Impact of Solidification on Inclusion Morphology in ESR and PESR Remelted Martensitic Stainless Steel Ingots

This study focuses on the impact of solidification on the inclusion morphologies in different sizes of production-scale electro-slag remelting (ESR) and electro-slag remelting under a protected pressure-controlled atmosphere, (PESR), ingots, in a common martensitic stainless steel grade. The investigation has been carried out to increase the knowledge of the solidification and change in inclusion morphologies during ESR and PESR remelting. In order to optimize process routes for different steel grades, it is important to define the advantages of different processes. A comparison is made between an electrode, ESR, and PESR ingots with different production-scale ingot sizes, from 400 mm square to 1050 mm in diameter. The electrode and two of the smallest ingots are from the same electrode charge. The samples are taken from both the electrode, ingots, and rolled/forged material. The solidification structure, dendrite arm spacing, chemical analyzes, and inclusion number on ingots and/or forged/rolled material are studied. The results show that the larger the ingot and the further towards the center of the ingot, the larger inclusions are found. As long as an ingot solidifies with a columnar dendritic structure (DS), the increase in inclusion number and size with ingot diameter is approximately linear. However, at the ingot size (1050 mm in diameter in this study) when the center of the ingot converts to solidification in the equiaxial mode (EQ), the increase in number and size of the inclusions is much higher. The transition between a dendritic and an equiaxial solidification in the center of the ingots in this steel grade takes place in the region between the ingot diameters of 800 and 1050 mm.


