Thermodynamics of Formation of Al₃Fe Inter-Metallic Compound for Fe Removal from Molten Al–Mg Alloy

Abstract

Recycling Al scraps is extremely important to achieve carbon neutrality. However, almost all impurity elements cannot be easily removed by oxidation and evaporation once dissolved in molten Al alloy. Therefore, it is necessary to develop an impurity removal process in molten Al. Iron is one of the impurity elements in Al, which easily forms intermetallic compounds with Al. On the contrary, Fe is immiscible with Mg in the liquid state. These facts suggest that the repulsive interaction between Fe and Mg in molten Al alloy enhances the precipitation of such intermetallic compounds as Al₃Fe. This study proposed the Fe removal method from molten Al alloy by the precipitation of the Al₃Fe inter-metallic compound by adding Mg. Molten Al–Mg alloy was equilibrated with Al₃Fe, and the Fe content of molten Al–Mg alloy was investigated. Fe content decreased with increasing Mg content, and this method was proved to be adequate for removing Fe from molten Al–Mg alloy. The standard Gibbs energy change for the precipitation of Al₃Fe at 873 K and the activity coefficient of Fe in molten Al–Mg alloy as a function of Mg content and temperature have been derived from the experimental results.

Using the derived values, the removal limit of Fe in molten Al–Mg alloy in the present method was discussed. Fe content can be decreased at higher Mg content and lower temperature, and the lowering limit of Fe is 0.0029 mass percent at the eutectic point of Al–Mg alloy.

Fig. 8 Contour lines of Fe content of molten Al–Mg alloy equilibrated with Al₃Fe.

1. Introduction

To achieve carbon neutrality, it is necessary to realize a resource-circulating society. Aluminum is the second most used metal material due to its low density and superior corrosion resistance.^1,2) The electric power required to produce Al from recovered Al scraps is about 3%^3–5) of that from the natural resources. Therefore, recycling Al scraps is extremely important. Al scraps are mainly recycled into cast Al alloy, used for automotive engines.⁶⁾ The electric vehicle will become increasingly popular, decreasing the demand for automotive engines and transmissions. As a result, Al scraps will be surplus in society.^7,8) Accordingly, it is necessary to produce wrought Al alloys from Al scraps.⁹⁾ Due to the low tolerance level of impurities in wrought Al alloys, the alloying elements must be controlled to be low concentration. Depending on the required function, various types of Al alloys are used for an automobile, which leads to the high cost of the separation at the end of their service life. Al scraps are recovered without regarding the types of the different alloys.^10,11) Iron is used for nut and bolt to fasten Al materials. Thus, it is contained in the Al scraps recovered from society. However, almost all impurity elements cannot be easily removed by the oxidation, evaporation, and flux treatment once dissolved in molten Al alloy.^12,13) Accordingly, a removal process of Fe by a pyrometallurgical process has not been developed due to higher chemical stability and lower vapor pressure of Fe. High purity Al can be produced by electrolysis, such as the trinal electrolytic process. However, this process is not used for refining scraps due to its large electric power consumption.¹⁴⁾ On the other hand, intermetallic compounds containing impurity elements precipitate in molten Al alloy at low temperatures. The compounds and molten Al can be separated by filtering and centrifugal separation.^14–17) However, when the impurity contents are low in molten Al, intermetallic compounds are not easily precipitated.¹⁸⁾ Moraes et al.¹⁹⁾ have reported that Fe can be removed by adding Mn in molten Al with high Si content. Grigorenko et al.²⁰⁾ have also reported the removal of Fe from Al–Si–Cu alloys. These facts suggest that the precipitation of intermetallic compounds is promoted by adding other elements. Fe easily forms intermetallic compounds with Al, although Fe is immiscible with Mg in the liquid state. These facts suggest that the repulsive interaction between Fe and Mg in molten Al alloy enhances the precipitation of intermetallic compounds such as Al₃Fe. Figure 1 shows the liquidus surface of the Al-rich corner in the Al–Fe–Mg ternary system.²¹⁾ As can be seen, Fe content becomes lower at lower temperature and higher Mg content on the liquidus surface of Al₃Fe at Al-rich corner. This study proposed the Fe removal method from molten Al–Mg alloy by the precipitation of the Al₃Fe inter-metallic compound by adding Mg. Mg is an essential element in 5000-series aluminum alloys, and Al–Mg alloys with a higher Mg content due to Fe removal may be used as an Mg source for producing such aluminum alloys. Furthermore, Mg in Al can be removed by oxidation or evaporation due to the lower partial pressure of oxygen under the equilibrium with its oxide and the higher vapor pressure of Mg. Molten Al–Mg alloy was equilibrated with Al₃Fe, and the Fe content of molten Al–Mg alloy was investigated. Thermodynamic data of Al₃Fe formation in Al–Mg alloy was derived, and the removal limit of Fe in molten Al–Mg alloy in the present method was discussed.

