Inﬂuence of Al and N Content and Cooling Rate on the Characteristics of Complex MnS Inclusions in AHSS

: This study focused on the characteristics of complex MnS inclusions in advanced high strength steels. The e ﬀ ect of metal chemistry (Al and N) and the cooling rate of steel were evaluated by analyzing the inclusions present in ﬁve laboratory produced steels. The observed complex MnS inclusions contained Al 2 O 3 -MnS, AlN-MnS, and AlON-MnS. An increase in Al content from 0.5% to 6% increased the number of complex MnS inclusions by ~4 times. In comparison, a decrease of ~80% was observed due to the increased N content of steel from < 10 ppm to ~50 ppm. MnS precipitation ratio was used to determine the potency of di ﬀ erent inclusions forming complex MnS inclusions due to heterogeneous nucleation. It was found that the MnS precipitation ratio of the observed inclusions was related to their misﬁt with MnS, and it decreased in the order of AlN > AlON > Al 2 O 3 . Moreover, it was determined that AlN particles could be easily engulfed at the solidiﬁcation front relative to Al 2 O 3 , which resulted in a higher MnS precipitation ratio for Al 2 O 3 under slow cooling conditions.


Introduction
An excellent combination of high strength, high toughness, and lightweight of advanced high strength steels (AHSS) makes them suitable for various applications. AHSS contain high concentrations of alloying elements such as Mn, Al, and Si. Therefore, these steels are likely to have a large number of inclusions having a range of compositions. There has been thorough research on the microstructure and mechanical properties of AHSS [1][2][3][4][5][6][7][8], while knowledge related to non-metallic inclusions in AHSS is still limited. Previous studies [9][10][11][12][13][14][15] on non-metallic inclusions have identified the primary type of inclusions as MnS, Al 2 O 3 , and AlN. The presence of Al 2 O 3 -MnO is also reported. In addition, a considerable amount of multiphase inclusions (complex), such as Al 2 O 3 -MnS, AlN-MnS, AlON, and AlON-MnS, has been observed.
During cooling of liquid steel, MnS precipitates on the existing inclusions resulting in the formation of complex MnS inclusions. In AHSS, these complex inclusions include Al 2 O 3 -MnS, AlN-MnS, and AlON-MnS. Park et al. [9] reported that in Fe-10Mn-yAl (y = 1, 3, and 6) steels~20% to 40% of the total inclusions were complex MnS. In the work of Gigacher et al. [15]~42% of the inclusions observed in TWIP steel with 25% Mn were duplex inclusions containing MnS. Alba et al. [12] studied the characteristics of inclusions in Fe-xMn-3Al-3Si (x = 2, 5 and 20) steels and found that the fraction of complex MnS inclusions increased from~10% to over 70% with an increase in Mn content. The characteristics of complex MnS inclusions are related to the potency of different inclusion to serve as a nucleation site for MnS. In previous work, the authors [16] reported that the order of effectiveness of inclusions for the precipitation of MnS in AHSS is AlN > AlON > Al 2 O 3 .
The current work investigates the influence of Al and N contents of steel on the characteristics of complex MnS inclusions. Moreover, the effect of the cooling rate of steel samples is considered. For this purpose, Al content is varied from 0.5% to 6% (0.5, 1, 3, and 6) in Fe-5Mn-3Si-Al alloy, and samples from Fe-5Mn-3Si-3Al alloy containing different levels of N (<10 ppm to 54 ppm) are analyzed. Moreover, air cooled and furnace cooled samples obtained from these alloys are investigated to study the effect of cooling rate.

