Electrolytic Reduction of Solid Al₂O₃ to Liquid Al in Molten CaCl₂

A schematic drawing of the experimental apparatus for molten CaCl₂ is shown in Fig. 2. In an open dry chamber (dew point < −65 °C, HRW-60AR, Daikin Co. Ltd.), 350 g of CaCl₂ (> 95.0%, Wako Pure Chemical Industries, Ltd.) was crushed and put in an alumina crucible (purity 99%, o.d.: 90 mm, height: 140 mm, As One Corp.). Then, the crucible containing CaCl₂ was dried at 453 K in a vacuum oven for more than 72 h. It was transferred to a stainless steel inner vessel settled in a Kanthal vessel, and vacuumed at 773 K for 24 h to further remove residual moisture. The experiments were conducted inside the Kanthal vessel at 1123 K under a dry Ar atmosphere. A chromel–alumel thermocouple was used for the temperature control. Electrochemical measurements and potentiostatic electrolysis were conducted by a three-electrode method using an electrochemical measurement system (HZ-3000, Hokuto Denko Corp.).

Zoom In Zoom Out Reset image size

Three types of working electrodes, shown in Fig. 3, were used depending on the purpose. An Al₂O₃-sealed electrode was prepared by inserting a W rod (> 99.95%, diameter: 2.0 mm, Nilaco Corp.) in an Al₂O₃ tube (99.6%, o.d.: 3.0 mm, i.d.: 2.0 mm, Nikkato Corp.), as shown in Fig. 3a. An Fe box-type electrode (6 mm × 6 mm × 10 mm) was prepared with an Fe sheet (99.5%, thickness: 0.10 mm, Nilaco Corp.). After spot-welding of a Mo wire current lead (Fig. 3b), approximately 100 mg of Al₂O₃ powder (99.9%, particle size: 0.212–0.5 mm, Kojundo Chemical Lab. Corp.) was charged in the Fe box-type electrode. A Mo box-type electrode (12 mm × 12 mm × 3 mm), shown in Fig. 3c, was fabricated from a Mo sheet (99.5%, thickness: 0.10 mm, Nilaco Corp.), for which small Fe sheets used to reinforce the corners of the box. The Mo box-type electrode was used for the reduction of an Al₂O₃ tube (99.6%, o.d.: 10 mm, i.d.: 6 mm, Nikkato Corp.). The counter electrode was a square rod of graphite (5 mm × 5 mm × 50 mm, IG-110, Toyo Tanso Co., Ltd.), and the reference electrode was a Ag⁺/Ag electrode.¹⁰

Zoom In Zoom Out Reset image size

After the potentiostatic electrolysis, the Al₂O₃-sealed electrodes were rinsed with distilled water to remove the residual salts and dried at room temperature. Then, they were cut by a desktop abrasive cutting machine (RC-120, As One Corp.) into sections of approximate length 10 mm. The samples prepared with Fe-box and Mo-box electrodes were washed with anhydrous ethylene glycol (99.5%, Wako Pure Chemical Industries Ltd.) using an ultrasonic washing machine and rinsed with anhydrous acetone (99.5%, Wako Pure Chemical Industries Ltd.). Ethylene glycol was used because it dissolves CaCl₂ without reacting with Ca alloy. X-ray diffractometry (XRD; Rigaku, Ultima IV, Cu-Kα ray, λ = 1.5418 Å, 40 kV, 40 mA) was used for sample identification. The samples were observed by using an optical digital microscope (Dino Lite PRO Polarizer DILITE30 AM-413ZT, Sanko Co., Ltd.) and scanning electron microscopy (SEM; VE-8800, Keyence Corp.). The elemental composition of the sample surfaces was analyzed by energy-dispersive X-ray spectroscopy (EDX; EDAX Genesis APEX2, AMETEK Co., Ltd., accelerating voltage: 15 kV). The aluminum droplets obtained by the electrolysis were dissolved into 0.528 mL of 30 wt% HCl solution. Then, pure water was added to a total volume of 50 mL. Inductively coupled plasma atomic emission spectrometry (ICP-AES; SPECTRO Blue, Hitachi High-Tech Science Group) was used for the elemental analysis.

