Identification of Intermetallic Compounds and Its Formation Mechanism in Boron Steel Hot-Dipped in Al-7 wt . % Mn Alloy

In laser welding and hot stamping Al-Si-coated boron steel, there is a problem that the strength of the joint is lowered due to ferrite formation in the fusion zone. The purpose of this study is to develop an Al-7 wt.% Mn hot-dip coating in which Mn, an austenite stabilizing element, replaces the ferrite stabilizing element Si. The nucleation and formation mechanism of the reaction layer was studied in detail by varying the dipping time between 0 and 120 s at 773 ◦C. The microstructure and phase constitution of the reaction layer were investigated by various observational methods. Phase formation is discussed using a phase diagram calculated by Thermo-CalcTM. Under a 30 s hot-dipping process, no reaction occurred due to the formation of a Fe3O4 layer on the steel surface. The Fe3O4 layer decomposed by a reduction reaction with Al-Mn molten alloy, constituent elements of steel dissolved into a liquid, and the reaction-layer nucleus was formed toward the liquid phase. A coated layer consists of a solidified layer of Al and Al6Mn and a reactive layer formed beneath it. The reaction layer is formed mainly by inter-diffusion of Al and Fe in the solid state, which is arranged on the steel in the order of Al11Mn4→ FeAl3 (θ)→ Fe2Al5 (η) phases, and the Fe3AlC (κ) in several nm bands formed at the interface between the η-phase and steel.


Introduction
In recent years, in the automobile industry, as the strength of the vehicle body has been increased to attain light weight and strength, there has been a problem in that spring-back phenomenon occurs due to insufficient ductility during the forming process [1,2].In order to overcome this challenge using Advanced High Strength Steel (AHSS), a hot stamping process has been developed that stabilizes both the formation and strength of boron steel [3][4][5].However, in this process, scale formation and surface decarburization occur, caused by surface oxidation of the steel when heated in excess of 900 • C. In turn, to solve this issue, Al-Si hot-dipped coating is typically performed [4][5][6][7].
However, when an Al-Si-coated boron steel sheet for hot stamping is laser-welded, a coated layer composed of the ferrite-forming elements is mixed into the fusion zone, and the δ-ferrite is formed in a band shape at the fusion zone boundary.As a large amount of ferrite is formed, the strength of the joint decreases, and fracture occurs in the segregation region [8][9][10].
To address these challenges, in this study, a coating layer was developed that suppresses the formation of ferrite in the fusion zone during laser welding by adding Mn, which is an austenite-forming element, to the Al-Si coating layer typically used.Al-7 wt.% Mn alloy was prepared by mixing 373.3 g of pure Al and 326.7 g of Al-15wt.%Mn alloy.The specimens were pre-heated at 150 • C for 20 s to remove moisture from the surface of the Al-7Mn (wt.%) molten alloy.The temperature of the molten metal was maintained at 773 • C, which is the lowest possible coating temperature, for 0-120 s to perform hot dip process then air cooling.
The coated specimens were polished by using 1 mm Diamond + 0.04 mm Silica after arranging the surfaces through several stages of SiC sand paper.After polishing, 50 mL CH 4 OH + 2 mL HNO 3 etching solution was used for the base material.100 mL distilled water + 1 g NaOH solution, which is the etching solution of the 3000 series Al alloy, was used for the coating layer [21].
To observe the microstructure of the coated layer after hot-dipping, the Olympus (Tokyo, Japan) BX-51M optical microscopy and the Carl Zeiss (Oberkochen, Germany) SUPRA 45 field-emission Scanning Electron Microscope (FE-SEM) were used.Phase identification of large areas of the oxide layer was performed using the Rigaku (Tokyo, Japan) Ultima IV X-ray diffraction (XRD) at 1 • /min.EDAX (Mahwah, NJ, USA).Energy-dispersive spectroscopy (EDS) was performed to observe the element distribution in the coating layer.In order to analyze the chemical composition distribution, the JEOL (Tokyo, Japan) JXA-8530F field-emission electron probe micro-analysis (FE-EPMA) was performed to analyze the area, with a 1 kV voltage, 100 nA current, and step sizes of 0.1-0.5 mm in a non-etching state.To analyze the phase of the reaction layer, specimens were deflected using the FEI (Hillsboro, OR, USA) Scios focused ion beam (FIB), and their chemical composition and diffraction pattern were then obtained using the FEI (Hillsboro, OR, USA) TALOS F200X field-emission transmission electron microscopy (FE-TEM).The EDAX Hikari Electron backscatter diffraction (EBSD) was used to observe the phase distribution and crystal orientation of the reaction layer during formation and growth, and the Media Cybernetics (Rockville, MD, USA) Image-plus pro image analysis program was used for analyzing images to obtain parameters such as the thickness and length of each phase.

Formation of the Reaction Layer at Initial Hot Dipping
Boron steel was hot-dipped in Al-7 wt.% Mn molten metal at 773 • C for various dipping times.Figure 1 shows the cross-sectional optical microstructure of hot dipped specimens for various times at 773 • C. In the specimens hot dipped for 10 s and 20 s, the reaction layer was not formed.As shown in Figure 1a, the localized reaction layer was formed on the surface of steel for 30 s.
Coatings 2017, 7,222 beam (FIB), and their chemical composition and diffraction pattern were then obtained using the FEI (Hillsboro, OR, USA) TALOS F200X field-emission transmission electron microscopy (FE-TEM).The EDAX Hikari Electron backscatter diffraction (EBSD) was used to observe the phase distribution and crystal orientation of the reaction layer during formation and growth, and the Media Cybernetics (Rockville, MD, USA) Image-plus pro image analysis program was used for analyzing images to obtain parameters such as the thickness and length of each phase.