Fig. 1

Liquidus surface in the Al–Fe–Mg ternary system at Al-rich corner.²¹⁾

2. Experimental Procedure

2.1 Preparation of sample

Al₃Fe and Al–Mg alloys were prepared by using a high-frequency induction furnace. To prepare Al₃Fe, Al (99.99%) and Fe (99.95%) were put in an Al₂O₃ crucible (38 mm o.d., 33 mm i.d., and 45 mm in height), and the crucible was put in the high-frequency induction furnace. The mass of Al and Fe was 26 g and 17 g, respectively. Ar gas was introduced in the furnace at a flow rate of 200 cm³/min (s.t.p.). The samples were heated from room temperature to 1823 K and held for 10 minutes at the temperature after melting. Subsequently, the samples were cooled in the furnace. To prepare Al–Mg alloys, Al and Mg (99.95%) were weighed and put in an Al₂O₃ crucible (38 mm o.d., 33 mm i.d., and 45 mm in height). The crucible was put in the high-frequency induction furnace. Ar gas was introduced in the furnace at a flow rate of 200 cm³/min (s.t.p.). The samples were heated from room temperature to 1073 K and held for 10 minutes at the temperature after melting. Subsequently, the samples were cooled in the furnace. In the case of adding Fe, prepared Al₃Fe were put in the Al₂O₃ crucible together with Al and Mg, and the samples were melted at 1473 K.

2.2 Equilibrium experiment

Equilibrium experiments were performed by an electric resistance furnace. Figure 2 shows a schematic illustration of the equilibrium experiment, and Table 1 shows the experimental conditions. Prepared Al₃Fe and Al–Mg(–Fe) alloys were put in an Al₂O₃ crucible (30 mm o.d., 26 mm i.d., and 50 mm in height), and the crucible was put in the electric resistance furnace. Ar gas was introduced in the furnace at a flow rate of 200 cm³/min (s.t.p.). The samples were heated up to 873 K at 10 K/min and equilibrated at the temperature. Preliminary experiments at different times revealed that 19 hours are enough for the equilibration. Afterward, the Al₂O₃ crucible was withdrawn from the furnace and cooled in the air. The content of Mg and Fe in molten Al–Mg alloys was analyzed by an inductively coupled plasma atomic emission spectrometry.

Fig. 2

Schematic of Experimental apparatus and arrangement of samples.

Table 1 Initial conditions of equilibrium experiments.

3. Results and Discussions

3.1 Equilibrium composition of iron in molten aluminum

Table 2 shows the Mg and Fe contents of molten Al–Mg alloy equilibrated with Al₃Fe. Figure 3 shows the relationship between Fe and Mg contents of molten Al–Mg alloy after experiments with their initial contents. In the samples of Nos. 1 to 5, the Fe contents were increased from the initial values, which means that Fe was dissolved into molten Al–Mg alloy from Al₃Fe. In sample No. 6, the Fe content was decreased from the initial value, which means that Al₃Fe was precipitated from molten Al–Mg alloy. The Fe contents were reduced at higher Mg content. In these experiments, the lowest value of the Fe content was 0.07 mass% when Mg content was 38.5 mass% at 873 K.

Table 2 Experimental results.

Fig. 3

Relationship between Fe and Mg contents of molten Al alloy equilibrated with Al₃Fe.

