Effect of La2O3 on the Microstructure and Grain Refining Effect of Novel Al-TiO2-C-XLa2O3 Refiners

New aluminum grain refiners, Al-TiO2-C-XLa2O3 refiners, were manufactured by the in situ reaction of Al powder, TiO2 powder, C powder, and La2O3 powder. The effects of La2O3 on the refiners’ microstructure and the grain refining effect of the Al-TiO2-C-XLa2O3 refiners on industrial aluminum were studied. The effect of the sintering process was also studied. The results show that the refiners contain Al3Ti, TiC, Al2O3, AlN, and Al20Ti2La. As the content of La2O3 increases, the amount of Al20Ti2La also increases. However, the amounts of Al3Ti, Al2O3, and TiC decrease. Al3Ti is a lath-like compound whose size becomes smaller. The distribution of Al2O3 and TiC, however, is more uniform. The Al-TiO2-C-0.2La2O3 refiner has the best grain refining effect on industrial aluminum. The size of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner is 320 μm; the refiner’s grain size is 8.4% that of the industrial aluminum without refiners 0 μm (380). After adding the novel Al-TiO2-C-0.2La2O3 refiner, the nucleation temperature TN reached 679.21 °C, which is 17 °C above the nucleation temperature of the industrial aluminum without refiners. The primary transformation time is the longest at 25.5 s, which is 4.6 s higher than that of industrial aluminum (20.9 s). Furthermore, the ΔT of the aluminum is 0.5 °C, which is the lowest value.


Introduction
Fine equiaxed crystals can effectively improve the mechanical properties of products. Improving mechanical properties by refining grains is an almost-essential step in the production of aluminum products. Adding refiners is one of the most effective and common methods for refining grains [1][2][3]. Prarad et al. realized Al-Ti-C grain refinement by spark plasma sintering and found that Al-Ti-C achieved a 25% reduction in the average grain size [4]. Zhang et al. proved that a Al-5Ti-0.2C master alloy produced via an improved self-propagating synthesis approach has better grain refining [5]. In industrial production, most refiners are master alloys, such as Al-Ti-B, Al-Ti-C, Al-Nb, etc. [6][7][8]. Moreover, some composite materials have been proven to be effective refiners for industrial aluminum, such as CeB 6 /Al composites and TiN/Ti composites [9,10]. The Al-TiO 2 -C system is one of the most common composites and can produce Al 3 Ti, TiC, and Al 2 O 3 in the system [11,12]. Al 3 Ti and TiC can act as heterogeneous nuclei of Al [13,14]. Al 2 O 3 , which is distributed in the solid-liquid interface, can inhibit grain growth, which has been proven in references [15,16]. However, an Al-TiO 2 -C system has rarely been used as a refiner. Studies on Al-TiO 2 -C systems have mainly focused on the preparation of composite materials and how to improve the properties of composites.
Many researchers have refined aluminum alloys by adding rare earth (RE) [17,18]. Zhang et al. discovered that a Al-Ti-C-Ce master alloy can refine industrial aluminum effectively [19]. Shi et al. proved that erbium (Er) can refine Al-Si-Mg alloys [20]. However, studies on RE oxides have mostly focused on the preparation of catalysts and glasses [21,22], but the application of RE oxides in the synthesis of Al-TiO2-C grain refiners has been less researched.
Computer-aided cooling curve analysis (CA-CCA) is often used to characterize the solidification parameters of metals during solidification. Due to its ease of use and low cost, it is the most popular thermal analysis technique [23][24][25]. Farahany et al. used CA-CCA to study the grain refining effect of Sb, Bi, and other elements on Al-Si alloys [26]. Osvaldo et al. studied the solidification process of Al-Cu alloy by CA-CCA [27]. In this paper, novel Al-TiO2-C-XLa2O3 refiners were prepared via an insitu reaction (where X is the weight fraction of La2O3, equal to 0, 0.1, 0.2, 0.3, and 0.4, as used in this study). The influence of La2O3 on the microstructures of refiners was analyzed. Then, adding different refiners to industrial aluminum, the grain refining effect of refiners on industrial aluminum was studied by CA-CCA. The influence of La2O3 in the sintering process is also discussed.

Materials and Methods
In this experiment, La2O3 powder (purity 99.9%, particle size 5 μm, Baotou Research Institute of Rare Earths, Baotou, China), Al powder (purity 99.9%, particle size 40-50 μm, Beijing General Research Institute of Nonferrous Metals, Beijing, China), C powder (purity 99.9%, particle size 30-95 μm, Beijing Dali Fine Chemical Plant, Beijing, China), and TiO2 powder (purity 99.9%, particle size 90 nm, Hangzhou Wanjing New Material Co., Ltd., Hangzhou, China) were mixed in certain proportions (Ti/C = 10:1) into a ball mill for milling. Then, the mixed powder was pressed into a cylindrical preformed block (30 × 10 mm 2 ), with a pressing pressure of 90 KN. The preformed blocks were sintered in a high-temperature sintering furnace, with a heating rate of 10 °C/min and a sintering temperature of 1250 °C. After reaching the target temperature, the preformed blocks were held for 2 h, and the furnace was cooled to prepare the Al-TiO2-C-XLa2O3 refiners.