Materials and Methods
The experimental steels were produced in a vertical resistance tube furnace. The steels contain 0.5%, 1%, 3%, or 6% Al, 5% Mn, 3% Si, and 0.1% C. They are labeled based on their Mn, Al, and Si contents, i.e., 50.53, 513, 533, and 563. The reagents were melted in an alumina crucible at 1873 K under Ar atmosphere (5N purity). The Ar gas was passed through Ti turnings at 973 K to reduce oxygen from it. The molten steel was homogenized for 30 minutes, and then the system was evacuated and backfilled with Ar. Several quartz tube samples (Dia. = 5 mm, Length ≈ 8 mm) were taken at different holding times and were air cooled (20-30 K/sec [17]). The remaining steel material (bulk steel) was cooled along with the furnace at a rate of 10 • C/min (0.167 K/sec). The remaining bulk steel had a diameter of 58 mm, and its height was 30 mm. In another experiment, N 2 gas was introduced into a 533 steel. In this experiment (533N-P), the furnace chamber was evacuated at 1873 K, and then it was backfilled by purging a mixture of Ar and N 2 gas (1:1). The mixing ratio was changed to Ar:N 2 = 1:2 after 180 min of purging. Quartz tube samples were collected before and after purging the mixture of Ar and N 2 gas.
Induction coupled plasma optical emission spectrometry (ICP-OES) was used to analyze the chemical composition of steel samples. The oxygen-nitrogen and carbon-sulfur contents of samples were measured by LECO O/N (ON736, LECO Co. St. Joseph, MI, USA) and LECO C/S (CS744, LECO Co. St. Joseph, MI, USA) analyzer, respectively. Both air cooled quartz tube samples and furnace cooled bulk steel samples were analyzed for inclusions by using Jeol 6610 (Jeol Co. Akishima, Tokyo, Japan) scanning electron microscope (SEM) equipped with the ASPEX inclusion analysis system. The inclusions having a maximum diameter (D max ) > 2 µm were considered for analysis. A detailed description of experimental materials and procedure is presented elsewhere [18].

Results
An insignificant difference was observed in the chemical composition of all the samples taken during an experiment except 533N-P. A variation in the N content was observed in the samples of 533N-P due to the purging of N 2 gas. The average composition of the air cooled samples obtained during an experiment is given in Table 1. For the case of 533N-P, three samples were selected to represent different N contents, which are low N, medium N, and high N. The low N, medium N, and high N samples were obtained before introducing N 2 gas to the steel, 45 min, and 270 min after purging N 2 gas to the steel, respectively. It can be seen in Table 1 that the N content of 50.53, 513, 533, and 563 steel samples fall in the low N category (<10 ppm). The detected values for total O content were scattered and less than 10 ppm, which is close to the detection threshold. Further, it is expected to have ±2 ppm in the calibration of N and O analyses. Therefore, <10 ppm is used for representation. The composition of furnace cooled steel was similar to that of air cooled samples. In the case of 533N-P, the furnace cooled steel, and air cooled samples containing high N had a similar composition.   Detailed inclusion analyses of the current experiments are presented and discussed elsewhere [19,20]. Figure 2 presents the number per unit area (N A ) of complex MnS inclusions observed in the air cooled samples of all experimental steels. As can be seen in Figure 2a, 50.53 steel contained~2 mm −2 of complex MnS inclusions. Which slightly increased to~2.7 mm −2 in 513 steel and then substantially increased to~10 mm −2 in 533 steel. This value decreased to~8 mm −2 with a further increase in Al content (563 steel). Figure 2b clearly indicates that an increase in the N content of steel resulted in a decrease in the amount of complex MnS inclusions from~3.5 mm −2 to less than 1 mm −2 .   This value increases to~54% and~68% as Al content increases to 1 and 3%, respectively. The fraction of AlON-MnS is lowest for 50.53 steel and is highest for 563 steel (31%), whereas that for 513 and 533 steels is 11% and 16%, respectively. In 533N-P steel, all samples with different N levels contain between~60% and~80% of AlN-MnS, see Figure 4b. The fraction of Al 2 O 3 -MnS and AlON-MnS varies from~12% to~22% without any clear trend. It can be inferred from Figure 4a that the contribution of Al 2 O 3 -MnS in complex MnS inclusions decreases with increasing Al content of steel, whereas that of AlN-MnS and AlON-MnS increases. However, it is debatable due to a low N A value of complex MnS inclusions in 50.53 and 513 steels. Figure 4b suggests that the composition of complex MnS is independent of the N content of the steel. Again it should be noted that the N A value of complex MnS inclusions in samples from 533N-P steel is small to confirm such a suggestion. Moreover, both Figure 4a,b indicate that in the air cooled samples, AlN-MnS inclusions make a higher portion of complex MnS inclusions as compared to that of Al 2 O 3 -MnS. The composition of complex MnS inclusions observed in furnace cooled samples is presented in Figure 5. The results show that an increase in the Al content of steel leads to a decrease and an increase in the fraction of AlN-MnS and Al 2 O 3 -MnS inclusions, respectively. This tendency is opposite to that observed in the air cooled samples. The fraction of AlN-MnS inclusions decreases from~50% to 22%, while that of Al 2 O 3 -MnS inclusions increases from~36% to over 60%. Moreover, Al 2 O 3 -MnS inclusions have a higher fraction in complex MnS as opposed to AlN-MnS in the air cooled samples. The fraction of AlON-MnS does not vary much with Al content; however, it was generally higher than that in air cooled samples. The complex MnS inclusions in furnace cooled sample from 533N-P have a similar composition as that of its air cooled samples.