3.2 Thermodynamics of formation of Al₃Fe inter-metallic compound

The equilibrium between molten Al–Mg alloy and an Al₃Fe is represented by eq. (1).   

\begin{align} &\text{3Al($l$, in Al–Mg alloy)} + \text{Fe($l$, in Al–Mg alloy)} \\ &\quad = \text{Al$_{3}$Fe(s)} \end{align} (1)

The system has four chemical species (Al, Mg, Fe, and Al₃Fe), three components (Al, Mg, and Fe), and two phases (molten Al–Mg alloy and Al₃Fe). Therefore, at constant temperature and pressure, the degree of freedom is calculated as 3 − 2 = 1. Thus, only one variable can be arbitrarily determined in this system. The precipitation of Al₃Fe in molten Al–Mg alloy was thermodynamically discussed, considering Mg content as a variable. The standard Gibbs energy change and the equilibrium constant of the reaction of eq. (1) are written as eqs. (2) and (3), respectively. Equation (2) was derived from the thermodynamic data of Al₁₃Fe₄²²⁾ and the heat of fusion of Al and Fe.²³⁾   

\begin{equation} \Delta G_{(1)}^{\text{o}} = -162{,}790 + 68.4T\ (\text{J/mol}) \end{equation} (2)

  

\begin{equation} \ln K_{(1)} = -\frac{\Delta G_{(1)}^{\text{o}}}{RT} = \ln\left(\frac{a_{\text{Al${_{3}}$Fe}}}{a_{\text{Al}}{}^{3} \cdot a_{\text{Fe}}}\right), \end{equation} (3)

where T is the absolute temperature (K), and R is the gas constant (J/mol·K). a_Al and a_Fe are the Raoultian activities of Al and Fe, respectively, relative to each pure liquid. $a_{\text{Al}_{3}\text{Fe}}$ is the activity of Al₃Fe relative to the pure solid and equal to unity in this work. Equation (4) is derived by arranging eq. (3) because the activity can be expressed by the product of the activity coefficient γ and molar fraction x.   

\begin{equation} \ln \gamma_{\text{Fe}}^{0} = -3\ln a_{\text{Al}} - \ln x_{\text{Fe}} - \ln K_{(1)}, \end{equation} (4)

where $\gamma_{\text{Fe}}^{0}$ is the activity coefficient of Fe in molten Al at an infinitely dilute solution. The activity coefficient of Fe was regarded to be the value for dilute solution $\gamma_{\text{Fe}}^{0}$ because the Fe contents of the molten Al–Mg alloy are low enough from the experimental results. The activity coefficient of Fe can be obtained from eq. (4) when the activity of Al and the molar fraction of Fe are known. The activity of Al was estimated from the Al–Mg binary data²¹⁾ because Fe in molten Al–Mg alloy is dilute. Equation (5) can be derived for a regular solution.   

\begin{equation} T\ln \gamma = \text{const.} \end{equation} (5)

Using eq. (5), the activity coefficient of a component at any temperature can be calculated from a known value at a temperature. Assuming the molten Al–Mg alloy is a regular solution, the activities of Al and Mg at 873 K were calculated from the values in Al–Mg binary system at 1000 K²¹⁾ and shown in Fig. 4. Furthermore, the activity coefficients of Al and Mg can be written as eqs. (6) and (7), where α is a constant.   

\begin{equation} \ln \gamma_{\text{Al (in Al–Mg)}} = \alpha x_{\text{Mg}}{}^{2} \end{equation} (6)

  

\begin{equation} \ln \gamma_{\text{Mg (in Al–Mg)}} = \alpha(1 - x_{\text{Mg}})^{2} \end{equation} (7)

From eqs. (6) and (7), the activities of Al and Mg are written as eqs. (8) and (9), in terms of the molar fraction of Mg.   

\begin{equation} a_{\text{Al}} = (1 - x_{\text{Mg}}) \cdot \exp (\alpha x_{\text{Mg}}{}^{2}) \end{equation} (8)

  

\begin{equation} a_{\text{Mg}} = x_{\text{Mg}} \cdot \exp (\alpha (1 - x_{\text{Mg}})^{2}) \end{equation} (9)

The activities of Al and Mg were calculated by eqs. (8) and (9) and were indicated by the solid curves in Fig. 4, where the value of α was adjusted to be −0.459 to minimize the difference between the solid line and the plot. Substituting eq. (8) into eq. (4), eq. (10) is derived.   

\begin{align} \ln \gamma_{\text{Fe}}^{\text{o}}& = -3\{-0.459 x_{\text{Mg}}{}^{2} + \ln (1 - x_{\text{Mg}})\} - \ln x_{\text{Fe}} \\ &\quad - \ln K_{(1)} \end{align} (10)

The activity coefficient of Fe was obtained by using eq. (10) and shown in Table 2. Figure 5 shows the dependence of the activity coefficient of Fe on Mg content. The activity coefficient of Fe increases at higher Mg content, which suggests that a repulsive force works between Mg and Fe in the molten Al–Mg alloy.