The samples of the Al-TiO2-C-XLa2O3 refiners were taken from the center of each sample in the transverse section and were etched using Keller's reagent (95% H2O + 2.5% HNO3 + 1.5% HCl + 1% HF). Scanning electron microscopy (SEM, QUANTA FEG 650, FEI Company, Hillsboro, OR, USA) and energy-dispersive X-ray spectroscopy (EDS) were used to analyze the microstructures and components. X-ray diffraction (XRD, Cu K-alpha, D/MAX-2500/PC, Japan Science Co., Ltd., Tokyo Metropolis, Japan) was used to identify the phases of the specimen. The area ratio of the Al20Ti2La and Al3Ti was measured by Image Pro Plus 6.0 by using three SEM images (50×) and taking the average. It should be pointed out that because the number of particles distributed in Al3Ti and Al20Ti2La (such as Al2O3 particles) is uniform, to improve precision, we have to remove the particles other than Al3Ti and Al20Ti2La when measuring the area ratio, as shown in Figure 1.  Table 1 shows the chemical composition of the industrial aluminum used in this experiment, which were purchased from Chinalco. The aluminum was melted in a resistance furnace at 750 °C.  Table 1 shows the chemical composition of the industrial aluminum used in this experiment, which were purchased from Chinalco. The aluminum was melted in a resistance furnace at 750 • C. After holding for 30 min, the blocks that were cut from the centers of the different refiners (0.3 wt %) were added, stirred thoroughly, and held again for 15 min. Then the mixture was poured into a preheated crucible (300 • C). A thermal analysis was carried out by attaching a K-type thermocouple (Anhui Beichen Electric Technology Co., Ltd., Wuhu, China) located in the middle of the preheated crucible, which was connected to a DAQ (Data Acquisition) central temperature acquisition device (OM-DAQ-USB-2400, Spectris, Shanghai, China). A DAQ central temperature acquisition device collects temperature data with a frequency of 20 times/s during the solidification of industrial aluminum. Figure 2 is a schematic diagram of the thermal analysis. At least two thermal analysis runs were made to ensure the reproducibility of the results. The remaining aluminum was poured into a preheated ring mold (300 • C). After air cooling, the aluminum was mechanically polished and etched by a reagent (60% HCl + 30% HNO 3 + 5% HF + 5% H 2 O), and then we observed a grain refining effect. A metallographic microstructural examination was conducted on the centers of the samples. The surface to be examined was ground using abrasive papers and then polished using an electrolyte (10% HClO 4 + 90% ethanol). The voltage and current density used were 30 V and 1.5 A, respectively. The time was 90 s. Electrochemical etching was applied to the polished surface to reveal the grain structure with the etching solution (98% H 2 O + 2% 50 vol% HBF 4 ). The voltage and current density used were 30 V and 0.1 A, respectively. The time was 240 s. Grain structure characterization was performed using an optical microscope (OM, DMI8M, Lecia, Wetzlar, Germany) in the polarized mode. After holding for 30 min, the blocks that were cut from the centers of the different refiners (0.3 wt %) were added, stirred thoroughly, and held again for 15 min. Then the mixture was poured into a preheated crucible (300 °C). A thermal analysis was carried out by attaching a K-type thermocouple (Anhui Beichen Electric Technology Co., Ltd., Wuhu, China) located in the middle of the preheated crucible, which was connected to a DAQ (Data Acquisition) central temperature acquisition device (OM-DAQ-USB-2400, Spectris, Shanghai, China). A DAQ central temperature acquisition device collects temperature data with a frequency of 20 times/s during the solidification of industrial aluminum. Figure 2 is a schematic diagram of the thermal analysis. At least two thermal analysis runs were made to ensure the reproducibility of the results. The remaining aluminum was poured into a preheated ring mold (300 °C). After air cooling, the aluminum was mechanically polished and etched by a reagent (60% HCl + 30% HNO3 + 5% HF + 5% H2O), and then we observed a grain refining effect. A metallographic microstructural examination was conducted on the centers of the samples. The surface to be examined was ground using abrasive papers and then polished using an electrolyte (10% HClO4 + 90% ethanol). The voltage and current density used were 30 V and 1.5 A, respectively. The time was 90 s. Electrochemical etching was applied to the polished surface to reveal the grain structure with the etching solution (98% H2O + 2% 50 vol% HBF4). The voltage and current density used were 30 V and 0.1 A, respectively. The time was 240 s. Grain structure characterization was performed using an optical microscope (OM, DMI8M, Lecia, Wetzlar, Germany) in the polarized mode.  Figure 3 shows the XRD patterns of the Al-TiO2-C-XLa2O3 refiners. It can be seen in Figure 3a that the refiners contain Al3Ti, TiC, Al2O3, and AlN, but no rare earth phase. It can be seen in Figure  3b that the diffraction peaks of TiC and Al2O3 become lower first and then higher as the content of La2O3 increases. The diffraction peaks are the lowest when the content of La2O3 is 0.3 wt %. The diffraction peaks of AlN remain very low.  