MnS Precipitation Ratio
The tendencies observed in Figures 3 and 4 are further quantified by determining the ratio of MnS precipitated on different types of inclusions. MnS precipitation ratio on a specific type of inclusion is defined as the ratio of the number of MnS containing particles to all of the particles in that type of inclusions. For example, the MnS precipitation ratio (PR MnS ) for AlN inclusions is calculated as follows: where, n AlN−MnS and n AlN (pure) represent the number of AlN-MnS inclusions and the number of AlN (pure) inclusions, respectively. Similar equations can be written for the MnS precipitation ratio for Al 2 O 3 and AlON inclusions. The calculated PR MnS for AlN, Al 2 O 3, and AlON inclusions observed in air cooled samples is given in Figure 6. As demonstrated in Figure 6a, the PR MnS for Al 2 O 3 inclusions is the least (<20%), and that of AlON is between 30% and 40%. The PR MnS for AlN inclusions is almost 100% for 50.53, 513, and 533 steels, and it is relatively less for 563 steel, i.e.,~63%. Figure 6b shows that the low N sample of 533N-P steel exhibits similar PR MnS values as observed for 533 steel, which also has a low N content. The PR MnS values for all three types of inclusions significantly decreased to less than 10%, with an increase in the N content of 533N-P steel samples. A high PR MnS value for a particular type of inclusions means that a higher fraction of those inclusions coexists with MnS. For instance, almost all the AlN type inclusions in a low N sample of 533N-P steel are AlN-MnS. Whereas, the very low PR MnS value for medium N and high N level samples indicates that almost all the AlN type inclusions in these samples are AlN (pure) . The calculated PR MnS values of the furnace cooled samples are shown in Figure 7. It is notable that in the furnace cooled samples, the PR MnS values for Al 2 O 3 inclusions are around 80% for all the samples, which is significantly higher than that observed in air cooled samples. Moreover, those for AlON have increased more than two times. On the contrary, the PR MnS values for AlN have decreased (except 533N-P) compared to those observed in Figure 6. For 50.53 and 513, the value dropped from 100% to~80%. A decrease of~37% and~25% is observed in the PR MnS values of 533, and 563 steel as compared to their respective values in the air cooled samples. The furnace cooled sample of 533N-P is to be compared to its high N level air cooled sample because of their similar N content. The furnace cooled sample has a higher PR MnS value for AlN as compared to that in air cooled one (compare Figures 6b and 7).  Figure 4 is not reflected in Figure 6.

Discussion
The following three factors are to be considered for understanding the tendencies observed in Figure 4 through Figure 7.

Thermodynamics of Inclusions Formation
Thermodynamic calculations were carried out by using Factsage 7.3 (CRCT, Montreal, QC, Canada and GTT-Technologies, Aachen, Germany) to determine the stable phases in the current steel compositions for a temperature range between 1473 K and 1873 K. It was found that at 1873 K Al 2 O 3 is the only stable phase for all steel compositions except for a high N sample of 533N-P steel. For a high N sample, both Al 2 O 3 and AlN are stable at working temperature. In addition to phase stability, the liquidus temperature (T liq ), solidus temperature (T sol ), AlN formation temperature (T AlN ), and MnS formation temperature (T MnS ) were also determined. The obtained data is presented in Table 2. The T AlN values for 563 and medium N sample of 533N-P are higher than their respective T liq values, indicating that for these two compositions, AlN starts to form before the solidification begins during cooling. Whereas, the T AlN values for the remaining four compositions suggest that AlN forms during the solidification of steel. This is because the driving force of AlN formation increases during solidification due to the enrichment of Al and N at the solidification front. The Scheil equation [11,13,21] where g s is the solid fraction, and  The relationship between the temperature at the solidifying front (T') and the solid fraction (g s ) is shown in Equation (5) [11].
where T m , T sol , and T liq are the melting temperature of pure Fe (1811 K, 1538 • C), the solidus temperature of steel, and the liquidus temperature of steel, respectively. Using Equations (2)-(5), a relationship between [%Al] × [%N] at solidification front and g s , and equilibrium value for AlN formation (K AlN ) and g s can be obtained. An intersection of both functions gives the g s value at which AlN formation starts. The obtained results for the current experimental steels are given in Table 3. For calculation of g s values, N content is assumed to be 10 ppm.
The results indicate the AlN starts to form at a g s of 0.85 for 50.053 steel. Moreover, AlN formation takes place at an early stage of solidification, i.e., at lower g s values in 513 and 533.

