2023 年 64 巻 2 号 p. 366-372
In this study, the effects of the roll load and superheating on the appearance of surface defects in Al–Mg and Al–Si alloy strips cast using a high-speed twin-roll caster were investigated. The band zone that existed in the center of the thickness direction was investigated. The alloy compositions were Al–4.8%Mg and Al–Si with Si contents of 1, 2, 3 and 11%. The diameter and width of the copper roll in the twin-roll caster were 300 and 50 mm, respectively. The superheating temperature of the molten metal was 5 and 50°C, and the roll load was varied from 2 to 88 N/mm. A roll speed of 30 m/min was utilized. The existence of surface cracks was investigated by penetrant inspection and bending test. The amount of surface cracking in the Al–Mg alloy was affected by the roll load, and the number of surface cracks decreased with decreasing roll load. Surface cracks occurred more easily in the Al–4.8%Mg alloy than in the Al–Si alloys. The degree of cracking could not be reduced by cold-rolling. In the Al–Si alloys, the ripple marks were affected by the Si content, degree of superheating of the molten metal and the roll load. In the Al–3%Si alloy, ripple marks were produced, and worsened as the roll load and the superheating of the molten metal were increased. A specimen for observation of the microstructure near the roll bite was obtained by performing a roll-stop during casting, and observed using Weck’s reagent. A band zone was found to exist. The band zone in Al–Si consisted of globular crystals, and that in Al–Mg consisted of equiaxed dendrites and Mg-rich areas.
The casting speed for strips using a conventional horizontal-type twin-roll caster for aluminum alloy (CHTRCA) is usually slower than 2 m/min.1–3) A vertical-type high-speed twin-roll caster (VHSTRC) has been proposed in order to increase the casting speed, and can cast aluminum alloy strip at speeds higher than 30 m/min.4–6) This means that the VHSTRC has excellent productivity compared with the CHTRCA. The high casting speed of the VHSTRC is made possible by the high thermal conductivity of the roll material. Steel rolls are used for the CHTRCA, and copper rolls are used for the VHSTRC. The thermal conductivity of copper is much higher than that of steels. The cooling ability of the copper roll is better than that of the steel roll, and so the casting speed (roll speed) of the VHSTRC can be increased. The roll load for a VHSTRC is much smaller than that for a CHTRCA, as the deformation stress for the copper roll is smaller than that for the steel roll. The roll load for a CHTRCA is usually larger than 1 kN/mm, while that for a VHSTRC is usually from 0.14 to 0.5 kN/mm.
It is known empirically that surface cracks form on Al–Si strips for some Si contents and for most Al–Mg strips cast using a VHSTRC. For Al–Si strips, cracking occurs when the Si content is less than 2%, and for Al–Mg strips, cracking occurs when the Mg content is from 3 to 10%. In Al–Si and Al–Mg alloys for sheet-forming, the Si content is usually less than 2%, and the Mg content is from 3.5 to 5%. This means that a VHSTRC cannot cast sound strips of Al–Si or Al–Mg within these composition ranges. Investigation of the relationship between the casting conditions and crack occurrence is very important, and preventing the formation of cracks is necessary for practical use of a VHSTRC. It is thought that crack suppression can be achieved by the appropriate choice of casting conditions. Cracking occurs at grain boundaries that may not be completely solidified, and may be caused by deformation due to the load induced by the rolls. Consequently, the normal roll load for a VHSTRC, which ranges from 140 to 540 N/mm, may be too large to prevent surface cracking. In this study, roll casting was conducted with much smaller roll loads than normally used for a VHSTRC to clarify the effect on crack occurrence. The effect of superheating of the molten metal on surface cracking was also investigated.
In Al–Si strips cast using a VHSTRC at a normal roll load, a band zone exists in the center of the thickness direction.7,8) Globular crystals are formed in the band zone in the Al–Si alloy. The reason for the existence of this band zone in Al–Si and Al–Mg strips cast using very small roll loads has not yet been clarified. Therefore, in the present study, the formation mechanism and the properties of these band zones were investigated.