The electrode potential and the activity of O²⁻ ions in the electrolyte are two major factors that determine the electrochemical reduction of oxides in molten salts. In several previous studies on the electrochemical reduction of solid oxides in molten salts, potential–pO²⁻ diagrams have been used to thermodynamically predict the chemical and electrochemical behaviors of the substances. A potential–pO²⁻ diagram was first applied to the electrowinning process by Littlewood.²¹ Potential–pO²⁻ diagrams have been reported for TiO₂ in molten CaCl₂,^22–25 NiTiO₃ in CaCl₂,²⁶ SiO₂ in CaCl₂,^10,20 Al₂O₃ in CaCl₂-based molten salts,^15,20 and for many other substances in various molten salts.^27–35 Yan and Fray constructed potential–pO²⁻ diagrams for the Al–Ca–O–Cl system in molten NaCl–CaCl₂ and LiCl–CaCl₂,¹⁵ and reported that the experimental results for the electrolytic reduction of Al₂O₃ were consistent with the predictions.

In the present study, potential–pO²⁻ diagrams for the Al–Ca–O–Cl system in molten CaCl₂ at 1123 K are created, as shown in Fig. 4, to predict suitable conditions for the production of Al metal and to interpret the electrolysis results. Fig. 4a depicts the diagram with wide potential and wide pO²⁻ regions (−0.5 V < E < +4.0 V vs. Ca²⁺/Ca, 0 < pO²⁻ < 14). To discuss the reduction of Al₂O₃, Fig. 4b focuses on the negative-potential and low-pO²⁻ regions (−0.1 V < E < +1.0 V, 0.5 < pO²⁻ < 5.0). The dotted lines indicate the frame of the potential–pO²⁻ diagram for CaCl₂ at 1123 K, i.e., deposition of Ca metal, evolution of Cl₂ and O₂ gases, and precipitation of solid CaO. Table I lists the thermochemical data used for the construction of the diagrams.^36–45 The diagrams in Fig. 4 are essentially similar to that reported in our previous study.²⁰ As for the differences, the detailed information on alloy formation and Ca content in the liquid phase has been added. Furthermore, the stable region for CaAl₁₂O₁₉ has appeared in the present version.

Zoom In Zoom Out Reset image size

According to the phase diagram of the binary Al–Ca system,³⁹ as shown in Fig. 5, the stable phases at 1123 K are Al-rich Al–Ca liquid alloy, solid Al₂Ca, and Ca-rich Al–Ca liquid alloy. Here, the Ca content, x_Ca, in the Al-rich Al–Ca liquid alloy that equilibrates with Al₂Ca is 0.16. Because the activity of Ca was reported to be 1.7 × 10⁻³ at this composition at 1373 K,⁴⁶ the activity coefficient, γ_Ca, and electrode potential, E, in the {Al-rich Al–Ca (l) and Al₂Ca (s)} two-phase state at 1123 K are calculated to be:

Here, the interaction parameter, Ω, is assumed to be independent of temperature (Ω = RT ln γ = const ∴ T ln γ = const.), where R and T are the gas constant and the absolute temperature, respectively. In the same manner, the electrode potential in the {Ca-rich Al–Ca (l) and Al₂Ca (s)} two-phase state at 1123 K is estimated from the reported activity of Ca, 0.313, at 1619 K.⁴⁷