Formation of the Reaction Layer at Initial Hot Dipping
Boron steel was hot-dipped in Al-7 wt.% Mn molten metal at 773 °C for various dipping times.Figure 1 shows the cross-sectional optical microstructure of hot dipped specimens for various times at 773 °C.In the specimens hot dipped for 10 s and 20 s, the reaction layer was not formed.As shown in Figure 1a, the localized reaction layer was formed on the surface of steel for 30 s. Figure 2 shows the results of the back-scattered electron (BSE) image in which the reaction phase is present (Figure 1a) and the Fe and O components of Al, Mn, and steel.The white line shows the original steel surface.A band structure with high concentrations of Fe (Figure 1d) and O (Figure 1e) was formed on the surface of the steel.From this result, this layer is considered to be a Fe-based oxide layer.On the other hand, the elliptical shape phase in the area indicated by γ on the surface coexisted with the coating components Al and Fe, but Mn and O are not detected.Thus, the reaction phase formed where the oxide layer broke, and the Al-Mn coating solution permeated, reacting with the steel to form the reaction layer.
Figure 3 includes an enlarged SEM microstructure and an EPMA surface analysis results of the area denoted by γ in Figure 1b, where specimens were dipped for 40 s.A reaction layer formed and increased in the absence of an oxide layer.The reaction layer can be classified into four phases-A, B, C, and D-from the comparison of the Al, Mn, and Fe concentrations with the formed positions, either at the steel surface or coating layer.The A phase contains a high concentration of Al and Fe but contains no Mn and was formed in the coating layer and the steel interior.The B phase formed in the coating layer and was a high concentration of Al and Mn, but was characterized by a small amount of Fe, which was connected to the A phase and the end.The C phase had a slightly higher in Mn concentration than the B phase.Phase D consisted almost entirely of Al. Figure 2 shows the results of the back-scattered electron (BSE) image in which the reaction phase is present (Figure 1a) and the Fe and O components of Al, Mn, and steel.The white line shows the original steel surface.A band structure with high concentrations of Fe (Figure 1d) and O (Figure 1e) was formed on the surface of the steel.From this result, this layer is considered to be a Fe-based oxide layer.On the other hand, the elliptical shape phase in the area indicated by γ on the surface coexisted with the coating components Al and Fe, but Mn and O are not detected.Thus, the reaction phase formed where the oxide layer broke, and the Al-Mn coating solution permeated, reacting with the steel to form the reaction layer.
Figure 3 includes an enlarged SEM microstructure and an EPMA surface analysis results of the area denoted by γ in Figure 1b, where specimens were dipped for 40 s.A reaction layer formed and increased in the absence of an oxide layer.The reaction layer can be classified into four phases-A, B, C, and D-from the comparison of the Al, Mn, and Fe concentrations with the formed positions, either at the steel surface or coating layer.The A phase contains a high concentration of Al and Fe but contains no Mn and was formed in the coating layer and the steel interior.The B phase formed in the coating layer and was a high concentration of Al and Mn, but was characterized by a small amount of Fe, which was connected to the A phase and the end.The C phase had a slightly higher in Mn concentration than the B phase.Phase D consisted almost entirely of Al.  Figure 4 shows the results of the XRD of the steel surface after dipping for 5 s at 773 °C, confirming that the oxide layer was comprised of Fe3O4.This oxide layer is the most thermodynamically stable Febased oxide at temperatures below the dipping temperature.When the boron steel was heated in an electric furnace maintained at 773 °C for 10 s in air, no oxide layer was formed.However, when the boron steel was dipped into the Al-7 wt.% Mn molten metal, an oxide layer formed.Thus, the Fe3O4 formed by the reaction of the steel surface with oxygen in the molten metal.
Figure 5 shows the formation rate which is defined as the ratio (%) of the formed reaction layer to the measurement length (10 mm) from the cross-sectional structure.The reaction area increased with an increase in the dipping time, and the reaction layer formed over the entire surface at 100 s.Moreover, the thickness (in the depth direction) distribution of the reaction layer was non-uniform, as shown in Figure 1e, but the specimen after hot-dipping for 120 s had a uniform reaction layer.Figure 4 shows the results of the XRD of the steel surface after dipping for 5 s at 773 °C, confirming that the oxide layer was comprised of Fe3O4.This oxide layer is the most thermodynamically stable Febased oxide at temperatures below the dipping temperature.When the boron steel was heated in an electric furnace maintained at 773 °C for 10 s in air, no oxide layer was formed.However, when the boron steel was dipped into the Al-7 wt.% Mn molten metal, an oxide layer formed.Thus, the Fe3O4 formed by the reaction of the steel surface with oxygen in the molten metal.
Figure 5 shows the formation rate which is defined as the ratio (%) of the formed reaction layer to the measurement length (10 mm) from the cross-sectional structure.The reaction area increased with an increase in the dipping time, and the reaction layer formed over the entire surface at 100 s.Moreover, the thickness (in the depth direction) distribution of the reaction layer was non-uniform, Figure 4 shows the results of the XRD of the steel surface after dipping for 5 s at 773 • C, confirming that the oxide layer was comprised of Fe 3 O 4 .This oxide layer is the most thermodynamically stable Fe-based oxide at temperatures below the dipping temperature.When the boron steel was heated in an electric furnace maintained at 773 • C for 10 s in air, no oxide layer was formed.However, when the Coatings 2017, 7, 222 5 of 15 boron steel was dipped into the Al-7 wt.% Mn molten metal, an oxide layer formed.Thus, the Fe 3 O 4 formed by the reaction of the steel surface with oxygen in the molten metal.
Figure 5 shows the formation rate which is defined as the ratio (%) of the formed reaction layer to the measurement length (10 mm) from the cross-sectional structure.The reaction area increased with an increase in the dipping time, and the reaction layer formed over the entire surface at 100 s.Moreover, the thickness (in the depth direction) distribution of the reaction layer was non-uniform, as shown in Figure 1e, but the specimen after hot-dipping for 120 s had a uniform reaction layer.