Fig. 4

Activities of Al and Mg in Al–Mg binary melt at 873 K estimated from thermodynamic data at 1000 K.²¹⁾

Fig. 5

Activity coefficient of iron in molten Al–Mg alloy at 873 K.

On the other hand, the activity coefficient of Fe can be estimated by Toop’s equation^24–26) in eq. (11) when Fe is dilute in molten Al–Mg alloy.   

\begin{align} \ln \gamma_{\text{Fe(${l}$ in Al–Mg alloy)}}^{0} & = \frac{x_{\text{Al}}}{x_{\text{Mg}} + x_{\text{Al}}} \cdot \ln \gamma_{\text{Fe (in Al)}}^{0} \\ &\quad + \frac{x_{\text{Mg}}}{x_{\text{Mg}} + x_{\text{Al}}} \cdot \ln \gamma_{\text{Fe (in Mg)}}^{0}\\ &\quad - (1 - x_{\text{Fe}})^{2} \cdot \frac{\varDelta G_{\text{Al–Mg}}^{Ex}}{RT}, \end{align} (11)

where $\gamma_{\text{Fe}(l\ \text{in}\ \text{Al–Mg}\ \text{alloy})}^{0}$ is the activity coefficient of Fe in molten Al–Mg alloy. $\gamma_{\text{Fe}\ \text{(in}\ \text{Al)}}^{0}$ and $\gamma_{\text{Fe}\ (\text{in}\ \text{Mg})}^{0}$ are the activity coefficients of Fe in molten Al and Mg, respectively. $\varDelta G_{\text{Al–Mg}}^{Ex}$ is the excess Gibbs energy of mixing in Al–Mg system. Because Fe is dilute in the Al–Mg alloy, x_Mg + x_Al ≈ 1. Considering molten Al–Mg alloy to be a regular solution, eq. (12) is obtained.   

\begin{equation} \varDelta G_{\text{Al–Mg}}^{Ex} = RT(x_{\text{Al}}\ln \gamma_{\text{Al}_{\text{(in Al–Mg)}}} + x_{\text{Mg}}\ln \gamma_{\text{Mg}_{\text{(in Al–Mg)}}}) \end{equation} (12)

Hence, eq. (11) can be rearranged as eq. (13).   

\begin{align} & \ln \gamma_{\text{Fe(in Al–Mg–Fe)}}^{0} = x_{\text{Mg}}\ln \gamma_{\text{Fe(in Mg)}}^{0} + x_{\text{Al}}\ln \gamma_{\text{Fe(in Al)}}^{0}\\ &\quad - x_{\text{Mg}}\ln \gamma_{\text{Mg(in Al–Mg)}} - x_{\text{Al}}\ln \gamma_{\text{Al(in Al–Mg)}} \end{align} (13)

When eqs. (6) and (7) were substituted into eq. (13), eq. (14) was derived for the activity coefficient of Fe as a function of Mg content.   

\begin{align} & \ln \gamma_{\text{Fe(in Al–Mg–Fe)}}^{0}\\ &\quad = -0.459x_{\text{Mg}}{}^{2} + (\ln \gamma_{\text{Fe(in Mg–Fe)}}^{0} - \ln \gamma_{\text{Fe(in Al–Fe)}}^{0} \\ &\qquad + 0.459)x_{\text{Mg}} + \ln \gamma_{\text{Fe(in Al–Fe)}}^{0} \end{align} (14)