Zoom In Zoom Out Reset image size

Based on the above calculations, the stable regions in the potential–pO²⁻ diagram are determined for Al-rich Al–Ca alloy (l), Al₂Ca (s), and Ca-rich Al–Ca alloy (l). There is a wide stable region of liquid Al in the high-pO²⁻ region, i.e., low activity of O²⁻ ions. Here, the content of Ca in the liquid Al is determined by the electrode potential. The potential values corresponding to several representative Ca contents in the liquid Al are plotted with dashed lines in Figs. 4a and 4b. Al₂Ca is predicted to be formed at potentials more negative than 0.36 V. Ca-rich Al–Ca(l) alloy forms at potentials more negative than 0.067 V. When the thermodynamic data obtained by FP-CALPHAD method⁴⁸ is adopted, the potentials of Al₂Ca formation and Ca-rich Al–Ca(l) alloy formation are calculated as 0.310 V and 0.101 V, respectively. The potential at the three-phase equilibrium between CaAl₁₂O₁₉, Al, and Al₂O₃ is 0.78 V, pO²⁻ = 4.31. A total of five stable regions for calcium aluminates appear in the low-pO²⁻ range: CaAl₁₂O₁₉ (3.04 ≤ pO²⁻ ≤ 4.31, E ≥ 0.62 V), CaAl₄O₇ (2.31 ≤ pO²⁻ ≤ 3.04, E ≥ 0.53 V), CaAl₂O₄ (1.41 ≤ pO²⁻ ≤ 2.31, E ≥ 0.40 V), CaAl₄O₇ (1.16 ≤ pO²⁻ ≤ 1.41, E ≥ 0.35 V), and Ca₁₂Al₁₄O₃₃ (1.08 ≤ pO²⁻ ≤ 1.16, E ≥ 0.34 V). It should be noted that Ca₁₂Al₁₄O₃₃ did not appear in the Al–Ca–O phase diagram at 1273 K published in 1990,⁴⁹ because its thermodynamic data was reported recently.

Cyclic voltammetry

The reduction behavior of Al₂O₃ in the three-phase zone is investigated by cyclic voltammetry using an Al₂O₃-sealed electrode. Fig. 6 shows cyclic voltammograms (CVs) measured in a potential range of 0.2–1.9 V (vs. Ca²⁺/Ca) for five consecutive cycles. The apparent current density is calculated by using the cross-sectional area of the W rod (dia. 2 mm). In the first cycle, when the scan starts from the rest potential (ca. 1.7 V) in the negative direction, the reduction current sharply increases from 0.6 V and gives a peak at ca. 0.45 V. According to the potential–pO²⁻ diagram (Fig. 4), the reaction is expected to be the electrochemical reduction of Al₂O₃ to Al metal:

The cathodic current further increases from 0.35 V, which would correspond to the formation of Al₂Ca:

In addition, the current increases with the cycle number, which indicates an increase in the effective reaction zone of conductor/Al₂O₃/molten salt by the formation of conductive products, such as metallic Al and Al₂Ca.

Zoom In Zoom Out Reset image size

Potentiostatic electrolysis of Al₂O₃-sealed electrodes

The reduction of Al₂O₃ is investigated by potentiostatic electrolysis of Al₂O₃-sealed electrodes at 0.30 V for 30 and 60 min. Microscope images of the samples before and after electrolysis are shown in Fig. 7. After 30 min, there are traces of reduction at the Al₂O₃ tube near the W rod. In the case of 60 min, the traces of reduction have spread in the outer direction. These results indicate that the reduction starts in the three-phase zone of Al₂O₃, the conductive W rod, and molten CaCl₂. From an EDX analysis of the 60-min sample, the composition of the reacted area is O 20 at%, Al 60 at %, which confirms the electrochemical reduction of Al₂O₃.

Zoom In Zoom Out Reset image size

Potentiostatic electrolysis of Al₂O₃ powder using Fe box-type electrodes

With reference to the CV results, potentiostatic electrolysis of Al₂O₃ powder is conducted at 0.50, 0.40, 0.30, 0.25, and 0.20 V for 6 h using Fe-box electrodes to investigate the potential dependence of reaction. Fig. 8a shows the current–time curves during the electrolysis. Larger cathodic currents are observed at more negative potentials. Cross-sectional optical images of the obtained samples are shown in Fig. 8b. The powders are found at the bottom and the solidified CaCl₂ are found above the powders in the Fe-box electrodes. The color of the powder changes from white to black at 0.40 V, to greenish-brown at 0.30 V and 0.25 V, and to metallic silver at 0.20 V.