Evolution of Microstructure in the Coating Layer
The reaction layer (Figure 1a) of the specimen dipped for 30 s was too small, the oxide layer was nearby, and TEM specimens could not be obtained by FIB.Among the reaction layers of the 40 s dipped specimens, the smallest, corresponding to the area in Figure 3a, was selected.TEM specimens were collected by FIB (focused ion beam) and observed by STEM to identify which phases were formed in the reaction layer.Figure 6a shows a specimen cut in the depth direction in γ area of Figure 3a, allowing observation of the phases of the liquid (area 1) and the steel (area 2).Additionally, (c,d) shows the STEM-HAADF (high-angle annular dark-field) microstructure observed in areas 1 and 2, respectively.

Evolution of Microstructure in the Coating Layer
The reaction layer (Figure 1a) of the specimen dipped for 30 s was too small, the oxide layer was nearby, and TEM specimens could not be obtained by FIB.Among the reaction layers of the 40 s dipped specimens, the smallest, corresponding to the area in Figure 3a, was selected.TEM specimens were collected by FIB (focused ion beam) and observed by STEM to identify which phases were formed in the reaction layer.Figure 6a shows a specimen cut in the depth direction in γ area of Figure 3a, allowing observation of the phases of the liquid (area 1) and the steel (area 2).Additionally, (c,d) shows the STEM-HAADF (high-angle annular dark-field) microstructure observed in areas 1 and 2, respectively.