Figure 3 shows the XRD patterns of the Al-TiO 2 -C-XLa 2 O 3 refiners. It can be seen in Figure 3a that the refiners contain Al 3 Ti, TiC, Al 2 O 3 , and AlN, but no rare earth phase. It can be seen in Figure 3b that the diffraction peaks of TiC and Al 2 O 3 become lower first and then higher as the content of La 2 O 3 increases. The diffraction peaks are the lowest when the content of La 2 O 3 is 0.3 wt %. The diffraction peaks of AlN remain very low.   Figure 4 shows SEM images, performed in the backscattered imaging mode, of the Al-TiO2-C-XLa2O3 refiners. In the Al-TiO2-C refiner, there are three phases: a white lath-like phase, a dark particle phase, and a white small particle phase. However, there are four phases in the Al-TiO2-C-XLa2O3 (X = 0.1, 0.2, 0.3, 0.4) refiners, as some white block phases also appear.   Figure 4 shows SEM images, performed in the backscattered imaging mode, of the Al-TiO 2 -C-XLa 2 O 3 refiners. In the Al-TiO 2 -C refiner, there are three phases: a white lath-like phase, a dark particle phase, and a white small particle phase. However, there are four phases in the Al-TiO 2 -C-XLa 2 O 3 (X = 0.1, 0.2, 0.3, 0.4) refiners, as some white block phases also appear.   Figure 4 shows SEM images, performed in the backscattered imaging mode, of the Al-TiO2-C-XLa2O3 refiners. In the Al-TiO2-C refiner, there are three phases: a white lath-like phase, a dark particle phase, and a white small particle phase. However, there are four phases in the Al-TiO2-C-XLa2O3 (X = 0.1, 0.2, 0.3, 0.4) refiners, as some white block phases also appear.   Figure 5 shows the EDS patterns of different phases. In Figure 5b, we can see that, at point A in the dark phases, the molar mass fraction of Al is 38.15%, while that of O is 61.85%,and the molar mass ratio is about 3:2. In Figure 5c, we can see that, at point B in the white block phases, the molar mass fraction of Al is 86.46%, that of Ti is 9.17%, that of La is 4.37%, and the molar mass ratio is about 20:2:1. In Figure 5d, we can see that, at point C in the lath-like phases, the molar mass fraction of Al is 74.73%, that of Ti is 25.27%, and the molar mass ratio is about 3:1. According to the analysis of the XRD pattern of the Al-TiO 2 -C-XLa 2 O 3 (X = 0-0.4) refiners, the dark phases are Al 2 O 3 , the lath-like phases are Al 3 Ti, and the small white phases are TiC. AlN is too small to appear, which has been proven in reference [11].

Microstructure Characterization
The white block phases are Al 20 Ti 2 La. A similar result can be found in the literature [28]. The formation process is [12,29] 2Al [La] + Al 3 Ti → Al 20 Ti 2 La (5) Metals 2019, 9, x FOR PEER REVIEW 5 of 14 Figure 5 shows the EDS patterns of different phases. In Figure 5b, we can see that, at point A in the dark phases, the molar mass fraction of Al is 38.15%, while that of O is 61.85%,and the molar mass ratio is about 3:2. In Figure 5c, we can see that, at point B in the white block phases, the molar mass fraction of Al is 86.46%, that of Ti is 9.17%, that of La is 4.37%, and the molar mass ratio is about 20:2:1. In Figure 5d, we can see that, at point C in the lath-like phases, the molar mass fraction of Al is 74.73%, that of Ti is 25.27%, and the molar mass ratio is about 3:1. According to the analysis of the XRD pattern of the Al-TiO2-C-XLa2O3 (X = 0-0.4) refiners, the dark phases are Al2O3, the lath-like phases are Al3Ti, and the small white phases are TiC. AlN is too small to appear, which has been proven in reference [11].
The white block phases are Al20Ti2La. A similar result can be found in the literature [28]. The formation process is [12,29] 2Al + 3TiO2 = 3[Ti] + 2Al2O3 (1) [Ti] + 3Al = Al3Ti  Table 2 shows the area ratio of the Al20Ti2La measured by Image Pro Plus 6.0 Image Pro Plus 6.0(Media Cybernetics, Rockville, MD, USA). The La2O3 content increased by 0.1%, and the Al20Ti2La area increased by about 0.4%. The amount of Al20Ti2La is the largest in the Al-TiO2-C-0.4La2O3 refiner, but its area ratio is only 1.7%. This value is not shown in the XRD pattern because of its lower amount.  Table 2 shows the area ratio of the Al 20 Ti 2 La measured by Image Pro Plus 6.0 Image Pro Plus 6.0(Media Cybernetics, Rockville, MD, USA). The La 2 O 3 content increased by 0.1%, and the Al 20 Ti 2 La area increased by about 0.4%. The amount of Al 20 Ti 2 La is the largest in the Al-TiO 2 -C-0.4La 2 O 3 refiner, but its area ratio is only 1.7%. This value is not shown in the XRD pattern because of its lower amount. It can be seen from Figure 4 that the amount of Al 2 O 3 decreases as the content of La 2 O 3 increases.
[La] can reduce the surface tension of the particles, so the surface activity of Al 2 O 3 will increase when La dissolves in the melt and promotes the reaction of Al 2 O 3 with the C powder and N 2 in the air to form AlN. However, because the amount and volume of AlN is small, it is not shown in Figure 4 (although these values have been proven in previous experiments [11]). Although the amount of Al 2 O 3 is small, the grain size of some Al 2 O 3 is 25 ± 5 µm in the Al-TiO 2 -C-0.4La 2 O 3 refiner, which is 11 ± 3 µm in the Al-TiO 2 -C-0.2La 2 O 3 refiner. Thus, the diffraction peaks become high again in the XRD pattern for the Al-TiO 2 -C-0.4La 2 O 3 refiner. Because the system is still oxygen-rich and hyperthermal, the carbon monoxide (CO) produced in reaction (3) will combust [30]. Al 2 O 3 diffusion can be promoted by the local high temperature region generated by CO combustion. Therefore, the amount of Al 2 O 3 decreases, but its distribution is more uniform.
