2021 Volume 61 Issue 1 Pages 335-342
In a molten zinc bath of a continuous galvanizing line, top dross particles crystallize as Fe2Al5, an intermetallic compound containing Zn. These particles readily adhere to the steel sheets, causing surface defects. Therefore, controlling the top dross particles is a key issue. The present study focused on determining the facet plane of top dross particles via three-dimensional analysis of morphology by serial sectioning and electron back scattering diffraction (EBSD). Furthermore, the crystallographic plane of the cleavage fracture surface of the top dross was determined by EBSD, after a cleavage crack was introduced by Vickers hardness indentation. The facet planes of the top dross consist of two planes of (001), four planes of {110}, and eight planes of {111}. In addition, the top dross particles grow fastest in the [001] direction. Consequently, the top dross particle was concluded to possess a polyhedron structure with 14 facet planes. Finally, the cleavage fracture surface of the main crack in the top dross is the (100) plane.
In a molten zinc bath of the Sendzimir-type continuous galvanizing line (CGL), Fe dissolved from the steel sheet reacts with Zn and Al to form crystallized products (dross) with a high melting temperature. These dross particles readily adhere to the steel sheets, causing surface defects. Despite the industrial importance of controlling dross particles to allow stable manufacture of high-quality galvanized steel sheets, much of the basic information on the details of dross particles, e.g., the three-dimensional shape and fracture morphology, remain unknown. Furthermore, various kinds of apparatuses used in a molten zinc bath should ideally not result in dross particles adhering to the steel sheets.1,2,3) To achieve this outcome, it is important to understand the exact three-dimensional shape of dross particles to design apparatus with this attribute. Moreover, understanding the detailed fracture morphology of dross particles is important to elucidate the mechanism underlying the formation of surface defects caused by dross particles. The top dross particles, that float on a molten zinc bath and are the majority of dross particles, crystallize as Fe2Al5, an intermetallic compound containing Zn (10.6–11.2 at%).4,5) Arioka et al.6) developed a method to extract the top dross particles from Zn and reported that they have facet planes and a polyhedron structure with 14 facet planes. In addition, our previous paper7) reported the structural and mechanical characteristics of the top dross particles and demonstrated the following: top dross particles containing Zn have a faceted interface with molten Zn. Moreover, the introduction of an indent by application of a Vickers hardness test on top dross particles results in cracks, depending on crystal orientation. Although Arioka et al.6) reported that the three-dimensional shape of top dross particles is a polyhedron with 14 facet planes, the concrete facet planes have never been reported. In addition, the growth rate of the crystal, which crystallizes from a liquid phase, depends on crystal orientation, leading to the anisotropic three-dimensional shape.8,9) However, the anisotropy of the three-dimensional shape has not been reported. Finally, our previous paper7) clarified that cracks introduced to the top dross particles can extend and widen, depending on crystal orientation. However, neither the fracture morphology nor the crystallographic characterization of the crack of the top dross particles have been reported.
In the present study, we focused first on the determination of the concrete facet planes to crystallographically elucidate the reason why the three-dimensional shape of the top dross particles becomes a polyhedron with 14 facet planes. Consequently, we determined the concrete facet planes by determining the exact three-dimensional placement of each facet plane from the perspective of the top surface. In addition, the anisotropy of the three-dimensional shape was clarified. Three-dimensional shape analysis by serial sectioning using ion milling (IM)10) and crystal orientation analysis by scanning electron microscopy-electron back scattering diffraction (SEM-EBSD) were utilized in this study. Second, the cleavage fracture plane of the top dross particles was determined after introducing an indent following a Vickers hardness test on top dross particles and crystal orientation of the cleavage fracture plane was analyzed by SEM-EBSD.