To calculate the activity coefficient of Fe in Al–Mg–Fe ternary system by eq. (14), those in Al–Fe and Mg–Fe binary systems are required. In the Al–Fe system, $\gamma_{\text{Fe}(\text{in}\ \text{Al–Fe})}^{0} = 0.054$²¹⁾ at 1900 K has been reported. The activity of Fe in Al at 873 K was obtained as $\ln \gamma_{\text{Fe}(\text{in}\ \text{Al–Fe})}^{0} = - 6.35$ by using eq. (5). In the Mg–Fe system, only immiscible two liquid phases exist at a high temperature according to the binary phase diagram.²¹⁾ Because the mutual solubilities of Fe and Mg in the liquid Mg phase and liquid Fe phases are low, Fe in the liquid Mg phase will obey Henry’s law, and that in the liquid Fe phase will obey Raoult’s law. Therefore, eq. (15) can be derived.   

\begin{equation} \gamma_{\text{Fe (in Mg)}}^{0} = \frac{1 - x_{\text{Mg (in Fe)}}}{x_{\text{Fe(in Mg)}}} \end{equation} (15)

Mg contents of the Fe phase and Fe contents of the Mg phase at 2500, 2300, 2100, and 1900 K were obtained from the phase diagram.²¹⁾ These values were substituted into eq. (15), and the activity coefficient of Fe in Mg at each temperature was obtained. $T\ln \gamma_{\text{Fe}(\text{in}\ \text{Mg})}^{0}$ was calculated at each temperature, and the mean value was obtained to be 7,581. From eq. (5) and the obtained value, the activity coefficient of Fe in the liquid Mg phase at 873 K was estimated as $\ln \gamma_{\text{Fe}(\text{in}\ \text{Mg–Fe})}^{0} = 8.68$. Substituting the derived values into eq. (14), the activity coefficient of Fe was calculated and shown in Fig. 5 by a dashed line. The slope of the dashed line calculated from the literature is close to the one from experimental values, although the absolute values are different. The difference between the experimental and literature values is caused by some error in referred standard Gibbs energy change of Al₃Fe precipitation. Accordingly, the standard Gibbs energy change of Al₃Fe precipitation was reassessed from the present results. The standard Gibbs free energy change at 873 K was calculated from $\Delta G^{\text{o}} = - RT\ln (\frac{1}{a_{\text{Al}}^{3}\cdot \gamma_{\text{Fe}(\text{in}\ \text{Al–Mg–Fe})}^{0}\cdot x_{\text{Fe}}})$, using the experimental results and the activity coefficients estimated by eq. (14). The mean value was obtained to be ΔG^o = −76,800 (J/mol). The enthalpy term of the standard Gibbs energy change in eq. (2) was adjusted, and eq. (16) was obtained.   

\begin{equation} \Delta G_{(1)}^{\text{o}} = -136{,}460 + 68.4T\ (\text{J/mol}) \end{equation} (16)

The activity coefficient of Fe was re-calculated from the experimental results using eq. (16) and shown in Fig. 6 with those calculated by eq. (14). The derived activity coefficients of Fe and used thermodynamic data are summarized in Table 3. The solid line in Fig. 6 indicates the fitted curve of the experimental results, considering the concentration dependence in eq. (14). From Fig. 6, the following equation was derived as the activity coefficient of Fe in the molten Al–Mg alloy.   

\begin{equation} \ln \gamma_{\text{Fe(in Al–Mg–Fe)}}^{0} = \frac{873}{T}(-0.459x_{\text{Mg}}{}^{2} + 12.54x_{\text{Mg}} - 5.68) \end{equation} (17)

Fig. 6

Corrected activity coefficient of iron in molten Al–Mg alloy at 873 K.

Table 3 Derived activity coefficient of Fe in molten Al–Mg alloy and used thermodynamic data.