The morphology of Al 3 Ti does not change significantly after adding La 2 O 3 . It remains lath-like, but the length of Al 3 Ti is shortened. Al 3 Ti is 1200 ± 150 µm in the Al-TiO 2 -C refiner but is only 800 ± 100 µm in the Al-TiO 2 -C-0.2La 2 O 3 refiner. Because Al 20 Ti 2 La will dissolve at a high temperature and release [La], RE elements usually aggregate at phase boundaries or grain boundaries [31]. Some [La] aggregates on the phase boundaries of Al 3 Ti and inhibits Al 3 Ti growth. It can be seen in Figure 6d and (e) that there is also a certain amount of La in the area where the titanium element is distributed. Moreover, researchers have proven that Al 3 Ti is greatly affected by the reaction temperature. After the addition of La 2 O 3 , the CO from reaction (3) combusts with the O 2 in the melt or air and produces local high temperature areas. Thus, the diffusion ability of [Ti] become stronger and fine Al 3 Ti is produced [29,30]. As seen in Table 1  It can be seen from Figure 4 that the amount of Al2O3 decreases as the content of La2O3 increases.
[La] can reduce the surface tension of the particles, so the surface activity of Al2O3 will increase when La dissolves in the melt and promotes the reaction of Al2O3 with the C powder and N2 in the air to form AlN. However, because the amount and volume of AlN is small, it is not shown in Figure 4 (although these values have been proven in previous experiments [11]). Although the amount of Al2O3 is small, the grain size of some Al2O3 is 25 ± 5 μm in the Al-TiO2-C-0.4La2O3 refiner, which is 11± 3 μm in the Al-TiO2-C-0.2La2O3 refiner. Thus, the diffraction peaks become high again in the XRD pattern for the Al-TiO2-C-0.4La2O3 refiner. Because the system is still oxygen-rich and hyperthermal, the carbon monoxide (CO) produced in reaction (3) will combust [30]. Al2O3 diffusion can be promoted by the local high temperature region generated by CO combustion. Therefore, the amount of Al2O3 decreases, but its distribution is more uniform.
The morphology of Al3Ti does not change significantly after adding La2O3. It remains lath-like, but the length of Al3Ti is shortened. Al3Ti is 1200 ± 150 μm in the Al-TiO2-C refiner but is only 800 ± 100 μm in the Al-TiO2-C-0.2La2O3 refiner. Because Al20Ti2La will dissolve at a high temperature and release [La], RE elements usually aggregate at phase boundaries or grain boundaries [31]. Some [La] aggregates on the phase boundaries of Al3Ti and inhibits Al3Ti growth. It can be seen in Figure 6d and (e) that there is also a certain amount of La in the area where the titanium element is distributed. Moreover, researchers have proven that Al3Ti is greatly affected by the reaction temperature. After the addition of La2O3, the CO from reaction (3) combusts with the O2 in the melt or air and produces local high temperature areas. Thus, the diffusion ability of [Ti] become stronger and fine Al3Ti is produced [29,30]. As seen in Table 1, there is a decrease in the content of Al3Ti as the content of La2O3 increases. Adding La2O3 can produce Al20Ti2La and reduce the Ti content in the melt, thereby decreasing the amount of Al3Ti.  It is difficult to see the effect of La 2 O 3 content on TiC in Figure 4, but it can be seen in Figure 3 that the diffraction peaks of TiC for the Al-TiO 2 -C-XLa 2 O 3 (X = 0.1-0.4) refiners are lower than those for the Al-TiO 2 -C refiner. The TiC diffraction peaks become extremely low for the Al-TiO 2 -C-0.3La 2 O 3 refiner. Based on reactions 1-5, adding La 2 O 3 will consume C powder and form the Al 20 Ti 2 La phase. Furthermore, the wettability of Al 2 O 3 and C is improved by [La]. Al 2 O 3 and C will react with N 2 in the air and form AlN, which also consumes C, so the amount of TiC is reduced. This phenomenon was proven in our previous experiment [11]. However, adding the rare element La 2 O 3 not only promotes the diffusion ability of [Ti] but also generates CO. CO oxidation will create a local high temperature and also facilitate TiC diffusion, making the distribution more uniform.  Figure 7a shows a macrograph of the industrial aluminum without a refiner. Here, the unrefined industrial aluminum is relatively coarse. Figure 7d presents a macrograph of the industrial aluminum refined by the Al-TiO 2 -C-0.2La 2 O 3 refiner. It can be seen that the coarse equiaxed crystals have all been replaced by fine crystals. It is difficult to see the effect of La2O3 content on TiC in Figure 4, but it can be seen in Figure 3 that the diffraction peaks of TiC for the Al-TiO2-C-XLa2O3 (X = 0.1-0.4) refiners are lower than those for the Al-TiO2-C refiner. The TiC diffraction peaks become extremely low for the Al-TiO2-C-0.3La2O3 refiner. Based on reactions 1-5, adding La2O3 will consume C powder and form the Al20Ti2La phase. Furthermore, the wettability of Al2O3 and C is improved by [La]. Al2O3 and C will react with N2 in the air and form AlN, which also consumes C, so the amount of TiC is reduced. This phenomenon was proven in our previous experiment [11]. However, adding the rare element La2O3 not only promotes the diffusion ability of [Ti] but also generates CO. CO oxidation will create a local high temperature and also facilitate TiC diffusion, making the distribution more uniform. Figure 7 shows macrographs of the industrial aluminum grains refined with different Al-TiO2-C-XLa2O3 refiners. The Al-TiO2-C-XLa2O3 refiners all show grain refining potency on industrial aluminum, although their efficiencies are quite different. Figure 7a shows a macrograph of the industrial aluminum without a refiner. Here, the unrefined industrial aluminum is relatively coarse. Figure 7d presents a macrograph of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner. It can be seen that the coarse equiaxed crystals have all been replaced by fine crystals.  