A schematic illustration of the VHSTRC is shown in Fig. 1. The rolls were rotated at the designated roll speed (30 m/min), and the molten metal was poured. The diameter and the width of the copper rolls were 300 mm and 50 mm, respectively. One of the two rolls was supported by coil springs, and the initial roll load was set by the spring compression amount. The roll load of this study was varied from 2 to 88 N/mm. The roll load was set smaller than normal roll load, which is ranging between the 140 N/mm to 540 N/mm, to clarify the roll load with which the crack does not occur. Initial gap between the rolls was 1 mm, and this gap varied depending on the strip thickness during the strip casting as one roll was supported by the springs. When the roll load was smaller than 2 N/mm, the strip broke and was not continuously cast as the strip did not solidify, so the minimum roll load was set at 2 N/mm. Two superheating temperatures of 5°C and 50°C were used for the molten metals. The rotating rolls were stopped during the strip casting. The cast strip was bitten by the rolls. The molten and the semisolid metal between the rolls were solidified. This technique is called “roll-stop” in this paper. The roll-stop was conducted to investigate the forming of the band zone and the globular crystal.
Schematic illustration of vertical-type high speed twin-roll caster.
Al–Si alloys with Si contents of 1, 2, 3, 6 and 11% were used. Surface cracking did not occur at the surface of Al–Si alloy strips with Si contents of 3, 6, 11%Si strips cast using the VHSTRC with normal roll loads. These two alloys were cast to investigate the influence of a small roll load on the occurrence of cracking. In the Al–Mg alloy strip cast using the VHSTRC with a normal load, surface cracks occurred when the Mg content was 3–10%. Surface cracking is less likely for Al–Si strips than for Al–Mg strips when the strip is cast using the VHSTRC with a normal roll load. It is thought that the roll load is one of the causes of cracking at the strip surface. The roll load may not necessarily be the cause of cracking for Al–Si and Al–Mg alloys, and the influence of the roll load may be different for each alloy. The Al–Si and the Al–Mg alloys were used for these reasons.
2.3 Surface and microstructure observationsThe surface of the as-cast strip was visually observed. A penetrant inspection was conducted on as-cast strips and cold-rolled strips. The as-cast strips were cold-rolled down to a thickness of 1 mm and the penetrant inspection was performed.
Specimens attained by the roll-stop were etched by Weck’s reagent.7) The chemical composition was 4 g KMnO4, 1 g NaOH, and 100 mL distilled water.
2.4 Mechanical propertiesTensile testing was conducted to investigate the influence of the surface cracks on the mechanical properties. The test samples had a thickness of 1 mm, a gage length of 9 mm and a width of 5 mm. The as-cast strips were cold-rolled down to a thickness of 1 mm, and then annealed at 380°C for 1.8 ks. A bending test was conducted to investigate the influence of the roll load on surface cracking, especially in the depth direction. A schematic illustration of the bending test is shown in Fig. 2.
Schematic illustration of bending test.
Strips could be cast continuously under the casting conditions shown above. The 2 N/mm roll load was much smaller than the normal load of 140 N/mm. The size of burrs at the edges of the strip decreased as the roll load decreased.9,10) It is estimated that the semisolid metal at the inside of the strip was squeezed to a lesser degree by the 2 N/mm roll load. The surfaces of as-cast strips are shown in Fig. 3. Fine pits and protrusions were observed on the strip surface, and the surface was dull when the strip was cast at the 2 N/mm roll load, except for Al–11%Si. Fine protrusions existed on the roll surface. It is thought that the metal from the fine pits on the strip surface stuck to the roll surface and became protrusions. The strip might remain semisolid including the surface under a 2 N/mm roll load. These results were not influenced by the superheating temperature of the poured molten metal. Fine pits and protrusions did not exist on the surface of the strips cast at a roll load of 88 N/mm, and the surface had a metallic luster. It is thought that the solid fraction of the strip increases as the roll load increases, because heat transfer between the strip and the rolls increase. The solid fraction of the surface of the strip increase and the protrusions did not stick to the roll surface. As result, pits do not exist on the surface of strips cast at the 88 N/mm of roll load. Ripple marks existed on the surface of the Al–3%Si strip cast at 88 N/mm with molten metal superheating temperatures of 5 and 50°C. For Al–4.8%Mg, ripple marks occurred for an 88 N/mm roll load and a superheating temperature of 50°C. In previous studies of strip casting using the VHSTRC, the roll load was usually larger than 140 N/mm, and ripple marks occurred on both Al–3%Si and Al–4.8%Mg strips. In this study, it became clear that the occurrence of ripple marks can be prevented by using a small roll load such as 2 N/mm. The reason that the occurrence of ripple marks is influenced by the roll load is explained below. It was reported that buckling occurs during strip casting using a CHTRCA,11) because roll position is fixed and the strip is rolled like a mill. During buckling, the solidified layer on the roll is pushed in the direction opposite to the casting direction by the deformation caused by rolling, and the solidified layer peels from the roll. Similar phenomena may take place in the VHSTRC despite the roll load being very small compared to that for the CHTRCA. In the VHSTRC, when the roll load is 88 N/mm, the force directed toward the direction opposite to the casting direction (rotation direction) might affect the occurrence of ripple marks. When the accumulated strain in the direction opposite to the casting direction reaches a limit, the solidified layer moves in the direction opposite to the casting direction. This means that the solidified layer periodically moves. The solidification starting position periodically moves, as shown in Fig. 4. This causes oscillation of the meniscus, and as a result, ripple marks occur. When the roll load is 2 N/mm, the force from the roll on the solidified layer is small enough that the accumulation of strain does not reach the limit at which the solidified layer moves.