Introduction
Uddeholms AB specializes in producing tool steels. The steels are manufactured in the Hagfors steel mill. The production is approximately 100,000 tons per year in the steel melting shop. The steel grades are mainly hot work steels, cold work steels, and plastic mould steels. About 55% of the steel grades are electro-slag remelted (ESR) or electro-slag remelted under a protected pressure-controlled atmosphere (PESR). The reason to choose ESR or PESR remelting processes could either be to reach higher mechanical properties (more homogenous material) or a better polishability of the steel.
In order to optimize the properties, (who are directly related to the microstructure), of a given steel grade, it is important to understand the evolution of the microstructure during the solidification and subsequent cooling to room temperature [1]. The main variables controlling the microstructure evolution include the interface velocity, thermal gradient, alloy composition, and nucleation potential [1]. These variables will change considerably during the columnar/equiaxial transition (CET). Therefore, it is important to know during which circumstances during remelting of an ingot, this transition will take place.
In this investigation, studies on the solidification structure, dendrite arm spacing, and inclusion number per area are made on an electrode (300 mm square) and ESR and PESR ingots (400 mm square-1050 mm in diameter) are performed. The steel grade is a commonly used martensitic stainless steel.
Previous studies on non-metallic inclusions (NMI) and cleanliness in ESR or PESR remelted ingots are presented in Table 1  . It can be seen that the majority of studies by other authors are executed in laboratory or pilot scales (from 0.8 up to 50 kg experimental trials) with focus on steel grades other than martensitic stainless tool steels. The inclusions found in laboratory-scale trials are usually in the size range of 1-5 µm, but more often ≤2 µm. However, in industrial-size ingots, significantly larger inclusions can be found, especially if larger sample areas are analyzed. Therefore, for reliable evaluations of the characteristics of NMI (especially larger size inclusions) in production-scale trials, only investigations of NMI in laboratory and pilot experiments are not sufficient. In this study of inclusions in production-scale ingots, the analyzed area is about 6000-6500 mm 2 per sampling position (about 2000 mm 2 for the ingots). On the forged or rolled areas, approximately 1-3 inclusions between 20 and 30 µm per sample are detected. The sample areas in laboratory-scale trials are usually about 100 mm 2 . According to Franceschini et al. [28], an analyzed surface of 5000 mm 2 is suitable for evaluating the cleanliness for remelting processed samples, to compare with air samples where 1000 mm 2 is sufficient. The earlier and most generally accepted theories assumes that the inclusions in the electrode will be rejected to the surface of the electrode tip to be incorporated into the liquid process slag [9,[21][22][23][24][25]29]. The steel melt will then seek equilibrium with the process slag, so that new, different oxygen levels in the liquid steel pool will be obtained. During solidification, new inclusions are re-precipitated as secondary inclusions in the ESR or PESR ingot.
Recently, Sjöqvist Persson et al. [26,27] stated that the inclusions in a remelted ingot can be primary, semi-secondary, or secondary inclusions. The primary inclusions are suggested to have been trapped in a fallen steel drop. Furthermore, the semi-secondary have a core of an un-melted Mg-spinel (Al-Mg oxide) that has survived from the electrode together with an outer layer with a composition corresponding to the ESR process slag [25][26][27]. Specifically, this could, e.g., be MgAl 2 O 4 inclusions (with a low content of MgO) and Al 2 O 3 inclusions formed in the liquid metal pool as a result of the reactions between alloying elements and dissolved oxygen.
The critical point for growing and clustering of the newly nucleated/formed secondary inclusion, as well as the growth and clustering of the primary and the semi-secondary inclusions, is the local solidification time (LST) [30,31]. This, in turn, depends on the melting rate, immersion depth of the electrode tip in the slag bath, the electrode/ingot ratio, and time for heat transfer from the center to the surface of the ingot. In reality, it means that the larger the ESR ingot size the longer the LST. At a critical LST value, the solidification will turn from a columnar dendritic (CD) to an equiaxial structure (EQ), which is called the columnar/equiaxial transition (CET). The parameters, which cause the transition, are the alloy segregation characteristics, freezing rate, undercooling, and the presence or absence of nucleating sites for the equiaxial crystals [32]. If the ingot structure changes from CD to EQ, the consequent competing growth and possible movement of the EQ crystals leads to a random formation of pockets of segregated inter-dendritic liquid. As a result, the homogenization temperature normally used for the steel grade may lie above the critical temperature where the process starts to produce porosities due to the lower melting point in the last solidified material. The result is that the ingot cannot be successfully homogenized [32]. Due to the same reason, the segregated liquid also leads to microstructure effects in many alloys that produce primary precipitates (NMI and intermetallic inclusions (IMI)), which are not changed by subsequent mechanical or thermal processing [32]. At this point (CET), the steel will be more segregated and the inclusions will have sufficient time to grow and agglomerate.
A calculation shows that an alloy with a LST value >3500 s is likely to solidify in the equiaxial mode [33]. In Figure 1, results from laboratory trials by Mitchell are displayed [33]. They indicate that the critical point could be reached at an ingot size of 1050 mm in diameter. As long as the ingot solidifies in a columnar-dendritic manner, the angle between the columnar dendrites direction and the bottom of the liquid steel pool is 90 • at the center position of the ingot [30,31,34].
An investigation of the solidification of a 750 mm in diameter ESR/PESR remelted ingot at different melt rates was made at Breitenfeldt Edelstahl AG in Austria [35]. The results showed that the higher the melt rate, the deeper liquid the steel pool. However, even at the highest used melt rate (standard melt rate +100 kg/h), the solidification was columnar-dendritic from the surface to center. At the same time, the number and size of the inclusion increased with a higher melt rate/deeper pool.
Borodin et al. [36] investigated the solidification of 400 mm in diameter ESR ingots remelted using different melt rates and on a 550 mm in diameter ingot remelted using one melt rate. They found disoriented dendrites (equiaxial solidification) in the 400 mm in diameter ingot, but not in the 500 mm in diameter ingot. However, the used melt rates were much higher (510 and 670 kg/h for the 400 mm in diameter ingot and 700 kg/h for the 500 mm in diameter ingot), than the "rule of thumb" of an estimated melt rate (kg/h) of ±80% of the ingot diameter (mm). The electrodes used were described to have a very convex shape (probably mostly due to the high melt rate and also depending on the use of a low electrode/ingot ratio), which not is the case in present high-quality ESR remelting processes.
According to Xuan et al. [37], the most important factors for an inclusion to be captured in a process slag are its size and flotation speed in the steel. According to this proposal, in the case of a vacuum-treated steel shop charge, with different slag compositions and movements of the steel, inclusions larger than approximately 20 µm in diameter should be captured in the process slag. The critical size for flotation of an inclusion from the molten steel pool in the ESR/PESR processes to the process slag, along with the chemical composition of the slag and the size of the inclusion will also depend on the movements in the liquid steel pool. The movements in the liquid steel pool depend on the size of the ingot and the depth of the steel pool. The depth of the steel pool is dependent of the melt rate (kg/h), the immersion depth of the electrode in the slag, and the electrode/ingot diameter relation. The terminal velocity of a Mg spinel inclusion 50 µm in diameter rising in a liquid steel is approximately 5 × 10 −4 m/s given the computed ingot pool pattern [38]. This implies that any inclusion removal is determined by a balance between the probabilities of being exposed to the slag/ingot interface or being trapped in the solidifying metal as dictated by the bulk metal flow of the liquid ingot pool. Inclusions (depending on their size and density) will be trapped in the flow pattern, many never reaching the steel/slag interface. According to this model, the contribution of inclusion refining though flotation is much smaller than what has previously been estimated. Further studies are needed to better understand this mechanism during ESR/PESR processes. However, it can be postulated, due to the shorter LST, i.e., shorter time for the steel pool to be liquid, that the smaller the ingot size, the less the overall possibility for flotation of inclusions. In contrast, a previous study [21,22] showed that the larger the ESR or PESR ingot, the more and larger inclusions could be found.
According to Fredriksson et al. [34], the surface of an as cast ingot solidifies in a different manner as compared to an ESR or PESR ingot. In an as-cast ingot, the melt close to the mould cools down rapidly to a temperature beneath the critical temperature for nucleation of crystals. This results in many nuclei, which form a surface zone consisting of equiaxial crystals. The dendrites grow <90 • up against the top center of the ingot, controlled by the temperature gradient. In ESR and PESR remelted ingots, the surface zone of equiaxial crystals does not usually exist. Instead, the surface zone consists of long radially oriented crystals. The long radial format of the dendrites is due to that the dendrites grow with the solidification front along the surface. Therefore, it can be stated that it in ESR and PESR processes, no strongly undercooled surface zones exist that result in the formation of equiaxial crystals. During solidification, however, due to the cooling effect from the mould, there is a finer dendrite structure in the surface than in the center of the ingot. The long crystals along the surface give the ESR and PESR ingots a smooth and good surface.
Electrode inclusions are modified by several mechanisms, i.e., dissolution in the steel as the electrode melts, dissolution in the slag at the electrode/slag interface, dissolution in the slag during the droplet fall, and flotation in the ingot pool followed by dissolution in the slag at the ingot/slag interface. This study does not define the role of each of these individual processes but presents the overall result of them as it pertains to the final inclusion content.
The current study focuses on the solidification process and the oxide inclusions in a production-scale electrode, relative to ESR and PESR ingots of different sizes in a stainless martensitic steel. The aim is to relate the solidification and ingot size to the number and type of inclusions found. In order to optimize process routes for different steel grades, it is important to define the inclusion characteristics in ESR and PESR materials.