3.3 Fe removal from molten Al–Mg alloy

Removal of Fe due to the precipitation of Al₃Fe, in the reaction of eq. (1) was assessed using the derived thermodynamic data with varying temperature and Mg content of a molten Al alloy. From eq. (4), eq. (18) is derived.   

\begin{equation} \ln x_{\text{Fe}} = -3\ln a_{\text{Al}} - \ln\gamma_{\text{Fe}}^{0} - \ln K_{(1)} \end{equation} (18)

The equilibrium constant can be obtained from eqs. (3) and (16), and the activity of Al can be calculated from eqs. (5) and (8). The activity coefficient of Fe can be calculated from eq. (17). Therefore, the solubility of Fe in molten Al–Mg alloy equilibrated with Al₃Fe can be calculated from Mg content when the temperature is determined. The calculated Fe solubility is shown in Fig. 7 with the experimental plots. There is a reasonable agreement between the experimental value and that calculated from the derived thermodynamic data. In addition, when the Fe content of Al–Mg alloy is determined, eq. (18) is a function of temperature and Mg content. Thus, the equilibrium content of Mg at each temperature can be obtained. Figure 8 shows the contour lines of Fe content near the liquidus in the Al-rich corner of the Al–Mg binary phase diagram, which indicates the effect of Mg content and temperature on the Fe content of molten Al–Mg alloy. In Fig. 8, the bold line indicates the liquidus of Al, and the plots are experimental results. From Fig. 8, the solubility of Fe is decreased at higher Mg content, resulting from the repulsive interaction between Fe and Mg in molten Al–Mg alloy. In addition, Fe removal is attained at a lower temperature. This is caused by the increase in the equilibrium constant of the precipitation of Al₃Fe, which is the exothermic reaction. From this analysis, Fe content can be decreased to 0.0029 mass% by increasing the Mg content and lowering the temperature to the eutectic point of the Al–Mg binary system at 733 K and 34 mass% Mg. This analysis revealed that the Fe content of molten Al–Mg alloy could be significantly decreased by adding Mg from a thermodynamic viewpoint. Mg is an essential element in 5000-series aluminum alloys, and Al–Mg alloys with a higher Mg content due to Fe removal may be used as an Mg source for producing such Al–Mg alloys. The results obtained in this work are helpful in optimizing the amount of adding Mg and the temperature of Fe removal treatment. Further investigations are required to separate Al₃Fe from the Al–Mg alloy to develop the practical process.

Fig. 7

Solubility of iron in molten Al–Mg alloy equilibrated with Al₃Fe at 873 K.

Fig. 8

Contour lines of Fe content of molten Al–Mg alloy equilibrated with Al₃Fe.

4. Conclusions

Molten Al–Mg alloy was equilibrated with Al₃Fe, and the effect of Mg content and temperature on the Fe content of molten Al–Mg alloy was investigated. A removal method of Fe as Al₃Fe intermetallic compounds by adding Mg was discussed. The following conclusions were obtained.

1. (1)    Iron content of molten Al–Mg alloy was decreased at higher Mg content. Under the experimental conditions of this work, the lowest value of the Fe content was 0.07 mass% at 873 K and 38.5 mass% Mg.
2. (2)    The standard Gibbs energy change for the precipitation of Al₃Fe from molten Al–Mg alloy at 873 K was obtained as follows:   
   \begin{align*} &\text{3Al($l$, in Al–Mg alloy)} + \text{Fe($l$, in Al–Mg alloy)} \\ &\quad = \text{Al$_{3}$Fe(s)} \end{align*}
     
   \begin{equation*} \Delta G_{\text{at 873$\,$K}}^{\text{o}} = -76{,}800\ (\text{J/mol}) \end{equation*}
3. (3)    The activity coefficient of Fe in molten Al–Mg alloy was derived as a function of Mg content and temperature.   
   \begin{align*} &\ln \gamma_{\text{Fe(in Al–Mg–Fe)}}^{0} \\ &\quad= \frac{873}{T}(-0.459x_{\text{Mg}}{}^{2} + 12.54x_{\text{Mg}} - 5.68) \end{align*}
   • This equation suggests the repulsive interaction between Fe and Mg in molten Al–Mg alloy.
4. (4)    The effect of Mg content and temperature on Fe content in Al–Mg alloy was discussed, and contour lines of Fe content in this removal process were described. Due to the repulsive interaction between Fe and Mg in molten Al–Mg alloy and the exothermic reaction for precipitation of Al₃Fe, Fe content can be decreased at higher Mg content and lower temperature. Fe content can be reduced to 0.0029 mass% at the eutectic point of the Al–Mg binary system at 733 K and 34 mass% of Mg.

