Precipitation Behavior of AlN Inclusions in Fe-0.5Al-2.0Mn Alloy Under Continuous Unidirectional Solidification Process

Medium Manganese Transformation Induced Plastic (Mn-TRIP) steels are expected to be a new generation of advanced high strength sheet steels due to their excellent balance between material cost and mechanical properties. During the solidification process, AlN precipitates at the grain boundary, which leads to the serious deterioration of hot ductility. However, the precipitation of AlN in Mn-TRIP steel has not been clear. In this study, the chemical compositions, morphology, size distribution, and the precipitation behavior of AlN inclusion in an Fe-0.5Al-2.0Mn alloy were studied under the continuous unidirectional solidification process. The results show that there are two types of nitride inclusions in the Fe-0.5Al-2.0Mn alloy: AlN inclusion and complex inclusion of Al2O3-AlN. The planar sections of most AlN particles are hexagonal. Based on the thermodynamic calculation, it was found that the content of Al has a large effect on the stability of Al2O3 and AlN. When the content of Al increases, the molten iron can be changed from saturated by Al2O3 to saturated by AlN. During the solidification process, the precipitation of Al2O3 inclusions occurred at the beginning of the solidification process. The precipitation of AlN inclusions occurred when the contents of Al and N exceeded the equilibrium value and grew until the end of the solidification. The precipitation conditions of AlN inclusion in the Fe-0.5Al-2.0Mn alloy during the solidification process were discussed. The precipitation and the amount of precipitate of AlN inclusions depend on the initial contents of Al, N, and O. It was found that the precipitation of AlN inclusions can be controlled by reducing the initial content of N to less than 0.0072 mass%.


INTRODUCTION
Nowadays, the global automotive industry has been attempting to address the challenge of weight reduction in automobiles to decrease greenhouse gas emissions and improve fuel efficiency. In the automotive industry, sheet steels are used as the major structural material, and thus the development of stronger and cheaper steels is the primary approach adopted by material scientists to meet the demands of the automotive industry. Throughout the world there is increasing interest in the development of new Advanced High Strength Steels (AHSS) with enhanced combinations of strength and ductility to provide sheet materials for demanding applications in future vehicles (Matlock et al., 2012). Transformation Induced Plastic (TRIP), Dual Phase (DP), and Complex Phase steels are AHSS grades that are referred to as the first generation of AHSS. The product of the ultimate tensile strength (UTS) and the total elongation (TE) of the first generation (UTS × TE) is usually less than 25 GPa%. The austenitic stainless steels, Twinning Induced Plasticity (TWIP) steels, and lightweight steels are referred to as the second generation of AHSS. The second of AHSS steel exhibits outstanding mechanical properties (UTS × TE ≥ 50 GPa %), but these austenitic grades are highly alloyed resulting in a significant cost increase (Matlock et al., 2012). In addition, the industrial processing of these alloys, specifically the TWIP steels with high manganese contents, has shown to be extremely challenging and pronely delayed cracking (Kim et al., 2008). Recent researches indicated that the embrittlement susceptibility can be reduced by aluminum alloying, although the exact mechanism involved is still under investigation (Kim et al., 2008). Because of restrictions of the first and second generation as mentioned above, in recent years, metallurgists have actively been developing a new generation of AHSS. Medium Mn TRIP steels have been actively investigated and expected to the third generation of AHSS due to their excellent balance between material cost and mechanical properties.
