Removal of Low-Content Impurities from Pure Al by Supergravity Combined with Semi-Solid Method

: Recently, the puriﬁcation technology for high-purity aluminum (Al) has become the focus and difﬁculty of the majority of researchers. In this study, a novel approach for removing iron (Fe) impurities from pure Al via combining the supergravity ﬁeld and semi-solid reﬁning was proposed. Various separation temperatures ( T ), holding times ( t h ), and separation times ( t s ) were applied within a gravitational ﬁeld to explore their impact on the puriﬁcation process and its underlying mechanisms. The optimal conditions were achieved at T = 653 ◦ C, t h = 40 min, t s = 3 min, and a gravity coefﬁcient G = 1000, with the loss rate of puriﬁed Al reaching up to 4.1% and the removal rate of Fe reaching 81.9%. The Fe content in pure Al was reduced from 0.32 wt.% to 0.06 wt.%. Moreover, the puriﬁed mechanism of supergravity in a semi-solid method was reported for the ﬁrst time. It was concluded that supergravity could decrease the value of the effective distribution coefﬁcient (k e ), thereby promoting the continuous migration of Fe impurities at the solidiﬁcation interface into the liquid phase. The Fe-rich phase in the Al melt was completely ﬁltered to the lower part of the crucible in the supergravity ﬁeld, completing the further puriﬁcation of the pure Al.


Introduction
In nature, the conductivity of metals from high to low is silver (Ag), copper (Cu), gold (Au), and aluminum (Al).However, Ag and Au, being precious metals, have higher prices, making them less suitable for wiring.Although the conductivity of Al is only 2/3 of Cu, its density is only 1/3 of Cu.Therefore, the weight of Al wire is only half that of Cu wire for the same amount of electricity.It has gradually become an inexpensive alternative material to replace Cu cables and is widely used to make wires, cables, and conductive components in the electrical industry [1][2][3][4][5].The electrical conductivity of Al is easily worsened by impurities in the matrix.In recent years, numerous researchers have dedicated substantial efforts to the research and development of the purification process for high-purity Al.Nevertheless, the capability to produce high-purity Al is currently limited to developed countries like the United States, Germany, and Japan.In China, only a handful of manufacturers possess the capacity to produce high-purity Al [6][7][8].Therefore, there is an urgent imperative to intensify research and development endeavors aimed at advancing production technologies for high-purity Al within China.
At present, the methods used to purify Al with low-content impurities mainly consist of chemical methods, such as electrolysis, and physical segregation purification methods [6,9,10].The electrolysis method primarily involves three-layer liquid electrolytic refining, while the segregation method includes vacuum evaporation, fractional crystallization, directional solidification, and zone melting methods [11].Although three-layer liquid electrolysis is effective for large-scale purification, it is expensive and harmful to the environment due to the use of fluoride electrolytes.Therefore, this study focuses on the segregation method, also Metals 2023, 13,1945.https://doi.org/10.3390/met13121945https://www.mdpi.com/journal/metalsMetals 2023, 13,1945 2 of 14 known as the solidification purification method, which purifies Al based on the solubility of each impurity element in the liquid phase and the solid phase.Currently, directional solidification and zone smelting methods are most commonly studied [9], but they have significant drawbacks, such as high cost and a long production cycle.Duan [10] et al. used zone melting purification for high-purity Al.This indicates that the sample needs 15 times zone melting with the travel speed of the molten area at 1.0 mm/min.The obtained contents of the impurities Cu, Si, Fe, and Ti in the middle of the sample (i.e., 140 mm from the head end) were 0.9057, 0.8116, 0.1202, and 0.4237 ppm, respectively.Fe is the most pervasive impurity element in Al alloys and is capable of forming Fe-rich phases with other elements, leading to detrimental conductivity and mechanical properties of the Al alloys [6,12].Trace amounts of Fe impurity elements in pure Al typically form low-melting-point eutectic compounds with the matrix, making them difficult to remove.At higher purification temperatures, trace Fe impurities are thoroughly dissolved in the matrix.At lower temperatures, solid-liquid separation is theoretically possible.However, due to the increase in the matrix Al solid phase fraction in pure Al, separating the low-melting-point eutectic Fe-rich phase from the matrix through conventional means becomes difficult.Therefore, it is particularly important to address how to remove the low-melting-point Fe-rich phases from pure Al samples at lower temperatures [13,14].Semi-solid processing for refining metal alloys was first introduced by Mehrabian [15] et al.This method involves heating the metal to the temperature range of the solid/liquid coexistence and separating the solid and liquid phases through filtration treatment [16,17], electromagnetic force [10,[18][19][20][21], or gravitational sedimentation.Cho [16] et al. investigated the industrial-scale purification of A7075 wrought Al alloy under semi-solid conditions using a servo press.It was confirmed that the purity of Al reached up to 95% at the semisolid temperature of 620 • C, with an extrusion ratio of 3 and an extrusion speed parameter of 25 mm/s.Additionally, supergravity can effectively enrich and segregate alloys with high levels of solid phase impurities at elevated temperatures due to its strong physical force, such as Al [22][23][24].Sun [22] et al. investigated the separation behavior of Al 2 O 3 particles and the prematurely precipitated Fe-rich phase around the Al 2 O 3 particles in the purification of primary Al liquid using supergravity technology.Although, Zhao [23] et al. used the supergravity field to enrich the low-content impurities from pure Al and efficiently sediment the Fe and Si impurities.However, it is noteworthy that the impurities were merely concentrated at the bottom of the samples and not separated from the Al matrix, which means the treated sample needed further treatment to separate these impurities using mechanical or chemical methods.
Herein, we investigated a novel method that combines the supergravity field with the semi-solid state to purify Al, reducing low-content impurities.This method enables the one-step removal of inclusions.Specifically, pure Al is heated to a semi-solid state, inducing the matrix Al to solidify, while the Fe-rich impurities exist in a liquid phase.Under the influence of a supergravity field, these impurities are expelled from the matrix, thereby accomplishing the purification of pure Al.The results show that under the conditions of G = 1000, T = 653 • C, t h = 40 min, and t s = 3 min, the Fe content decreased from 0.32 wt.% to 0.06 wt.%, with the removal rate of Fe reaching 81.9%.