Zoom In Zoom Out Reset image size

XRD patterns of the powders are shown in Fig. 9. The distinct peak pattern for Ca₁₂Al₁₄O₃₃ is observed for the samples at 0.25, 0.30, 0.40, and 0.50 V. Yan and Fray also reported that Ca₁₂Al₁₄O₃₃ was formed in the electrochemical reduction of Al₂O₃ in CaCl₂-based melt.¹⁵ The patterns for 0.30, 0.40, and 0.50 V indicate the existence of unreacted Al₂O₃, where the peak intensities are smaller at more negative potentials. For the sample at 0.20 V, only the Al₂Ca phase is confirmed. Incidentally, the inner surface of the Fe-box electrode had changed to Al₅Fe₂ (PDF # 00-047-1435) after electrolysis of Al₂O₃ powder at 0.4 V for 12 h, which is confirmed by its appearance and XRD analysis (Fig. 10). Since the Fe-box alloyed with the produced Al, the Al₅Fe₂ formed at the surface. The more internal parts of Fe walls may have contained other Al–Fe alloy phases with higher Fe content. However, they could not be detected by XRD due to the limited penetration depth of X-ray.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Ca₁₂Al₁₄O₃₃ has been reported to form free electrons in the cage framework of a crystal structure by a deoxidation reaction, showing a green color owing to the transition of free electrons to the conduction band and metallic conductivity at room temperature.^50–54 The samples obtained at 0.25 and 0.30 V, which are composed of Ca₁₂Al₁₄O₃₃, exhibit a greenish color, suggesting that they have high conductivity after the electrochemical deoxidation. After Al₂O₃ powder in an Fe-box electrode is immersed in CaCl₂ for 6 h without electrolysis, the formation of Ca₁₂Al₁₄O₃₃ is not confirmed. Thus, the Ca₁₂Al₁₄O₃₃ at 0.25 and 0.30 V is considered to be formed by the reaction of Al₂O₃, Ca²⁺, and electrochemically formed O²⁻ ions.

To summarize these results, the mechanism of the reduction of Al₂O₃ powder in an Fe-box electrode is described as follows.

• (1)  
  At first, Al₂O₃ in contact with the Fe-box electrode is reduced to Al₅Fe₂ alloy.

  
• (2)  
  In the vicinity of the reduced Al₂O₃ powder, Ca₁₂Al₁₄O₃₃ is formed by the reaction of Al₂O₃, Ca²⁺, and the electrochemically produced O²⁻ ions.

  
• (3)  
  The Ca₁₂Al₁₄O₃₃ near the Fe-box electrode is reduced to form Al₅Fe₂ alloy.

  
• (4)  
  When the electrode potential is below 0.20 V, Ca₁₂Al₁₄O₃₃ is electrochemically reduced to form Al₂Ca.

  

Considering that the obtained phases are Ca₁₂Al₁₄O₃₃ and Al₂Ca alloy for the Fe-box electrodes, the concentration of O²⁻ ions in the three-phase zone of Al₂O₃/molten salt/conductor is high (pO²⁻ < 1.41) during the electrolysis. The high concentration of O²⁻ ions is believed to be brought about by the stagnation of O²⁻ ions in the Al₂O₃ powder. The diffusion of O²⁻ ions from the electrochemical reaction interface to the bulk molten salt is not easy, owing to the long diffusion paths passing through the gaps between the Al₂O₃ particles. In order to obtain Al metal with low Ca content, the concentration of O²⁻ ions must be kept low (pO²⁻ > 2.86 for Ca content < 0.1 at%) To realize this, the diffusion of O²⁻ ions from the reaction interface to the bulk molten salt has to be facile. For this purpose, a new electrode structure, in which a dense Al₂O₃ tube is placed on a flat metal plate, is considered to be effective.