Figure 8 shows macrographs of the industrial aluminum grain refined with different refiners. Figure 8a is a macrograph of industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner, while Figure 8b shows a macrograph of the industrial aluminum refined by the Al-5Ti-B refiner. As can be seen, the grain refining effects of the two refiners are similar.  Figure 8 shows macrographs of the industrial aluminum grain refined with different refiners. Figure 8a is a macrograph of industrial aluminum refined by the Al-TiO 2 -C-0.2La 2 O 3 refiner, while Figure 8b shows a macrograph of the industrial aluminum refined by the Al-5Ti-B refiner. As can be seen, the grain refining effects of the two refiners are similar. Figure 9a shows the microstructure of the industrial aluminum. Here, the grain size is very big. Figure 9d shows the microstructure of the industrial aluminum refined by the Al-TiO 2 -C-0.2La 2 O 3 refiner. Compared with Figure 9a, the grain size is very small. Table 3 shows the grain size of the industrial aluminum refined by different refiners. The grain size of the aluminum refined by the Al-TiO 2 -C-XLa 2 O 3 (X = 0.2, 0.3) refiners was measure by the line intercept method in the corresponding microstructure. To improve accuracy, we selected more than 50 lines randomly. The grain size of the industrial aluminum refined by other refiners is very big, so we have to measure the corresponding macrostructure. To improve accuracy, we selected more than 50 grains randomly and then took the average. The grain of the industrial aluminum was 3800 µm, but the grain size of the industrial aluminum refined by the Al-TiO 2 -C-0.2La 2 O 3 refiner was 320 µm, with a duction of 91.6%. Here, the grain refinement is the most effective.  Figure 9a shows the microstructure of the industrial aluminum. Here, the grain size is very big. Figure 9d shows the microstructure of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner. Compared with Figure 9a, the grain size is very small. Table 3 shows the grain size of the industrial aluminum refined by different refiners. The grain size of the aluminum refined by the Al-TiO2-C-XLa2O3 (X = 0.2, 0.3) refiners was measure by the line intercept method in the corresponding microstructure. To improve accuracy, we selected more than 50 lines randomly. The grain size of the industrial aluminum refined by other refiners is very big, so we have to measure the corresponding macrostructure. To improve accuracy, we selected more than 50 grains randomly and then took the average. The grain of the industrial aluminum was 3800 μm, but the grain size of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner was 320 μm, with a duction of 91.6%. Here, the grain refinement is the most effective.    Figure 9a shows the microstructure of the industrial aluminum. Here, the grain size is very big. Figure 9d shows the microstructure of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner. Compared with Figure 9a, the grain size is very small. Table 3 shows the grain size of the industrial aluminum refined by different refiners. The grain size of the aluminum refined by the Al-TiO2-C-XLa2O3 (X = 0.2, 0.3) refiners was measure by the line intercept method in the corresponding microstructure. To improve accuracy, we selected more than 50 lines randomly. The grain size of the industrial aluminum refined by other refiners is very big, so we have to measure the corresponding macrostructure. To improve accuracy, we selected more than 50 grains randomly and then took the average. The grain of the industrial aluminum was 3800 μm, but the grain size of the industrial aluminum refined by the Al-TiO2-C-0.2La2O3 refiner was 320 μm, with a duction of 91.6%. Here, the grain refinement is the most effective. Table 3. Grain size of the industrial aluminum refined by the Al-TiO2-C-XLa2O3 refiners

Refiners
No   Table 3. Grain size of the industrial aluminum refined by the Al-TiO 2 -C-XLa 2 O 3 refiners.

Thermal Analysis
The conclusion is usually drawn by analyzing the relationship between temperature and time when evaluating the grain refining effect by computer-aided cooling curve analysis (CA-CCA). However, temperature variations are usually not obvious in the cooling curve. Therefore, the phase transition temperature, the solidification interval, and other characteristic parameters are determined by the first derivative curve (dT/dt). Then, we take the cooling curve of the industrial aluminum after adding 0.3 wt % of the Al-TiO 2 -C-0.2La 2 O 3 refiner and its first derivative curve; the characteristic parameters are described in Figure 10.
In the initial stage, the point on dT/dt where the slope of dT/dt changes abruptly is marked as 1, which means that the cooling rate changes obviously at point 1. The latent heat of crystallization is released during the crystallization of the aluminum and leads to a change in the cooling rate, so point 1 is considered an accurate starting moment for primary nucleation (α-Al). The corresponding temperature on the cooling curve is T N , and the corresponding time is t N .
when evaluating the grain refining effect by computer-aided cooling curve analysis (CA-CCA). However, temperature variations are usually not obvious in the cooling curve. Therefore, the phase transition temperature, the solidification interval, and other characteristic parameters are determined by the first derivative curve (dT/dt). Then, we take the cooling curve of the industrial aluminum after adding 0.3 wt % of the Al-TiO2-C-0.2La2O3 refiner and its first derivative curve; the characteristic parameters are described in Figure 10.