Surfaces of as-cast strips. Roll speed: 30 m/min, RL: Roll load, SH: Superheating. Arrows show ripple marks.
Schematic illustration showing oscillation of meniscus during solidified layer movement, which occurs due to the roll load. Distance between the solidification starting position and the back dam plate in (b) is smaller than in (a). The meniscus in (b) is smaller than that in (a). The distance between the solidification starting position and the back dam plate in (c) is further in (c). The meniscus in (c) is larger than that in (b). A ripple mark is formed on the strip surface by oscillation of the meniscus.
The results of penetrant inspections are shown in Fig. 5. For Al–1%Si and Al–4.8%Mg, cracking was observed for the 88 N/mm roll load but not for the 2 N/mm roll load. This means that the roll load plays a dominant role in surface cracking for both the Al–Si and Al–Mg strips. It appears that surface cracking remarkably increased as the superheating temperature increased. It is considered that the grain boundaries were weaker for a superheating temperature of 50°C than for 5°C.
Surfaces of as-cast strips after penetrant inspection. RL: Roll load, SH: Superheating. Arrows show cracks.
The upper limit of the roll load to cast strips without surface cracking was investigated for Al–1%Si and Al–4.8%Mg. The results are shown in Fig. 6. The limiting roll loads for Al–1%Si and Al–4.8%Mg were about 5 and 10 N/mm, as shown in Figs. 6(a) and 6(b), respectively. The ranges of the roll load at which the Al–1%Si strip and the Al–4.8%Mg strip can be cast without cracking are from 2 to 5 N/mm and from 2 to 10 N/mm, respectively.
Surfaces of as-cast strips after penetrant inspection. RL: Roll load, SH: Superheating. Arrows show cracks.
In the Al–2%Si strip, cracking did not occur for a roll load of 88 N/mm, as shown in Fig. 6(c). When the roll load ranged from 140 to 540 N/mm, cracking occurred on the surface of the Al–2%Si alloy strip. 2%Si is the content at the boundary where the cracking occurs.
3.2 Effect of surface cracking on bent surfaceThe outer surfaces of bent as-cast strips are shown in Fig. 7. The Al–4.8%Mg strip cast at 88 N/mm with 5°C superheating was broken, but the other strips were not. The cracks in the broken strip might be deeper than in other non-broken strips with cracks. In the cases of Al–1%Si and Al–4.8%Mg, surface cracking occurred during bending for strips cast at 2 N/mm. Small cracks that are not visible during the penetrant inspection or ductility weak points might exist, and cracking may occur at these points.
Outer surfaces of as-cast strips after bending. RL: Roll load, SH: Superheating. Arrows show cracks.
As-cast strips with and without cracks could be cold-rolled to 1 mm-thick plate without breaking. The results of penetrant inspections of the surface of cold-rolled strips are shown in Fig. 8. Cracking was confirmed at the surfaces of the Al–1%Si strip and the Al–4.8%Mg strip cast at 88 N/mm. The amount of cracking was somewhat reduced, but some cracks were still present. The number of shallow cracks might be reduced by compression. The pits on the strip cast at 2 N/mm were reduced by cold-rolling, and the lumpy surface formed by the 2 N/mm roll load could be flattened. This may mean that the use of a small roll load such as 2 N/mm does not harm final plate surface if the roll-cast strip is subjected to cold-rolling.
Surfaces of cold-rolled to 1 mm-thick plates after penetrant inspection. RL: Roll load, SH: Superheating. Arrows show cracks.
The tensile strength and elongation measured by tensile testing are shown in Figs. 9 and 10, respectively. Both the tensile strength and elongation in the rolling direction were greater than those in the transverse direction when cracking did not occur. This may be due to the fibrous structure in the rolling direction caused by the cold-rolling.