Potentiostatic electrolysis of Al₂O₃ tube in a Mo box-type electrode

On the basis of the above considerations, the electrochemical reduction of Al₂O₃ tube is carried out using a Mo-box electrode with a wide bottom and a shallow depth (Fig. 3c). Mo has been used instead of Fe, because it is not easily alloyed with Al.

Fig. 11 shows a current–time curve during the potentiostatic electrolysis at 0.25 V for 12 h. A nearly constant current of −200 mA is observed. Fig. 12a shows an optical image of the sample after the electrolysis. Black products are found around the Al₂O₃ tube. After washing the sample with ethylene glycol (Fig. 12b), evident decreases in height and wall thickness are observed for the Al₂O₃ tube. Moreover, metallic droplets with an approximate maximum diameter of 3 mm are obtained from the bottom of the crucible (Fig. 12c). The total mass of the droplets is 32 mg. Fig. 13 shows an XRD pattern of the droplets, which confirms the formation of Al metal. An ICP-AES analysis of the obtained Al droplets reveals that the composition is 95.4–98.0 at% Al and 2.0 at%–4.5 at% Fe. The concentrations of Ca and Mo are less than the detection limit. In this experiment, the level of detection limit for ICP-AES is 1–2 × 10⁻² at %. The current efficiency is calculated to be approximately 50% from the mass loss of the Al₂O₃ tube, 605 mg, and the charge during the electrolysis, 3431 C. The background current of the Mo-box electrode is likely the main reason for the low efficiency. In addition, formation of dissolved Ca in molten CaCl₂, which induces shuttle current between the anode and the cathode, is another cause of the low current efficiency; the solubility of metallic Ca in molten CaCl₂ is reported to be ca. 3 mol% at 1123 K.⁵⁵ The black deposits in the Mo-box electrode are identified as Al₂Ca by XRD analysis. The Al₂Ca phase is considered to be precipitated from liquid Al–Ca alloy upon cooling.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The above results have demonstrated that the new electrode structure is effective. Fig. 14 compares the diffusion paths of O²⁻ ions for Al₂O₃ powder and Al₂O₃ tube. Electrochemical reduction of Al₂O₃ occurs at the three-phase zone to produce O²⁻ ions, which decrease the value of pO²⁻. In the case of powder, the diffusion of O²⁻ ions is slow due to the small space among particles (left in Fig. 14), which causes the very low pO²⁻ in the three-phase zone. The deeper color means the low pO²⁻ region, i.e., the high O²⁻ ion concentration region. On the contrary, for the tube, the facile diffusion of O²⁻ ions enables the higher pO²⁻ in the three-phase zone. According to the potential–pO²⁻ diagram, the production of Al metal is possible only when the pO²⁻ is not very low. Thus, even for Al₂O₃ powder, the production of Al metal is expected by using a larger particle size and/or by performing agitation.

Zoom In Zoom Out Reset image size

The electrochemical reduction of solid Al₂O₃ has been investigated in molten CaCl₂ at 1123 K. Cyclic voltammetry using an Al₂O₃-sealed electrode indicates that the reduction of Al₂O₃ proceeds at potentials more negative than 0.6 V. When Al₂O₃ powder is electrolyzed in an Fe-box electrode, Ca₁₂Al₁₄O₃₃ is formed at 0.25 and 0.30 V, and Al₂Ca is formed at 0.20 V. However, Al droplets are obtained from the bottom of a crucible by the electrolysis of an Al₂O₃ tube in a Mo-box electrode at 0.25 V. The difference in the reduction behavior is explained by the diffusivity of O²⁻ ions from the reaction interface to the bulk molten salt and thermodynamic considerations using the potential–pO²⁻ diagrams.

Toshiyuki Nohira 0000-0002-4053-554X