REFERENCES

• 1)   H.  Matsumura: Materia Japan 60 (2021) 399–403. doi:10.2320/materia.60.399
• 2)   T.  Tsukeda: J. JILM 62 (2012) 493–501. doi:10.2464/jilm.62.493
• 3)   T.  Ohnishi: J. JFS 69 (1997) 995–1002. doi:10.11279/jfes.69.995
• 4)   M.  Okubo,  S.  Kumai and  K.  Ueda: J. MMIJ 123 (2007) 850–854. doi:10.2473/journalofmmij.123.850
• 5)   T.  Ohnishi: J. JILM 46 (1996) 557–563. doi:10.2464/jilm.46.557
• 6)   Y.  Isobe: Zairyo-to-Kankyo 48 (1999) 469–475. doi:10.3323/jcorr1991.48.469
• 7)   A.  Takasugi: J. JILM 59 (2009) 87–93. doi:10.2464/jilm.59.87
• 8)   Y.  Harada,  H.  Yamamoto,  S.  Nagano,  M.-S.  Kim and  S.  Kumai: J. JFS 87 (2015) 772–781. doi:10.11279/jfes.87.772
• 9)   T.  Yamamoto: Materia Japan 60 (2021) 347–350. doi:10.2320/materia.60.347
• 10)   S.  Aoki: J. JILM 63 (2013) 260–270. doi:10.2464/jilm.63.260
• 11)   T.  Sakurai: J. JILM 71 (2021) 569–573. doi:10.2464/jilm.71.569
• 12)  T. Hiraki, L. Xin, K. Nakajima, K. Matsubae, S. Nakamura and T. Nagasaka: Proceedings of the Annual Conference of Japan Society of Material Cycles and Waste Management, (2012).
• 13)   L.  Xin: Materia Japan 59 (2020) 495–499. doi:10.2320/materia.59.495
• 14)   T.  Nakamura: Materia Japan 35 (1996) 1290–1293. doi:10.2320/materia.35.1290
• 15)   S.  Nishi,  M.  Nagao,  K.  Yoshikawa,  R.  Masuda and  K.  Ogawa: R&D Kobe Steel Engineering Reports 47 (1997) 21–25.
• 16)   F.  Murata: J. JILM 46 (1996) 551–556. doi:10.2464/jilm.46.551
• 17)   M.  Nagao,  K.  Oosumi and  T.  Nakamura: J. JILM 46 (1996) 588–591. doi:10.2464/jilm.46.588
• 18)   M.  Ohtaki,  T.  Soutome,  K.  Mori,  H.  Kudoh and  S.  Tanaka: The Furukawa Electric Review 104 (1999) 25–30.
• 19)   H.L.  de Moraes,  J.R.  de Oliveira,  D.C.R.  Espinosa and  J.A.S.  Tenório: Mater. Trans. 47 (2006) 1731–1736. doi:10.2320/matertrans.47.1731
• 20)  V.M. Grigorenko, V.A. Popov and A.A. Ofengenden: Sovjet Non-Ferrous Metals Research, (1997) 281–282.
• 21)  B. Predel: Phase Equilibria, Crystallographic and Thermodynamic Data of Binary Alloys, Landolt-Börnstein, Numerical Data and Functional Relationships in Science and Technology, New Series/Editors in Chief: O. Madelung and W. Martienssen, Group IV: Macroscopic Properties of Matter, Vol. 5, (Springer-Verlag, Berlin, 1991) pp. 1–2, 143–147, 160–163, 285–291.
• 22)  I. Ansara, A.T. Dinsdale and M.H. Rand (eds.): COST 507, “Thermochemical database for light metal alloys”, Vol. 2, EUR 18499, (1998) pp. 34–39.
• 23)   A.T.  Dinsdale: Calphad 15 (1991) 317–425. doi:10.1016/0364-5916(91)90030-N
• 24)   M.  Zheng and  Z.  Kozuka: J. Japan Inst. Metals 51 (1987) 44–50. doi:10.2320/jinstmet1952.51.1_44
• 25)   J.  Moriyama,  Z.  Kozuka and  T.  Oishi: Bull. Jpn. Inst. Met. 9 (1970) 764–776. doi:10.2320/materia1962.9.764
• 26)   H.  Ono-Nakazato,  D.  Hirai and  T.  Usui: J. High Temp. Soc. 26 (2000) 139–144.