Materials and Methods
In this study, 1060 industrial pure Al was selected as the research subject.It contains a small amount of Fe (0.32 wt.%) element, which is a typical conductive material.The composition of the raw material was determined by ICP-OES analysis, and the results are listed in Table 1.The supergravity field generated by a centrifugal apparatus is illustrated in Figure 1, along with the schematic of the experimental apparatus.This supergravity metallurgical experimental setup was adapted from the DL-8M model of a medical centrifuge manufactured by Shanghai Luxiangyi Centrifuge Instrument Co., Ltd.(Shanghai, China), comprising a heating furnace and a counterweight furnace.Both furnaces were fixed into the centrifugal rotor.A composite graphite crucible containing a porous support plate was Metals 2023, 13,1945 3 of 14 applied to accomplish the separation process.The graphite felt, with a thickness of 5 mm, a diameter of 20 mm, and a volume density of 0.12-0.14, was employed as the filter.China), comprising a heating furnace and a counterweight furnace.Both furnaces were fixed into the centrifugal rotor.A composite graphite crucible containing a porous support plate was applied to accomplish the separation process.The graphite felt, with a thickness of 5 mm, a diameter of 20 mm, and a volume density of 0.12-0.14, was employed as the filter.The gravity coefficient (G) was calculated as the ratio of the centrifugal acceleration to the normal gravity acceleration (as shown in Equation ( 1)).where w is the angular velocity (in rad•s -1 ), N is the rotating speed of the centrifuge (in r•min -1 ), r is the distance from the axis to the center of the sample (r = 0.25 m), and g is the normal gravity constant.The experimental procedure is as follows.Blocks of 1060 industrial pure Al with a diameter of less than 20 mm and a height of 30 mm were selected.The blocks were heated at 648 °C, 653 °C, or 658 °C for a duration of 20-60 min.Following the heat preservation, the supergravity apparatus was turned on and operated at an angular velocity equal to 1892 rpm (G = 1000).After the supergravity apparatus was shut off, prompt water quenching was employed to maintain its as-cast structure.Reference samples were obtained using the same method but under normal gravity conditions (acceleration 1 g).The morphology of the samples was investigated through scanning electron microscopy coupled with an energy-dispersive spectroscope (SEM-EDS, MLA 250, New York, NY, USA).The chemical composition of the inclusions was characterized by inductively coupled plasmaoptical emission spectroscopy (ICP-OES, Optima 7000DV, New York, NY, USA).The solidification path of 1060 industrial pure Al was simulated by the Scheil solidification model in Thermo-Calc software (USTB MatCom of Beijing Advanced Innovation Center for Materials Genome Engineerin).The gravity coefficient (G) was calculated as the ratio of the centrifugal acceleration to the normal gravity acceleration (as shown in Equation ( 1)).
where w is the angular velocity (in rad•s -1 ), N is the rotating speed of the centrifuge (in r•min -1 ), r is the distance from the axis to the center of the sample (r = 0.25 m), and g is the normal gravity constant.The experimental procedure is as follows.Blocks of 1060 industrial pure Al with a diameter of less than 20 mm and a height of 30 mm were selected.The blocks were heated at 648 • C, 653 • C, or 658 • C for a duration of 20-60 min.Following the heat preservation, the supergravity apparatus was turned on and operated at an angular velocity equal to 1892 rpm (G = 1000).After the supergravity apparatus was shut off, prompt water quenching was employed to maintain its as-cast structure.Reference samples were obtained using the same method but under normal gravity conditions (acceleration 1 g).The morphology of the samples was investigated through scanning electron microscopy coupled with an energy-dispersive spectroscope (SEM-EDS, MLA 250, New York, NY, USA).The chemical composition of the inclusions was characterized by inductively coupled plasma-optical emission spectroscopy (ICP-OES, Optima 7000DV, New York, NY, USA).The solidification path of 1060 industrial pure Al was simulated by the Scheil solidification model in Thermo-Calc software (USTB MatCom of Beijing Advanced Innovation Center for Materials Genome Engineerin).

Separation of Fe Impurity from Pure Al
Figure 2 presents the SEM images of the original material with marked analyzed phases, and their corresponding elemental compositions are listed in Table 2.It can be seen that the gray phase represents the Al matrix, while the gray and white spherical eutectic particles correspond to the Fe-rich phase.At the grain boundary, the white particles were the Al 13 Fe 4 phase, composed of 76.80 at.%Al and 23.20 at.%Fe.Additionally, the black areas denote the presence of shrinkage holes in the raw material.