Materials and Methods
The material studied was from one ingot-cast consumable electrode 300 × 300 mm 2 (denoted as the CE-300 sample), one ESR remelted ingot 400 × 400 mm 2 (denoted as the ESR-400 sample), and three PESR remelted ingots 500, 800, and 1500 mm in diameter (denoted below as the PESR-500, -800, and -1050 samples, respectively). The electrode, the ESR-400, and the PESR-500 were cast from the same initial steel charge. The difference between the ESR and PESR process is that the ESR here represents a multiple-electrode remelting process (involving electrode changes), which is performed in a moving mould, in an open furnace under an air atmosphere, using a continuous aluminum deoxidation. In contrast, the PESR process is a single-electrode remelting process using a static mould and an inert pressure-controlled atmosphere.
A typical composition of martensitic stainless steel used in this study is as follows: C 0.38%, Si 0.9%, Mn 0.45%, Cr 13.6%, and V 0.28%. In the ESR and PESR trials, a common process slag was used containing about one-third each of CaO, CaF 2 , and Al 2 O 3 , and including ≈3% MgO and ≈1% SiO 2 .
The steel samples were taken from horizontal slice/slices of the electrode and ingots as follows: top (T), middle (M), and bottom (B) for the electrode, two samples taken at positions in between top and bottom (1,2) for the ESR-400, and one middle sample (M) for the PESR ingots, respectively, see Figure 2a,b. The cross-sections have been cut in order to study the solidification structures and dendrite arm spacings. The samples from the electrode and ingots cover the full distance/radius from the corner/surface to the center. A schematic presentation of the positions of the macro samples for both the square electrode, ESR ingot, and PESR ingots can be seen in Figure 2d,e. Each sample with a dimension of ≈150 × 100 × 15 mm 3 was ground and polished using a 3 µm diamond finish and etched in a solution consisting of either 50% HCl in water at 50 • C or using Björk's etchant followed by rinsing in ammonia. Thereafter, the secondary arm spacing (SDAS), at different positions on the cross-sections was measured, as described below. Longitudinal sections were cut out from the cross-sections from the 500, 800, and 1050 mm in diameter PESR ingots. They have been examined with respect to the angle between the dendrite arms and axial planes, as seen in Figure 3. Chemical compositions of the samples of the cross-sections were made using LECO CS600 (LECO Corporation, St. Joseph, MI, USA) for the element carbon (C) and sulfur (S) and X-ray Thermo 9800 for the other elements.
Both the SDAS at different positions and the angles between the dendrite arms and the axial plane were measured using a light optical microscope (LOM, Leica Microsystems, Wetzlar, Germany), Leica MZ6 equipped with Leica Qwin Runner V3.5.1 software. The investigations on the cross and longitudinal sections were made at former Exova AB, Karlskoga, Sweden.
Inclusion measurement were made on the cross-sections on the PESR-800 and PESR-1050 ingots. The field of view per sample was about 2000 mm 2 . Inclusion measurements were been made on bottom, middle, and top samples on forged/rolled materials. The field of view per sample was about 6000 mm 2 . Both the samples from the ingots and from the forged/rolled materials were studied using a scanning electron microscopy (FEI Quanta 600 Mark II, Thermo Fisher Scientific, Waltham, MA, USA. The number, size and, chemical composition of the inclusions larger than 8 µm were determined using the software Inca features from Oxford Instruments (ETAS Group, Stuttgart, Germany). Afterwards, the inclusions were divided into four size classes, namely, (i) 8-11.2 µm, (ii) 11.2-22.4 µm, (iii) 22.4-44.8 µm, and (iv) larger than 44.8 µm.

