Materials Chemistry
Yttriothermic Reduction of TiO2 in Molten Salts
2020 Volume 61 Issue 10 Pages 1967-1973
Details
2020 Volume 61 Issue 10 Pages 1967-1973
A new reduction process for producing titanium (Ti) with an ultra-low oxygen concentration directly from TiO2, employing yttrium (Y) as the reductant, was developed in this study. Several methods for the direct reduction of TiO2 have been proposed to lower the cost of production of Ti. However, none of them have yet been applied industrially. In addition, Y has never been used as a reductant for reducing TiO2 although its reducing ability is the highest among the reductants capable of reducing TiO2 to Ti with low oxygen concentration. In this study, the reduction reactions of TiO2 using Y/Y2O3 equilibrium and Y/YOCl/YCl3 equilibrium, employing various molten salts as solvents, were investigated. TiO2 pellets and metallic Ti pieces were placed in several types of solvents along with sufficient Y and heated at 1300 K for 86 ks. When YCl3 or CaCl2 was used as the solvent, the TiO2 pellets were reduced to metallic Ti. The oxygen concentrations in the Ti pieces after heating in YCl3 and CaCl2 were 90 ± 40 mass ppm O and 350 ± 60 mass ppm O, respectively. However, when NaCl or KCl was used as the solvent, a small amount of metallic Ti and a large amount of complex oxides were obtained. It is considered that the reduction reaction did not proceed sufficiently owing to the low solubility of the oxide ions in molten salts. It was experimentally demonstrated that Ti with an ultra-low oxygen concentration (100 mass ppm O or less) can be directly produced from TiO2 by using Y as the reductant in an appropriate solvent. This method is expected to lead to the development of a new industrial process for the production of Ti with an ultra-low oxygen concentration directly from its ore.
Titanium (Ti) is known as a rare metal, more expensive than commonly used structural metals. Therefore, despite its excellent properties, such as its high specific strength and corrosion resistance, Ti is not used commercially as a structural material, primarily because of the low productivity of the current Ti smelting processes. The development of new smelting processes for Ti, with enhanced productivity, can reduce its price drastically.
Methods for the direct reduction of TiO2 have recently attracted considerable attention owing to the efforts for lowering the cost of Ti. Direct reduction techniques can simplify and accelerate the smelting process by removing the chlorination process of the oxide feed materials from the smelting process. However, the affinity between Ti and O, and the solubility of O in Ti are both high, and very few metals can reduce TiO2 to low-oxygen-concentration Ti.1–5)
Figure 1 is the Ellingham diagram showing the relationship between ΔGr° (= RT ln $p_{\text{O}_{2}}$) and temperature, either in metal/metal oxide equilibrium or metal/metal oxychloride/metal chloride equilibrium.4,6–11)
At present, the industrial requirement of the oxygen concentration in Ti metal is approximately 500 mass ppm O. According to Fig. 1, the metals which can reduce the oxygen concentration in Ti to the required industrial level in metal/metal oxide equilibrium are calcium (Ca) and yttrium (Y). In addition, the oxychloride formation reaction has been experimentally proved to enhance the deoxidation ability of several rare earth metals (REMs) whose deoxidation ability in their metal/metal oxide equilibrium is low.4,5,10,12–17) Figure 1 shows that Y has the highest deoxidation ability among the reactive metals. Recently, Iizuka et al. experimentally demonstrated the deoxidation of metallic Ti down to approximately 30–60 mass ppm O using Y/YOCl/YCl3 equilibrium.15)
Direct methods for the reduction of TiO2 using Ca as the reductant have been studied earlier by many researchers. The calciothermic reduction methods,18–28) as well as methods combining electrochemical methods with calciothermic reduction methods (e.g., FFC method29–31) and OS method32–34)), have been developed. However, these methods have not been industrialized on a large scale mainly because of the inefficient regeneration process of Ca. Although methods for the reduction of TiO2 using Mg or Al as the reductant have also been developed, it is currently difficult to produce low-oxygen-concentration Ti with high reproducibility by these methods.35–40)
In this study, we focused on using Y as a reductant. In an attempt to produce Ti from TiO2 using REMs as reductants for TiO2, Vogel et al. tried to reduce TiO2 using mischmetal as a reductant.41) However, a large amount of Ti nitride was produced along with metallic Ti owing to a technical problem in controlling the atmospheric conditions in those days. Recently, our group studied a new process for the reduction of TiO2 using lanthanum (La) as the reductant.16) The low deoxidation ability of La in La/La2O3 equilibrium necessitates the use of the La/LaOCl/LaCl3 equilibrium to produce Ti with a low oxygen concentration (cf. Fig. 1). Conversely, when Y is used as the reductant, several solvents that do not contain the Y halide salt (e.g., YCl3) can be used because the deoxidation ability of the Y/Y2O3 equilibrium is high enough to lower the oxygen concentration in metallic Ti.