Top dross particles were prepared using the same method as described in our previous paper.7) Approximately 10 kg of Zn (99.99 wt%) was dissolved in a graphite crucible (top diameter: 135 mm, height: 100 mm, bottom diameter: 85 mm) and held in a bath at a temperature of 460°C. Pure Al balls with 0.5% of the total weight with a diameter of approximately 20 mm were added to this Zn bath. After confirming the dissolution of Al, a sample containing 0.5% of the total weight of Fe (extremely low carbon steel mass-produced as galvannealed steel sheet) finely cut into 0.5 mm squares was added. Table 1 shows the chemical compositions of extremely low carbon steel. The above conditions were maintained for one day to prepare top dross particles. The concentration of the liquid phase in this Zn bath after 1 day was Al: 0.140 wt%, Fe: 0.32 wt%, which was in good agreement with the triple point of the phase diagram by Tang.5) Therefore, it can be concluded that Al and Fe were supersaturated within the Zn bath and the top dross particles were easily crystallized. Moreover, since the state of the top dross particles changes very quickly during cooling, close attention was paid to the method of freezing the samples. First, using a glass pipette, the top dross was collected via suction from the surface of the Zn bath, in a laboratory. Thereafter, the collected melt was dropped into a copper mold and rapidly cooled. The region of the contact surface between the droplet and the copper mold, which experienced the most rapid cooling, was ground and subjected to metallographic observations.
| wt% | ||||
|---|---|---|---|---|
| C | Si | Mn | P | S |
| ≦0.01 | 0.01 | 0.14 | 0.01 | 0.01 |
In the present study, in analyzing the facet planes and the anisotropy of the three-dimensional shape of the top dross particles, the basic method was to employ serial sectioning because this does not fundamentally affect the shape of the top dross particles. However, for purposes of comparison, the method of extracting the top dross particles, which preferentially dissolves Zn, was also adopted, followed by SEM observation. In the extraction of the top dross particles, the same method reported by Arioka et al.6) was used (Fig. 1). A top dross (sample) was placed on a glass petri dish, and fuming nitric acid was added dropwise from a 25 mL burette. Only the Zn on the sample surface was preferentially dissolved, and finally, the top dross particle was separated. After separation, the top dross particles were quickly filtered and washed with water. For filtration, an omnipore membrane filter (1.0 μm JA) was used. The top dross particles thus extracted were observed by SEM, and their three-dimensional shape was determined.
Extraction method for top dross.
To analyze the facet plane and the anisotropy of the three-dimensional shape of top dross particles, the serial sectioning method using IM,10) with repeated polishing of the target sample and SEM-EBSD measurement, was used. Figure 2 is a schematic drawing of the three-dimensional shape analysis method using serial sectioning. In addition, a hole was made vertically in the sample, and the accurate alignment was adjusted using the hole as a reference, and then the SEM-EBSD measurement was performed to prevent displacement. As described below, this method is excellent for the purpose of accurately analyzing the facet plane. That is, when trace analyses of the facet planes are performed by the ordinary SEM-EBSD measurement without grinding the sample, the facet plane of top dross may enter from the top surface diagonally, leading to EBSD measurement as shown in Fig. 3(a). In contrast, when the shape and position does not change with grinding, the facet plane of the top dross can be judged to be perpendicular to the observation plane (Fig. 3(b)). In this study, after confirming that the facet planes were perpendicular to the observation plane, trace analysis was conducted to determine the concrete facet planes. In addition, the anisotropy of the three-dimensional shape of the top dross particles were evaluated by determining the length of the top dross, which presumably grew onto a specific crystal plane, based on the results of the serial sectioning. Here, length refers to the length in the depth direction of sectioning. For quantification of the length, the required number of slices until individual top dross particles were completely cut, was used. In performing serial sectioning, the sectioning method is a key issue. We compared several mechanical polishing methods, including the focused ion beam (FIB) method11) and the ultra-microtome method. Finally, the IM method,10) a general-purpose sectioning method, was selected as meeting the criteria of the processing area, the hardness of the material to be processed, and the positional accuracy of the three-dimensional processing. In the IM method, sectioning was performed using an Ar ion beam.10) Although the sectioning accuracy of IM is approximately several μm and not extremely good, it has the following advantages: the sample is hardly damaged, and simultaneous equal sectioning of a sample having different hardness is possible. The maximum processing area is 1000 μm × 500 μm, and we considered that this method was optimal for the purpose of the present study, which targets top dross particles with a size exceeding 100 μm.