Macrostructures of the Electrode and ESR/PESR Ingots
The macrostructure of the electrodes cross-sections at radial bottom, middle, and top positions is analyzed from the cross-section samples. The electrode was cast with the "big end down." The macrostructure of the ESR ingot cross-section at two axial positions and the PESR ingots at the axial middle section was studied. No clear surface columnar zone could be observed in the radial ESR or PESR samples as shown in Figure 4a-c. The columnar zone in the electrode measured from the corner was found to be approximately 60 mm for the cross-sections from the radial bottom, see   A coarser dendritic pattern can be seen starting from 80 to 90 mm from the surface on the PESR-500 and 100 to 110 mm from the surface on the PESR-1050, see Figure 4b,c. Furthermore, white etching concentric bands are weakly visible at a distance approximately 15-30 mm below the surface for the samples PESR-500, -800, and -1050, see Figure 4b,c.

Dendrite Arm Spacing of the Electrode and ESR/PESR Ingots
The secondary dendrite arm spacing (SDAS) was measured on the cross-sections of the electrode and the ESR and PESR ingots. The measurement positions are shown in Table 2, the measurement principle is shown in Figure 5, which shows the primary arms, and their connecting secondary arms are marked in red. The measurements were made in-between the red secondary arms. Two to four measurements on each position were used to determine an average SDAS value per position. Table 2. Secondary dendrite arm spacing (SDAS, ‫ג‬ 2 ) per distance from the surface for the electro de 300 × 300 mm 2 (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 (ESR-400 1,2), and the pressure electro-slag remelted ingots with 500-1050 mm diameter (PESR-500, PESR-800, and PESR-1050).  Difficulties in resolving the fine surface structure (except for the PESR-800) resulted in measurements of secondary arm spacing (SDAS) made at position T/6 (≈quarter radius, near corner), instead of at the surface, as seen in Table 2 and Figure 6, in which the SDAS versus the axial positions on the electrode and the ingots are displayed. The curve for the PESR-1050 deviates from the other curves. The data and statistics for the SDAS measurements are shown in Table 3. The SDAS value per ingot and measured distance from the surface (mm) are displayed in Figure 7. It can be seen that the relation is approximately linear for the smaller ingots but more polygonal for the PESR-1050. The data behind Figure 7 is shown in Tables 2 and 3. Figure 6. Secondary dendrite arm spacing, (SDAS, ‫ג‬ 2 ), (µm), for electrode 300 × 300 mm 2 (C-300), electro-slag remelted 400 × 400 mm 2 (ESR-400), and electro-slag remelted under pressure-controlled atmosphere pressure electro-slag remelted ingots 500-1050 mm in diameter (PESR-500, -800, and-1050). Cross-section positions: T/6-quarter radius, T/4-half radius, and T/2-center. Table 3. Data and statistics for the secondary arm spacing, (SDAS, ‫ג‬ 2 ), measurements for electrode 300 × 300 mm 2 (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 (ESR-400 1,2), and the pressure electro-slag remelted ingots 500-1050 mm in diameter (PESR-500, PESR-800, and PESR-1050).