Evolution of Microstructure in the Coating Layer
The reaction layer (Figure 1a) of the specimen dipped for 30 s was too small, the oxide layer was nearby, and TEM specimens could not be obtained by FIB.Among the reaction layers of the 40 s dipped specimens, the smallest, corresponding to the area in Figure 3a, was selected.TEM specimens were collected by FIB (focused ion beam) and observed by STEM to identify which phases were formed in the reaction layer.Figure 6a shows a specimen cut in the depth direction in γ area of Figure 3a, allowing observation of the phases of the liquid (area 1) and the steel (area 2).Additionally, (c,d) shows the STEM-HAADF (high-angle annular dark-field) microstructure observed in areas 1 and 2, respectively.
Figure 7 is the surface analysis by EDS for Al, Mn, Fe, and C in Figure 6c and Table 2 shows the results of EDS point analysis of the phases 1-5 in Figure 6c.From the distributions of Mn and Fe, the areas of 1, 2, and 3 had high Mn concentrations but low Fe concentration.Region 5 contains high Fe and low Mn concentrations, and 4 region contains a high Fe concentration and negligible Mn. pecimens, the smallest, corresponding to the area in Figure 3a, was selected.TEM specimens lected by FIB (focused ion beam) and observed by STEM to identify which phases were the reaction layer.Figure 6a shows a specimen cut in the depth direction in γ area of Figure 3a, observation of the phases of the liquid (area 1) and the steel (area 2).Additionally, (c,d) e STEM-HAADF (high-angle annular dark-field) microstructure observed in areas 1 and 2, ely. Figure 7 is the surface analysis by EDS for Al, Mn, Fe, and C in Figure 6c and Table 2 shows th results of EDS point analysis of the phases 1-5 in Figure 6c.From the distributions of Mn and Fe, th areas of 1, 2, and 3 had high Mn concentrations but low Fe concentration.Region 5 contains high F and low Mn concentrations, and 4 region contains a high Fe concentration and negligible Mn.  Figure 8 shows the results of the diffraction pattern in the area indicated by 1-5 in Figure 6c an the diffraction pattern analysis obtained the composition analysis result in Table 2. From th diffraction pattern analysis, Point 1 phase is Al6Mn (orthorhombic), Point 2 and 3 phases are Al11Mn (triclinic) phase, point 4 phase is Fe2Al5 (η, orthorhombic) phase, and point 5 phase is FeAl3 (θ monoclinic) phase.Figure 7 is the surface analysis by EDS for Al, Mn, Fe, and C in Figure 6c and Table 2 shows the results of EDS point analysis of the phases 1-5 in Figure 6c.From the distributions of Mn and Fe, the areas of 1, 2, and 3 had high Mn concentrations but low Fe concentration.Region 5 contains high Fe and low Mn concentrations, and 4 region contains a high Fe concentration and negligible Mn.  Figure 8 shows the results of the diffraction pattern in the area indicated by 1-5 in Figure 6c and the diffraction pattern analysis obtained the composition analysis result in Table 2. From the diffraction pattern analysis, Point 1 phase is Al6Mn (orthorhombic), Point 2 and 3 phases are Al11Mn4 (triclinic) phase, point 4 phase is Fe2Al5 (η, orthorhombic) phase, and point 5 phase is FeAl3 (θ, monoclinic) phase.Figure 8 shows the results of the diffraction pattern in the area indicated by 1-5 in Figure 6c and the diffraction pattern analysis obtained the composition analysis result in Table 2. From the diffraction pattern analysis, Point 1 phase is Al 6 Mn (orthorhombic), Point 2 and 3 phases are Al 11 Mn 4 (triclinic) phase, point 4 phase is Fe 2 Al 5 (η, orthorhombic) phase, and point 5 phase is FeAl 3 (θ, monoclinic) phase.Figure 9 shows the result of EDS analysis of the reaction zone Figure 6d in steel, and Table 3 shows the result of EDS analysis of the composition of each phase.The lower left part where Al was not detected was steel.On the top of the steel, the reactive layer was much thinner than the other, and as a result the intensity detected during the EDS was too low and not measured.In regions 6 and 7, the concentration of Al was high, but the concentrations of Mn, Fe, and C were low.Thus, it is a reaction layer.In the band with a high concentration of C in Figure 9d as shown in Table 3, Al was detected in addition to Fe, and this band was expected to be composed of Fe, Al, and C.However, the phases of regions 9 and 10 were expected to be steel structures because almost no Al was detected, yet the concentration of C was different.Figure 9 shows the result of EDS analysis of the reaction zone Figure 6d in steel, and Table 3 shows the result of EDS analysis of the composition of each phase.The lower left part where Al was not detected was steel.On the top of the steel, the reactive layer was much thinner than the other, and as a result the intensity detected during the EDS was too low and not measured.In regions 6 and 7, the concentration of Al was high, but the concentrations of Mn, Fe, and C were low.Thus, it is a reaction layer.In the band with a high concentration of C in Figure 9d as shown in Table 3, Al was detected in addition to Fe, and this band was expected to be composed of Fe, Al, and C.However, the phases of regions 9 and 10 were expected to be steel structures because almost no Al was detected, yet the concentration of C was different.