Schematic drawing of serial sectioning process.
Schematic illustration showing cross sectional view of top dross. (a) Facet planes of top dross are not perpendicular to the surface. (b) Facet planes of the top dross are perpendicular to the surface.
SEM-EBSD was used for the crystal orientation analysis of the top dross particles. SEM was operated with an accelerating voltage of 25 kV, and the crystal orientation analysis of the top dross particles was performed using the Kikuchi pattern. The software OIM Analysis ver.7 of TSL Solutions was used for the analysis. X-ray profiles of Fe2Al5 have been analyzed based on either the oC16 structure12) or the oC24 structure13) in the past. Recently, the crystal structure analysis of Fe2Al5 has been actively performed, and it has been reported14,15,16) that the partial occupation site of Al atoms along the c-axis is ordered and there exist multiple superlattice structures as a parent structure of the conventional oC24 structure, particularly in the low temperature range of the Al excess or the Fe excess compositions. Therefore, in our previous report,7) the exact lattice constant of the top dross (orthorhombic: a = 7.61 Å b = 6.48 Å c = 4.23 Å) was determined based on both the conventional space group (Cmcm)13) and the peak position obtained from X-ray diffraction using the Pawlley method.17) In this report, the Kikuchi pattern obtained by EBSD was fitted using these newly determined lattice constants.
2.4. Method of Introducing a Crack into Top Dross ParticlesThe same method reported by Tsukahara et al.18) was used to introduce cracks into the top dross. That is, the Vickers hardness indent was pushed into the top dross to introduce cracks. When a load was small (for example, 98 mN), cracks did not open. Therefore, as described in our previous paper,7) the Vickers hardness indent was pushed into the top dross with a load of 588 mN, after confirming that stable Palmqvist-type cracks were generated.
First, Zn was dissolved preferentially using fuming nitric acid. Immediately thereafter, washing with water and filtration were undertaken to extract only the top dross particles. Figure 4 shows the SEM observation of the extracted top dross particles. When the amount of fuming nitric acid was excessively large relative to the amount of Zn in the sample, the shape of the extracted top dross particle collapsed (Fig. 4(a)). In contrast, when the amount of fuming nitric acid was too small relative to the amount of Zn in the sample, Zn remained undissolved and the top dross could not be separately extracted. Therefore, the optimal amount of fuming nitric acid was investigated, and found to be about 15 mL of fuming nitric acid for 3 g of sample. As a result, several top dross particles with clear facet planes (Fig. 4(b)) were extracted. However, the shape of many top dross particles collapsed, even under the optimized condition, (Fig. 4(a)). This suggests that it is not easy to extract only top dross particles using fuming nitric acid and top dross particles themselves may be dissolved, together with Zn.
SEM images of extracted top dross particle. (a) Partly dissolved shape and (b) normal shape.