Dendrite Arm Angles towards the Cross-Section of the Ingot
Longitudinal sections were cut from the cross-sections of three PESR ingots. The samples have been examined with respect to the angle between the dendrite arms and axial plane. The angles between dendrite arms and axial plane are shown in Table 4. The angle between the dendrite arm and the axial plane for the PESR-500 and PESR-800 is exactly 90 • in position T/2, at the center of the ingot. For the PESR-1050, the angle could not be measured at T/2, as the macrostructure had no distinct direction at this position. According to theory [30,31,34], the angle of the dendrites towards the bottom of the liquid steel pool is 90 • . This is valid as long as the solidification results in a columnar-dendritic structure. However, if the solidification results in an equiaxial structure, no distinct direction of the dendrites exists. Table 4. Data and statistics for the angle between the dendrite arms and the axial plane. Electrode 300 × 300 mm 2 (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 (ESR-400 1,2), and the pressure electro-slag remelted ingots 500-1050 mm in diameter (PESR-500, PESR-800, and PESR-1050). Pictures of the measured angles for the PESR-800 and PESR-1050 are shown in Figure 8a,b. It can clearly be seen that the angle in the center is 90 • for PESR-800 and that no angle could be measured for the PESR-1050.

Chemical Compositions over the Cross-Sections
The deviations between the amount of carbon (C) in the electrode and ingots and their electrode charge are displayed in Figure 9a,b. The radial segregation of the element C between the top and bottom sample of the electrode, as well as the larger axial segregation in the larger relative to the smaller ingots, can be seen clearly. The axial segregation of carbon in the ESR and PESR ingots is larger as the larger ingot diameter increases. The amount of sulfur (S) in the remelted ingots is approximately 25% of the amount in their corresponding electrode charges. The amounts of manganese (Mn), chromium (Cr), and vanadium, (V) are relatively stable compared to the amounts in their corresponding electrode charges. However, the elements Cr and Mn can have a slight decrease in the axial center of the larger ingots. Silicon is 0-5% lower in the remelted ingots compared with the corresponding electrode charges. This decrease is due to the reaction with the ESR or PESR process slag.

Number of Oxide Inclusions per Size, Position, and Ingot Type
The number of oxide inclusions per size class and ingot type from the electrode and the rolled/forged and heat-treated material (after approximately 75% reduction) at different axial positions is displayed in Figure 10. The samples are taken at position T/2, at the center of the ingot. The ingots used in this investigation are the same as in the solidification structure samples, supplemented with results from a 650 mm in diameter ingot. The ESR-400 mm and the PESR-500Ø ingots are from the same electrode charge from the steel shop. Normally, the number of inclusions increase almost linear with the ingot size, specifically when a larger statistical basis is studied. No inclusions > 44.8 µm were detected in the processed material. The number of oxide inclusions per size class and ingot position measured in percentage of the largest number of inclusions per size class, measured over the cross-section for the unprocessed ingot types PESR-800 and PESR-1050, is displayed in Figure 11. It can be seen that the numbers of inclusions increase with distance from the surface. In laboratory and/or pilot scale, this phenomenon may be neglected. No inclusions > 44.8 µm were detected in the ingots. The difference between ESR and PESR remelted ingots are probably mostly due to the continuous Al de-oxidation during ESR remelting in air [27].

Solidification
The solidification of an ingot either starts with a surface zone with small equiaxial dendrites or long radially orientated crystals [34]. At a certain distance from the mould when the cooling is smaller, the columnar dendritic structure (CD) appears. The reason for that no clear visible columnar zone was found on the ESR and PESR samples is that remelted ingots solidify with a dendrite angle of 90 • towards the bottom of the liquid pool unlike the conventional cast electrode that solidifies with a dendrite angle slightly above 90 • towards the top center. The larger equiaxial structure in the center were only found in the CE-300 and the PESR-1500. This is due to their higher Longest Solidification Time (LST).
The relationship between the dendrite arm distance and the growth rate can be described as follows

Chemical Analyzes over the Cross-Sections
The axial segregations of C correspond to both the me dendrites orientations towards the bottom of the liquid pool the larger the SDAS value and the larger the orientation ang segregation and the size of the SDAS and orientation angle Solidification Time (LST) which increases towards the cente tial element and will, therefore, segregate more than the mor Mo. The decrease in the elements S and Si is due to reactions

Number of Oxide Inclusions Per Size, Position, and Ingot Ty
In a previous investigation, Sjöqvist Persson et al. [21] inclusion morphologies between different ESR/PESR ingot ference between the electrode heats. The tool steel grade cont where Figure 14. Growth rate Ʋgrowth (m/s) (calculated on primary dendrite arm spacings) versus the temperature gradient G (°C/m) around the solid/liquid tip. The black line is plotted data from Kurz solidification diagram [40]. The area above the line solidifies in an equiaxial manner and the area below solidifies in a columnar-dendritic manner [40]. Data from Uddeholms martensitic stainless steel are included, the electrode 300 × 300 mm 2 , (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 , (ESR-400), and the pressure electro-slag remelted ingots 500-1050 mm in diameter, (PESR-500, PESR-800, and PESR-1050). All data are from the center position of the ingots. The electrode and the PESR-1050, which solidify equiaxial dendritic in the center, are marked (•), because the equations used are not valid for the equiaxial solidification structure.