In the initial stage, the point on dT/dt where the slope of dT/dt changes abruptly is marked as 1, which means that the cooling rate changes obviously at point 1. The latent heat of crystallization is released during the crystallization of the aluminum and leads to a change in the cooling rate, so point 1 is considered an accurate starting moment for primary nucleation (α-Al). The corresponding temperature on the cooling curve is TN, and the corresponding time is tN. The point of dT/dt = 0 during the rise of the first derivative curve is marked as 2, which corresponds to Tmin. At this time, the latent heat of crystallization is equal to the amount of heat lost in the melt. Point 3 is the peak of dT/dt, which is the largest number of nucleation, corresponding to Tmax.
The growth temperature of primary nucleation is deduced when dT/dt drops to point 4 (where dT/dt = 0 again). Point 4′s corresponding temperature on the cooling curve is TG, and the corresponding time is tG. ΔtN is the primary transformation time (ΔtN = tG−tN). Both the crystallization and growth of grains excrete heat, but the heat generated by crystallization is greater than that generated by grain growth before point 4. However, there will be almost no new nucleation when the nucleation rate reaches a certain level during the solidification process after point 4, because the driving force cannot reach critical nucleation. Therefore, the grain growth will be the main source of heat. Point 5 is a new peak of the cooling curves that has not been documented previously. Figure 11 shows the cooling curve of industrial aluminum refined by Al-TiO2-C-XLa2O3 refiners (X = 0, 0.1, 0.2, 0.3, 0.4). There is only one exothermic peak visible in the cooling curves of the aluminum without any refiners, but it can be seen from the first derivative that an exothermic peak appears in the region (2) after the addition of the Al-TiO2-C-XLa2O3 refiners. The peak value is the highest after adding 0.3 wt % Al-TiO2-C-0.2La2O3 refiner.
This study shows that some Al3Ti dissolves and produces [Ti] when the temperature is high [32][33][34]. It can be seen from Figure 11a that there is no peak in the cooling curve of the industrial aluminum without adding refiners, which indicates that a peak is not generated during the aluminum crystallization process. After adding a Al-TiO2-C refiner (without rare earth oxides), an exothermic peak is generated in region 2. It means that this peak does not originate from the rare earth oxides, so there are only two potential sources for this peak, Al3Ti and TiC. However, TiC is stable, and there is no C in industrial aluminum, so it this peak cannot be caused by the generation of TiC. Therefore, Al3Ti(II) is the source of the peak. In this paper, Al3Ti is produced in Al-TiO2-C- The point of dT/dt = 0 during the rise of the first derivative curve is marked as 2, which corresponds to T min . At this time, the latent heat of crystallization is equal to the amount of heat lost in the melt. Point 3 is the peak of dT/dt, which is the largest number of nucleation, corresponding to T max .
The growth temperature of primary nucleation is deduced when dT/dt drops to point 4 (where dT/dt = 0 again). Point 4 s corresponding temperature on the cooling curve is T G , and the corresponding time is t G . ∆t N is the primary transformation time (∆t N = t G −t N ). Both the crystallization and growth of grains excrete heat, but the heat generated by crystallization is greater than that generated by grain growth before point 4. However, there will be almost no new nucleation when the nucleation rate reaches a certain level during the solidification process after point 4, because the driving force cannot reach critical nucleation. Therefore, the grain growth will be the main source of heat. Point 5 is a new peak of the cooling curves that has not been documented previously. Figure 11 shows the cooling curve of industrial aluminum refined by Al-TiO 2 -C-XLa 2 O 3 refiners (X = 0, 0.1, 0.2, 0.3, 0.4). There is only one exothermic peak visible in the cooling curves of the aluminum without any refiners, but it can be seen from the first derivative that an exothermic peak appears in the region (2) after the addition of the Al-TiO 2 -C-XLa 2 O 3 refiners. The peak value is the highest after adding 0.3 wt % Al-TiO 2 -C-0.2La 2 O 3 refiner.
This study shows that some Al 3 Ti dissolves and produces [Ti] when the temperature is high [32][33][34]. It can be seen from Figure 11a that there is no peak in the cooling curve of the industrial aluminum without adding refiners, which indicates that a peak is not generated during the aluminum crystallization process. After adding a Al-TiO 2 -C refiner (without rare earth oxides), an exothermic peak is generated in region 2. It means that this peak does not originate from the rare earth oxides, so there are only two potential sources for this peak, Al 3 Ti and TiC. However, TiC is stable, and there is no C in industrial aluminum, so it this peak cannot be caused by the generation of TiC. Therefore, Al 3 Ti (II) is the source of the peak. In this paper, Al 3 Ti is produced in Al-TiO 2 -C-XLa 2 O 3 refiners, and Al 3 Ti (II) is produced in the industrial aluminum refined by the Al-TiO 2 -C-XLa 2 O 3 refiners. To make this easier to understand, we add (II) 3Al + [Ti] = Al 3 Ti (II) Al reacts with [Ti] to form Al 3 Ti (II) , according to reaction (6). The standard Gibbs generation free energy of reaction (6) can be expressed as ∆G 0 T = −144.242 + 0.021T (kJ/mol) [35]. When the temperature is below 700 • C, ∆G 0 T < 0. This means the reaction can react spontaneously. Moreover, the distribution of [Ti] produced by the dissociation of Al 3 Ti is uneven. In some positions, the amount of Ti is higher than the average level, which satisfies the nucleation condition of Al 3 Ti. Therefore, we believe that the peak is caused by the generation of Al 3 Ti (II) . The intensity of the peak is determined by the heat released in the system, and the key to this release is the amount of Al 3 Ti (II) . Compared with the industrial aluminum refined by the Al-TiO 2 -C refiner, [La] will be produced in the Al melt after adding Al-TiO 2 -C-0.2La 2 O 3 refiners. On the one hand, [La] can improve the wettability of [Ti] and Al and promote the production of Al 3 Ti (II) . Moreover, adding [La] can improve the melt fluidity and increase the possibility of Al contacting [Ti] [36]. On the other hand, [La] will promote Al nucleation, and the latent heat is greater during nucleation. Thus, the intensity of the peak is highest in the cooling curve of the aluminum refined by the Al-TiO 2 -C-0.2La 2 O 3 refiners.