Continuous casting (CC) is one of the key technologies and an indispensable manufacturing process for the production process of Mn-TRIP steel. However, the fundamental research on the industrial CC technology of Mn-TRIP steel has been quite limited, thus the fundamental research on CC technology of Mn-TRIP steel should be carried out. In the CC process, the surface crack is one of the main problems affecting the quality of the slab. The research on the control of the slab cracks has attracted more and more attention. The hot ductility is an important parameter that reflects the crack sensitivity of the slab. During the solidification process, AlN precipitates at the grain boundary, which leads to the serious deterioration of hot ductility (Yang et al., 2018). Tuling et al. (2011) reported that when the content of Al in TRIP steel was 0.05% (in this paper, the alloy composition unit is mass%), no AlN was found; when the content of Al was increased to 1%, dendritic AlN was precipitated on the austenite grain boundary; and when the content of Al was up to 1.5%, the AlN in TRIP steel was lath-like, without obvious dendritic AlN. Due to the peritectic reaction during the solidification of 1% Al TRIP steel, the precipitation of dendritic AlN on the austenite grain boundary and coarse (∼800 μm) austenite grain leads to the deterioration of hot ductility. However, the precipitation of AlN inclusions in Mn-TRIP steel during the solidification process has not been clear. Therefore, the research on precipitation of AlN inclusions in Mn-TRIP steel is urgently needed.
In this study, the precipitation behavior of AlN inclusions in an Fe-0.5Al-2.0Mn alloy under the continuous unidirectional solidification process was investigated. The chemical compositions, morphology, and size distribution of AlN inclusions at different positions were observed and analyzed by a Scanning Electron Microscope with Energy Dispersive Spectroscopy (SEM-EDS) and an image processing ImageJ program. In addition, under the thermodynamic calculation, the precipitation conditions of AlN inclusion in the Fe-0.5Al-2.0Mn alloy during the solidification process were discussed, which provides theoretical support for controlling the precipitation of AlN inclusions.

EXPERIMENTAL PROCEDURES Experimental
The Fe-0.5Al-2.0Mn alloy was produced in an induction furnace under an atmosphere of nitrogen (99.999% purity) with high purity materials, and the initial chemical composition of the alloy is shown in crucible was used as an outer crucible. The unidirectional solidification experiment was carried out in an electric resistance furnace under an atmosphere of nitrogen (99.999% purity) and a temperature gradient of 13.73 K/cm. The temperature gradient was measured in advanced without the samples by a thermocouple at a set temperature of 1900 K. A schematic diagram of the experimental apparatus is shown in Figure 1. Firstly, the air in the furnace was evacuated by a vacuum pump and nitrogen gas was introduced into the furnace. Subsequently, the alloy was heated to 1873 K with a heating rate of 10 K/min. The temperature of the furnace during the unidirectional solidification experiment was measured directly by an attached thermocouple as shown in Figure 1. After the surface temperature of the sample reached 1873 K, the alloy was held for 2 h, then cooled down to 1773 K with a cooling rate of 4 K/ min. The cooling rate was determined as the multiplication of the temperature gradient and solidification rate which was chosen to be 0.3 cm/min. Finally, when the temperature cooled down to 1773 K, the sample was withdrawn from the furnace and quenched intermediately into the cool water.

Methods of Analysis
The sample was longitudinally cut at the middle, on which the morphology and chemical compositions of AlN inclusions were observed and analyzed by a Scanning Electron Microscope with Energy Dispersive Spectroscopy (SEM-EDS) after grinding and polishing. The sample was observed from the bottom (low temperature) to the top (high temperature) at 4 mm intervals and the same surface area. The number and size of AlN inclusions were measured by the image processing ImageJ program (Igathinathane et al., 2008;Mazzoli and Favoni, 2012). The contents of nitrogen and oxygen at different positions of the sample were measured by the Inert Gas Fusion-Infrared Absorption method. The initial contents of Al and Mn were analyzed by Inductively Coupled Plasma Atomic Emission Spectroscopy.