To determine the facet plane of the top dross particles, three-dimensional shape analysis by simultaneous serial sectioning and crystal orientation analysis by SEM-EBSD was applied. Representative results are shown in Figs. 5, 6, 7, 8, 9, 10. Figure 5 shows the results of an analysis by EBSD just below the surface of the solidified sample. Image quality (IQ) maps of the top dross particles and the solidified structure of the liquid phase of Zn are presented. Images of both the top dross particles and the solidified structure of the liquid phase of Zn are very clear and indicate high crystallinity. Figure 5(a) is an IQ map of the initial surface before grinding, while Figs. 5(b)–5(g) show IQ maps of top dross particles ground from the top surface every 15 μm. Figure 6 shows the inverse pole figure (IPF) maps of the ND of top dross particles in the same field as Fig. 5. Figures 7(a)–7(d) show the pink colored top dross particles having an orientation within 20° from ND//[001], ranging in depth from 45 to 90 μm from the top surface (Figs. 6(d)–6(g)). Figures 8(a)–8(d) are (001) pole figures of the top dross particles colored in Figs. 7(a)–7(d). Here, the pole figures were prepared using ND-A1-A2 as a sample coordinate axis, where the A1 and A2 axes indicate the vertical and horizontal directions of the sample, respectively. Relatively large top dross particles exceeding 50 μm within 20° from ND//[001] exhibited a square shape and clear facet planes (Figs. 7(a)–7(d)). Comparing images of Figs. 7(a)–7(d), the size and shape of the top dross particles indicated by the arrows hardly changed regardless of the observation depth, measured from the top surface. Furthermore, as designated by the arrows in the pole figures in Figs. 8(a)–8(d), the pole of the top dross particles changed less than 1°, even after grinding. Thus, the top dross particle is a single crystal, the sample was ground parallel, and the facet planes are suggested to exist perpendicular to the observation plane. By selecting such a representative top dross particle, whose shape hardly changed in the depth direction as indicated by the arrow in Fig. 7, detailed trace analysis of the facet planes was conducted, and the results are exhibited in Figs. 11(a), 11(b). In reference to the crystal model illustrated in Fig. 11(a), the facet planes of the top dross particle observed was found to be parallel to the planes containing the diagonal line of the crystal model. The plane of {110} is therefore concluded to be the facet plane. Moreover, if the facet planes of the top dross particle exist perpendicular to the observation plane, the exact index of the facet plane was obtained by trace analysis, and the result is plotted in an IPF of Fig. 11(b). Figure 11(b) clearly shows that all four facet planes of the top dross particle shown in Fig. 11(a) were parallel to the {110} plane, within an error of 15°. The error is presumed to be because the facet plane was not completely vertical to the observation plane. Figure 9 shows another relatively large top dross particles with blue color within 20° from the direction perpendicular to the (110) plane in Figs. 6(a)–6(c) in a depth range from the initial top surface to 30 μm. Figs. 10(a)–10(c) are the (110) pole figures corresponding to each of Figs. 9(a)–9(c). The top dross particles, whose ND was within 20° from the direction perpendicular to (110) plane, exhibited a polygonal shape with facet planes (Figs. 9(a)–9(c)). Figures 9(a)–9(c) clearly show that the size and shape of the top dross particle indicated by the arrows hardly changed, even when the observation plane became deeper. Furthermore, as shown by the arrows in the pole figures in Figs. 10(a), 10(b) and 10(c), the poles of the top dross particle indicated by the arrows in Fig. 9 changed less than 1°, even after grinding. Thus, the top dross particle is a single crystal and the sample was ground in parallel. Therefore, it is suggested that the facet planes exist perpendicular to the observation plane. By selecting a representative top dross particle, whose shape hardly changed in the depth direction indicated by the arrow in Fig. 9, it was possible to undertake a detailed trace analysis. (Figs. 11(c), 11(d)). By referring to the crystal model illustrated in Fig. 11(a), the facet planes of the top dross particle observed in the field were (001), {110}, and {111}, which have low Miller indices. Moreover, by assuming that the eight facet planes of the top dross particle observed in the field of Fig. 11(c) exist perpendicular to the observation plane, it was possible to plot the exact plane indices obtained by trace analysis in the IPF of Fig. 11(d). It is also evident from Fig. 11(d) that all the facet planes of the top dross particle shown in Fig. 11(c) are any of the above-mentioned planes. Furthermore, the same was confirmed in many other large top dross particles within 20° from the direction perpendicular to (110) plane. Therefore, the facet planes of the top dross particle consist of two planes of (001), four planes of {110}, and eight planes of {111}.
SEM-EBSD image quality maps of top dross particles at (a) the initial surface, at the depth of (b) 15 μm, (c) 30 μm, (d) 45 μm, (e) 60 μm, (f) 75 μm, and (g) 90 μm from the initial surface.
EBSD inverse pole figure maps of top dross particles at (a) the initial surface, at the depth of (b) 15 μm, (c) 30 μm, (d) 45 μm, (e) 60 μm, (f) 75 μm, and (g) 90 μm from the initial surface.
