2.1. In Situ MgAl2O4 Particles
The similar crystal structure between Al and MgAl
2O
4 provides an opportunity to form a low-energy interface between them and any of their orientations. Meanwhile, MgAl
2O
4 displays nice wettability with Al alloys, and there are no chemical reactions between Al and MgAl
2O
4 at low temperatures. All of these characteristics probably make MgAl
2O
4 particles act as potential heterogeneous nucleation sites and further enhance the grain refinement of alloys. MgAl
2O
4 particles can be synthesized in situ in Al–Mg melts through the addition of oxides or other oxygen sources, such as SiO
2, TiO
2, B
2O
3, and H
3BO
3 [
7,
19,
20]. Negative Gibbs free energy in the reactions between oxides and Al–Mg melts at processing temperatures of 650–900 °C suggests that the formation of an MgAl
2O
4 phase is thermodynamically stable. However, it is difficult to obtain a complete reaction of parent oxides and to disperse the in situ MgAl
2O
4 particles into the alloy melt, thus leading to the poor nucleation ability of MgAl
2O
4. In order to resolve this question, physical fields such as mechanical stirring, ultrasonic treatment, and intensive melt shearing have been introduced, which has resulted in in situ MgAl
2O
4 forming with, on average, small particle sizes and a diffuse distribution in the melt.
Using ultrasonic treatment, one reaction resulted in the massive growth of MgAl
2O
4 particles from an Al–Mg melt, and the oxides could be attributed to the constant removal of MgAl
2O
4 particles on the oxides’ surface and to the pushing of MgAl
2O
4 particles into the melt, thus exposing fresh oxide surfaces and allowing for the advancement of the bulk oxide reaction [
13]. Ultrasonic treatment is used as an effective physical tool to enhance the heterogeneous nucleation of an Al–Mg melt from three fields. Firstly, ultrasonic treatment introduces ultrasonic cavitation, producing intense local hotspots of temperature (5000 °C), high pressure (100 MPa), and microjets (100 m/s) [
21]. A high local temperature and pressure assist with the reaction of Al and Mg atoms on the surfaces of oxides, leading to the in situ formation of MgAl
2O
4 crystals. Secondly, microjets aid in the fragmentation of oxides as well as in the removal of MgAl
2O
4 crystals, exposing fresh oxide surfaces for further reactions. Lastly, acoustic streaming and ultrasonic treatment aid in the dispersion of small MgAl
2O
4 crystals, which are bonded by a van der Waals force. These in situ MgAl
2O
4 crystals grow with low index faces, such as {220}, {311}, and {400}, and have low interfacial energy with Al [
12,
13].
The experimental parameters, including the melting and solidification methods, the alloy compositions and melt treatments, the compositions of the Al–MgAl
2O
4 master, the addition contents and modes of the master, and the ultrasonic treatment processes, can affect the size of the microstructural components. Therefore, studies with similar experimental methods and parameters were selected to assess the grain refinement effects of MgAl
2O
4 on Al alloys. The refinement degrees of the average alloy grain sizes in References [
7,
12,
13,
18] (with different contents of MgAl
2O
4 and different ultrasonic treatment (UT) processes) are summarized in
Table 1.
In the table, it can be seen that a significant grain size reduction was observed in pure Al and Al–Mg alloys with in situ MgAl2O4 and ultrasonic treatment, while few grain refinements (a reduction of 2–3-fold) were obtained with just in situ MgAl2O4 and hardly any grain refinements were obtained (a reduction of ~1.1-fold) with just ultrasonic treatment. This implies that the combination of in situ MgAl2O4 and ultrasonic treatment contributes to a significant grain size reduction in alloys. The presence of MgAl2O4 particles acting as nucleated sites is critical, and ultrasonic treatment results in the in-situ-formed MgAl2O4 being nanoscale-sized, with a uniform dispersion and a 2–4-fold increased number density when compared to previous formulations. The mean grain size of MgAl2O4 was almost equivalent with different UT times and temperatures. It is worth noting that the grain refinement effects with the above melting and solidification methods, UT temperatures and times, and mean grain sizes of MgAl2O4 in the different experiments were similar except for the larger grain refinement effect obtained with a short UT time (30 s) with pure Al compared to more UT time (5 min) with the Al–Mg alloy. This suggests that UT time and temperature do not play decisive roles in grain refinement. The same refinement degree for average alloy grain sizes of the Al–4Mg alloy with different mass percents of MgAl2O4 (0.58 wt % and 3 wt %) was achieved due to the different lengths of the samples (Φ 20 × 80 mm and Φ 20 × 120 mm ) in the experiment. Therefore, the effect of the mass percent of MgAl2O4 on grain refinement was then mainly studied and discussed in terms of another physical force, intensive melt shearing.