Morphology of AlN Inclusions
According to SEM-EDS observations, AlN was the dominant inclusion in the experimental alloy. The planar sections of most of the AlN particles were hexagonal. The results of SEM-EDS show that there were two types of nitride inclusions in the experimental Fe-0.5Al-2.0 alloy: AlN inclusion and complex inclusion of Al 2 O 3 -AlN. Other inclusions such as MnS or the complex inclusions of AlN-MnS were not observed because the content of Sulphur consisting in the alloy was as extremely low as 10 ppm. The diameter of the AlN particle was determined as 8.9 μm. The morphology and elemental mappings of complex inclusion of Al 2 O 3 -AlN are shown in Figure 3. As can be seen from Figure 3, there are two phases of the complex inclusion of Al 2 O 3 -AlN with the spherical shape of inner Al 2 O 3 and the hexagonal shape of outer AlN inclusion. The diameter of the inner Al 2 O 3 particle was about 2.5 μm and the diameter of the outer AlN particle was 7.4 μm. The compositions of the complex inclusion of Al 2 O 3 -AlN determined by EDS were Al 52.7 at %, N 42.4 at % at spot 1, and Al 41.9 at %, O 53.7 at % at spot 2 as shown in Figures 3A,E. According to the results of EDS analysis, the compositions of the inclusion were heterogeneous, Al 2 O 3 can be considered as the nucleation core of the AlN inclusion.   O was increased after the solidification experiment compared with the initial content. In contrast to the average content of O, the average content of N was decreased after the solidification experiment compared with the initial content. A small amount of O 2 in N 2 gas used for the solidification experiment was considered as the reason for the increased amount of O in the experimental alloy. During the solidification process, the content of N in the liquid phase was enriched due to the segregation phenomenon that is, discussed in Precipitation Behavior of AlN Inclusions. When the content of N exceeded the nitrogen solubility in the liquid phase, N 2 gas was formed which leads to a decrease in the content of N after the solidification experiment. As can be seen from Figure 4, the distribution of O in the experimental alloy was irregular. Meanwhile, the contents of N showed a gradient distribution along the direction of the solidification process. Figure 5A,B show the size and number density of AlN inclusions at different positions from the bottom of the experimental alloy. Due to the limitation of the measurement by image processing ImageJ program, the complex inclusions of Al 2 O 3 -AlN were considered as AlN inclusions. AlN inclusions were observed from the bottom to the position of 30 mm. Above 30 mm, the existence of AlN inclusions could not be confirmed by SEM-EDS and instead of the AlN inclusions, Al 2 O 3 inclusions were observed. As can be seen from Figure 5A, the average sizes of AlN inclusions at different positions were largest at 2 mm (10.8 μm), gradually decreased, and smallest at 30 mm (2.17 μm). The largest AlN particle was observed at the position of 10 mm with a diameter of 26.65 μm. In contrast to the average size, the number density of AlN inclusions at 2 mm was the smallest, 6.76 mm −2 . The differences in the average size from 6 to 22 mm were around 1 μm and this was the region where AlN inclusions were mostly distributed as can be seen from Figure 5B. The AlN inclusions were divided by sizes: 0-5, 5-10, 10-15, 15-20, 20-25, and larger than 25 μm. It can be seen that the large AlN inclusions (larger than 5 μm) were mostly distributed from 6 to 22 mm where AlN inclusions were also mostly distributed. At the position of 30 mm, the diameters of all AlN particles were less than 5 μm. There were gradient distributions of the average size and number density of AlN inclusions along the direction of the solidification in the experimental alloy. Shibata et al. (1998) reported that smaller and larger inclusions are likely to be pushed and engulfed by a solidified shell, respectively. Accordingly, smaller inclusions might transfer along the direction of solidification. This might be the cause of the decrease in the size of inclusions. Moreover, only small inclusions would form at the initial stage of solidification, such inclusions might be pushed to the upper region. Therefore, only a small amount of inclusions remained at the bottom of the sample. On the surface, the liquid phase was exposed directly to a nitrogen atmosphere, and thus the top region of the sample can be saturated by nitrogen during the unidirectional solidification experiment. As the content of nitrogen increases and exceeds the solubility in the liquid phase due to the segregation phenomenon, nitrogen gas was formed and became bubbles. The non-metallic inclusions can be entrapped by the bubbles and moved to the surface. As can be seen from Figure 4B, the content of nitrogen in the solidified sample significantly increases at higher than 24 mm from the bottom where the amount of AlN inclusions larger than 10 μm also significantly decreases. Therefore, the nitrogen bubbles can be considered as a reason for the decrease in the number density of AlN inclusions at the top of the solidified sample. It was also found that the size and amount of inclusions are almost constant in the range from 6 to 22 mm where AlN inclusions and complex inclusions of Al 2 O 3 -AlN were observed. In the region, the effect of pushing and nitrogen bubbles may be negligible. Therefore, the formation of the AlN inclusions and complex inclusions of Al 2 O 3 -AlN were discussed from the viewpoints of thermodynamics and microsegregation in Thermodynamic Analyses of Formation of Al 2 O 3 and AlN through Precipitation Conditions of AlN Inclusions in Fe-0.5Al-2.0Mn. The diameter distribution of AlN inclusions is shown in Figure 5C. The diameter of AlN inclusions was divided at 2 μm intervals. As can be seen from Figure 5C, there were many fine inclusions less than 4 μm, the cumulative frequency of which was 49.2%. The frequencies of AlN inclusions size in the range of 4-16 μm at 2 μm intervals were not significantly different, the cumulative frequency of which was 42.8%, whereas the cumulative frequency of AlN inclusions greater than 16 μm was 8.0%.