Heterogeneous Nucleation of Inclusions
Inclusions in steel can homogeneously nucleate in the melt or heterogeneously on the existing solid particles or matrix. Heterogeneous nucleation is more likely to take place as it requires a less nucleation energy. The morphology of complex MnS inclusions (Figure 1) observed in the current study suggests they are formed due to heterogeneous nucleation of MnS on other inclusions. The heterogeneous nucleation is influenced by several factors, among them, the lattice misfit or disregistry between the substrate and nucleated phase has a significant contribution [24,25]. The lattice misfit or disregistry determines the effectiveness of an inclusion to act as a nucleus for another. It is suggested that an inclusion can act as an effective nucleation site for a solid phase when the lattice misfit between the substrate inclusion and nucleated solid is less than 12% [24]. A higher value means less effectiveness as nucleant.
The lattice misfit can be calculated as follows [26]: where a n and a s are the lattice parameters of nucleated solid and substrate, respectively. Table 4 presents the lattice parameters used for misfit calculations. The calculated misfit between MnS and Al 2 O 3 (substrate) is~10% along a axis. Whereas, MnS nucleation on the c axis of AlN gives a misfit of~5%. These values are in agreement with those reported by Ohta and Suito [27]. A lower mismatch value for AlN means that less interfacial energy is required for nucleation of MnS on AlN compared to that on Al 2 O 3 ; hence, MnS would preferentially nucleate on AlN. This explains a significantly higher MnS precipitation ratio for AlN inclusions in the experimental steels.

Similar to Al 2 O 3 -MnS and AlN-MnS inclusions, AlON-MnS complex inclusions are formed by heterogeneous nucleation of MnS on AlON inclusions.
Here it is important to mention that the AlON inclusions observed in the current study are not single AlON phase inclusions rather they consist of distinct Al 2 O 3 and AlN phases (see Figure 1c). Moreover, it is reported that the AlON phase is stable only above 1923 K (1650 • C), and below this temperature, it dissociates into AlN+Al 2 O 3 [30]. Based on the thermodynamic stability of AlON (single phase), Al 2 O 3, and AlN phases, it can be implied that the observed AlON inclusions are a result of AlN nucleation on existing Al 2 O 3 inclusions. The calculated lattice misfit between (a axis of) AlN and Al 2 O 3 is about 35%. However, a 30 • in-plane rotation of AlN with respect to Al 2 O 3 reduces this misfit to~12% [31,32]. A 12% misfit means that Al 2 O 3 is not a very effective substrate for nucleation of AlN, but it can still occur particularly in the presence of high Al content of the steel. The effect of Al content on the formation of AlON is reflected in Figure 4b, which shows that the fraction of AlON-MnS in complex MnS inclusions increases with Al content of the steel. Hence, the formation of AlON-MnS inclusions takes place by duplex heterogeneous nucleation. First, AlN nucleates on Al 2 O 3, then MnS heterogeneously nucleates on AlN (or Al 2 O 3 but preferentially on AlN).

