The solidification microstructure evolution of Ni-25 at.% Cu alloys under different undercooling degrees were studied by the cladding method and cyclic superheating method. Two grain refinement phenomena were observed in the obtained undercooling. In the low undercooling condition, dendrite remelting is the main reason for grain refinement in the recalescence process, while in the high undercooling condition, the stress accumulated in the recalescence process leads to recrystallization in the later stage of recalescence. Under the condition of high undercooling, the solidification structure is composed of complete equiaxed grains with relatively uniform grain size, which indicates that grain boundary migration occurs during grain growth.
Compared with the traditional process, the grain refinement in supercooled melt is spontaneous. Because of this, more and more attention has been paid to this phenomenon since Walker  obtained the grain refinement structure in supercooled melt in 1959. Researchers have studied many alloy systems, such as pure Ni [2, 3] and Ni–Zr . At the same time, the methods used by researchers to obtain undercooling of alloys or metals are also inconsistent (such as melt flux method [5, 6], dropper method , and melt spinning method ). However, molten coating and cyclic superheating techniques have been widely accepted and used as methods to obtain three–dimensional bulk alloys [9, 10, 11, 12, 13, 14].
In the studied alloy system, fine grain structure can be obtained under high undercooling, and the formation mechanism of fine grains has been widely studied. At present, the main mechanisms include growth instability, recrystallization, dendrite remelting and dendrite fracture induced by critical growth rate . In single-phase binary alloys, dendrite remelting debris caused by chemical overheating at low undercooling is the main reason for grain refinement. However, the stress-induced recrystallization mechanism proposed in this paper is used to verify the grain refinement phenomenon at high undercooling. Recrystallization usually occurs in highly deformed metals or alloys. The main process of recrystallization is the nucleation and subsequent growth of plastic deformed metals and alloys under annealing conditions. In high undercooling metals, with the increase of initial undercooling, stress accumulation occurs in the solidification structure, and recrystallization occurs spontaneously at the recrystallization temperature, so as to obtain the grain refinement structure.
In this paper, binary single phase Ni-25 at.% Cu alloy was used as the research object, and B2O3 was used as the purification agent to adsorb impurities. The solidification structures under different undercooling were studied systematically. The results show that there are two kinds of grain refinement structure in the whole solidification process. Electron backscattering diffraction (EBSD) images of the two kinds of grain refinement samples show that the dislocation angles of the two kinds of grain refinement samples are obviously different, which also confirms the different mechanism of the two kinds of grain refinement.
2 Experimental procedure
The Ni-25 at.% Cu alloys were prepared by melting pure nickel (99.99%) and copper (99.99%) particles under the protection of high-purity Ar gas. Before melting,the pure nickel and copper were pre-polished with sandpaper to remove surface oxidants and etched with HCI solution. Before the undercooling experiment, the quartz crucible glass and the sample were first cleaned in an ultrasonic cleaning machine containing alcohol for at least 6 min to remove the impurities on the surface. The whole undercooling experiment was carried out under vacuum condition, and the vacuum pressure was 10–3 Pa. Under the heating of induction heating coil, B2O3 was first heated to the molten state and kept warm for 10 min to absorb impurities on the surface of the material. Then gradually increasing the temperature until the temperature was 100 K to 150 K higher than the melting point of the alloy and keeping the temperature for about 15 min, each sample will be continuously heated and cooled by circulation, then turnning off the high-frequency heating power supply, and the undercooled sample will spontaneously cool to room temperature. Undercooling was the liquidus temperature minus the starting temperature point of recalescence (solidification). We used a temperature detector to measure the cooling-recalescence temperature curve of the alloy sample, and the undercooling value was calculated by using the liquidus temperature minus the starting temperature point of recalescence(solidification). In order to get a more detailed rule of solidification microstructure evolution, the difference of undercooling degree of each supercooled sample was about 10 K. The whole temperature change process was collected by an infrared thermometer with an accuracy of ±5 K and a response time of 10 ms, and transmitted to a computer in real time.
3 Results and discussion
In the range of supercooling, some typical microstructural transformation processes were observed. Figure 1 shows the solidification structure of undercooled Ni-25 at.% Cu under typical undercooling conditions. The results show that there are several obvious microstructural transformation stages in the whole solidification process.
Figure 1a shows the solidification structure at 30 K undercooling. The results show that the whole solidification structure is coarse dendrite with large grain size and a large number of secondary dendrite arms.