Thermodynamic Analyses of Formation of Al 2 O 3 and AlN
To describe the precipitation behavior of AlN inclusions during the solidification process, the formations of Al 2 O 3 , AlN, and N 2 in liquid iron were calculated thermodynamically. The formation reactions of Al 2 O 3 , AlN, and N 2 can be written as follows: where ΔG o Al2O3 (Itoh et al., 1997), ΔG o AlN (Paek et al., 2013a), and ΔG o N (Ishii et al., 1982) are the standard Gibbs free energy changes of reactions Eqs. 1, 3, 5, respectively. The equilibrium constants K Al 2 O 3 , K AlN , and K N can be written as follows: The activity coefficients in the above equations can be calculated by using Wagner's formalism (Wagner, 1952) (13) where e i j and r j, k i are the first and second-order interaction parameters of elements in the liquid iron (Gordon and Breach, 1988;Itoh et al., 1997;Hino and Ito, 2010;Paek et al., 2013a;Paek et al., 2013b;Jang et al., 2017), respectively. The values of e i j and r j, k i are listed in Table 2. Using the interaction parameters in Table 2 and Eqs. 1-13, the equilibrium relations between Al, O, and N in the Fe-Al-2.0Mn system at 1873 K and P N 2 1 atm were calculated and shown in Figure 6. It can be seen that Al 2 O 3 is the stable phase and AlN is the unstable phase at the initial chemical composition of the experimental alloy. However, AlN inclusions were observed in the alloy after the unidirectional continuous solidification experiment.
As shown in Morphology of AlN Inclusions, the complex inclusions of Al 2 O 3 -AlN were also observed in the alloy. To describe the formation of the complex inclusions of Al 2 O 3 -AlN, the equilibrium relations between Al 2 O 3 and AlN were calculated. Combining reactions Eqs. 1, 3, the following equations were obtained: The equilibrium constant of reaction Eq. 14 can be rewritten similarly as the following equation: Using Eqs. 10-13, 15, 17, the equilibrium relations between Al 2 O 3 and AlN in the Fe-Al-2.0Mn system at different contents of Al were calculated and shown in Figure 7. The calculations showed that the content of Al has a large effect on the equilibrium between Al 2 O 3 and AlN. At (mass %Al) 0.5, there is not a stable phase of AlN, and Al 2 O 3 is the stable phase at the initial conditions as can be seen from Figure 7A. When the content of Al is increased to 1.5% and the contents of O and N are the same as the initial contents, AlN is the stable phase as can be seen from Figure 7B. It is suggested that AlN inclusions may precipitate during the solidification process due to the enrichment of solutes. Both segregation and drop in temperature can affect the AlN formation during solidification. From Eq. 15, reaction Eq. 14 is an endothermic reaction. Therefore, the reaction behavior proceeds spontaneously to the left-hand side of the reaction formula at lower temperatures, which suggests that Al 2 O 3 becomes more stable than AlN. Therefore, the formation of AlN can be promoted by rather the microsegregation than by the drop in temperature. In order to clarify the conditions of AlN formation, the formation of AlN inclusions, as well as the formation of complex inclusion Al 2 O 3 -AlN, were considered from microsegregation's view in the following discussion.