In this study, a new process for the production of Ti with a low oxygen concentration directly from TiO2, using Y as a reductant, was developed. We discuss the differences in the progress of the reduction reactions of TiO2 observed when different molten salts were used as solvents facilitating the reduction reactions.
Thermodynamic considerations were used to determine whether TiO2 could be reduced to metallic Ti using Y as a reductant. The reduction of oxides to metals is often hindered by the intermediate compounds formed during the reaction. In particular, the existence of complex oxides in the TiO2–Y2O3 system has been reported earlier, as shown in Fig. 2.42)
TiO2–Y2O3 pseudo-binary phase diagram.42)
Figure 3(a) shows Ti–Y–O ternary chemical potential diagram at 1300 K containing complex oxides of TiO2–Y2O3 system. The thermodynamic data in Tables 1 and 2 were used to construct the diagram shown in Fig. 3(a).42) Figure 3(b) depicts the Ti–Y–O isothermal phase diagram at 1300 K constructed using the thermodynamic data in Tables 1 and 2. A Ti–Y binary phase diagram43) shows that the mutual solubility between Ti and Y is low. The Ti/Y/Y2O3 three-phase equilibrium point observed in Fig. 3(a) implies that from the thermodynamic point of view, TiO2 can be reduced to Ti when a sufficient amount of metallic Y is present in the system.
(a) Chemical potential diagram of the Ti–Y–O ternary system at 1300 K and (b) the isothermal ternary phase diagram for the Ti–Y–O system at 1300 K. The end point of reaction (eq. (2)), which was calculated based on the initial quantities of chemicals in the Exp. no. R1 listed in Table 4 is shown as a white circle.
The molten salts available for the yttriothermic reduction of TiO2 were selected. Figure 4 shows the $p_{\text{O}_{2}}$–$p_{\text{Cl}_{2}}$ diagram reported by Iizuka et al.15) which indicates that LiCl, NaCl, KCl, and CaCl2 are thermodynamically stable in Y/Y2O3 equilibrium. These salts containing YCl3 or pure YCl3 can be used to establish Y/YOCl/YCl3 equilibrium as shown in Fig. 4. Metallic Y cannot coexist with MgCl2, because they react according to eq. (1). Consequently, MgCl2 is unsuitable as a solvent for the yttriothermic reduction of TiO2.
| \begin{equation} \text{Y ($s$)} + \text{3/2 MgCl$_{2}$ ($l$)} = \text{YCl$_{3}$ ($l$)} + \text{3/2 Mg ($l$)} \end{equation} | (1) |
| \begin{equation} \text{TiO$_{2}$ ($s$)} + \text{4/3 Y ($s$)} = \text{Ti ($s$)} + \text{2/3 Y$_{2}$O$_{3}$ ($s$)} \end{equation} | (2) |
| \begin{align} &\text{TiO$_{2}$ ($s$)} + \text{4/3 Y ($s$)} + \text{2/3 YCl$_{3}$ ($l$)} \\ &\quad = \text{Ti ($s$)} + \text{2 YOCl ($s$)} \end{align} | (3) |
Thus, in principle, Ti with a low oxygen concentration can be produced directly from the oxide feed material using the yttriothermic reduction. In this work, experimental studies of the yttriothermic reduction of TiO2 were conducted, and the differences observed in the reduction reactions in the solvents NaCl, KCl, CaCl2, and YCl3 were discussed.