Separation of Fe impurity from pure Al
Figure 2 presents the SEM images of the original material with marked analyzed phases, and their corresponding elemental compositions are listed in Table 2.It can be seen that the gray phase represents the Al matrix, while the gray and white spherical eutectic particles correspond to the Fe-rich phase.At the grain boundary, the white particles were the Al13Fe4 phase, composed of 76.80 at.%Al and 23.20 at.%Fe.Additionally, the black areas denote the presence of shrinkage holes in the raw material.The Scheil solidification model in Thermo-Calc was employed to obtain the solidification path of 1060 industrial pure Al in this study, as shown in Figure 3.The results indicate that the α-Al phase started to solidify at 658 °C, followed by the Al13Fe4 phase at 648 °C .Subsequently, the Al8Fe2Si phase precipitated at 646 °C, and the temperature decreased to 629 °C, at which point the industrial pure Al was completely solidified.Therefore, traditional supergravity separation techniques (utilizing density differences between different phases in a high-temperature molten state for impurity removal) have limited efficacy and cannot accomplish one-step separation.To achieve the solid-liquid separation of the sample and, thereby, purify the industrial pure Al, the subsequent experimental temperatures were set between 648 °C and 658 °C.Specifically, the Al matrix serves as the solid phase, while the Fe-rich impurities act as the liquid phase.However, as observed in Figure 3, within this temperature range, the solid phase in the sample is relatively high and the natural settling time is long.Previous studies [15] have revealed that during the solidification of the sample when the solid phase fraction reaches 0.2, dendritic networks begin to form (interpreted as a continuous solid skeleton).The permeability and fluidity of the mushy zone decrease.When the solid phase fraction exceeds 0.2, it results in the hard separation of the liquid phase at the grain boundaries from the matrix under normal gravity.Techniques such as squeezing or other methods are necessary to allow the interdendritic liquid to flow out from the dendritic network.Therefore, this section proposes a semi-solid supergravity method for purifying industrial pure Al, that is, the liquid Fe-rich impurity phase is separated from the Al sample under the supergravity field.Table 2. EDS analysis of components marked with numerical symbols in Figure 1.

No. Al (at.%) Fe (at.%) Phase
Pt.The Scheil solidification model in Thermo-Calc was employed to obtain the solidification path of 1060 industrial pure Al in this study, as shown in Figure 3.The results indicate that the α-Al phase started to solidify at 658 • C, followed by the Al 13 Fe 4 phase at 648 • C. Subsequently, the Al 8 Fe 2 Si phase precipitated at 646 • C, and the temperature decreased to 629 • C, at which point the industrial pure Al was completely solidified.Therefore, traditional supergravity separation techniques (utilizing density differences between different phases in a high-temperature molten state for impurity removal) have limited efficacy and cannot accomplish one-step separation.To achieve the solid-liquid separation of the sample and, thereby, purify the industrial pure Al, the subsequent experimental temperatures were set between 648 • C and 658 • C. Specifically, the Al matrix serves as the solid phase, while the Fe-rich impurities act as the liquid phase.However, as observed in Figure 3, within this temperature range, the solid phase in the sample is relatively high and the natural settling time is long.Previous studies [15] have revealed that during the solidification of the sample when the solid phase fraction reaches 0.2, dendritic networks begin to form (interpreted as a continuous solid skeleton).The permeability and fluidity of the mushy zone decrease.When the solid phase fraction exceeds 0.2, it results in the hard separation of the liquid phase at the grain boundaries from the matrix under normal gravity.Techniques such as squeezing or other methods are necessary to allow the interdendritic liquid to flow out from the dendritic network.Therefore, this section proposes a semi-solid supergravity method for purifying industrial pure Al, that is, the liquid Fe-rich impurity phase is separated from the Al sample under the supergravity field.
To evaluate the effects of these conditions on the experimental efficiency, several parameters, including the experimental liquid fraction that can be separated (F L ), the loss rate of the purified Al (R Al ), and the removal rate of Fe (η Fe ) were defined (as shown in Equations ( 2)-( 4)).To evaluate the effects of these conditions on the experimental efficiency, several parameters, including the experimental liquid fraction that can be separated (FL), the loss rate of the purified Al (RAl), and the removal rate of Fe (ηFe) were defined (as shown in Equations ( 2)-( 4)).
where mr is the mass of the remaining Al left in the upper crucible, mf is the mass of the filtered Al collected in the lower crucible, and ωFe_r and ωFe_f are the Fe content (wt.%) in the remaining and the filtered Al, respectively.
The 1060 industrial pure Al was purified under conditions including the holding time (th), separation temperature (T), and separation time (ts) under the gravity coefficient of 1000.It is worth noting that the holding time was the sum of the holding time in normal gravity plus the separation time in the supergravity field.

Effect of Separation Temperature
The alloy can be maintained in a solid/liquid coexistence state through semi-solid isothermal treatment during the melting of Al samples [25,26].In this study, separation temperatures of 648 °C , 653 °C , and 658 °C were applied under the gravity coefficient G = 1000.This process allows for the preservation of the Fe-rich phase in the liquid phase while keeping the Al matrix predominantly in the solid phase.The macroscopic images of the samples obtained from these experiments are presented in Figure 4. Table 3 lists their corresponding experimental parameters, the chemical compositions of the filtered Fe contents, and the separation efficiencies.It is worth noting that even with the addition of the supergravity field, the Fe-rich impurities could not be filtered to the lower part of the crucible because the liquid fraction was too small, and the flow was significantly restricted through the mushy zone (boundary layer + Liquid phase) at T = 648 °C [16].The 1060 industrial pure Al was purified under conditions including the holding time (t h ), separation temperature (T), and separation time (t s ) under the gravity coefficient of 1000.It is worth noting that the holding time was the sum of the holding time in normal gravity plus the separation time in the supergravity field.

