Microstructure of Materials
Effect of Ultrasonic Radiation on the Grain Refinement of High Purity Aluminum
2020 Volume 61 Issue 3 Pages 444-448
Details
2020 Volume 61 Issue 3 Pages 444-448
Effect of high-intensity ultrasonic irradiation on solidified structure of high purity aluminum was studied. The results reveal that the solidified structure was obviously divided into two regions: an effectively refined region I and a poorly refined region II. Cavitation induced nucleation was considered acting as a leading role in causing the final refinement. Re-distribution of cavitation induced nuclei by acoustic flow resulted in the cone-shape of the region I filled with remarkable columnar crystals.
Microstructural refinement has proved to be crucial for industrially used metals and alloys to improve their mechanical performances. The current and major practices for many commercial alloys are chemical refining methods via refiners addition. Actually, the refiners are inefficient under typical casting conditions1) and just few of the added refiners participate in nucleating grains,2) which not only increases the cost, but modifies the original composition: inactive particles segregate along grain boundaries.3) More embarrassedly, many alloy systems have not established chemical refinement.4) Ultrasonic refinement has been approved as a simple and sound physical method of greatly minimizing metal grains.5–10) In light of the refining mechanisms of cavitation-induced dendrite fragmentation, cavitation induced nucleation,11–13) extensively previous researches boldly shown that the introduction of a high intensity ultrasonic vibration in solidification processes effectively refines the final structure.3,14,15) However, most of the previous work conducted within solute contained melts, which promotes growth restriction during the solidification and contributes to the final refinement, although declared that there is a poor influence on resultant refinement when ultrasonication applied.16) Thus caused a rare opportunity of studying ultrasonic assisted microstructural refinement directly. In present work, high purity aluminum was selected as the aimed metal to investigate the entire attribution of ultrasonication on resultant refinement.
A 20 kHz ultrasonic system (Shanghai Sonxi Ultrasonic Instrument CO., LTD.) was utilized, of which consists of 1 kW acoustic generator, a water-cooled transducer made of piezoelectric lead zirconate titanate crystals, a sonotrode made of titanium alloy with 20 mm in diameter and a resistor furnace. A manual controlled device was installed to precisely move the sonotrode.
Experiments started with pre-heated 150 ml clean arc corundum crucibles at 700°C placed in the center of the ultrasonic system equipped furnace. Then a charge of 360 g high purity aluminum (99.999 mass%) was melt in the crucible for 2 hours. Ultrasonic treatment (UT) under a power input of 1000 W for 2 min was introduced by the sonotrode as immersed 25 mm from the melt surface. To undertake isothermal holding (IH), the melt was kept in the crucible at 670°C for 10 min after UT. Finally, the melts along with the crucible were cooled down in ambient temperature, of which the whole operation was conducted within 1 min. For comparison, a counterpart was produced with the same process but an idle sonotrode.
All the obtained samples were longitudinally sectioned and one entire half section for each sample was polished and examined. Simultaneously, the other half of the samples were crossly sectioned and each part with the same thickness of roughly 15 mm was examined as well. It is noted that the quantification and of grains on the sections is an average based on three identical specimens under each treatment, and the error bars were provided as deviations from the maximal values.
The pressure distribution in the melt was numerically computed to measure the influence of UT. As the geometry and the boundaries shown in Fig. 1, the radiation face of the sonotrode was set as a pressure input, which is 25 mm away from the melt surface according to the experiment. The input pressure, pin, is written as $p_{in} = p_{a}\cos (\omega t)$, where $p_{a} = \sqrt{2\rho cW/(\pi R^{2})} $, and W and R are the applied ultrasonic power and the radius of the ultrasound radiation face, respectively. The air-melt interface is a soft-sound boundary and the melt-crucible interface is a sound-impedance wall. The Helmholtz equation was solved in the computation domain by COMSOL Multiphysics, which is expressed as
| \begin{equation*} \nabla \left( \frac{1}{\rho}\nabla p \right) + \frac{\omega^{2}}{\rho {c^{2}}} p = 0, \end{equation*} |
Geometry of the commutation domain (a) and boundary conditions (b).
Figure 2 exhibits the structures of nUT (Fig. 2(a), (b)), UT (Fig. 2(c), (d)), and UT+IH (Fig. 2(e), (f)), respectively. In their longitudinal sections, the nUT characterizes coarse crystals (Fig. 2(a)) while the UT shows appreciable finer ones. Specially, two noticeable regions are distinguished in the longitudinal section of the UT (Fig. 2(c)): a cone-shape refined region below the sound radiation face (denoted I) and a poorly refined region around the sonotrode (denoted II). Within the region I, the refined grains exhibit columnar morphologies with a larger length/diameter ratio as close to the radiation face. In their cross sections (5 mm away from the radiation face), fine grains gather at up-center positions and the 2D-contours alike equiaxial crystals for the UT, which are consistent with the longitudinal section of columnar grains located in the refined region I; comparatively, the cross section of the nUT represents several coarse grains. Once the following IH implemented, the refining effect of the UT deteriorates as indicated in Fig. 2(e)–(f).
Macrostructures of high purity aluminum ingots with (a), (b) non-UT (nUT), (c), (d) UT, and (e), (f) UT+IH. Grains are contoured red in their longitudinal (a), (c), (e) and half cross (b), (d), (f) sections, and the cross sections are 5 mm away from radiation face.