Chemical Analyzes over the Cross-Sections
The axial segregations of C correspond to both the measured SDAS values and th dendrites orientations towards the bottom of the liquid pool. The more the C segregation the larger the SDAS value and the larger the orientation angle. The reason is that both th segregation and the size of the SDAS and orientation angle correspond with the Longes Solidification Time (LST) which increases towards the center of the ingot. C is an interst tial element and will, therefore, segregate more than the more stable elements, like Cr an Mo. The decrease in the elements S and Si is due to reactions with the process slag.

Number of Oxide Inclusions Per Size, Position, and Ingot Type
In a previous investigation, Sjöqvist Persson et al. [21] found that the difference i inclusion morphologies between different ESR/PESR ingot sizes was larger than the di ference between the electrode heats. The tool steel grade contains a relatively high amoun growth is the growth rate, ň den is the distance between the secondary or primary dendrite arms, and C is a constant [30]. It is established that the constant is 10 −6 cm 3 /s for low carbon iron base alloys and that it should (information given to the author by Professor Emeritus Hasse Fredriksson) be about the same for this steel grade, as seen in Figure 12 and Table 5. It can be established that the growth rates for the ingots that solidify entirely as columnar-dendritic structure correlates best with a linear equation, as seen in Equation (1). However, the PESR-1050 mm ingot, which has an equiaxial solidification structure in the center, correlates, according to the results in this investigation, more with a polygonal equation. The reason for that is that Equation (1) only is valid for the columnar dendritic zone and that it, therefore, should not be used for the equiaxial zone.  Table 5. Distance from ingot surface (mm) and growth rate, Figure 14. Growth rate Ʋgrowth (m/s) (calculated on primary de (°C/m) around the solid/liquid tip. The black line is plotted data line solidifies in an equiaxial manner and the area below solidi deholms martensitic stainless steel are included, the electrode 3 400 × 400 mm 2 , (ESR-400), and the pressure electro-slag remelte and PESR-1050). All data are from the center position of the ingo axial dendritic in the center, are marked (•), because the equatio ture.

Chemical Analyzes over the Cro
The axial segregations of C dendrites orientations towards th the larger the SDAS value and th segregation and the size of the S Solidification Time (LST) which i tial element and will, therefore, s Mo. The decrease in the elements

Ingot/Growth Rate
Metals 2021, 11, x FOR PEER REVIEW Table 5. Distance from ingot surface (mm) and growth rate, ʋgrowth (m/s), per electrode 300 × mm 2 (CE-300), electro-slag remelted ingot (ESR-400), and pressure electro-slag remelted ing 500-1050 mm in diameter (PESR-500, PESR-800, and PESR-1050. According to Imagumbai [39], the relation between the growth rate of the pr dendrite tip (R) and the temperature gradient around the solid/liquid tip (G) correl the primary dendrite spacing (ℷ 1), which can be expressed using the following re ship: Imagumbai [39] estimated the secondary dendritic arm spacing at a complete lidified zone (ℷ ʃ 2) to be approximately half of the primary dendritic arm spacing unidirectionally solidified C-Mn steel at steady state.