Top dross particles within 20° from ND//[001] are colored together with the crystal model. Top dross particles at the depth of (a) 45 μm (b) 60 μm, (c) 75 μm, and (d) 90 μm from the initial surface.
(a) (001) pole figure of top dross particles shown in Figs. 7(a), 7(b) that in Figs. 7(b), 7(c) that in Figs. 7(c) and 7(d) that in Fig. 7(d). Arrows designate poles of top dross particle indicated by arrows in Fig. 7.
Top dross particles within 20° from the direction perpendicular to (110) plane are colored together with the crystal model. Top dross particles at (a) the initial surface, at the depth of (b) 15 μm and (c) 30 μm.
(a) (110) pole figure of top dross particles shown in Figs. 9(a), 9(b) that of Figs. 9(b) and 9(c) that of Fig. 9(c). Arrows designate poles of top dross particle indicated by arrows in Fig. 9.
(a) Trace analyses of a top dross particle with near ND//[001] showing four {110} facet planes and (b) their orientations in inverse pole figure. (c) Trace analyses of a top dross particle with normal direction perpendicular to (110) plane showing two (001) facet planes, two {110} facet planes and four {111} facet planes and (d) their orientations in inverse pole figure.
The facet planes of top dross particles were (001), {110}, and {111} (Section 3.1.2). The anisotropy of the three-dimensional shape of top dross particles was examined, taking into account (001), {110} and {111} as the facet plane, together with the (010) plane, which is not a facet plane but one of the representative low index planes. The anisotropy of the crystal shape was evaluated by determining the length of the top dross particles grown in a specific crystal direction, based on the results of the three-dimensional EBSD analysis of top dross particles, by grinding 15 times, every 15 μm. Here, length refers to the direction of depth. For the quantification of length, the required number of slices until an individual top dross particle completely disappeared was used (Fig. 12). The sectioning interval was assumed to be 15 μm, and the angle allowance of the facet plane was set to be within 30°, to increase the number of the targeted top dross particles, which was supposed to exceed 50. The arrow shown in Fig. 12 indicates the length that was most frequently present for each growth direction. In addition, in order to exclude small top dross particles that had just formed as nuclei, the dross particles that had disappeared after only one sectioning were neglected from the evaluation. By focusing on top dross particles whose observation plane had a surface close to the facet plane, the crystal length in the depth direction was compared. The top dross particles, whose observation plane was close to the (001) facet plane, appeared to have significantly longer length in the depth direction than the top dross particle with other oriented facet planes. The top dross particles with the second longest length in the depth direction have almost the (111) plane. In contrast, the top dross particles with the shortest length in the depth direction exhibited almost the (110) plane. The growth rate of a crystal, which crystallizes from a liquid phase, depends on the crystal orientation leading to an anisotropic three-dimensional shape.8,9) The length of crystallized top dross particles from liquid Zn, which is the subject of the present study, was also shown to reveal an anisotropy, and the direction perpendicular to the (001) plane was longer in the depth direction than other planes. Therefore, it is expected that the top dross particles have slender shapes with a high growth rate in the [001] direction. Furthermore, by comparing the top dross particles whose observation plane are close to the (110) plane with the ones close to the (010) plane, the top dross particles close to the (010) plane were longer in the depth direction. These results can be understood as follows. The dross particles observed from the ND//[001] direction have square shapes having the (110) plane as the facet plane (Fig. 11(a)). The length in the [010] direction, which is the diagonal of the square of the top dross particles, was longer than the length in the direction perpendicular to the (110) plane (Fig. 11(a)). This fact supports the results shown in Fig. 12 in terms of the shape anisotropy obtained from three-dimensional analysis of top dross particles.
Number – length frequency profiles of top dross particles in (a) (010), (b) (001), (c) (110) and (d) (111) planes.