Precipitation Behavior of AlN Inclusions
As mentioned in Thermodynamic Analyses of Formation of Al 2 O 3 and AlN, during the solidification process, the contents of solutes in the liquid phase are enriched due to the segregation   phenomenon. The enriched contents of solutes can lead to the precipitation of new inclusions. In the present study, Clyne-Kurz's equation (Clyne and Kurz, 1981) was used to calculate the contents of Al, O, and N in the liquid iron during the solidification process.
where C L , C S , C o , k, and f s are the content of solute in the liquid phase, the content of solute in the solid phase, the initial content of solute, equilibrium distribution coefficient, and the fraction of solid, respectively. Ω is a function of the back diffusion coefficient α that is, defined as the following equation: where D s , t f , and λ s are the diffusion coefficient of solute in solid (m 2 s −1 ), local solidification time (s), and second dendrite arm spacing (m), respectively. The diffusion coefficients (Ueshima et al., 1986;Yamada et al., 1990;Won and Thomas, 2001) and the equilibrium distribution coefficients (Choudhary and Ghosh, 2009) of Al, O, and N are listed in Table 3. The local solidification time can be calculated by the following equations: The secondary arm spacing λ s was calculated as a function of cooling rate for steels containing less than 0.53% carbon content and the following equation (El-Bealy and Thomas, 1996) was employed: where C R 4 K/min is the cooling rate, which was determined from the experimental conditions. According to Diederichs and Bleck's report (Diederichs and Bleck, 2006), the liquidus temperature (K) and solidus temperature (K) can be estimated by the following equations: During the solidification process, when the contents of Al, O, and N are enriched in the liquid phase, the precipitation of Al 2 O 3 and AlN may occur simultaneously. Therefore, the order of precipitation of Al 2 O 3 and AlN needs to be considered. The order of precipitation of Al 2 O 3 and AlN 3 | Partition coefficients (k) and diffusion coefficients (D) in delta phase of solute elements.
Based on the conditions of thermodynamic equilibrium, the order of precipitation of Al 2 O 3 and AlN was considered as 3 cases as follows: • ΔG Al2O3−AlN 0: the equilibrium between Al 2 O 3 and AlN is achieved. • ΔG Al2O3−AlN < 0: AlN is assumed to precipitate before Al 2 O 3 .
During the solidification process, as the dendrite grows into the melt, and as secondary arms spread from the main dendrite stem, the solute is rejected (Campbell, 2015). The solute is concentrated in tiny regions between secondary dendrite arms due to the microsegregation phenomenon. In the tiny regions, the microsegregation of solute can be calculated without considering the effect of the drop in temperature. Therefore, in the present study, the microsegregation of solutes was calculated at a constant temperature T 1873 K. The content of solute in the liquid phase during the solidification process was calculated in a differential solid fraction, df s , which was taken to be 0.0001, until the end of the solidification process. A schematic illustration of the microsegregation model is shown in Figure 8. The initial contents in the alloy were employed in the calculation model. For each step of the calculation, the precipitations of AlN and Al 2 O 3 inclusions were calculated by considering equilibrium conditions of Al, O, and N in liquid iron and the order of precipitation of Al 2 O 3 and AlN. The calculation was repeated until the total of the solid fraction, Al 2 O 3 fraction, and AlN fraction reached one, which means the solidification process was completed. Figure 9 shows the variations of the concentrations of Al, O, and N in the liquid phase and solid phase during the solidification process. A path of variations of the concentrations of Al, N, and O in the liquid phase is shown in Figure 10A. It can be seen that the content of O in the liquid phase was strongly decreased to the equilibrium value due to the precipitation of Al 2 O 3 at the beginning and increased gradually until the end of the solidification process. In contrast to the content of O, the content of N was increased gradually at the beginning. When the content of N exceeded the nitrogen solubility, N 2 gas was formed in the liquid phase and the content of N in the liquid phase was decreased slightly due to the effect of the segregation of Al on the nitrogen solubility. When the content of Al was enriched high enough at f s 0.76, the precipitation of AlN occurred. Subsequently, the content of N in the liquid phase was decreased significantly, the content of Al was increased gradually until the end of the solidification process.