Figure 10 shows the results of the diffraction pattern analysis based on the diffraction pattern obtained in the area indicated by 6-10 in Figure 9 and the composition analysis result (Table 3).The phases of 6 and 7 are Fe 2 Al 5 (η, orthorhombic) phase in the same pattern.The phases 9 and 10 are α-Fe (ferrite) and Fe 3 C (cementite), which are steel structures.On the other hand, Fe 3 AlC (κ, Cubic) formed as a band-like phase at the interface between the steel and the reaction layer.and as a result the intensity detected during the EDS was too low and not measured.In regions 6 and 7, the concentration of Al was high, but the concentrations of Mn, Fe, and C were low.Thus, it is a reaction layer.In the band with a high concentration of C in Figure 9d as shown in Table 3, Al was detected in addition to Fe, and this band was expected to be composed of Fe, Al, and C.However, the phases of regions 9 and 10 were expected to be steel structures because almost no Al was detected, yet the concentration of C was different.Figure 10 shows the results of the diffraction pattern analysis based on the diffraction pattern obtained in the area indicated by 6-10 in Figure 9 and the composition analysis result (Table 3).The phases of 6 and 7 are Fe2Al5 (η, orthorhombic) phase in the same pattern.The phases 9 and 10 are α-Fe (ferrite) and Fe3C (cementite), which are steel structures.On the other hand, Fe3AlC (κ, Cubic) formed as a band-like phase at the interface between the steel and the reaction layer.The microstructure of the coating layer was classified into a reaction layer (reactive phase) formed while maintaining the dipping temperature and a solidified structure formed during cooling of the coating liquid.Figure 11 shows the microscopic optical structure of the Al-7 wt.% Mn cast alloy, which consists of matrix Al and Al6Mn.The microstructure of the coating layer was classified into a reaction layer (reactive phase) formed while maintaining the dipping temperature and a solidified structure formed during cooling of the coating liquid.Figure 11 shows the microscopic optical structure of the Al-7 wt.% Mn cast alloy, which consists of matrix Al and Al 6 Mn. Figure 12 shows the EBSD analysis of the reaction layer (also shown in Figures 1c, 4a, and 5a) formed at the initial coating process, 40 s. Figure 12a is an IQ (image quality) map; Figure 12b   The microstructure arrangement from the coating layer toward the steel is, in order, Al + Al6Mn (coated alloy structure) → Al11Mn4 → FeAl3 (θ) → Fe2Al5 (η) → Fe3AlC (κ) → αFe (steel).In addition, a Fe3O4 oxide layer formed on the surface of the steel, but Al2O3 formed around the reaction layer.
Compared with the sample maintained for 120 s the reaction layer formed over the entire surface with an initially formed reaction layer appearing after 40 s. Figure 13 shows the SEM microstructure and EPMA analysis of the area denoted as γ in Figure 1f, which was hot-dipped for 120 s. (a-e) show the solidification layer near the liquid phase and reaction layer, and (f-k) show the analysis of the reaction layer near the base material.Table 4 shows the results of the EDS point analysis for areas 1-8 in Figure 13a,f.Figure 14 shows the EBSD analysis of this area.Compared with the solidification layer that formed on the surface of the steel, the reaction layer after 120 s (Figures 13 and 14) and at the initial 40 s (Figures 3 and 10), the shape and distribution are similar.From the EDS, EPMA, and EBSD analysis results, the solidification layer consisted of Al (1) and Al6Mn (2) on the surface of the steel, the reaction layer was composed of Al11Mn4 (3) below Al6Mn (2).FeAl3 (θ) phase ( 4) is formed on the Al11Mn4 (3) and Fe2Al5 (5) interfaces.The Fe2Al5 (η) phase (5, 6) grew on the inside of the steel, and Fe3AlC (7) formed at the interface between the steel (8) and the η (6). Figure 12 shows the EBSD analysis of the reaction layer (also shown in Figures 1b and 6a) formed at the initial coating process, 40 s. Figure 12a is an IQ (image quality) map; Figure 12b  Figure 12 shows the EBSD analysis of the reaction layer (also shown in Figures 1c, 4a, and 5a) formed at the initial coating process, 40 s. Figure 12a is an IQ (image quality) map; Figure 12b   The microstructure arrangement from the coating layer toward the steel is, in order, Al + Al6Mn (coated alloy structure) → Al11Mn4 → FeAl3 (θ) → Fe2Al5 (η) → Fe3AlC (κ) → αFe (steel).In addition, a Fe3O4 oxide layer formed on the surface of the steel, but Al2O3 formed around the reaction layer.
Compared with the sample maintained for 120 s the reaction layer formed over the entire surface with an initially formed reaction layer appearing after 40 s. Figure 13 shows the SEM microstructure and EPMA analysis of the area denoted as γ in Figure 1f, which was hot-dipped for 120 s. (a-e) show the solidification layer near the liquid phase and reaction layer, and (f-k) show the analysis of the reaction layer near the base material.Table 4 shows the results of the EDS point analysis for areas 1-8 in Figure 13a,f.Figure 14 shows the EBSD analysis of this area.Compared with the solidification layer that formed on the surface of the steel, the reaction layer after 120 s (Figures 13 and 14) and at the initial 40 s (Figures 3 and 10), the shape and distribution are similar.From the EDS, EPMA, and EBSD analysis results, the solidification layer consisted of Al (1) and Al6Mn (2) on the surface of the steel, the reaction layer was composed of Al11Mn4 (3) below Al6Mn (2).FeAl3 (θ) phase ( 4) is formed on the Al11Mn4 (3) and Fe2Al5 (5) interfaces.The Fe2Al5 (η) phase (5, 6) grew on the inside of the steel, and Fe3AlC (7) formed at the interface between the steel (8) and the η (6).Compared with the sample maintained for 120 s the reaction layer formed over the entire surface with an initially formed reaction layer appearing after 40 s. Figure 13 shows the SEM microstructure and EPMA analysis of the area denoted as γ in Figure 1f, which was hot-dipped for 120 s. (a-e) show the solidification layer near the liquid phase and reaction layer, and (f-k) show the analysis of the reaction layer near the base material.Table 4 shows the results of the EDS point analysis for areas 1-8 in Figure 13a,f.Figure 14 shows the EBSD analysis of this area.Compared with the solidification layer that formed on the surface of the steel, the reaction layer after 120 s (Figures 13 and 14) and at the initial 40 s (Figures 3 and 10), the shape and distribution are similar.From the EDS, EPMA, and EBSD analysis results, the solidification layer consisted of Al (1) and Al 6 Mn (2) on the surface of the steel, the   Consequently, the same microstructure arrangement formed initially (40 s).The arrangement from the coating layer toward the steel was, in order, Al + Al6Mn (coated alloy structure) → Al11Mn4 → FeAl3 (θ) → Fe2Al5 (η) → Fe3AlC (κ) → αFe (steel).The η phase dipped for 40 s (see Figure 11) mainly grew toward the liquid phase, whereas the specimen dipped for 120 s (Figure 14) grew very deep into the steel interior.The growth direction of the η phase was the same in the [001] direction.