The cracks in the top dross particles were crystallographically analyzed using EBSD. In our previous report,7) it was demonstrated that a brittle cracking of the top dross particle occurred when a Vickers hardness indent was applied and this might be uniform at the four corners, or might be anisotropic, To determine the crystallographic cleavage plane, an orientation analysis using EBSD of the cleavage plane generated by pushing the Vickers hardness indent from the [100], [010], and [001] directions, was undertaken (Fig. 13). When pushing the Vickers hardness indent from the [010] direction (Fig. 13(a)), clear cracks from two of the four corners of the indent were generated. As shown in the IPF map of the top dross particle (Fig. 13(d)), the cleavage fracture plane was demonstrated to be the (100) plane. Here, considering the stress state when the Vickers hardness indent is pushed in, the fractured surface of the crack is assumed to be perpendicular to the observation plane, and cleavage fracture is considered to have taken place on this plane. When the Vickers hardness indent was pushed in from the [001] direction, cracks were generated from four corners (Fig. 13(c)). However, the crack parallel to the (100) plane was the principle one and the crack parallel to the (010) plane was minor, with a relatively small crack length. In contrast, when the Vickers hardness indent was pushed in from the [100] direction (Fig. 13(b)), the crack was unclear, and a small crack was generated from any one of the four corners. Moreover, other cracks were present at locations unrelated to the corners. In this case, the cleavage fracture plane was also different from that originating from the corner. Orientation analysis of the crack generated from the corner suggests that the (001) plane may also become a cleavage fracture plane, but it does not become a main crack (Fig. 13(e)). Three samples were measured in each direction, and similar results were obtained. In addition, the relationship between the indent direction and the manner of crack opening in the present study reproduced the results of our previous report.7)
Secondary electron images showing the crack occurrence on the surface of top dross particles after applying a 588 mN Vickers hardness indentation force in the direction of (a) [010], (b) [100], and (c) [001]. (d) EBSD inverse pole figure map of (a), (e) that of (b), and (f) that of (c).
The facet planes of the top dross particle consist of two planes of (001), four planes of {110}, and eight planes of {111}. The facet planes of a relatively small top dross particle of approximately 10 μm (Fig. 14(a)) obtained by the extraction method are now discussed.
(a) SEM image of extracted top dross particles having facets, (b) conceptual figure of the vertical section B, and (c) that of the transverse section C together with the (d) crystal model.
Since the top dross particle has two (001) planes, the (001) plane is most likely to be the plane indicated by the blue arrow in Fig. 14(a). Since the top dross particle has four {110} planes, the {110} plane is most likely to be the plane indicated by the red arrow in Fig. 14(a). Furthermore, the top dross particle has eight {111} planes; therefore, the {111} plane is most likely to be the plane indicated by the green arrow in Fig. 14(a). Figure 14(d) shows a crystal model for discussion. The cross section of B is an octagon (Fig. 14(b)); here, the (001), {111}, and {110} planes are the facet planes. It is noteworthy that the {111} and {110} planes are wider than the (001) plane (Fig. 14(a)). This suggests that the {111} and {110} planes have a lower interfacial energy with the liquid phase of Zn than the (001) plane. Therefore, they are more likely to become the facet plane. Consequently, the growth rate of the (001) plane was likely to have been the highest, resulting in the <001> direction becoming the longest, as schematically shown in Fig. 14(b). This mechanism is supported by the results of the length data for each crystal orientation shown in Fig. 12. In contrast, in cross section C, the {110} plane becomes a facet plane. Therefore, the rectangle grows into a diamond shape, as shown in Fig. 14(c). The EBSD analyses (Figs. 7 and 11(a)) reveal the diamond shape. Thus, when the top dross particles grow, they tend to have two facet planes of (001), four facet planes of {110}, and eight facet planes of {111}, constructing the polyhedron structure with these 14 facet planes. Moreover, they grow fastest in the [001] direction. Figure 14(a) shows an example of a top dross particle with a relatively small size, of approximately 10 μm. A detailed study on the change in shape due to growth is required in the future. Presumably, crystal growth proceeds to minimize the total interfacial energy.