Ingot/Growth
In Table 6 and Figure 13a, the temperature gradient is calculated by combining tions (2)  The figure shows that the PESR-1050 deviate from a linear relationship, which to the equiaxial solidification, structure in the center of the ingot. The growth rate (m plotted as a function of the temperature gradient (°C/m), and its correlation is sho Figure 13b (the result for position T2 for the PESR-1050 is excluded). The primary growth rate needed for the CET transition is less than 4 × 10 −7 m/s. In order to und the transition, the temperature gradient needs to be lower than approximately 10  According to Imagumbai [39], the relation between the growth rate of the primary dendrite tip (R) and the temperature gradient around the solid/liquid tip (G) correlates to the primary dendrite spacing ‫ג(‬ 1 ), which can be expressed using the following relationship: Imagumbai [39] estimated the secondary dendritic arm spacing at a completely solidified zone ‫ג(‬ S 2 ) to be approximately half of the primary dendritic arm spacing in an unidirectionally solidified C-Mn steel at steady state.
In Table 6 and Figure 13a, the temperature gradient is calculated by combining Equations (2) and (3) to give Equation (4).
The figure shows that the PESR-1050 deviate from a linear relationship, which is due to the equiaxial solidification, structure in the center of the ingot. The growth rate (m/s) is plotted as a function of the temperature gradient ( • C/m), and its correlation is shown in Figure 13b (the result for position T2 for the PESR-1050 is excluded). The primary arms growth rate needed for the CET transition is less than 4 × 10 −7 m/s. In order to undertake the transition, the temperature gradient needs to be lower than approximately 10 3 • C/m. Table 6. Temperature gradient ( • C/mm) and distance from surface (mm) for electrode 300 × 300 mm 2 (CE-300), electro-slag remelted ingot 400 × 400 mm 2 (ESR-400), and pressure electroslag remelted ingots 500-1050 mm in diameter (PESR-500, PESR-800, and PESR-1050).

Chemical Analyzes over the Cross-Sections
The axial segregations of C correspond to both the measured SDAS values and the dendrites orientations towards the bottom of the liquid pool. The more the C segregation, the larger the SDAS value and the larger the orientation angle. The reason is that both the growth (m/s), towards temperature gradient, G ( • C/m). Figure 14 displays data from the electrode and ingots (center position) in this investigation together with plotted data from Kurz solidification diagram [40]. The figure shows the growth rate  [40]. The area above t line solidifies in an equiaxial manner and the area below solidifies in a columnar-dendritic manner [40]. Data from U deholms martensitic stainless steel are included, the electrode 300 × 300 mm 2 , (CE-300), the electro-slag remelted ingo 400 × 400 mm 2 , (ESR-400), and the pressure electro-slag remelted ingots 500-1050 mm in diameter, (PESR-500, PESR-80 and PESR-1050). All data are from the center position of the ingots. The electrode and the PESR-1050, which solidify equ axial dendritic in the center, are marked (•), because the equations used are not valid for the equiaxial solidification stru ture.

Chemical Analyzes over the Cross-Sections
growth (m/s) (calculated on primary dendrite arm spacings) vs. the temperature gradient G ( • C/m). According to Kurz [40], the area above the black curve solidifies in an equiaxial manner and the area below the line solidifies in a columnar-dendritic manner. The red dots for the electrode (CE-300) and the 10.500 mm ingot (PESR-1050) indicates that they both solidify in an equiaxial manner and that the used Equation (1) is not valid for them.  [40]. The area above the line solidifies in an equiaxial manner and the area below solidifies in a columnar-dendritic manner [40]. Data from Uddeholms martensitic stainless steel are included, the electrode 300 × 300 mm 2 , (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 , (ESR-400), and the pressure electro-slag remelted ingots 500-1050 mm in diameter, (PESR-500, PESR-800, and PESR-1050). All data are from the center position of the ingots. The electrode and the PESR-1050, which solidify equiaxial dendritic in the center, are marked (•), because the equations used are not valid for the equiaxial solidification structure.

Chemical Analyzes over the Cross-Sections
The axial segregations of C correspond to both the measured SDAS values and the dendrites orientations towards the bottom of the liquid pool. The more the C segregation, the larger the SDAS value and the larger the orientation angle. The reason is that both the segregation and the size of the SDAS and orientation angle correspond with the Longest Solidification Time (LST) which increases towards the center of the ingot. C is an interstitial element and will, therefore, segregate more than the more stable elements, like Cr and Mo. The decrease in the elements S and Si is due to reactions with the process slag.

Number of Oxide Inclusions Per Size, Position, and Ingot Type
In a previous investigation, Sjöqvist Persson et al. [21] found that the difference in inclusion morphologies between different ESR/PESR ingot sizes was larger than the difference between the electrode heats. The tool steel grade contains a relatively high amount growth (m/s) (calculated on primary dendrite arm spacings) versus the temperature gradient G ( • C/m) around the solid/liquid tip. The black line is plotted data from Kurz solidification diagram [40]. The area above the line solidifies in an equiaxial manner and the area below solidifies in a columnar-dendritic manner [40]. Data from Uddeholms martensitic stainless steel are included, the electrode 300 × 300 mm 2 , (CE-300), the electro-slag remelted ingots 400 × 400 mm 2 , (ESR-400), and the pressure electro-slag remelted ingots 500-1050 mm in diameter, (PESR-500, PESR-800, and PESR-1050). All data are from the center position of the ingots. The electrode and the PESR-1050, which solidify equiaxial dendritic in the center, are marked (•), because the equations used are not valid for the equiaxial solidification structure.