Point 6 is considered to be the accurate end point of primary nucleation; the corresponding temperature on the cooling curve is T E . The characteristic parameters obtained in the cooling curve and the first derivative curve are shown in Table 4. Table 4. List of characteristic parameters.

Temperature Labels Characteristic Description Corresponding Characteristic Points
Al reacts with [Ti] to form Al3Ti(II), according to reaction (6). The standard Gibbs generation free energy of reaction (6) can be expressed as Δ = −144.242 + 0.021T (kJ/mol) [35]. When the temperature is below 700 °C, Δ ﹤ 0. This means the reaction can react spontaneously. Moreover, the distribution of [Ti] produced by the dissociation of Al3Ti is uneven. In some positions, the amount of Ti is higher than the average level, which satisfies the nucleation condition of Al3Ti. Therefore, we believe that the peak is caused by the generation of Al3Ti(II). The intensity of the peak is determined by the heat released in the system, and the key to this release is the amount of Al3Ti(II). Compared with the industrial aluminum refined by the Al-TiO2-C refiner, [La] will be produced in the Al melt after adding Al-TiO2-C-0.2La2O3 refiners. On the one hand, [La] can improve the wettability of [Ti] and Al and promote the production of Al3Ti(II). Moreover, adding [La] can improve the melt fluidity and increase the possibility of Al contacting [Ti] [36]. On the other hand, [La] will promote Al nucleation, and the latent heat is greater during nucleation. Thus, the intensity of the peak is highest in the cooling curve of the aluminum refined by the Al-TiO2-C-0.2La2O3 refiners.
Point 6 is considered to be the accurate end point of primary nucleation; the corresponding temperature on the cooling curve is TE. The characteristic parameters obtained in the cooling curve and the first derivative curve are shown in Table 4.   Table 5 shows the solidification characteristic parameters of the industrial aluminum with different refiners. The starting nucleation temperature (TN) of the α-Al without any refiners is 662 °C. However, after adding the Al-TiO2-C-XLa2O3 refiners, The TN becomes larger. When adding the Al-  Table 5 shows the solidification characteristic parameters of the industrial aluminum with different refiners. The starting nucleation temperature (T N ) of the α-Al without any refiners is 662 • C. However, after adding the Al-TiO 2 -C-XLa 2 O 3 refiners, The T N becomes larger. When adding the Al-TiO 2 -C-0.2La 2 O 3 refiner, the α-Al nucleation temperature was 679.38 • C, which is 17 • C higher than the industrial aluminum without an added refiner. Homogeneous nucleation is the main nucleation mode in the melt without refiners, so the nucleation temperature is low. After adding refiners, heterogeneous nucleation is promoted. Heterogeneous nucleation occurs earlier than homogeneous nucleation, so its nucleation temperature is higher. T min is the highest (660.15 • C) after the addition of the Al-TiO 2 -C-0.2La 2 O 3 refiner. ∆T is the recalescence temperature, which is smallest (0.52 • C) after adding the Al-TiO 2 -C-0.2La 2 O 3 refiner. In addition, the primary transformation time (∆t N ) is the longest after adding the Al-TiO 2 -C-0.2La 2 O 3 refiner at 25.5 s.
This phenomenon corresponds to Figure 7. The grain size is the smallest after adding the Al-TiO 2 -C-0.2La 2 O 3 refiner. The longer the nucleation time, the more crystal nuclei are formed. Thus, the longer the nucleation time, the better.
The heterogeneous nucleation of α-Al is promoted significantly by adding Al-TiO 2 -C-XLa 2 O 3 refiners, as it has been proven that Al 3 Ti, TiC, and Al 20 Ti 2 La are the most significant heterogeneous nuclei of α-Al [29]. The quantity and morphology of Al 3 Ti are important factors that affect the nucleation ability of Al-TiO 2 -C-XLa 2 O 3 refiners. After adding 0.2 wt % La 2 O 3 , the morphology of Al 3 Ti becomes fine and its quantity becomes large. La 2 O 3 not only greatly promotes α-Al nucleation but also forms a large amount of Al 3 Ti (II) . In addition, the process of forming Al 3 Ti (II) also generates a large amount of heat, thereby providing a large amount of energy and again promoting the nucleation of α-Al.