Effect of Separation Temperature
The alloy can be maintained in a solid/liquid coexistence state through semi-solid isothermal treatment during the melting of Al samples [25,26].In this study, separation temperatures of 648 • C, 653 • C, and 658 • C were applied under the gravity coefficient G = 1000.This process allows for the preservation of the Fe-rich phase in the liquid phase while keeping the Al matrix predominantly in the solid phase.The macroscopic images of the samples obtained from these experiments are presented in Figure 4. Table 3 lists their corresponding experimental parameters, the chemical compositions of the filtered Fe contents, and the separation efficiencies.It is worth noting that even with the addition of the supergravity field, the Fe-rich impurities could not be filtered to the lower part of the crucible because the liquid fraction was too small, and the flow was significantly restricted through the mushy zone (boundary layer + Liquid phase) at T = 648 • C [16].Therefore, no experimental liquid fraction could be separated (F L = 0%) at this temperature [27].As the separation temperature increased, the sample that was filtered to the lower part of the crucible increased significantly due to the elevated liquid fraction and flowability of the samples as shown in Figure 4b,c.At T = 653 • C, the F L increased to 4.3%.However, when the separation temperature was increased to 658 • C, the Al phase began to melt, and the surface tension and the filtration resistance of the sample decreased, resulting in the mass of the Al liquid flowing into the lower part of the crucible with F L = 50.2%.Furthermore, a cross-sectional comparison of the macroscopic morphology of the samples obtained from the lower crucible at different separation temperatures reveals interesting insights.At lower temperatures, the sample exhibits higher viscosity, retaining the characteristic trajectory of liquid phase filtration to the lower crucible.Conversely, at higher temperatures, the sample's viscosity decreases, resulting in a smoother disc-shaped appearance.
sulting in the mass of the Al liquid flowing into the lower part of the crucible with FL = 50.2%.Furthermore, a cross-sectional comparison of the macroscopic morphology of the samples obtained from the lower crucible at different separation temperatures reveals interesting insights.At lower temperatures, the sample exhibits higher viscosity, retaining the characteristic trajectory of liquid phase filtration to the lower crucible.Conversely, at higher temperatures, the sample's viscosity decreases, resulting in a smoother disc-shaped appearance.In order to observe the distribution of the composition in the filtered sample, the microstructure of the sample was studied under different separation temperatures, as shown in Figure 6. Figure 7a illustrates the microstructure of the sample filtered to the lower crucible at 653 °C. Figure 7b,c correspond to the microstructures of the upper and lower positions of the sample filtered to the lower crucible at 658 °C.It can be seen that the sample was collected in the lower crucible, which mainly consisted of the Fe-rich phase [14].As the separation temperature was increased, the amount of liquid phase in the treated sample increased, and the liquid of the Fe-rich phase carrying more and more Al matrix was filtered down into the lower part of the crucible, resulting in a decrease in the In order to observe the distribution of the composition in the filtered sample, the microstructure of the sample was studied under different separation temperatures, as shown in Figure 6. Figure 7a illustrates the microstructure of the sample filtered to the lower crucible at 653 • C. Figure 7b,c correspond to the microstructures of the upper and lower positions of the sample filtered to the lower crucible at 658 • C. It can be seen that the sample was collected in the lower crucible, which mainly consisted of the Fe-rich phase [14].As the separation temperature was increased, the amount of liquid phase in the treated sample increased, and the liquid of the Fe-rich phase carrying more and more Al matrix was filtered down into the lower part of the crucible, resulting in a decrease in the mass fraction of the Fe phase.Meanwhile, the Fe-rich phase was enriched along the direction of supergravity, forming a continuous network structure at the bottom, as shown in Figure 6b,c.This was because the density of the Fe-rich phase was greater than that of the Al melt.Thus, the separation temperature of 653 • C was chosen as the optimum temperature.In order to observe the distribution of the composition in the filtered sample, the microstructure of the sample was studied under different separation temperatures, as shown in Figure 6. Figure 7a illustrates the microstructure of the sample filtered to the lower crucible at 653 °C. Figure 7b,c correspond to the microstructures of the upper and lower positions of the sample filtered to the lower crucible at 658 °C.It can be seen that the sample was collected in the lower crucible, which mainly consisted of the Fe-rich phase [14].As the separation temperature was increased, the amount of liquid phase in the treated sample increased, and the liquid of the Fe-rich phase carrying more and more Al matrix was filtered down into the lower part of the crucible, resulting in a decrease in the mass fraction of the Fe phase.Meanwhile, the Fe-rich phase was enriched along the direction of supergravity, forming a continuous network structure at the bottom, as shown in Figure 6b,c.This was because the density of the Fe-rich phase was greater than that of the Al melt.Thus, the separation temperature of 653 °C was chosen as the optimum temperature.

Effect of Holding Time
Due to the high purity of metallic Al, a substantial number of impurities (Fe-rich phase) are present in the Al matrix.However, the diffusion rate of Fe in solid Al is relatively slow.To facilitate the movement of Fe-rich impurities from the matrix to the grain boundary and their separation out of the sample, the holding time is particularly crucial [28].Figure 7 illustrates a macro diagram of the samples obtained with different holding times at T = 653 °C, ts =3 min.Table 4 lists the corresponding experimental parameters, chemical compositions of the filtered Fe contents, and the separation efficiencies.Notably, the results of the experiment at the holding time of 20 min were similar to those in Figure 4a, where the Fe-rich phase could not be filtered to the lower part of the crucible, as increasing the holding time from 30 min to 60 min resulted in more and more samples being filtered into the lower part of the crucible.The FL increased from 3.4% to 4.5%.The results indicate that irrespective of the holding time, the morphology of the sample in the upper crucible remains unchanged, displaying no noticeable collapse.Additionally, upon comparing the morphology of the samples obtained in the lower crucible, it is evident that with a prolonged holding time, there is a substantial increase in the quantity of samples collected in the lower crucible.Furthermore, these samples gradually transition from an irregularly stacked configuration to a disc-shaped form.This transformation suggests that as the liquid phase content increases in the lower crucible samples over time, their viscos-