Discussion
Analyzing the reaction layer microstructure formed during initial hot dipping (within 40 s), the formation mechanism of the reaction layer its phase are as follows.The Fe3O4 oxide layer is formed on the surface of the steel when the steel sheet is dipped in the Al-7 wt.% Mn molten alloy.However, the Fe3O4 oxide layer is formed on the surface of the steel, but since Al2O3 is formed around the reaction layer, Fe3O4 is destroyed as it reacts with Al-7 wt.% Mn molten alloy and is substituted with   Consequently, the same microstructure arrangement formed initially (40 s).The arrangement from the coating layer toward the steel was, in order, Al + Al6Mn (coated alloy structure) → Al11Mn4 → FeAl3 (θ) → Fe2Al5 (η) → Fe3AlC (κ) → αFe (steel).The η phase dipped for 40 s (see Figure 11) mainly grew toward the liquid phase, whereas the specimen dipped for 120 s (Figure 14) grew very deep into the steel interior.The growth direction of the η phase was the same in the [001] direction.Consequently, the same microstructure arrangement formed initially (40 s).The arrangement from the coating layer toward the steel was, in order, Al + Al 6 Mn (coated alloy structure) → Al 11 Mn 4 → FeAl 3 (θ) → Fe 2 Al 5 (η) → Fe 3 AlC (κ) → αFe (steel).The η phase dipped for 40 s (see Figure 11) mainly grew toward the liquid phase, whereas the specimen dipped for 120 s (Figure 14) grew very deep into the steel interior.The growth direction of the η phase was the same in the [001] direction.