Adding La 2 O 3 can help form TiC by improving the wettability between C and [Ti]. However, La 2 O 3 also consumes C and forms Al 20 Ti 2 La, which is the main reason for the decrease in the amount of TiC. Researchers have proven that TiC surrounded by a titanium-rich layer can promote its nucleation abilities [37]. There is a large amount of Al 3 Ti in Al-TiO 2 -C-0.2La 2 O 3 refiners. After dissolution at high temperatures, a large amount of [Ti] can promote the nucleation of α-Al with TiC. Moreover, the content of La 2 O 3 is small, so its influence on the amount of TiC is small. Al 20 Ti 2 La can only be formed by adding La 2 O 3 in Al-TiO 2 -C refiners. Its crystal index number is 8 (matching Al), and its mismatch degree with Al is very small. Further, its density is close to that of industrial aluminum, and it is difficult to sink. Al 20 Ti 2 La will continue to dissolve and produce [La].
[La] and [Ti] aggregate to the TiC's surface and form layers comprised of [La] and [Ti], respectively [28]. These layers will make TiC become heterogeneously nucleated [29]. As seen in Figure 2, some La elements are distributed in the area where Ti is also distributed. This means that Al 3 Ti may be covered by Al 20 Ti 2 La. Al 20 Ti 2 La dissolves and produces [La], which will distribute around Al 3 Ti preferentially. Moreover, [La] is a surfactant, so it can promote the wettability of Al 3 Ti and Al.
[La] can also enhance the wettability of Al 3 Ti, TiC, Al 2 O 3 , and the aluminum melt [29,38]. However, its grain refining ability will decrease when the La 2 O 3 content exceeds 0.2% in the Al-TiO 2 -C-XLa 2 O 3 refiners. As the content of La 2 O 3 increases, C will be consumed and cause the amount of TiC and Al 3 Ti to decrease sharply. Moreover, other rare earth phases may be formed and weaken the nucleation ability when the content of La 2 O 3 is greater than 0.2% in the Al-TiO 2 -C-XLa 2 O 3 refiners.

Effect of La 2 O 3 on Sintering Process
The influence of La 2 O 3 on the sintering process is mainly reflected in three aspects: (1) improving the wettability between raw materials and promoting the reaction, (2) improving the sintering driving force, and (3) promoting material transfer.
In this experiment, the TiO 2 particle size is 90 nm. Moreover, the particle size is very small, and the surface energy is large, so the surface will absorb the surrounding particles to form secondary particles. However, compared with TiO 2 , the particle sizes of the Al powder and the C powder are large, and there are many voids in the secondary particles. The particle size of La 2 O 3 (5 µm) is small and can almost reach a lower level after milling. Nano-sized La 2 O 3 can enter these voids. Furthermore, La 2 O 3 is a surfactant, so it can promote the reaction.
Because the sintering temperature is 1250 • C, and the melting point of Al is 660 • C, the sintering occurs in the liquid-phase. In liquid-phase sintering, if the wettability of the liquid phase and solid phase is great, the liquid phase can enter the cracks and voids of the particles. After adding La 2 O 3 , the wettability between Al with TiO 2 and C is improved, and the reaction will be promoted. By adding La 2 O 3 , the surface energy is increased. Surface energy is the driving force of sintering, so the driving force is increased during sintering, which promotes sintering.
Liquid-phase sintering can be divided into three stages. Adding La 2 O 3 mainly affects the initial stage of sintering (i.e., the liquid phase and particle redistribution stage). Adding La 2 O 3 will produce CO, and solid phase particles, such as TiO 2, will be redistributed under the capillary force of the liquid when CO burns with the gas in the prefabricated blocks or escapes from the melt. In this way, the possibility of TiO 2 and C contacting Al increases. In addition, the atomic motion is more intense and tends to have more defects, such as vacancies, under high temperature. Diffusion through vacancies is also an important method for material transfer during sintering. The radius of a Ti atom is 145 pm, and the radius of a La atom is 146 pm. Ti can utilize the vacancies of La to achieve material transfer and promote sintering.
However, there are adverse effects after adding La 2 O 3 . After addingLa 2 O 3 , the wettability between Al 2 O 3 and C will increase and may produce unwanted new phases, such as AlN. No research has yet proven that AlN can be used as a heterogeneous nucleus of α-Al. Also, La 2 O 3 consumes a great deal of C, and the amount of TiC decreases sharply. Thus, the amount of La 2 O 3 must be suitable.

Conclusions
The principal conclusions of this study are summarized as follows.
(1) The content of La 2 O 3 has a significant effect on the microstructure of the novel Al-TiO 2 -C-XLa 2 O 3 refiner. As the La 2 O 3 content increases, the amount of Al 20 Ti 2 La increases, and the content of Al 3 Ti decreases as the size becomes fine. The amount of Al 2 O 3 and TiC decreases, but their distribution is more uniform.
(2) The Al-TiO 2 -C-0.2La 2 O 3 refiner has the best grain refining effect on industrial aluminum and can refine industrial aluminum to 320 µm, which is 8.4% of the grain size of the industrial aluminum without refiners. After adding the novel Al-TiO 2 -C-0.2La 2 O 3 refiner, the nucleation temperature T N is 679.21 • C, which is 17 • C higher than the nucleation temperature of the industrial aluminum without a refiner. The primary transformation time is the longest (25.5 s), 4.6 s longer than that of the industrial aluminum (20.9 s), and the ∆T is 0.52 • C, which is the lowest value.