Effect of Holding Time
Due to the high purity of metallic Al, a substantial number of impurities (Fe-rich phase) are present in the Al matrix.However, the diffusion rate of Fe in solid Al is relatively slow.To facilitate the movement of Fe-rich impurities from the matrix to the grain boundary and their separation out of the sample, the holding time is particularly crucial [28].Figure 7 illustrates a macro diagram of the samples obtained with different holding times at T = 653 • C, t s =3 min.Table 4 lists the corresponding experimental parameters, chemical compositions of the filtered Fe contents, and the separation efficiencies.Notably, the results of the experiment at the holding time of 20 min were similar to those in Figure 4a, where the Fe-rich phase could not be filtered to the lower part of the crucible, as increasing the holding time from 30 min to 60 min resulted in more and more samples being filtered into the lower part of the crucible.The F L increased from 3.4% to 4.5%.The results indicate that irrespective of the holding time, the morphology of the sample in the upper crucible remains unchanged, displaying no noticeable collapse.Additionally, upon comparing the morphology of the samples obtained in the lower crucible, it is evident that with a prolonged holding time, there is a substantial increase in the quantity of samples collected in the lower crucible.Furthermore, these samples gradually transition from an irregularly stacked configuration to a disc-shaped form.This transformation suggests that as the liquid phase content increases in the lower crucible samples over time, their viscosity decreases, and their fluidity is enhanced.Figure 8 shows a line chart of the separation results at T = 653 • C, t s = 3 min for different holding times.It can be seen that as the holding time was increased, the ω Fe_f decreased significantly.The η Fe increased significantly.However, when the holding time was greater than 30 min, the ω Fe_f and η Fe of the sample basically remained unchanged.The high-purity Al collected in the upper part from the different holding times was subjected to electron microscopic analysis at typical locations, as shown in Figure 9.At th = 20 min (Figure 9a), the results showed that during the sample melting process, the Ferich phase was essentially present as spherical eutectic particles in the sample compared to the raw material (Figure 2).This indicates that during the sample melting process, the Fe-rich phase dissolved, was broken, and spheroidized into small droplets before solidifying into the Al matrix at T = 653 °C.However, the Fe-rich phase could not be removed.When the holding time was increased, the amount of liquid phase in the grain boundaries increased and filtered to the lower part of the crucible, forming holes, as shown in Figure 9b-d.Meanwhile, it can be seen that the Fe-rich phase distributed in the Al matrix decreased, indicating that the Fe-rich phase in the Al matrix migrated to the grain boundary.At th = 40 min and 60 min (Figure 9c,d), only a very small amount of residual Fe-rich phase was distributed in the Al matrix, and almost all of the spheroidized Fe-rich phase in the Al matrix had migrated to the grain boundary.This is because the equilibrium distribution coefficient (k0) value of Fe impurities is 0.03, which is defined as the ratio of impurity concentration in the solid (CS) to that in the liquid (CL).This means that the solubility of the Fe-rich phase in the solid phase is much smaller than that in the liquid phase [29,30].It has been shown that Fe impurity elements at the solidification interface of the Al melt can be continuously moved to the liquid.Furthermore, the greater the difference in the k0 value of the impurity from 1, the more significant the purification effect.Here, the k0 value of Fe impurities is much smaller than 1.Thus, the Fe-rich impurities in the Al matrix continuously migrate towards the liquid phase near the grain boundaries [29,31].Ultimately, there is almost no presence of Fe-rich impurities in the Al matrix at th = 40 min and 60 min.The high-purity Al collected in the upper part from the different holding times was subjected to electron microscopic analysis at typical locations, as shown in Figure 9.At t h = 20 min (Figure 9a), the results showed that during the sample melting process, the Fe-rich phase was essentially present as spherical eutectic particles in the sample compared to the raw material (Figure 2).This indicates that during the sample melting process, the Fe-rich phase dissolved, was broken, and spheroidized into small droplets before solidifying into the Al matrix at T = 653 • C.However, the Fe-rich phase could not be removed.When the holding time was increased, the amount of liquid phase in the grain boundaries increased and filtered to the lower part of the crucible, forming holes, as shown in Figure 9b-d.Meanwhile, it can be seen that the Fe-rich phase distributed in the Al matrix decreased, indicating that the Fe-rich phase in the Al matrix migrated to the grain boundary.At t h = 40 min and 60 min (Figure 9c,d), only a very small amount of residual Fe-rich phase was distributed in the Al matrix, and almost all of the spheroidized Fe-rich phase in the Al matrix had migrated to the grain boundary.This is because the equilibrium distribution coefficient (k 0 ) value of Fe impurities is 0.03, which is defined as the ratio of impurity concentration in the solid (C S ) to that in the liquid (C L ).This means that the solubility of the Fe-rich phase in the solid phase is much smaller than that in the liquid phase [29,30].It has been shown that Fe impurity elements at the solidification interface of the Al melt can be continuously moved to the liquid.Furthermore, the greater the difference in the k 0 value of the impurity from 1, the more significant the purification effect.Here, the k 0 value of Fe impurities is much smaller than 1.Thus, the Fe-rich impurities in the Al matrix continuously migrate towards the liquid phase near the grain boundaries [29,31].Ultimately, there is almost no presence of Fe-rich impurities in the Al matrix at t h = 40 min and 60 min.Considering the separation efficiency and energy consumption comprehensively, t h = 40 min was selected as the optimal holding time.5 lists the corresponding experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies.When the separation time increased, more and more samples were filtered to the lower part of the crucible because the convection in the melt was enhanced, and the heat exchange was sufficient under the supergravity field, accelerating the filtering of the Fe-rich phase.The FL was 3.9%, 4.3%, and 8.4% for separation times of 1 min, 3 min, and 5 min, respectively.Similarly, upon comparing the morphologies of the purified aluminum samples obtained in the upper crucible at different separation times, it is observed that with an increase in the separation time, a partial collapse occurs in the bottom part of the upper sample.Upon contrasting the morphology of the samples in the lower crucible, it is evident that with prolonged supergravity action, the filtered liquid, initially in a drip-like stacked configuration, gradually undergoes compression under supergravity, transforming into a disc-shaped structure.This indicates that prolonged exposure to supergravity can result in the overcoming of intermolecular forces, leading to the formation of strong convection and the attainment of a relatively dense solidified structure.5 lists the corresponding experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies.When the separation time increased, more and more samples were filtered to the lower part of the crucible because the convection in the melt was enhanced, and the heat exchange was sufficient under the supergravity field, accelerating the filtering of the Fe-rich phase.The F L was 3.9%, 4.3%, and 8.4% for separation times of 1 min, 3 min, and 5 min, respectively.Similarly, upon comparing the morphologies of the purified aluminum samples obtained in the upper crucible at different separation times, it is observed that with an increase in the separation time, a partial collapse occurs in the bottom part of the upper sample.Upon contrasting the morphology of the samples in the lower crucible, it is evident that with prolonged supergravity action, the filtered liquid, initially in a drip-like stacked configuration, gradually undergoes compression under supergravity, transforming into a disc-shaped structure.This indicates that prolonged exposure to supergravity can result in the overcoming of intermolecular forces, leading to the formation of strong convection and the attainment of a relatively dense solidified structure.
Figure 11 shows a line chart of the separation results at T = 653 • C and t h = 40 min for different separation times.It can be observed that as the separation time increased from 1 min to 5 min, the ω Fe_f decreased from 0.15 wt.% to 0.06 wt.%.Additionally, the values of η Fe and R Al increased from 55.0% to 83% and from 3.7% to 8.1%, respectively.
served that with an increase in the separation time, a partial collapse occurs in the bottom part of the upper sample.Upon contrasting the morphology of the samples in the lower crucible, it is evident that with prolonged supergravity action, the filtered liquid, initially in a drip-like stacked configuration, gradually undergoes compression under supergravity, transforming into a disc-shaped structure.This indicates that prolonged exposure to supergravity can result in the overcoming of intermolecular forces, leading to the formation of strong convection and the attainment of a relatively dense solidified structure.Figure 12 illustrates the microstructure of the sample at T = 653 °C and th = 40 min under different separation times, where (a-l) correspond to Figure 10 at different positions marked "A-L" in red.After separation at T = 653 °C , th = 40 min for 1 min, along the supergravity direction, the grain boundary pores of the sample in the upper crucible initially increase and then decrease.This phenomenon can be attributed to the enrichment of the Fe-rich phase in the supergravity direction.Because the separation time was too short, the Fe-rich phase at the bottom did not have sufficient time to filter to the lower part of the crucible, resulting in smaller bottom pores.Moreover, a few residual spherical eutectic particles of the Fe-rich phase in the Al matrix can be observed.Upon increasing the separation time to 3 min, the grain boundary pores of the purified Al sample in the upper part of the crucible widened along the supergravity direction, as depicted in Figure 12e-g.This indicates that the separation time was appropriate for the Fe-rich phase along the grain boundary under these experimental conditions.The Fe-rich phase neither accumulates at the bottom of the Al sample (Figure 12c) nor results in excessively wide grain boundary pores due to excessive loss of the Al matrix (Figure 12i-k).Additionally, there was a noticeable decrease in the number of spherical eutectic particles in the Al matrix compared to Figure 12a-c, demonstrating that increasing the separation time was beneficial for the immigration of the Fe-rich phase to the grain boundary from the Al matrix.Compared to Figure 12d,h,l, the concentration of the Fe-rich phase in the filtered sample initially in-   C, t h = 40 min for 1 min, along the supergravity direction, the grain boundary pores of the sample in the upper crucible initially increase and then decrease.This phenomenon can be attributed to the enrichment of the Fe-rich phase in the supergravity direction.Because the separation time was too short, the Fe-rich phase at the bottom did not have sufficient time to filter to the lower part of the crucible, resulting in smaller bottom pores.Moreover, a few residual spherical eutectic particles of the Fe-rich phase in the Al matrix can be observed.Upon increasing the separation time to 3 min, the grain boundary pores of the purified Al sample in the upper part of the crucible widened along the supergravity direction, as depicted in Figure 12e-g.This indicates that the separation time was appropriate for the Fe-rich phase along the grain boundary under these experimental conditions.The Fe-rich phase neither accumulates at the bottom of the Al sample (Figure 12c) nor results in excessively wide grain boundary pores due to excessive loss of the Al matrix (Figure 12i-k).Additionally, there was a noticeable decrease in the number of spherical eutectic particles in the Al matrix compared to Figure 12a-c, demonstrating that increasing the separation time was beneficial for the immigration of the Fe-rich phase to the grain boundary from the Al matrix.Compared to Figure 12d,h,l, the concentration of the Fe-rich phase in the filtered sample initially increased and then decreased.This trend indicates that when the separation time exceeded 3 min, the proportion of the Al matrix in the sample filtered to the lower part of the crucible increased relative to that of the Fe-rich phase, resulting in an increase in the loss of Al, consistent with the results presented in Table 5.Thus, 3 min was chosen as the optimum separation time.