Discussion
Analyzing the reaction layer microstructure formed during initial hot dipping (within 40 s), the formation mechanism of the reaction layer its phase are as follows.The Fe 3 O 4 oxide layer is formed on the surface of the steel when the steel sheet is dipped in the Al-7 wt.% Mn molten alloy.However, the Fe 3 O 4 oxide layer is formed on the surface of the steel, but since Al 2 O 3 is formed around the reaction layer, Fe 3 O 4 is destroyed as it reacts with Al-7 wt.% Mn molten alloy and is substituted with Al 2 O 3 .It is well known that Al 2 O 3 is a more stable oxide than Fe 3 O 4 , and therefore a substitution reaction occurred.When the Fe 3 O 4 was destroyed by the substitution reaction, the melt flowed into the solid, reacting with the steel to form a reaction layer.The reaction layer formed mainly in the liquid phase, with a thickness of about 1 µm or less inside the steel.The microstructure of the coating layer formed at the initial stage (40 s) of the coating process as shown in Figure 12 was in the order of Al + Al 6 Mn (coated alloy structure) → Al 11 Mn 4 → FeAl 3 (θ) → Fe 2 Al 5 (η) → Fe 3 AlC → αFe(steel).The microstructure of the coated layer can be classified into a solidification structure in which Al-7 wt.% Mn melts are formed during cooling and a reaction layer is formed while holding.As shown in Figure 11, the solidification structure of Al-7 wt.% Mn is composed of Al with Al 6 Mn as the matrix.Therefore, Al and Al 6 Mn were solidified structures and the remaining Al 11 Mn 4 , FeAl 3 (θ), Fe 2 Al 5 (η) are the reaction phases formed during the coating process.
Reaction layers or reaction phases have been studied for a long time in the diffusion reaction or coating process of Fe and pure Al [22][23][24], Fe and Al alloys [25,26], steel and pure Al [27][28][29][30][31], and steel and Al alloys [32,33].The reaction layers formed in these studies were classified into the following three types: (1) Al/FeAl 3 (θ)/Fe 2 Al 5 (η)/Fe [22][23][24]33] where the intermetallic layer is composed of an outer minor FeAl 3 (θ) layer and an inner major Fe 2 Al 5 (η) layer; (2) Al/Fe 4 Al 13 (θ)/Fe 2 Al 5 (η)/Fe [31,32] where the intermetallic layer is composed of an outer minor Fe 4 Al 13 layer and an inner major Fe 2 Al 5 layer; and (3) Al/Fe 2 Al 5 /Fe [33][34][35] where the only reaction product was Fe 2 Al 5 .Most studies have reported that θ phase (Fe 4 Al 13 or FeAl 3 ) is formed on Al side and η phase is formed on Fe side or steel side.There is no study on the reaction of Al-Mn coated layer in these studies.In the Al-Mn coating process of this study, Al 11 Mn 4 is formed before θ phase formation.
As seen in Figure 12b, all reaction phases except for Fe 3 AlC (κ) phase formed on the liquid side of the steel surface, the elements such as Fe dissolved and formed in the Al-7 wt.% Mn molten alloy.Thus, an (Al-7 wt.% Mn)-Fe quasi-binary phase diagram was calculated using Thermo-Calc TM (Ver.6.0,Stockholm, Sweden) to analyze phase transformation during and while maintaining the dipping temperature.Figure 15 shows the phase diagram.When Fe is added into the melts at a dipping temperature of 773 • C at 0.6 wt.% or more, solid Al 11 Mn 4 (A) formed within the liquid phase.However, the phase in the coating layer composed of Al and Mn had two phases, Al 11 Mn 4 and Al 6 Mn (B).In the phase diagram, Al 6 Mn does not exist at the dipping temperature, but does exist below temperatures of 725 • C. In particular, Al 6 Mn can form on Al 11 Mn 4 or can form in the coating layer.With this distribution, the Al 11 Mn 4 formed as the Fe concentration in the melts increased while maintaining the dipping temperature.And Al 6 Mn can crystallize in liquid phase during cooling or form on Al 11 Mn 4 which is nucleation site during cooling.Therefore, Al 11 Mn 4 and Al 6 Mn may be connected to each other.The reaction layer, except for the coating structure (Al + Al 6 Mn), was formed as Al 11 Mn 4 → FeAl 3 (θ) → Fe 2 Al 5 (η) → Fe 3 AlC (κ) in the coating process, maintaining the dipping temperature.Al 11 Mn 4 formed by reacting with the liquid phase, but the remaining FeAl 3 (θ), Fe 2 Al 5 (η), and Fe 3 AlC (κ) did not react directly with the liquid phase but instead would be formed by a solid-phase diffusion mechanism between Al 11 Mn 4 and steel.
Figure 16 is a schematic diagram of the processes formed by inter-diffusion in the order of phases.At interface Al 11 Mn 4 /liquid phase (L) formed first, and when Fe diffused toward the liquid phase and Al diffused toward the solid phase Al 11 Mn 4 , the Al 11 Mn 4 phase grew.At the Al 11 Mn 4 phase/steel interface, Al diffused toward the steel, and when Fe diffused into the A phase, the Al concentration decreased in the region close to the steel.On the other hand, since the Fe concentration increased, the Fe-rich phase formed.The θ phase (FeAl 3 ) formed when the Fe concentration increased from 0 to 4 wt.% or more in the (Al-7 wt.% Mn)-x wt.% Fe quasi-binary diagram in Figure 15.
11 maintaining the dipping temperature.And Al6Mn can crystallize in liquid phase during cooling or form on Al11Mn4 which is nucleation site during cooling.Therefore, Al11Mn4 and Al6Mn may be connected to each other.The reaction layer, except for the coating structure (Al + Al6Mn), was formed as Al11Mn4 → FeAl3 (θ) → Fe2Al5 (η) → Fe3AlC (κ) in the coating process, maintaining the dipping temperature.Al11Mn4 formed by reacting with the liquid phase, but the remaining FeAl3 (θ), Fe2Al5 (η), and Fe3AlC (κ) did not react directly with the liquid phase but instead would be formed by a solid-phase diffusion mechanism between Al11Mn4 and steel.However, in Tables 2 and 3, the actual Mn concentrations of the θ and η phases are lower than 7 wt.%.That is, the amounts of Mn of θ and η phases were 6.32 wt.% and 0.22 wt.%, respectively, and the Mn content of η was much lower than the Mn concentration in 1.13 wt.% of steel.Therefore, phase transformation by inter-diffusion while considering Mn content can be interpreted as Al-Mn-Fe ternary phase diagram.Figure 17 shows the Al-Mn-Fe ternary phase diagram calculated by Thermo-Calc TM .In the phase diagram, when the amount of Mn is 5% or less, Al11Mn4 did not form and instead the θ phase formed.Also, as Fe increased, it transformed from FeAl3 (θ) to Fe2Al5 (η).As a result, the amount of Fe increased by diffusion at the A11Mn4 phase/steel interface, and when the amount of Al decreased, θ would form.At the θ phase/steel interface (c), when the amount of Al decreased and the amount of Fe further increased, η (Fe2Al5) would form as shown in the phase diagram.As shown in Figure 14c, large columnar η grains grow along the c-axis of the phase.Morphology of the η phase has been studied and explained as a result of the 30% vacancy rate in the c-axis of the crystal structure, which can be as a rapid diffusion tunnel.It causes η phase to be oriented by the fixed c-axis of the crystal structure, and grows preferentially along the diffusion direction during hot-dipping.However, in Tables 2 and 3, the actual Mn concentrations of the θ and η phases are lower than 7 wt.%.That is, the amounts of Mn of θ and η phases were 6.32 wt.% and 0.22 wt.%, respectively, and the Mn content of η was much lower than the Mn concentration in 1.13 wt.% of steel.Therefore, phase transformation by inter-diffusion while considering Mn content can be interpreted as Al-Mn-Fe ternary phase diagram.Figure 17 shows the Al-Mn-Fe ternary phase diagram calculated by Thermo-Calc TM .In the phase diagram, when the amount of Mn is 5% or less, Al 11 Mn 4 did not form and instead the θ phase formed.Also, as Fe increased, it transformed from FeAl 3 (θ) to Fe 2 Al 5 (η).As a result, the amount of Fe increased by diffusion at the A 11 Mn 4 phase/steel interface, and when the amount of Al decreased, θ would form.At the θ phase/steel interface (c), when the amount of Al decreased and the amount of Fe further increased, η (Fe 2 Al 5 ) would form as shown in the phase diagram.As shown in Figure 14c, large columnar η grains grow along the c-axis of the phase.Morphology of the η phase has been studied and explained as a result of the 30% vacancy rate in the c-axis of the crystal structure, which can be as a rapid diffusion tunnel.It causes η phase to be oriented by the fixed c-axis of the crystal structure, and grows preferentially along the diffusion direction during hot-dipping.in Figure 14c, large columnar η grains grow along the c-axis of the phase.Morphology of the η phase has been studied and explained as a result of the 30% vacancy rate in the c-axis of the crystal structure, which can be as a rapid diffusion tunnel.It causes η phase to be oriented by the fixed c-axis of the crystal structure, and grows preferentially along the diffusion direction during hot-dipping.The formation mechanism of the κ (Fe3AlC) phase at the interface between the steel and the η phase is as follows.As shown in Figure 3f, the concentration of C in the region transformed from α Fe (Steel) to η phase is lower than that of the steel.This phenomenon coincides with the maximum solubility of 0.07 wt.% of η phase (see Table 5) calculating by using Thermo-Calc TM , and the solubility is very low.Therefore, C in the region transformed to η phase will be diffused and discharged toward The formation mechanism of the κ (Fe 3 AlC) phase at the interface between the steel and the η phase is as follows.As shown in Figure 3f, the concentration of C in the region transformed from α Fe (Steel) to η phase is lower than that of the steel.This phenomenon coincides with the maximum solubility of 0.07 wt.% of η phase (see Table 5) calculating by using Thermo-Calc TM , and the solubility is very low.Therefore, C in the region transformed to η phase will be diffused and discharged toward the steel that can be soluted by increasing the concentration of Al and C in the steel interface.The content of C in the Fe 3 AlC (κ) is 5.05 wt.%, and the average concentration of steel is 0.23 wt.%.Since the amount of C for forming this phase is absolutely insufficient, the thickness is formed in nm.

