2019 Volume 60 Issue 3 Pages 386-390
In order to check the solubility of CaS during sulfide reduction in molten CaCl2, the mixture of CaCl2 and a small amount of CaS was melted in Ar at the range from 1123 K to 1223 K. The melt was sampled by quartz tube and rapidly solidified. The solidified samples showed a lamella structure with CaS particles, which results in a simple eutectic reaction between CaCl2 and CaS. Using ICP-AES analysis, the saturation of CaS was found to be completed within 1.8 ks, and 1.77 ± 0.1 mol%CaS at 1173 K was measured as the solubility limit at the initial composition of 3.0 mol%CaS. A tentative phase diagram of CaCl2–CaS binary system was proposed based on the solubility analysis in these temperatures and eutectic structure.
Fig. 8 Phase diagram of CaCl2–CaS system estimated by CaS solubility in CaCl2. Closed and open circles show the initial conditions and the analyzed liquidus position, respectively.
Molten calcium chloride with addition of small amount of CaO has been often used to prepare pure metallic titanium from its oxides. Using this molten CaCl2–CaO, the direct reduction from TiO2 is extensively studied under the proposals such as FFC-Cambridge process1) and OS process.2,3) For example, when only 3.0 V was applied between a carbon anode and cathodic oxide powder, CaO dissolved in the CaCl2 melt is electrochemically decomposed to metallic Ca, which can react with TiO2 at the cathode to form Ti. A sequence of this CaO decomposition and reduction of TiO2 can be written as,3)
| \begin{equation} \text{CaO}=\text{Ca$^{2+}$}+\text{O$^{2-}$ (in the melt)} \end{equation} | (1) |
| \begin{equation} \text{Ca$^{2+}$}+\text{2 e$^{-}$}=\text{Ca (at cathode)} \end{equation} | (2) |
| \begin{equation} \text{Ca}+\text{TiO$_{2}$}=\text{Ti}+\text{CaO (near cathode)} \end{equation} | (3) |
| \begin{equation} \text{$m$ O$^{2-}$}+\text{C}=\text{CO$_{m}$}+\text{$2m$ e$^{-}$ (at anode, $m=1$ or 2)} \end{equation} | (4) |
Chen et al. proposed another mechanism as known as FFC-Cambridge process1) that the oxide at the cathode releases O2− to form the metal via the lower oxides. The oxide ions diffuse out from the oxides to the CaCl2 melt. In case of titanium oxide, their proposal can be shown as,
| \begin{equation} \text{TiO$_{n}$}+\text{$2n$ e$^{-}$}=\text{Ti}+\text{$n$ O$^{2-}$ ($1\leq n\leq 2$)} \end{equation} | (5) |
The oxygen anion both from OS process and FFC process dissolves in the chloride melt and reacts with carbon anode to remove oxygen as CO or CO2 gas out from the reaction vessel. Therefore, the behavior of oxygen dissolving into CaCl2 is critical for successive reactions.