For convenient discussion, grains initialized from the crucible/melt and atmosphere/melt interfaces are termed wall crystals, and the others are denoted as inner crystals. Apparently, the nUT consists of coarse wall crystals whist the UT is constituted of both fine wall crystals and inner crystals. An explicit comparison of grain number among the nUT, the UT and the UT+IH is exhibited in Fig. 3. There is a significant increment of crystal number after UT from the nUT: the wall crystal increases almost four times and the inner crystal gains more than two times. The following IH results in a number reduction involved in both wall and inner crystals. Further statistics in Fig. 4(a) reveals the number increment largely attributes to a contribution from the region I even experienced the IH. It is interesting that the IH brings with a marginal influence on the crystal number in region II. Figure 4(b) indicates that the length/diameter ratio of the columnar crystals resided in the region I increases more obvious than in the region II as applied UT, and the IH affect much in the region I. Figure 5 gives the crystal number distribution in the cross sections as distanced from the radiation face. It is apparent that the UT possesses more than the UT+IH and then the nUT. The crystal numbers of the UT and the UT+IH gradually reduce as away from the radiation face, while that of the nUT have not an obvious variation. These statistics manifest that the UT takes effect largely in the region I, and effectively refines the structure into fine columnar crystals there, characterized larger length/diameter ratios.
Numbers of different grains appeared in the longitudinal sections.
Comparisons of grain number (a) and mean length/diameter ratio (b) between region I and region II.
Distribution of grain number against the distance as away from the radiation face.
Figure 6 shows the calculated pressure field in the melt during UT. Higher pressure, typically 2 MPa, concentrates closely beneath the radiation face and exhibit a semi-ellipsoid like shape of the equal-strength contours, which predicts an intense cavitation locally. As distancing away from the radiation face, the pressure steadily decreases. Figure 6(b) elaborates the pressure distribution along the melt-crucible interface, and illustrates that the cavitation happened region almost appears in the volume below the radiation face.
The calculated pressure field (a) and the pressure distribution along the melt-crucible interface (b). Black line in (a) delineates the cavitation boundary of an ideal cavitation pressure threshold of 1 MPa.19)
It is well known that cavitation events will happen when ultrasonic intensity exceeds the threshold in metal melts. Collapses of cavitation bubbles introduce abundant nucleations on basis of two main mechanisms:11) cavitation induced nucleation (including homogeneous/heterogeneous nucleation and activated nucleation), and cavitation caused dendrite fragmentation. According to the studies of ice sonocrystallization under power ultrasound,17) crystallization occurs immediately in the vicinity of the cavitation bubble walls, which indicates that the locations of the solidification embryos could be detected within the cavitation bubble existed region (cavitation region). Considering the as-performed UT prior to the solidification of the extremely high purity aluminum without crystallization interval, homogeneous nucleation would be provoked due to a pressure spike governed supercooling, originated from the cavitation bubble collapse (high pressure results in the melting-point elevation, and thus leads to a supercooling which could not happen in atmosphere pressure). The literature17) have extensively validated that the cavitation bubble cloud (a volume of numerical cavitation bubbles gathered) of a small volume appears right below the radiation face. Taken by the acoustic stream, cavitation induced nucleus (in metal melts) then tend to move toward the bottom and further circled within entire melt,18) and cause refinement within the whole volume. However, molten aluminum differentiates from water and other polymer materials on liquid properties, such as a strong cohesion and high viscosity which hinder an intense cavitation occurring under lower power inputs. Therefore, it is reasonable that cavitations occurred violently within a limited volume beneath the radiation face, which is verified by the semi-ellipsoid liked high pressure volume as indicated in Fig. 6. Accordingly, a host of nuclei were induced by the cavitation and then fine inner crystals developed there. Considering a weak acoustic flow (flows downward to the bottom of the crucible and turns horizontally then up along the inner face of the crucible), the nuclei would re-distribute during the movement following the acoustic flow and gradually remelt in the path (low/no supercooling or even overheating owing to a decreasing pressure distribution locally, typically shown in Fig. 6(b)). During the journey, once the nuclei were arrested by the crucible/melt interface, wall crystals would be developed. However, the nucleus number dramatically decreased after the flow turned up, with a long-period journey and decelerated flow speed caused a prolonged remelting. As a result, the cone-shape refined region I dominated by inner crystals and poorly refined region II characterized by wall crystals, thus formed (Fig. 1(c)–(d)). It is noted that wall crystals also could form owing to a heat flux out from the crucible wall, but the number is far less than the cavitation developed and should maintain constant after the IH.
From the crystallization point of view, more nucleation sites within a unit melt develop more grains and a finer structure finally, thus reversely reveals that a grain concentrated region with more active nuclei should exist prior to solidification. Owing to the directional transfer of heat flux, the nuclei grew up into columnar crystals. Consequently, a finer columnar grain structure characterized by the larger length/diameter ratio appeared in a region of more active nuclei per unit. Therefore, an intense homogeneous nucleation was responsible for the remarkable columnar grain structure in the longitudinal section (Fig. 4(c)).
Unfortunately, the homogeneous nuclei were highly thermodynamically unstable in a superheated melt and would easily remelt to deteriorate the final refinement. To confirm and validate this, the IH for 10 min were taken. As shown in Fig. 2(e)–(f) and Fig. 3–4, the UT+IH sample represents a less crystal number and a smaller length/diameter ratio of columnar crystals in the refined region I as expected.
Cavitation induced nucleation took a leading role in refining high purity aluminum. The resultant structure is characterized by two distinct regions: an effectively refined region I of a cone-shape right below the radiation face and a poorly refined region II aside of sonotrode. The region I is dominated by columnar inner crystals while the region II is constituted by coarser wall crystals.
The authors gratefully acknowledge the financial supports from the National Science Foundation of China (Grants No. 51704196) and the Zhejiang Provincial Natural Science Foundation (Grant No. LY15E050001).