Intensive melt shearing is another approach used in grain refinement treatments of Al and Al–Mg alloys. When given enough oxidation time, porous MgO initially forms on the surface of the Al–4Mg alloy, and MgAl
2O
4 particles later form, covered with a thin layer composed of Al
2O
3 from the reaction of liquid aluminum and oxygen, which is introduced from the air through the porous MgO. Therefore, it seems that MgAl
2O
4 generated naturally during oxidation has difficulty serving as a direct substrate for the nucleation of Al due to the separation of Al
2O
3 [
11]. Due to the similar crystal structure of Al and MgAl
2O
4 and due to the MgAl
2O
4 solid phase forming in the melt prior to the solidification of Al, it has been proven that intensive melt shearing breaks up the MgAl
2O
4 oxide films and disperses the potent oxide particles, which leads to potent heterogeneous nucleation and a grain refinement effect in Al–Mg alloys. Detailed research has been carried out on the effect of the Mg content and intensive melt shearing on grain refinement in Al–Mg alloys [
10]. Al–Mg alloys were prepared in an electric resistance furnace using commercial purity Al and Mg. After Mg was dissolved under the protection of Ar gas, the melt was held at a constant temperature of 700 °C for 4 h to assist in the natural reaction of MgAl
2O
4, with an average grain size of hundreds of nanoscales. The sheared samples were prepared through sheared melting for 60 s at 700 °C and then poured into a copper mold at a consistent cooling rate of 3.5 K s
−1 in the central region. The dimensions of the mold were about 25 mm in diameter at the bottom, 60 mm in diameter at the top, and 65 mm in length. The nonsheared samples were treated in the same way without the melting and shearing. Significant grain refinement was achieved with intensive melt shearing or by increasing the content of Mg when it was less than 1 wt %. When comparing the grain sizes, there was a critical Mg content observed, around 2 wt %, regardless of whether the Al–Mg alloys sheared or not, as they were nearly equivalent and almost constant. A similar result was observed in the Al–5Mg alloy with the addition of Ti. Intensive melt shearing introduced further grain refinement when the content of Ti was less than 0.05 wt %, while the grain sizes of the nonsheared and sheared Al–5Mg alloys were constant with more than a critical content of 0.05 wt %. The effect of the mass percent of MgAl
2O
4 or Mg with UT or an intensive melt shearing process on grain refinement is comprehensively considered and illustrated in
Figure 1 with data from References [
7,
10,
12,
13,
18].
It is interesting to note that in
Figure 1a, the average grain sizes of the pure Al and Al–Mg alloys obviously decrease with an increase in the content of MgAl
2O
4 or Mg, which is in the low range of 0 to 1 wt %. At the same time, UT and intensive melt shearing further highly strengthen the grain size reduction in the low content range of MgAl
2O
4 or Mg, which suggests that physical force is an effective method in breaking up the MgAl
2O
4 film into nanoparticles, dispersing MgAl
2O
4 particles into melts, and then enhancing the heterogeneous nucleation effect of MgAl
2O
4 particles. However, with a continuing increase in the content of MgAl
2O
4 or Mg, the average grain size decreases slowly and has a gradual leveling tendency: meanwhile, the effect of the size reduction is small under physical force when the content of MgAl
2O
4 or Mg is more than the critical value of around 2 wt %. In order to understand the respective contributions of endogenous MgAl
2O
4 particles and the physical force of UT or shearing, the reduction degree of grain size with respect to the mass percent of MgAl
2O
4 or Mg with UT or a shearing process was converted into
Figure 1b from
Figure 1a. It can be seen that the contribution of endogenous MgAl
2O
4 to grain size reduction is 2–3-fold, while that of physical force is about 5–8 and up to 11–12-fold when the content of MgAl
2O
4 or Mg is in the low range, less than 1 wt %. The contribution of physical force to grain size reduction becomes stable after reaching the critical content of MgAl
2O
4 or Mg.