The solubility limits of CaO vary depending on solvent and temperature.9–14) As a typical condition of oxide reduction, the experiments were often operated in CaCl2 melt at 1173 K in Ar gas atmosphere. This condition is taken for comparison of solubility. Neumann et al. reported 22 mol%CaO9) as the solubility limit, while Wenz et al. and Perry showed 1912) and 20 mol%,10) respectively. Wang et al. reported the largest value, about 26 mol%.14) Although a large scatter can be found in the previous papers,9–15) the solubility of CaO in CaCl2 melt is much larger than the other combination of solid oxide-chloride melt.14) Combining this wide solubility of CaO and a strong thermochemical reducibility of Ca in CaCl2 melt,16) the by-product CaO from the oxide reduction as shown in eq. (3) can be apart from the reaction site immediately, and promote the reduction toward the right side direction in eq. (3) without any interfere on the successive reduction.16) The ternary phase equilibrium in Ca–CaO–CaCl2 region was examined16) and it is known that the solubility of Ca becomes smaller at the higher concentration range of CaO. This means that the supply of reductant Ca is weaker as the residual CaO increases.3,16)
On the other hand, the oxygen solubility in the solid valve metals such as Ti, V, Nb and Ta is generally very large; for example, 14 mass%O in α-Ti.17) Titanium metal with such a high oxygen content is very brittle, and the complete removal of oxygen from Ti–O solid solution is required as proposed by Okabe et al.18,19)
| \begin{equation} \text{O (in Ti)}+\text{Ca}=\text{CaO} \end{equation} | (6) |
As an approach to avoid the oxygen contamination, the authors proposed a method as a kind of modification of OS process:20–22) Once TiO2 is converted to TiS2, and the sulfide is reduced to metallic Ti using CaCl2–CaS mixed liquid. This idea is inspired that sulfur locates below oxygen in periodic table, and that the chemical properties may be similar. However, sulfur has larger atomic radius than oxygen, and it is not easy to dissolve sulfur atoms in the metallic lattice. Therefore, the solubility of sulfur in any metal is generally much smaller than that of oxygen.17) The authors found that sulfur removal from the Ti and V sulfides occurred more rapidly to form their metals, and that the concentrations of residual sulfur in the obtained metals were very low.20–22)
V3S4 is one of the recycled products from the waste of secondary battery. In the example of vanadium metal formation, it was considered that the below-listed reactions are operated simultaneously.22)
| \begin{equation} \text{V$_{3}$S$_{4}$}+\text{4 Ca}=\text{3 V}+\text{4 CaS} \end{equation} | (7) |
| \begin{equation} \text{CaS}=\text{Ca$^{2+}$}+\text{S$^{2-}$ (in molten salt)} \end{equation} | (8) |
| \begin{equation} \text{Ca$^{2+}$}+\text{2 e$^{-}$}=\text{Ca} \end{equation} | (9) |
| \begin{equation} \text{2 S$^{2-}$}=\text{S$_{2}$}+\text{4 e$^{-}$} \end{equation} | (10) |
The mechanism consisting of these reactions is constructed using the analogy that the behavior of oxygen in CaCl2–CaO melt is similar with that of sulfur in CaCl2–CaS. The authors previously observed that solid particles of CaS floated on the top surface of CaCl2 when 3 mol%CaS was added in the melt,21) and they expected the added CaS in CaCl2 will dissolve like CaO in CaCl2. The solubility was considered less than 3 mol%CaO21) in contrast with 19–26 mol%CaO.9–15) There is no quantitative study on CaS solubility in CaCl2.
The purpose of this work is to observe the dissolution behavior of CaS in liquid CaCl2 (melting point is 1053 K23)) when a small amount of CaS is added in CaCl2, and to determine the saturation values of CaS concentration, i.e., the concentration of liquidus line. This information assists the basic understanding of electrochemical reaction mechanism of sulfides20–22) in CaCl2 melt.
The apparatus used for this experimental work is illustrated in Fig. 1. High purity of anhydrous CaCl2 powder (Wako-Chemicals Co., Japan) and CaS (99%, Furuuchi Chemical Co., Japan) were used. The samples with various composition of CaCl2–CaS mixture (about 300 g) were filled in a dense and high purity Al2O3 crucible, and set in the reaction vessel. The temperature of salt mixture was measured by inserting a thermocouple sheathed in Al2O3 protection tube. The vessel made of stainless steel SUS316L was well evacuated and heated to 873 K at the rate of 8.33 × 10−2 K s−1. Then it was well dried for 25.2 ks in vacuum. The environment was replaced by 1 atm of highly purified Ar gas and the crucible was heated to a desired temperature. The holding time was counted after the stabilization of temperature. A part of the melted salt was soaked by a quartz tube (6 mm in outside diameter) and rapidly cooled without water, i.e., by blowing air from the outside of tube. The solidified salt was picked up after breaking the quartz tube.
Schematic diagram of solubility measurement of CaS in molten salt.
The terminal parts (about 30 mm) of the sample solidified in the quartz tube were cut off, and the central part with homogeneous color (milky white with light pink) was taken as the sample for analysis. The cross-section of the sample was polished without water, and immediately observed by optical microscope. The sample without coating could not be observed in scanning electron microscope (SEM) because of electrostatic charge. The platinum coating and successive SEM observation damaged the sample due to strong hygroscopic property. The optical microscope was useful to observe the sample quickly.