Figure 5 shows a schematic representation of the experimental apparatus. TiO2 powder (≥98%, Anatase, Kanto Chemical Co., Ltd.) was sintered into pellets to ensure that the reduction of TiO2 could be easily observed. TiO2 powder was pressed at approximately 600 MPa, heated at 1300 K for 86 ks in air, and sintered. The TiO2 pellets, Y shot (≥99%, Shot, Santoku Co., Ltd.), YCl3 powder (≥99.9% (REO), anhydrous, Alfa Aesar Co., Ltd.), and metallic Ti pieces were placed in a Ti crucible (CP-Ti, 25.4 mm diameter × 1 mm thickness × 80 mm length, Shinkinzoku Industry Co., Ltd.). Three types of Ti pieces with different shapes and oxygen concentrations (“Ti–1.2U”: CP-Ti wire, 1.2 mm diameter × 20 mm length, 1300 mass ppm O, Shinkinzoku Industry Co., Ltd.; “Ti–2.0I”: CP-Ti wire, 2.0 mm diameter × 10 mm length, 900 mass ppm O, Shinkinzoku Industry Co., Ltd.; “Ti–6N”: Ti cube, 2 mm in length, 150 mass ppm O, Electron-beam melted high purity Ti) were used. As shown in Fig. 5, Y was placed at the bottom of the Ti crucible, followed by the Ti foil. The TiO2 pellets and Ti pieces were placed on the Ti foil to avoid direct contact with Y. The amount of Y placed was proportionate to the amount of TiO2 (cf. Fig. 3(b)).
Schematic representation of the experimental apparatus (Exp. condition: T = 1300 K, t = 86 ks).
The Ti crucibles, along with a small amount of Ti sponge (≥99%, Toho Titanium Co., Ltd.), were sealed in a stainless-steel crucible (SUS-316, 88.9 mm diameter × 3 mm thickness × 100 mm length, Azabu Industry Co., Ltd.) by welding. The reduction experiment was performed by placing the stainless-steel crucible in a muffle furnace at 1300 K for 86 ks, and terminated by quenching in water. The stainless-steel and Ti crucibles were then opened mechanically, and the samples were extracted and analyzed.
The Ti pieces were cleaned by leaching, first with distilled water and then with diluted HCl solution (1 + 10) (HCl: 35.0–37.0%, Kanto Chemical Co., Ltd.). The Ti pieces were first chemically polished using a 1:4:10 mixture of HF–HNO3–H2O (HF: 46.0–48.0%, Morita Chemical Industry Co., Ltd.; HNO3: 60.0–62.0%, Wako Industry Co., Ltd.) and then rinsed using distilled water and acetone (≥99%, Kanto Chemical Co., Ltd.). The oxygen concentrations of the Ti pieces, which had different shapes and initial oxygen concentrations, were analyzed using the inert gas fusion method (TC-600, LECO Co., Ltd.). The achievement of equilibrium was confirmed by verifying that the concentrations of oxygen in Ti pieces with different initial oxygen concentrations and shapes reached approximately the same value after reduction.
The pellets extracted were also cleaned by leaching with distilled water and diluted HCl (1 + 10) solution. A scanning electron microscope with an energy dispersive X-ray spectrometer (SEM-EDS) (JSM-65101LA, JEOL Co., Ltd.) was used for the surface observation and composition analysis of the cleaned pellets. X-ray diffraction (XRD) analyses (D2 Phaser, Bruker) of the salt used for reduction and the leached pellets were also conducted.
Figure 6 shows the cross-sectioned Ti crucible and the pellets removed after the experiment. The pellets that reacted in YCl3 (Exp. no.: R1) and CaCl2 (Exp. no.: R4) had a shiny silver appearance after removal and leaching with pure water, as shown in Fig. 6(d) and (e). Conversely, the pellets that reacted in NaCl (Exp. no.: R2) and KCl (Exp. no.: R3) spontaneously disintegrated into powder upon leaching with pure water, as shown in Fig. 6(f) and (g).
Photographic images of the samples: (a) Sectioned Ti crucible and samples obtained after the reduction experiment (T = 1300 K, t = 86 ks), (b) TiO2 pellets and Ti foil before the experiment, (c) Ti pieces after the experiment (Ti–1.2U, Ti–2.0I, and Ti–6N), (d) R1: Pellets after reacting in YCl3 and leaching with distilled water, (e) R4: Pellets after reacting in CaCl2 and leaching with distilled water, (f) R2: Pellets after reacting in NaCl and leaching with distilled water, and (g) R3: Pellets after reacting in KCl and leaching with distilled water.
Figure 7 shows the SEM image of the pellets after the reaction, while Table 3 shows the analytical results obtained using EDS. Figure 7(a) also shows that the R1 pellets which reacted in YCl3 were reduced to high purity Ti pellets. Figure 7(b) shows that the R4 pellets which reacted in CaCl2 transformed into Ti pellets with a relatively high concentration of impurities such as oxygen. Conversely, Fig. 7(c) shows that the R2 pellets which reacted in NaCl contained some powdered Ti as well as a large amount of powder consisting mainly of Na–Ti–O based complex oxides. Metallic Ti powder was not obtained from the R3 pellets that reacted in KCl, instead, K–Ti–O based complex oxides were found in large quantities, as shown in Fig. 7(d).