Tensile strength results from tensile testing. Specimen: cold-rolled to 1 mm-thick plate and annealed at 380°C for 3.6 ks. RD and TD indicate rolling and transverse directions, respectively. 2 N/mm and 88 N/mm are the roll loads. 5°C and 50°C are the superheating temperatures.
Elongation results from tensile testing. Specimen: cold-rolled to 1 mm-thick plate and annealed at 380°C for 3.6 ks. RD and TD show rolling direction and transverse direction, respectively. 2 N/mm and 88 N/mm show roll loads. 5°C and 50°C are the superheating temperatures.
For the Al–1%Si and Al–4.8%Mg alloys, surface cracking occurred at a roll load of 88 N/mm. The tensile strength and elongation for strips with cracks were lower than those for strips without cracks, as shown in Fig. 9 and Fig. 10, respectively. The influence of the cracks on the tensile strength shown in Fig. 9 was less than that on the elongation, as shown in Fig. 10. In the strain-stress curve, the rate of increase of the tensile strength, which is greater than the yield strength, was lower than that of the elongation. This means that the change in the tensile strength is smaller than that of the elongation in the strain-stress curve.
For the Al–1%Si and Al–4.8%Mg alloys, surface cracks had a remarkable effect on elongation. In the cold-rolled plate including surface cracks, which was fabricated from the strip produced at 88 N/mm of roll load, the elongation in the transverse direction became much smaller than that in the rolling direction. The fibrous structure in the rolling direction formed by the cold rolling. This might be the cause of the smaller decrease of the elongation in the rolling direction than in the transverse direction when the surface crack was remarkable.
For Al–11%Si, the tensile strength and elongation for the strip cast at a roll load of 88 N/mm were superior to those for the strip cast at 2 N/mm roll load. The Al–11%Si strip did not crack. The eutectic Si in the strip cast at 88 N/mm was finer than that in the strip cast at 2 N/mm, as the strip cooled faster when the roll load was 88 N/mm.
3.5 Band zone in cast strip at roll biteA roll-stop was conducted during roll casting, and specimens were obtained. The casting conditions for the roll-stop were a roll speed 30 m/min, a roll load 2 N/mm and a superheating temperature of 5°C. A cross-section of the specimen is shown in Fig. 11. Weck’s reagent was used to etch the sample. This reagent causes regions rich in the alloying element to become bright, and regions rich in Al to become dark.7) A band zone existed in the center of the thickness direction; this is characteristic of high-speed roll casting.7) There was a tendency for the band zone to form under high-speed roll casting at smaller roll loads than for the CHTRCA. The band zone formed up to the roll bite. For Al–2%Si and Al–6%Si, there was a band zone consisting of globular crystals at the center. It was clear that the band zone and the globular crystals formed at a very small roll load. The globular crystals in the band zone resemble those observed in thixocasting. In the thixocasting of Al–Si alloy, the secondary arms break off from the dendrites, and these separated arms become globular. When roll-cast Al–Si strip was heated to a semisolid condition, globular crystals were formed.12) It appears that the globular crystals in the band zone in Al–Si alloys could be formed from broken-off secondary arms of dendrites in the solidification layers on both sides of the band zone. The latent heat of the band zone might cause the secondary arms of the dendrites to separate and form globular crystals.
Cross-section of strip near roll bite. Casting specimen was obtained after roll-stop while roll casting. Roll speed: 30 m/min, A Roll load: 2 N/mm, Superheating: 5°C. Weck’s reagent was used.
For Al–4.8%Mg, the band zone consisted of equiaxed dendrites and Mg-rich areas (bright areas). The Mg-rich area may arise from the segregation of Mg, and might be formed due to the low roll load. The Mg content of the equiaxed dendrites in the band zone was less than in the areas on both sides of the band zone because the equiaxed dendrites appeared darker. Precipitates containing Mg in the equiaxed dendrites may not diffuse to the sides of the band zone, but instead may remain near the dendrites.
It was reported that equiaxed dendrites may flow out from solidified layers on both sides of the band zone into the band zone.13) For Al–4.8%Mg, the equiaxed dendrites in the band zone might flow out of the solidified layers on both sides of the band zone.
The latent heat for Si is larger than that for Al, while that for Mg is smaller than that for Al. In the Al–4.8%Mg strip, the dendrites in the band zone might not be heated enough for the secondary arms of the equiaxed dendrites to be separated by the latent heat of the band zone. On the other hand, for the Al–Si strip, such separation may be possible, causing globular crystals to be formed.