4.2. Cleavage Fracture Plane of Top Dross ParticlesThe cleavage fracture plane of top dross particles containing Zn is the (100) plane (Section 3.2). This suggests that the surface energy of the (100) plane is low and the bonding force between atoms in the [100] direction is weak. Here, our crystallographic understanding that the cleavage fracture plane of the top dross becomes the (100) plane is discussed, considering the crystal structure of the Fe2Al5 phase, which is the parent structure. Figure 15(a) shows a schematic illustration of the crystal structure of the Fe2Al5 phase reported by Burkhardt et al.13) (rhombic: a = 7.656 Å b = 6.415 Å c = 4.218 Å). Figure 15(b) shows the original atomic arrangement of the crystal model of Burkhardt et al.13) viewed along the [001] direction, where the vertical axis is the [100] direction and the horizontal axis is the [010] direction. Similarly, Fig. 15(c) is the atomic arrangement viewed along the [100] direction, where the vertical axis is the [010] direction and the horizontal axis is the [001] direction, and Fig. 15(d) is the atomic arrangement viewed along the [010] direction, where the vertical axis is the [001] direction and the horizontal axis is the [100] direction. Figures 15(b), 15(c) and 15(d) show the planes with the highest atomic density among the planes perpendicular to the viewing direction, respectively. In the plane perpendicular to the [100] direction, the plane, which has two Fe atoms and four Al partial occupancy sites with 0.25 probability and two Al partial occupancy sites with 0.32 probability in a unit cell, has the highest atomic density with 0.133 (unit/Å2). Furthermore, as shown in Fig. 15(b), the interplanar spacing between the packed plane, which is expected to be the cleavage fracture plane, and the adjacent plane, is 1.276 Å. In the plane perpendicular to the [010] direction, the plane, which has one Fe atom and two Al atoms per unit cell, exhibits the highest atomic density (0.093 (unit/Å2)). As shown in Fig. 15 (c), the interspacing between the packed plane and the adjacent plane is 1.069 Å. In the plane perpendicular to the [001] direction, the plane, which has two Fe atoms and four Al atoms per unit cell, exhibits the highest atomic density (0.122 (unit/Å2)). Finally, as shown in Fig. 15(d), the interspacing between the packed plane and the adjacent plane is 0.352 Å. Consequently, it is concluded that in the Fe2Al5 phase, the (100) plane is the closest packed plane and the interplanar spacing in the [100] direction is the longest. Therefore, the (100) plane is the plane with the lowest surface energy, and, at the same time, the bonding force between atoms in the [100] direction is presumed to be the weakest.19,20) Therefore, even in the top dross particle having a crystal structure similar to that of the Fe2Al5 phase, it may be presumed that the (100) plane also became a cleavage fracture plane. It should be noted that the top dross has a different chemical composition from that of the Fe2Al5 phase and contains approximately 11 at% of Zn. In order to compare the anisotropy of the surface energy and interatomic bonding force of the top dross particle containing Zn with those of the Fe2Al5 phase by utilizing first-principles calculations as discussed in our previous report,7) information on the possible positions of Zn atoms in the Fe2Al5 structure is required, which will be the subject for future study.
(a) Schematic illustration showing crystal structure of Fe2Al5 phase reported by Burkhardt et al.13) and that viewed along (b) [001] direction, (c) [100] direction and (d) [010] direction.
The present study focused on the determination of the facet planes of the top dross based on the three-dimensional analysis of morphology of the top dross in a molten Zn bath. Serial sectioning and EBSD were simultaneously applied. Furthermore, the crystallographic cleavage fracture plane of the top dross was determined by EBSD, after introducing a cleavage crack by a Vickers hardness indent. The following results were obtained:
(1) The facet planes of the top dross consist of two planes of (001), four planes of {110} and eight planes of {111}. In addition, top dross particles grow fastest in the [001] direction. Consequently, the top dross particle is concluded to have a polyhedron structure with 14 facet planes.
(2) The cleavage fracture plane of the main crack in the top dross is the (100) plane.
The authors are grateful for valuable opinions provided by Profs. Ryoichi Monzen, Chihiro Watanabe and Tomotsugu Shimokawa of Kanazawa University.