The precipitated amounts of Al 2 O 3 and AlN were calculated by considering the mass conservation of solutes in the liquid phase. The precipitated amounts of inclusions are proportional to the decreased amounts of solutes in the liquid phase. Figure 10B shows the precipitated amounts of Al 2 O 3 and AlN during the solidification process. It was found that Al 2 O 3 precipitated at the beginning of the solidification process. However, the precipitated amount of AlN was more than Al 2 O 3 at the end of the solidification process.
According to the SEM-EDS observations, the complex inclusions of Al 2 O 3 -AlN were observed in the experimental alloy. The nucleation of AlN on Al 2 O 3 can be understood by considering the lattice disregistry concept. Bramfiit (Bramfitt, 1970) suggested that inclusion can act as an effective nucleation site for a solid phase when the disregistry between the substrate and nucleated solid is less than 12%. Dovidenko et al. (1996) reported that the disregistry between (0001) AlN plane and (0001) Al 2 O 3 is about 12%, which means that AlN can nucleate on Al 2 O 3 . Through the calculation of microsegregation shown above, the formation of the complex inclusions of Al 2 O 3 -AlN was described. A schematic illustration of the formation of the complex inclusions of Al 2 O 3 -AlN during the solidification process is shown in Figure 11. At the beginning of the

Precipitation Conditions of AlN Inclusions in
Fe-0.5Al-2.0Mn To control precipitation of AlN inclusions in the Fe-0.5Al-2.0Mn alloy, the thermodynamic analyses of the precipitations of AlN and Al 2 O 3 inclusions were carried out by the method as shown in Precipitation Behavior of AlN Inclusions at the contents of Al 0.5% and Mn 2.0%. The content of O was varied in a range of 0-0.0015% at intervals of 0.00001% and the content of N was varied in a range of 0-0.06% at intervals of 0.0002%. Figure 12 shows the precipitation conditions of AlN inclusions in the Fe-0.5Al-2.0Mn alloy under the experimental conditions used in the present study. It was found that the precipitation of AlN inclusions can be controlled by reducing the initial content of N to less than 0.0072 mass%. The region without precipitation of AlN is represented by the dashed boundary in Figure 12. In this region, the only precipitation of Al 2 O 3 inclusions may occur during the solidification process. When the contents of N are greater than 0.0072% and the contents of O are less than 0.00003%, the only precipitation of AlN may occur without the precipitation of Al 2 O 3 represented by the dotted boundary in Figure 12. The precipitations of Al 2 O 3 and AlN may occur simultaneously when the contents of O and N are greater than 0.00003 and 0.0072%, respectively.

CONCLUSION
The precipitation behavior of AlN inclusions during the solidification process in the Fe-0.5Al-2.0Mn alloy under an atmosphere of nitrogen was investigated. The conclusions are summarized as follows: 1) There are two types of nitride inclusion in the Fe-0.5Al-2.0Mn: AlN inclusion and complex inclusion of Al 2 O 3 -AlN. The planar sections of most AlN particles are hexagonal. 2) Based on the thermodynamic calculation, it was found that the content of Al has a large effect on the stability of Al 2 O 3 and AlN. When the content of Al increases, the molten iron can be changed from saturated by Al 2 O 3 to saturated by AlN. During the solidification process, Al 2 O 3 inclusions precipitated at the beginning of the solidification process. The precipitation of AlN inclusions occurred when the contents of Al and N exceeded the equilibrium value and grew until the end of the solidification. Al 2 O 3 inclusions can be considered as the nucleation core of the AlN inclusions during the solidification process.
3) The precipitation conditions of AlN inclusions in an Fe-0.5Al-2.0Mn alloy under the experimental conditions used in the present study were discussed. The precipitation and the amount of precipitate of AlN inclusions depend on the initial contents of Al, N, and O. It was found that the precipitation of AlN inclusions can be controlled by reducing the initial content of N to less than 0.0072 mass%.