With the increase of undercooling, the morphology of the solidified structure changes. It can be seen from the optical micrographs that coarse dendrites are replaced by refined equiaxed crystals (Fig. 1b). This is the first time to refine the grain at low undercooling. The solidification structure has smooth grains and wide boundary. This is the result of dendrite remelting.
With the increase of undercooling, it can be seen from Fig. 1c that dendrites gradually replace refined equiaxed grains. However, compared with the dendrite under low undercooling, the dendrite under undercooling is obviously smaller, the dendrite spacing is closer, and the secondary dendrite arm is developed.
With the increase of undercooling, the solidification structure was also observed (Fig. 1d and e). It is obvious that dendrites are replaced by equiaxed grains, which is the second grain refinement. Compared with the first refinement, the solidification structure presents complete equiaxed grains with relatively uniform grain size, which indicates that grain boundary migration occurs during grain growth. At the same time, we can see the presence of annealing twins, which is important evidence of recrystallization.
According to the solidification structure diagram in Fig. 1, the structures obtained from the two kinds of grain refinement are different. At low undercooling, the grains are smooth and the grain boundaries are wide, which is called a spherical crystal structure. When the undercooling is high, the solidification structure is equiaxed, the grain size is uniform, and there are annealing twins in the grains. It can be seen that the two grain refinement mechanisms are necessarily different.
The grain refinement at low undercooling can be attributed to the dendrite remelting at the later stage of recalescence. Dendrite remelting mainly includes dendrite remelting fragments driven by solid–liquid interfacial tension and dendrite remelting fragments caused by chemical overheating proposed by Karma . Firstly, a model describing the dendrite fragmentation at low undercooling was proposed. The physical mechanism of grain refinement is described. Grain refinement occurs when the time of complete solidification of interdendritic melt is longer than the time of dendrite fracture. At the same time, both of these times depend on the undercooling degree, so the grain refinement corresponds to the critical undercooling degree. In order to obtain fine grain structure, undercooling must reach a certain critical value.
In addition, the effect of chemical superheating on the solidification structure also plays an important role. When the temperature of undercooled melt is lower than the liquidus temperature, the alloy will obtain the driving force of thermodynamic crystallization, which leads to the rapid growth of nucleating crystals. At the same time, a large amount of latent heat of crystallization will be released during the solidification process of undercooled metal. Because the rate of rapid solidification is much higher than normal, the released latent heat of crystallization mainly acts on the dendrites, which directly leads to overheating and remelting of solidified dendrites, and leads to the transformation of solidification morphology of primary dendrites. Figure 2 shows the relationship between the initial undercooling and the remelting rate of dendrites. The results show that when the undercooling is less than 50 8C, the remelting rate of dendrites is very high, which proves that the remelting of dendrites has an important effect on the transformation of solidification structure at low undercooling. Compared with low undercooling, the dendrite remelting fraction at high undercooling is smaller, and it is not easy to cause dendrite remelting.
With the increase of initial undercooling, the solidification rate of undercooled melt also increases. Therefore, under the condition of high undercooling, the transformation rate of undercooled melt to solid will be very high. Because primary dendrites are very fragile, they have no ability or strength to resist deformation. When the residual undercooled melt acts on the dendrite, there will be a certain amount of stress accumulation. When the accumulated stress is greater than the maximum stress that the primary dendrite can bear, the dendrite will fracture and a large amount of stress will accumulate in the dendrite. The energy is stored in the deformed dendrite fragments in the form of strain energy, which pats the dendrite fragments in a metastable thermodynamic state. The formation of dense crystal defects leads to the occurrence and development of high temperature recovery and recrystallization.
Typical grain refinement structures were detected using the EBSD technique for both low and high undercooling conditions. Firstly, we define the orientation difference as the crystal orientation difference between two adjacent grains. The frequency of a certain orientation angle is defined as the length of the grain boundary with the same orientation angle divided by the total length of the grain boundary in the EBSD diagram. We define a low angle grain boundary as a grain boundary angle less than 158and high angle grain boundary with grain boundary angle more than 158. Figure 3 shows the EBSD diagram of Ni-25 at.% Cu alloy undercooled at 50 K. The grain boundary is smooth and wide, which indicates that the grain has been coarsened. As can be seen from the inverse pole graph of Fig. 3c, the texture is almost random. Figure 3b shows the orientation angle distribution of grain boundaries in alloy Ni-25 at.% Cu undercooled at 50 K. It can be seen that the proportion of low angle grain boundary is 60.7%. In addition, at low undercooling, the proportion of annealing twins in the whole grain boundary is very small. Figure 4 shows the EBSD pattern of Ni-25 at.% Cu undercooled at 260 K, which is a typical grain refinement structure under high undercooling condition. The orientation angle distribution of grain boundaries in undercooled alloy Ni-25 at.% Cu at 260 K shows that the proportion of high angle grain boundary is about 88.4%, and the proportion of orientation angle is about 60%. The recognition rate of twin boundary is about 14.5%. According to the above data, recrystallization is the main cause of grain refinement under deep undercooling. As shown in Fig. 4c, high density dislocations can be observed in the specimen. These lattice defects indicate that plastic deformation should occur during rapid solidification, leading to highly defective structures.