Conclusions
Nucleation and growth of the reaction layer formed on boron steel hot-dipped in an Al-7 wt.% Mn molten alloy at 773 • C for various times was investigated.

1.
The layer formed on the surface of the steel was classified into a reaction layer formed at the coating temperature and a solidification layer in which the liquid phase solidified during cooling.The solidification layer was composed of Al and Al 6 Mn phases, and the reaction layer was composed of Al 11 Mn 4 , FeAl 3 (θ) and Fe 2 Al 5 (η) phases.In particular, the η phase grew long inside the steel.The Fe 3 AlC (κ) phase was formed at the interface between the η phase and the steel in a very thin band of several nanometers.

2.
The solid phase Al 11 Mn 4 was eluted when Fe was dissolved in Al-Mn molten alloy at 773 • C at 0.6 wt.% or more from the (Al-7 wt.% Mn)-x wt.% Fe quasi phase diagram (Thermo-Calc TM ).At the interface between Al 11 Mn 4 phase and the steel, as Fe diffused toward the Al 11 Mn 4 and Al diffused toward the steel, at the interface, θ phase was formed as the Fe concentration increased and the Al concentration decreased.In addition, η phase was formed due to inter-diffusion of Fe, Al between the θ phase and the steel.In other words, Al 11 Mn 4 → θ → η was formed in which the Fe content was higher toward the steel by inter-diffusion in the solid phase state.

3.
The η phase formed by the reaction with the liquid phase initially had a fine polygonal structure.However, it formed inside the steel, growing to the order of Al 11 Mn 4 → θ → η in the liquid phase, and grew in a long columnar form because the Al diffusion was along the c-axis and the <100> direction was fast, as already known.4.
Fe 3 AlC (κ) phase was formed as the concentration of Al and C is increasing at the steel interface because the η phase diffused C toward steel due to the fact that η hardly contains C.

Figure 4 .
Figure 4. X-ray diffraction (XRD) analysis of the oxide layer formed on the steel surface upon initial dipping.

Figure 5 .
Figure 5. Change in formation ratio of reaction layer on boron steel with dipping time (s).

Figure 4 .
Figure 4. X-ray diffraction (XRD) analysis of the oxide layer formed on the steel surface upon initial dipping.

Figure 4 .
Figure 4. X-ray diffraction (XRD) analysis of the oxide layer formed on the steel surface upon initial dipping.

Figure 5 .
Figure 5. Change in formation ratio of reaction layer on boron steel with dipping time (s).

Figure 5 .
Figure 5. Change in formation ratio of reaction layer on boron steel with dipping time (s).

Figure 6 . 6 Figure 6 .
Figure 6.(a) Back-scattered electron (BSE) structure enlarged of area denoted as γ in Figure 3a,b TEM specimen took at area denoted as γ in (a) using FIB; (c,d) HAADF image of area 1 and area 2 in (b), respectively.
Figure12shows the EBSD analysis of the reaction layer (also shown in Figures1c, 4a, and 5a) formed at the initial coating process, 40 s.Figure12ais an IQ (image quality) map; Figure12bis a phase map; and Figure 12c is an IPF (inverse pole figure).The phases present in the EBSD are the same as identified in the TEM observation; Figure 12d is a schematic diagram of a reaction layer formed on a steel surface based on the above described TEM and EBSD analysis results.

Figure 12 .
Figure 12. Results of EBSD analysis hot-dipping for 40 s.(a) Image Quality (IQ) map; (b) Phase map; (c) Inversed Pole Figure (IPF) map and (d) the schematic diagram of the reaction layer nucleated in liquid (40 s).

Figure 11 .
Figure12shows the EBSD analysis of the reaction layer (also shown in Figures1b and 6a) formed at the initial coating process, 40 s.Figure12ais an IQ (image quality) map; Figure12bis a phase map; and Figure 12c is an IPF (inverse pole figure).The phases present in the EBSD are the same as identified in the TEM observation; Figure 12d is a schematic diagram of a reaction layer formed on a steel surface based on the above described TEM and EBSD analysis results.
Figure12shows the EBSD analysis of the reaction layer (also shown in Figures1c, 4a, and 5a) formed at the initial coating process, 40 s.Figure12ais an IQ (image quality) map; Figure12bis a phase map; and Figure 12c is an IPF (inverse pole figure).The phases present in the EBSD are the same as identified in the TEM observation; Figure 12d is a schematic diagram of a reaction layer formed on a steel surface based on the above described TEM and EBSD analysis results.

Figure 12 .
Figure 12. Results of EBSD analysis hot-dipping for 40 s.(a) Image Quality (IQ) map; (b) Phase map; (c) Inversed Pole Figure (IPF) map and (d) the schematic diagram of the reaction layer nucleated in the liquid (40 s).

Figure 12 .
Figure 12. Results of EBSD analysis hot-dipping for 40 s.(a) Image Quality (IQ) map; (b) Phase map; (c) Inversed Pole Figure (IPF) map and (d) the schematic diagram of the reaction layer nucleated in the liquid (40 s).

Table 1 .
Chemical composition of Boron steel.