*2.3. Experimental Process*

The technological process of this ultrasonic casting experiment was as follows: The first step was to bake the smelting furnace, and then about 10 tons of commercial purity aluminum were put into the furnace. The material was heated and melted, and then fully stirred by electromagnetic stirrer. After the surface scum of the aluminum liquid was removed, metal additives, like Cu and Mg, as well as master alloy were added into the aluminum melt. Then, the sample was sent to the analysis room for composition detection and analysis. After the compositions of the molten aluminum alloy reached the required standard, Al-Ti-B wire rod was added. The molten aluminum was drained into the hot top mold through the launder, and the casting started after the hydrogen measurement of the melt was qualified. As shown in Figure 2a,b, ultrasonic wave was introduced into the aluminum melt from the top directly above the center to realize ultrasonic casting. Finally, the large ingot, as shown in Figure 2c, was transferred to the soaking pit furnace for homogenization treatment for 72 h. After annealing treatment, the chill layer of about 10 mm thick on the surface of ingots was removed by lathe. The specific casting parameters are shown in Table 2. In order to evaluate the effect of ultrasonic guided wave device ultrasonic treatment, the specific experimental conditions and variables were set as follows: (1) no ultrasonic treatment; (2) the pre-heated L-shaped ultrasonic wave guide rod was immersed 70 mm below the liquid level of the melt, and the ultrasonic treatment with a power of 500 W was applied; (3) the pre-heated the straight-rod wave guide rod was immersed 70 mm below the liquid level of the melt, and the ultrasonic treatment with a power of 500 W was applied.

**Figure 2.** 2A14 aluminum alloy ultrasonic semi-continuous casting: (**a**) schematic diagram of casting; (**b**) photo of casting site; and (**c**) ingot product.

**Table 2.** Casting process parameters of the 2A14 aluminum ingots.


A circular section with a thickness of 20 mm was cut at 400 mm away from the head of ingots with a vertical band saw. A quarter of the section was taken for macrostructure analysis, and a long strip with a width of 100 mm was cut along the radius direction for spectral analysis. The sampling locations is shown in Figure 3. The specific analysis contents were as follows:

**Figure 3.** Sampling locations of ingot samples.

Firstly, the macrostructure of the samples was analyzed. After milling, the samples were polished with fine sandpaper, then the surface was cleaned with anhydrous ethanol, and the surface was treated with Keller's solution for 10–20 s to observe the corrosion effect. Then, the strip sample was milled, and the compositions were detected along the radius direction by the SPECTRO MAXx direct reading spectrometer. A group of data were measured every 40 mm. Each group of data measured three points and took the average value of them. Finally, the distribution curve of the alloy elements in the ingots along the radial direction was drawn by using the measured data. Moreover, the microstructure analysis was carried out. The sampling positions are shown in Figure 3. The positions A, B, and C

in the figure correspond to the structure of the center, the one half radius, and the edge of the ingot samples, respectively. After the selected samples were polished, they were wiped with a cotton ball. The secondary phase structure and energy spectrum analysis were carried out using a Phenom fully automatic scanning electron microscope (Phenom-world BV, Holland). Then, their Vickers hardness values were measured using a HV-1000A microhardness tester. The applied load was 1 N with the dwell time of 10 s. Five results were tested for every samples, and after the maximum and minimum values were discarded, an average of the remaining three values was applied. Finally, the metallographic microstructure of the samples corroded by Keller's solution was observed using the OLYMPUS DSX500 metallurgical microscope (OLYMPUS Corporation, Japan), and the photos were taken, and the grain sizes were measured by the OLYMPUS Stream image analysis software. The three-circle intercept procedure was adopted. The measuring grid consists of three concentric circles of equal distance. Their total perimeter is 500 mm. This grid was used to measure at least five different fields of view of any choice for each specimen. The intercept points were counted manually with a mouse, and the average of the line intercept was calculated automatically by the analysis software.