Removal Mechanism
According to the above experimental results, it can be seen that the method of purifying pure Al is similar to the mechanism of purification by unidirectional solidification/segregation, which is based on the fact that impurities have different solubility in the liquid or solid phases of metal [12,16].The schematic diagram of the purification process using supergravity in the semi-solid state is shown in Figure 13.Impurity elements at the solidification interface of the melt can be continuously moved to the liquid or solid phase depending on their equilibrium distribution coefficient (k0), that is, the greater the difference of the k0 value of the impurity from 1, the more significant the purification effect.However, is a variance between the k0 and the actual distribution coefficient.Thus, the actual distribution coefficient is defined as the effective distribution coefficient (ke).The melting-solidification equilibrium equation of the impurity elements in the purification process is as follows [13]: where D is the diffusion coefficient of the Fe, δ is the thickness of boundary layer, and R is the speed of the crystal growth.The supergravity field can effectively increase the diffusion coefficient of the Fe (D) in the liquid phase, reduce the crystal growth rates (R), and decrease the boundary layer thickness (δ) [32].By combining the above formulas, it can be observed that the effective distribution coefficient (ke) is reduced, thereby improving the efficiency of Fe element migration from the solid phase to the liquid phase, as shown in Figure 13a.Additionally, the supergravity field increases the convection in the melt, facilitating the expulsion and