Inclusion Behavior at Solid/Liquid Interface
It is established in the preceding sections that for most of the investigated steel compositions, complex MnS inclusions are formed during the solidification of steel and by heterogeneous nucleation of MnS on already existing inclusions. Therefore, the behavior of nucleant inclusions at the solidification front is considered. During the solidification of liquid steel, inclusions are either engulfed or pushed by the solid/liquid interface. There are several factors that determine the inclusion behavior at solid/liquid interface, such as solidification velocity, interfacial energy between inclusion and liquid steel, the viscosity of steel, size of the particle, and thermal conductivity of inclusion and steel. Stefanescu et al. [33] derived a relationship describing the critical velocity, V CR , of solid/liquid interface to determine engulfment/pushing transition. A solidification velocity higher than V CR would result in the engulfment of inclusions of a specific size. The relationship is expressed as follows: where a o is the atomic distance (m), η is the viscosity of liquid steel (kg/ms), and R is the radius of particle (m). k * is the ratio of the thermal conductivities of the particle (k P ) and liquid steel (k L ), i.e., k * = k P k L . ∆γ o is given as follows: where γ PS is the interfacial energy between particle and liquid steel (J/m 2 ). γ P , γ L , and γ S are the surface energies of the particle, liquid steel, and solid steel, respectively (J/m 2 ). θ PL is the contact angle and W AD is the work of adhesion.
In the current study, Equation (7) is adopted to determine the relationship between V CR and the size of inclusion for Al 2 O 3 and AlN. The parameters used to determine the relationship are given in Table 5. The data used for calculating ∆γ o of AlN comes from the work of Y. Luo [34], where the AlN substrate was produced using Y 2 O 3 as a binder. The a o value for both Al 2 O 3 and AlN is taken as the atomic diameter of aluminum [33]. The obtained results are plotted in Figure 8. It is evident from the figure that there exists a difference in the behavior of Al 2 O 3 and AlN inclusions at the solid/liquid interface. Figure 8 suggests that AlN inclusions are more likely to be engulfment by the solidification front as compared to Al 2 O 3 inclusions at a certain solidification velocity. For instance, AlN inclusions with diameter >~1 µm would be engulfed at a solidification velocity of 4 × 10 −6 m/s, and only inclusions with diameter < 1 µm would be pushed. Whereas, Al 2 O 3 inclusions with diameter up tõ 3.5 µm can be pushed by the solidification front for the same solidification velocity. The values without references are calculated using other data given in Table 5.  Figure 6. AlN inclusions have higher PR MnS values due to a low lattice misfit between AlN and MnS. Considering this and the stability of AlN in the medium N and high N containing samples of 533N-P, it can be argued that these samples should also have high PR MnS values for AlN. However, AlN inclusions have very low PR MnS values (<5%) in these samples (Figure 6b). This phenomenon suggests that only the AlN inclusions which are formed at the solidification front, or closer to the formation of MnS, participate in heterogeneous nucleation of MnS. This behavior is also reflected in the air cooled sample of 563 steel (Figure 6a), where the PR MnS value for AlN is relatively lower because the formation of AlN starts above T liq (Table 2).
In all the furnace cooled samples (except for 533N-P), the PR MnS values for AlN decrease in comparison to those of air cooled samples. This can also be explained by considering the formation of AlN-MnS inclusions at the solidification front. However, due to the slow cooling rate, the AlN inclusions which form at the early stage of solidification are engulfed by the solidification front as AlN (pure) inclusions. Hence, lower PR MnS values for AlN are observed in furnace cooled samples. This reasoning also explains relatively lower PR MnS values for AlN in 533 (~60%) and 563(~40%) furnace cooled samples as compared to those of 50.53 and 533 (~80%), see Figure 7. A higher number of AlN (pure) inclusions would form in 533 and 563 due to their low g s values for AlN formation; hence, low PR MnS values for AlN in these steels. An even lower PR MnS value for furnace cooled sample 533N-P affirms this logic, as AlN (pure) inclusions start to form in liquid steel for this sample.
Moreover, Al 2 O 3 inclusions have significantly higher (~80%) PR MnS values in furnace cooled samples as compared to less than 10%, as observed in air cooled samples. As shown in Figure 8

Conclusions
In the present study, laboratory experiments were carried out to investigate the characteristics of complex MnS inclusions in AHSS and their formation behavior. The following observations were made.
The number of complex MnS inclusions increases with an increase in Al content of steel, and it decreases with an increase in N content.
Under air cooling condition, an increase in the Al content of steel results in an increase and a decrease in Al 2 O 3 -MnS and AlN-MnS inclusions, respectively. Whereas, N content does not have any specific relation to the composition of complex MnS inclusions.
Under furnace cooled conditions, the complex MnS inclusions become rich in Al 2 O 3 -MnS as Al content of steel increases; meanwhile, the portion of AlN-MnS decreases.
The effectiveness of inclusions for the precipitation of MnS is related to their lattice misfit with MnS, and its order is AlN > AlON > Al 2 O 3 .
The precipitation ratio of MnS for AlN is influenced by the stability of AlN in liquid steel, i.e., it decreases when AlN can form in liquid steel before solidification starts.
Slow cooling significantly increases the MnS precipitation for Al 2 O 3 and AlON, and slightly decreases that for AlN. It is attributed to the difference in the pushing and engulfment behavior of Al 2 O 3 and AlN.