Combining the above experimental results and analysis, it is clear that physical force plays a more important role when there is an assured existence of endogenous MgAl2O4 in a low content range. This implies that physical force enhances the heterogeneous nucleation and grain refinement of Al by tuning the wetting and distribution of MgAl2O4 particles. In addition, and more importantly, it is clear that the MgAl2O4 content is a critical factor in the grain refinement of Al and Al–Mg alloys, since similar grain sizes are obtained with or without physical force when the content of MgAl2O4 or Mg exceeds 2 wt %. This also proves that the endogenous MgAl2O4 phase is a potent heterogeneous core for the nucleation of Al and Al–Mg alloys and plays a vital role in grain size reduction under the following conditions: a uniform distribution, hundreds of nanoscale-sized to several microscale-sized particles, and a mass percent of about 2 wt % (whether there is physical force or not).
2.2. Exogenous MgAl2O4
Both MgO and MgAl
2O
4 are common oxides formed during the preparation and remelting processes of Al–Mg alloys and are regarded as being effective heterogeneous nucleating agents for Al-based alloys due to their similar lattice structures and small lattice misfits. However, some researchers have reported that liquid Al reacts with MgO substrates in various exposed crystal planes over a wide temperature span [
16,
17,
22,
23]. In these research results, Sun [
16] found MgAl
2O
4 is a final product at a normal casting temperature between 700 °C and 800 °C with a holding time of 3 min. MgAl
2O
4 products are straight, and distinct layers or small islands with thicknesses of 10–40 nm form between Al and MgO. This can be referred to as exogenous MgAl
2O
4 in this case due to the reaction product MgAl
2O
4 always being retained outside of the Al melt during the nucleation process. The cube-on-cube ORs between Al, MgAl
2O
4, and MgO have also been confirmed by Sun, such as {200} <001>Al||{200} <001>MgAl
2O
4||{200} <001>MgO, {220} <001>Al||{220} <001>MgAl
2O
4||{220} <001>MgO, and {111} <110>Al||{111} <110> MgAl
2O
4||{111} <110>MgO. Both the cube-on-cube ORs and small nucleated undercooling (3–8 °C) suggest that MgAl
2O
4 is a perfect catalyzer for the nucleation of liquid Al.
J. Morgile et al. [
22,
23] have clarified that MgAl
2O
4 is an intermediate product and Al
2O
3 is a final product of the reaction at a high temperature between the Al melt and MgO substrate. Similarly to Morgile’s results, the reaction product MgAl
2O
4 (with a thickness between 100 and 300 nm) was also seen at the Al/MgO interface in our research. High-purity Al (>99.999%) was melted at 1027 °C with a holding time of 5–10 s and then cooled on the selected terminated planes of the MgO substrate from 1027 °C, with a cooling rate of 20 K/s in a high-vacuum chamber. The interface between MgAl
2O
4 and MgO was approximately straight, while the interface between MgAl
2O
4 and Al was a zigzag, as seen in
Figure 2. To further confirm the morphology of MgAl
2O
4, a high-angle annular dark-field (HAADF) Z-contrast image of the Al/MgO interface was obtained (
Figure 2b, corresponding to the frame of
Figure 2a). Because the image contrast is proportional to the atomic mass, the contrast of the Al area is bright, while that of the MgO area is dark. The contrast of MgAl
2O
4 is distinct from Al and MgO, with a straight interface with MgO and a zigzag interface with Al.
The morphology and thickness of MgAl2O4 at the Al/MgO interface were different from the straight and distinct layer or small islands of MgAl2O4 formed at a normal casting temperature (between 700 and 800 °C). This suggests that the different reactions of Al and MgO occur at different heating temperatures, leading to a different morphology and thickness of the MgAl2O4 product. It can be concluded that the exogenous MgAl2O4 layer formed at the interface of pure Al and MgO can act as a new heterogeneous nucleation substrate for Al when the heating temperature is at a normal casting temperature or at a higher temperature with a holding time of a few seconds.