Chemical Analyzes over the Cross-Sections
The axial segregations of C correspond to both the measured SDAS values and the dendrites orientations towards the bottom of the liquid pool. The more the C segregation, the larger the SDAS value and the larger the orientation angle. The reason is that both the segregation and the size of the SDAS and orientation angle correspond with the Longest Solidification Time (LST) which increases towards the center of the ingot. C is an interstitial element and will, therefore, segregate more than the more stable elements, like Cr and Mo. The decrease in the elements S and Si is due to reactions with the process slag.

Number of Oxide Inclusions per Size, Position, and Ingot Type
In a previous investigation, Sjöqvist Persson et al. [21] found that the difference in inclusion morphologies between different ESR/PESR ingot sizes was larger than the difference between the electrode heats. The tool steel grade contains a relatively high amount of silica. Silica has a strong affinity to oxygen, which contributes to the de-oxidation process during ladle refining. This reaction results in a stable dissolved oxygen content in the electrode heats. Therefore, it is appropriate to compare the small ingots from the same electrode heat with the larger ingots from different electrode heats.
As expected (except the 800Ø mm ingot in this investigation), both this and the previous investigation [21] showed that larger ESR and PESR ingots contained a higher number of inclusions than the smaller ingots. This corresponds to the former universal theory that the majority of inclusions in an ESR or PESR ingot are secondary in nature and that they are formed due to slag/metal and de-oxidation reactions [9,[21][22][23][24][25]29]. The larger the LST, the longer the time for the inclusions to precipitate and grow. This finding is further supported by the presence of a large number of small inclusions that presumably were precipitated during solidification. However, Sjöqvist Persson et al. [26,27] stated that the inclusions in a remelted ingot can both be primary, semi-secondary, or secondary inclusions. The result also corresponds to the new theory, since even the semi-secondary inclusions will have a longer time to grow. Furthermore, the primary inclusion maybe more numerous due to the probable differences in the liquid film formation on the electrode s melting surface.
Previous investigations have shown that the amount of larger inclusions (22.4-44.8 µm) is higher for the 1050 mm in diameter ingot (PESR-1050) than for the smaller PESR ingots. In order to improve the statistical accuracy, the results from the previous presented investigation [21], unpublished results from six PESR-1050, and the new results from this investigation are put together in Figure 15. As seen in the picture, the number of the larger inclusions (22.4-44.8 µm) is nearly doubled from the PESR-800 to the PESR-1050 ingots. This result demonstrates that the relation between the amount of larger inclusions (here 22.4-44.8 µm) in different ingot sizes is strongly related to whether or not the material has solidified with a columnar-dendritic or equiaxial structure. The PESR-1050 ingot is no longer in production use for this steel grade.

Conclusions
The focus of the study was the impact of the solidification on the inclusion morphologies in different sizes of electro-slag, (ESR) and pressure electro-slag, (PESR) remelted ingots in a common martensitic steel grade. The inclusion characteristics were studied using LOM and SEM. The main conclusions may be summarized as follows: • The transition between a dendritic and an equiaxial solidification in the center of the ingots in this steel grade takes place in the region between the ingot diameters of 800 and 1050 mm. The primary arms growth rate needed for the CET transition is less than 4 × 10 −7 m/s. In order to undertake the transition, the temperature gradient must be less than approximately 10 3 • C/m.

•
The transition can readily be seen by measuring the angle towards the surface on the side of cross-sections from the remelted ingots. Up to the ingot 800 mm in diameter, the dendritic growth direction in the center of the ingot is parallel to the ingot axis. In the case of the 1050 mm in diameter ingot, it was not possible to determine a dendrite growth direction in the center of the material.

•
The axial segregations of carbon, (C), correspond both to the measured secondary dendritic arm spacing, (SDAS), values and the dendrites angles towards the bottom of the liquid pool. The more C segregation, the larger the SDAS value and the larger, (or no angle as for the PESR-1050), the angle.

•
The larger the ingot and the further towards the center of an ingot, the larger the inclusion sizes. As long as an ingot solidifies with a columnar dendritic structure, the increase in the inclusion numbers and sizes for larger ingots are almost linear. However, at the ingot size (1050 mm in diameter in this study) when the center of the ingot solidifies in an equiaxial mode, the increase in number and size of the inclusions is much higher.

•
The measurement of the amount of larger inclusions (here >22.8 µm) on different ingot sizes will give a good indication of where the transition from a dendritically to equiaxial solidification takes place.