SEM images and the EDS analyzed areas of the reduction products: (a) R1: reduced in YCl3, (b) R4: reduced in CaCl2, (c) R2: reduced in NaCl, and (d) R3: reduced in KCl.
Figure 8 shows the results of the XRD analysis.44–46) Metallic Ti was detected in the R1 pellets after the experiment, whereas in the R4 pellets, metallic Ti was detected in addition to Y2O3. Many peaks, including the Ti and Y2O3 phases, were detected from the R2 and R3 pellets. Numerous peaks, including large peaks located at low angles, did not match the reported compound data available to the authors such as Y(OH)3 and YCl3·6H2O. These could correspond to the complex oxides detected during the SEM-EDS analysis.
XRD profile of the experimental samples: (a) Rutile-type TiO2 sample before heating, (b) R1: reduced in YCl3 at 1300 K for 86 ks using Y, (c) R4: reduced in CaCl2 at 1300 K for 86 ks using Y, (d) R2: reduced in NaCl at 1300 K for 86 ks using Y, and (e) R3: reduced in KCl at 1300 K for 86 ks using Y.
Table 4 shows the analytical results for the oxygen concentration in the Ti pieces. The oxygen concentration in the Ti pieces deoxidized in YCl3 (R1) was 90 ± 40 mass ppm O, whereas the concentration in the Ti pieces deoxidized in CaCl2 (R4), NaCl (R2) and KCl (R3) were 350 ± 60 mass ppm O, 160 ± 40 mass ppm O, and 130 ± 40 mass ppm O, respectively.
The above results show that the reduction reaction of TiO2 progressed and metallic Ti was obtained when YCl3 or CaCl2 was used as solvent, whereas the reduction reaction of TiO2 did not progress sufficiently when NaCl or KCl were used. In addition, the oxygen concentration in the Ti pieces reduced using YCl3 in Y/YOCl/YCl3 equilibrium was lower than that of the Ti pieces in the other molten salts.
We now explain the reasons behind the insufficient progress of the reduction reaction of TiO2 and the formation of complex oxides only when NaCl or KCl was used. The depletion of metallic Y (reductant) in the system cannot be held responsible for the incomplete reduction since the oxygen concentrations in the Ti pieces decreased to a value close to the deoxidation limit in Y/Y2O3 equilibrium.
As shown in Fig. 9, YOCl and CaO have relatively high solubility in YCl3 and CaCl2, respectively,47,48) indicating that a relatively large amount of oxide ions dissolve in the molten YCl3 and CaCl2 salts. Conversely, the solubility of the oxide ions in NaCl and KCl is low.49) Thus, it can be assumed that under the experimental conditions, the oxide ions produced during the reduction were not removed at a sufficient rate from the surface of the TiO2 pellets in NaCl and KCl; hence, the reduction reaction did not proceed as desired.
In this study, a new method for the production of Ti by the yttriotherimic reduction of TiO2 was developed. In addition, the reduction experiments of TiO2 in several molten salts using Y as a reductant were conducted, and the progress of the reduction reaction was observed. Metallic Ti was obtained experimentally when YCl3 and CaCl2 were used as solvents. When NaCl and KCl were used, the reduction reaction did not proceed satisfactorily and complex oxides were formed. The difference in the results probably originates from the solubility of the oxide ions in the molten salts.
In particular, we have shown that in principle, Ti with an ultra-low oxygen concentration (100 mass ppm O or less) can be produced directly from TiO2 using YCl3 as the solvent. We have experimentally shown, for the first time, the effectiveness of the yttriothermic reduction method for the direct reduction of TiO2 to Ti with an ultra-low oxygen concentration.
With the introduction of a technique for regenerating Y and YCl3 from YOCl, such as an electrochemical process, the new process flow—including the developed yttriothermic reduction—will achieve zero net consumption of Y. This study is expected to lead to the development of an industrial process for the production of Ti with an ultra-low oxygen concentration from Ti ore without chlorination.
We are grateful to Mr. Akihiro Iizuka and Dr. Lingxin Kong of the University of Tokyo for their various suggestions. This work was financially supported by the Japan Society for the Promotion of Science (JSPS) through a Grant-in-Aid for Scientific Research (S) (KAKENHI grant No. 26220910 and 19H05623).