The heterogeneous nucleation effect of exogenous MgAl
2O
4 oxides acting as direct nucleated substrates was investigated at a high heating temperature by Zhang et al. [
17] in an Al/MgAl
2O
4 system at 1027 °C with a holding time of 30 s. An Al
2O
3 dendritic reaction from the Al melt and MgAl
2O
4 substrate was noted, and meanwhile, part of the MgAl
2O
4 substrate (with a new crystal plane) was exposed to the Al melt. As a result, differently from the above three cube-on-cube ORs, the OR between the Al and freshly exposed MgAl
2O
4 substrate was {111} <011>Al||{200} <013>MgAl
2O
4 with a large misfit of about 8.36% and a distorted layer about 1 nm thick, which relieved the strain between the matching planes. This indicates that MgAl
2O
4 substrates easily form alumina and are not potent nucleation substrates for Al alloys at a high heating temperature and long holding time.
To understand the direct effect of the exogenous MgAl
2O
4 substrate on the heterogeneous nucleation of Al at 1027 °C with a short holding time of 3–5 s, high-resolution transmission electron microscopy (TEM) was carried out at the Al/MgAl
2O
4 nucleation interface in the <001> MgAl
2O
4-zone direction.
Figure 3b is a selected area electron diffraction (SAED) pattern taken from both MgAl
2O
4 and the adjacent Al matrix along their <001> zone axes across the MgAl
2O
4/Al interface in
Figure 3a. The SAED pattern on the MgAl
2O
4/Al interface shows overlapped spots. Through an analysis of the SAED pattern, we found two sets of diffraction patterns. One set was a face-centered cubic (FCC) structure with a d-spacing of 0.196 nm and 0.278 nm, which corresponded to {400} and {2
0} planes of MgAl
2O
4 (as indexed with small solid circles in
Figure 3c). The other set was also an FCC structure, with d-spacing of about 0.196 nm and 0.139 nm, which corresponded to {200} and {2
0} planes of Al (as indexed with large open circles).
Figure 3c gives a schematic indexing of the SAED pattern, indicating that the same crystal planes and the same crystal directions were parallel to each other in the MgAl
2O
4 and Al crystals. From the evidence of the SAED, it is clear that there was a cube-on-cube OR between the MgAl
2O
4 and Al matrices, which was {200} <001>Al||{400} <001>MgAl
2O
4.
It is worth noting that both MgAl
2O
4 and Al had an FCC crystal structure and that the lattice parameter for MgAl
2O
4 was 0.80831 nm, about double that of Al (0.40494 nm) [
2]. The theoretical crystal plane spacings of {400} MgAl
2O
4 and {200} Al were 0.2021 nm and 0.2025 nm, respectively. Theoretically, the lattice misfits for the parallel crystal planes between Al and MgAl
2O
4 in
Figure 3a were small, and the calculation results can be seen in
Table 2.
The calculated lattice misfits of the different parallel planes between Al and MgAl
2O
4 were near those (0.09%) of the Al and Al
3Ti monolayer between the Al and TiB
2 substrate [
24], suggesting that MgAl
2O
4 acts as a potent heterogeneous substrate that can be considered for actual industrial applications. Actually, well-defined atomic rows and lattice planes can be identified in
Figure 3a. As can be seen in
Figure 3, the {200} Al plane was parallel to the {400} MgAl
2O
4 plane, and the plane spacing of {200} Al was the same as that of {400} MgAl
2O
4. Meanwhile, {020} Al was parallel to {040} MgAl
2O
4, and their d-spacings were also equal to each other. The {2
0} Al was parallel to the {2
0} MgAl
2O
4, while the
d-spacing (0.278 nm) of the {2
0} MgAl
2O
4 was double that of (0.139 nm) {2
0} Al. The parallel crystal planes, the crystal directions, and the same or integral multiple-crystal-plane spacing indicated that the cube-on-cube OR {200} <001>Al||{400} <001>MgAl
2O
4 was the expected value.
According to theoretical lattice misfit calculations, there should be small lattice misfits in Al/MgAl
2O
4 along three sets of parallel crystal planes. However, the crystal plane spacings between the three sets of parallel crystal planes were the same or double. This means that there was lattice distortion at the interface, which was consistent with our previous study, where the nucleated-phase Al fit the substrate with limited lattice distortion within a small lattice misfit (
f < 3.1%) [
25]. These studies prove that both in situ MgAl
2O
4 (using physical force) and endogenous MgAl
2O
4 (controlling for the heating temperature) are possibly potent heterogeneous nucleation substrates in the heterogeneous nucleation of Al alloys.