Removal Mechanism
According to the above experimental results, it can be seen that the method of purifying pure Al is similar to the mechanism of purification by unidirectional solidification/segregation, which is based on the fact that impurities have different solubility in the liquid or solid phases of metal [12,16].The schematic diagram of the purification process using supergravity in the semi-solid state is shown in Figure 13.Impurity elements at the solidification interface of the melt can be continuously moved to the liquid or solid phase depending on their equilibrium distribution coefficient (k 0 ), that is, the greater the difference of the k 0 value of the impurity from 1, the more significant the purification effect.However, there is a variance between the k 0 and the actual distribution coefficient.Thus, the actual distribution coefficient is defined as the effective distribution coefficient (k e ).The melting-solidification equilibrium equation of the impurity elements in the purification process is as follows [13]: where D is the diffusion coefficient of the Fe, δ is the thickness of the boundary layer, and R is the speed of the crystal growth.The supergravity field can effectively increase the diffusion coefficient of the Fe (D) in the liquid phase, reduce the crystal growth rates (R), and decrease the boundary layer thickness (δ) [32].By combining the above formulas, it can be observed that the effective distribution coefficient (k e ) is reduced, thereby improving the efficiency of Fe element migration from the solid phase to the liquid phase, as shown in Figure 13a.Additionally, the supergravity field increases the convection in the melt, facilitating the expulsion and transport of solutes and squeezing the liquid phase out of the Al matrix near the grain boundaries [22,23].This reduces the Fe-rich phase in the liquid phase, ensuring a significant concentration difference between the solid and liquid phases.This, in turn, promotes the migration of the Fe-rich phase from the solid phase to the liquid phase near the grain boundaries, reducing the solute content (Fe-rich phase) in the solid phase.Therefore, this achieves the goal of purifying pure Al, as shown in Figure 13b.However, heightened metal purity corresponds to a more restricted semi-solid holding temperature interval, which means higher requirements for equipment temperature control.

Conclusions
This study investigated the impact of the separation temperature, holding time, and separation time on the loss rate of purified Al and the removal ratio of Fe elements.Simultaneously, the mechanism of supergravity in the semi-solid method of Al melt was discussed.The conclusions are summarized as follows.
1.The separation temperature significantly influences the loss rate of purified Al (R Al ), with a loss of only 4.1% at 653 • C and a substantial increase to 49.4% at 658 • C. The holding time has a notable impact on the removal rate of Fe (η Fe ) from the Al matrix.Specifically, at holding times of 20 min, 30 min, 40 min, and 60 min, the corresponding removal rates of Fe are 0%, 55.4%, 81.9%, and 81.6%, respectively.The separation time predominantly governs the removal rate of Fe (η Fe ) at the grain boundaries.At separation times of 1 min, 3 min, and 5 min, the corresponding removal rates of Fe are 55%, 81.9%, and 83%, respectively.
2. In the supergravity field, the Fe-rich phase at the grain boundary of the Al melt is preferentially filtered to the lower part of the crucible.At G=1000, T=653 • C, t h =40 min, and t s = 3 min, the loss rate of the purified Al reached 4.1%, and the Fe content was only 0.06 wt.% (the Fe content was 0.32 wt.% in the original material of pure Al).The removal rate of Fe was 81.9%.
3. The effective distribution coefficient (k e ) plays a crucial role in the purification of Al.Under the influence of the supergravity field, the value of k e experiences reduction, thereby facilitating the migration of Fe impurities from the Al matrix to the grain boundary liquid phase.This process aids in the efficient separation of Fe impurities from the Al matrix under the supergravity field, consequently enhancing the overall purity of the final product.

Figure 2 .
Figure 2. SEM images of the original material of pure Al at different positions (a) and (b)

Figure 2 .
Figure 2. SEM images of the original material of pure Al at different positions (a) and (b).

15 Figure 3 .
Figure 3. Solid fraction vs. temperature for pure Al, calculated using the Thermo-Calc Scheil solidification module.

Figure 3 .
Figure 3. Solid fraction vs. temperature for pure Al, calculated using the Thermo-Calc Scheil solidification module.

Figure 5
Figure5shows a line chart of the separation results at different experimental temperatures.It can be observed that as the temperature increased, the Fe content (ωFe_f) in the remaining Al decreased significantly, while the Fe removal rate (ηFe) and the loss rate of the purified Al (RAl) increased significantly.At T = 653°C , the ωFe_r, ηFe and RAl achieved 0.06 wt.%, 81.9%, and 4.1%, respectively.Although the ωFe_f and ηFe reached 0.05 wt.% and 92.0% at 658°C, the RAl reached 49.9%.

Figure 5 15 Figure 5 .
Figure 5 shows a line chart of the separation results at different experimental temperatures.It can be observed that as the temperature increased, the Fe content (ω Fe_f ) in the remaining Al decreased significantly, while the Fe removal rate (η Fe ) and the loss rate of the purified Al (R Al ) increased significantly.At T = 653 • C, the ω Fe_r, η Fe and R Al achieved 0.06 wt.%, 81.9%, and 4.1%, respectively.Although the ω Fe_f and η Fe reached 0.05 wt.% and 92.0% at 658 • C, the R Al reached 49.9%.Metals 2023, 13, 1945 7 of 15

Figure 5 .
Figure 5. Line chart of the separation results at t h = 40 min, t s = 3 min for different experimental temperatures.

Figure 5 .
Figure 5. Line chart of the separation results at th = 40 min, ts = 3 min for different experimental temperatures.

Figure 6 .
Figure 6.Microstructure of sample at th = 40 min, ts = 3 min for different separation temperatures (corresponding to Figure 4 marked as "A-C"): (a) the sample filtered to the lower crucible at 653 °C , and the upper (b) and lower (c) positions of the sample filtered to the lower crucible at 658 °C .