The microstructural evolution of Ni-25 at.% Cu alloy under different degrees of undercooling was systematically studied by means of cladding and cyclic superheating, and the typical solidification structures were detected in detail by EBSD. The main conclusions are as follows
Two kinds of grain refinement were observed in the whole solidification structure, one occurred in the range of low undercooling, the other occurred in the range of high undercooling. According to the evolution of the solidification structure, the whole solidification process can be divided into four stages: coarse dendrite → equiaxed crystal → fine dendrite → equiaxed crystal.
Through the EBSD analysis of two typical grain refinement structures, it can be more clearly observed that there are differences between the two kinds of grain refinement, which also shows that the mechanism of the two kinds of grain refinement is inconsistent
At low undercooling, dendrite remelting is the main reason for grain refinement. At high undercooling, the stress accumulated during rapid solidification provides the driving force for the subsequent recrystallization, resulting in grain refinement.
Funding statement: This work was supported by National Natural Science Foundation of China (Nos. 51701187). Basic Applied Research Projects in Shanxi Province (201801D221151). Key research and development and promotion projects in Henan Province (212102210267).
 P. Haasen: Physical Metallurgy, third ed., Cambridge university press. 1996. DOI:10.1017/CBO9781107295551 Search in Google Scholar
 F.J. Humphreys, M. Hatherly: Recrystallization and Related Annealing Phenomena, (2004), second ed., New York, Oxford. DOI:10.1016/B978-008044164-1/50002-5 Search in Google Scholar
 J.W. Christian: The Theory of Transformation in Metals and Alloys: Recovery, recrystallization and Grain Growth, (2002), Oxford, Pergamon Press. DOI:10.1016/B978-008044019-4/50024-6 Search in Google Scholar
 J.L. Walker: The Physical Chemistry of Process Metallurgy, Part 2, New York, Interscience (1959). Search in Google Scholar
 F. Liu, G.C. Yang: J. Cryst. Growth 231 (2001) 295. DOI:10.1016/S0022–0248(01)01438–5 Search in Google Scholar
 W.W. Mullins, R.F. Sekerka: Dyna. Curv. fron. (1988) 444. DOI:10.1016/B978-0-08-092523-3.50037-X Search in Google Scholar
 X.L. Xu, Y.Z. Chen, F. Liu: Mater. Sci. Technol. 28 (2012) 1492. DOI:10.1179/1743284712Y.0000000055 Search in Google Scholar
 X.L. Xu, Y.Z. Chen, F. Liu: Mater. Sci. Technol. 29 (2012) 117. DOI:10.1179/1743284712Y.0000000102 Search in Google Scholar
 X.L. Xu, Y.Z. Chen, F. Liu: Mater. Sci. Technol. 28 (2012) 886. DOI:10.1179/1743284712Y.0000000049 Search in Google Scholar
 X.L. Xu, Y.Z. Chen, F. Liu: Mater. Lett. 81 (2012) 73. DOI:10.1016/j.matlet.2012.04.042 Search in Google Scholar
 X.L. Xu, Y.Z. Chen, F. Liu: J. Cryst. Growth 377 (2013) 153. DOI:10.1016/j.jcrysgro.2013.05.016 Search in Google Scholar
 X.L. Xu, F. Liu: J. Alloys Compd. 597 (2014) 205. DOI:10.1016/j.jallcom.2014.01.228 Search in Google Scholar
 S.Y. Lu, J.F. Li, Y.H. Zhou: J. Cryst. Growth. 309 (2007) 103. DOI:10.1016/j.jcrysgro.2007.09.005 Search in Google Scholar
 G. Wilde, G.P. Gorler, R. Willnecker: Appl. Phys. Lett. 69 (1996) 2995. DOI:10.1063/1.117755 Search in Google Scholar
© 2021 Walter de Gruyter GmbH, Berlin/Boston, Germany