner (Chair of General and Analytical Chemistry, Technical University of Leoben) for inductively coupled plasma mass spectrometry is greatly appreciated.

Funding

The financial support by the Austrian Federal Ministry of Economy, Energy and Tourism, the National Foundation for Research, Technology and Development, the Christian Doppler Research Association and voestalpine Stahl GmbH is gratefully acknowledged.

Conflicts of interest

The authors declare no conflict of interest regarding this article.

Data availability statement

The raw data of the laboratory trials presented in this article will be made available by the corresponding author on request.

Author contribution statement

Conceptualization, B.S.; Methodology, B.S., K.T., R.R., S.I., and H.K.; Software, B.S.; Validation, B.S. and K.T.; Formal Analysis, B.S..; Investigation, B.S., K.T., R.R., S.I., and H.K.; Resources, B.S., R.R., S.I., and H.K.; Data Curation, B.S.; Writing – Original Draft Preparation, B.S.; Writing – Review & Editing, K.T., R.R., S.I., H.K and S.K.M.; Visualization, B.S.; Supervision, S.K.M.; Project Administration, K.T., R.R., S.I., H.K and S.K.M; Funding Acquisition, S.K.M.

References

All Tables

All Figures

	Fig. 1 Steel production process including continuous casting of non-oriented electrical steel.
In the text

	Fig. 2 Sampling positions throughout the manufacturing process.
In the text

	Fig. 3 Richardson-Ellingham diagrams of a) oxides and b) sulfides.
In the text

	Fig. 4 Number of NMIs per mm2 and mean ECD of investigated industrial samples.
In the text

	Fig. 5 Relative number per mm2 of different NMI classes.
In the text

	Fig. 6 Main NMI types after casting, a) AlN, b) (Al,Mg)NS, c) (Ca,Mg,Mn)S, d) (Al,Ca,Mg,Mn)OS.
In the text

	Fig. 7 Evolution of number per mm2 and mean ECD during the laboratory trials.
In the text

	Fig. 8 Typical NMIs observed in the laboratory trials, a) heterogenous (Al,La)NS, b) heterogenous (Al,Ca,La)OS, c) homogenous (Al,Ca,Ce)OS and d) heterogenous (Al,Ca,Ce)ONS.
In the text