Figure 6 . 15 Figure 7 .
Figure 6.Microstructure of sample at t h = 40 min, t s = 3 min for different separation temperatures (corresponding to Figure 4 marked as "A-C"): (a) the sample filtered to the lower crucible at 653 • C, and the upper (b) and lower (c) positions of the sample filtered to the lower crucible at 658 • C. Metals 2023, 13, 1945 8 of 15

Figure 8 .
Figure 8. Line chart of the separation results at T = 653 °C , ts = 3 min for different holding times.

Figure 8 .
Figure 8. Line chart of the separation results at T = 653 • C, t s = 3 min for different holding times.

Metals 2023, 13 , 1945 10 of 15 Figure 9 .
Figure 9. Microstructure of sample at T = 653 °C , ts=3 min for different holding times: (a) 20 min, (b) 30 min, (c) 40 min, and (d) 60 min.3.1.3.Effect of Separation Time Figure 10 shows a macro diagram of the samples obtained at T = 653 °C and th = 40 min for different separation times.Table5lists the corresponding experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies.When the separation time increased, more and more samples were filtered to the lower part of the crucible because the convection in the melt was enhanced, and the heat exchange was sufficient under the supergravity field, accelerating the filtering of the Fe-rich phase.The FL was 3.9%, 4.3%, and 8.4% for separation times of 1 min, 3 min, and 5 min, respectively.Similarly, upon comparing the morphologies of the purified aluminum samples obtained in the upper crucible at different separation times, it is observed that with an increase in the separation time, a partial collapse occurs in the bottom part of the upper sample.Upon contrasting the morphology of the samples in the lower crucible, it is evident that with prolonged supergravity action, the filtered liquid, initially in a drip-like stacked configuration, gradually undergoes compression under supergravity, transforming into a disc-shaped structure.This indicates that prolonged exposure to supergravity can result in the overcoming of intermolecular forces, leading to the formation of strong convection and the attainment of a relatively dense solidified structure.

Figure 9 .
Figure 9. Microstructure of sample at T = 653 • C, t s =3 min for different holding times: (a) 20 min, (b) 30 min, (c) 40 min, and (d) 60 min.3.1.3.Effect of Separation Time Figure 10 shows a macro diagram of the samples obtained at T = 653 • C and t h = 40 min for different separation times.Table5lists the corresponding experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies.When the separation time increased, more and more samples were filtered to the lower part of the crucible because the convection in the melt was enhanced, and the heat exchange was sufficient under the supergravity field, accelerating the filtering of the Fe-rich phase.The F L was 3.9%, 4.3%, and 8.4% for separation times of 1 min, 3 min, and 5 min, respectively.Similarly, upon comparing the morphologies of the purified aluminum samples obtained in the upper crucible at different separation times, it is observed that with an increase in the separation time, a partial collapse occurs in the bottom part of the upper sample.Upon contrasting the morphology of the samples in the lower crucible, it is evident that with prolonged supergravity action, the filtered liquid, initially in a drip-like stacked configuration, gradually undergoes compression under supergravity, transforming into a disc-shaped structure.This indicates that prolonged exposure to supergravity can result in the overcoming of intermolecular forces, leading to the formation of strong convection and the attainment of a relatively dense solidified structure.Figure11shows a line chart of the separation results at T = 653 • C and t h = 40 min for different separation times.It can be observed that as the separation time increased from 1 min to 5 min, the ω Fe_f decreased from 0.15 wt.% to 0.06 wt.%.Additionally, the values of η Fe and R Al increased from 55.0% to 83% and from 3.7% to 8.1%, respectively.

Figure 11
Figure11shows a line chart of the separation results at T = 653 °C and th = 40 min for different separation times.It can be observed that as the separation time increased from 1 min to 5 min, the ωFe_f decreased from 0.15 wt.% to 0.06 wt.%.Additionally, the values of ηFe and RAl increased from 55.0% to 83% and from 3.7% to 8.1%, respectively.

Figure 11 .
Figure 11.Line chart of the separation results at T = 653 °C , th = 40 min for different separation times.

Figure 11 .
Figure 11.Line chart of the separation results at T = 653 • C, t h = 40 min for different separation times.

Figure 12
Figure 12 illustrates the microstructure of the sample at T = 653 • C and t h = 40 min under different separation times, where (a-l) correspond to Figure 10 at different positions marked "A-L" in red.After separation at T = 653• C, t h = 40 min for 1 min, along the supergravity direction, the grain boundary pores of the sample in the upper crucible initially increase and then decrease.This phenomenon can be attributed to the enrichment of the Fe-rich phase in the supergravity direction.Because the separation time was too short, the Fe-rich phase at the bottom did not have sufficient time to filter to the lower part of the crucible, resulting in smaller bottom pores.Moreover, a few residual spherical eutectic particles of the Fe-rich phase in the Al matrix can be observed.Upon increasing the separation time to 3 min, the grain boundary pores of the purified Al sample in the upper part of the crucible widened along the supergravity direction, as depicted in Figure12e-g.This indicates that the separation time was appropriate for the Fe-rich phase along the grain boundary under these experimental conditions.The Fe-rich phase neither accumulates at the bottom of the Al sample (Figure12c) nor results in excessively wide grain boundary pores due to excessive loss of the Al matrix (Figure12i-k).Additionally, there was a noticeable decrease in the number of spherical eutectic particles in the Al matrix compared

Figure 13 .
Figure 13.Schematic diagram of the purification process using supergravity in the semi-solid state: (a) sample melting process, (b) supergravity separation process.

Table 1 .
Chemical composition of 1060 industrial pure Al.

Table 1 .
Chemical composition of 1060 industrial pure Al.

Table 2 .
EDS analysis of components marked with numerical symbols in Figure1.

Table 3 .
Experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation results at different temperatures.

Table 3 .
Experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation results at different temperatures.

Table 4 .
Experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies at different holding times.

Table 5 .
Experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies at different separation times.

Table 5 .
Experimental parameters and the corresponding chemical compositions of the filtered Fe contents and separation efficiencies at different separation times.