About 100 mg of the sample was dissolved in ethylenediaminetetraacetic acid (EDTA) solution. The EDTA-glycerin method is popular for fixing sulfuric ions in an aqueous solution to analyze hydrogen sulfide concentration in water.24–26) 2 g NaOH, 25 mL glycerin and 0.2 g EDTA.2Na were diluted in 500 g distilled water purified highly. EDTA is active as masking regent of iron which may act as catalysis of oxidization of S2−.24–26) Basic solution using NaOH was taken to suppress the possible scattering due to H2S gas evolution from the acidic solution.
The sulfur concentration was measured by inductively coupled plasma-atomic emission spectroscopy (ICP-AES) using the standard solutions.26) The emission at 180.7 nm was taken because of the best reproducibility and linearity in the measurable 5 strong peaks. The analytical values of ionic sulfur concentration in aqueous solution, CS2−, was converted to the sulfur concentration in the solidified sample, CCaS using
| \begin{equation} C_{\text{CaS}}=C_{\text{S${^{2-}}$}}\frac{M_{\text{CaS}}}{M_{\text{S${^{2-}}$}}}\frac{V_{\text{solvent}}}{W_{\text{sample}}} \end{equation} | (11) |
Figure 2 shows (a) the appearance of solidified sample and (b) the cross-sectional view of the solidified sample using the optical microscope. Homogeneous white color was observed over the cross sections, as shown in Fig. 2(a), although the top surface of the sample was hygroscopic. A white lath-like structure in the microscopic cross-sectional view was often found in addition to the polishing damages, especially in the samples with high concentration of CaS. The matrix shows the characteristic eutectic structure, as shown in Fig. 2(b).
(a) Photograph and (b) optical microscopic image of sample soaked from CaCl2–2.0 mol%CaS melt after melting at 1173 K for 21.6 ks, where the initial concentration was 2.0 mol%CaS.
Figure 3 shows the time dependency of CaS concentration in the rapidly solidified sample. The analytical values approached to the constant values within 3.6 ks in the samples that 1 and 2 mol%CaS were initially added. These steady values of CaS concentration are close to the initial values, respectively. In contrast, in the series of 3 mol%CaS, the CaS concentration in the bath did not arrive at the initial concentration (3.0 mol%CaS), and saturated at about 2 mol%. In the sampling on CaS analysis, if the undissolved CaS particles were picked up in the soaked samples homogeneously, the measured value should be constant at 3.0 mol%. However, the analyzed value (2.0 mol%) was lower than the initial concentration. This means that the undissolved CaS does not affect the bulk concentration when the initial concentration exceeds a saturation value. The maximum concentration of CaS dissolution could be reproducibly measured at 3.0 mol% when the equilibrium was achieved at 1173 K. Therefore, the solubility of CaS in CaCl2 melt at 1173 K was concluded as 1.77 ± 0.1 mol% by considering a strong weight on the concentration analysis at 3.0 mol%.
Time dependency of CaS concentration in the samples quenched from the molten CaS–CaCl2 at 1173 K. Three initial concentrations, XCaS, were used as the starting compositions.
In order to examine the solubility at 1123 and 1223 K, the concentration of the quenched samples was analyzed only at 3.0 mol%CaS (initial), assuming that the solid CaS coexisted in equilibrium with the saturated melt at this concentration. Figure 4 shows the time dependency of analyzed concentration of the samples quenched from various temperatures using CaCl2–3.0 mol%CaS mixture. The concentrations approached to a certain value at any conditions within a short time.
Time dependency of CaS content in the quenched samples at various temperatures. The initial concentration of CaCl2–3 mol%CaS was melted.
Figure 5 shows the appearance of quenched samples from 1223 K. The inner parts of solidified samples were homogeneously white solid, but turned gray at the longer holding time. The sample quenched after holding for 25.2 ks was analyzed by X-ray diffraction (XRD). CaCl2 and CaS were detected and the dissolution of Al2O3 crucibles and SiO2 tubes was not found, and any phases containing Al and Si were not detected in the solidified salt by XRD. It is noted that the excess amount of CaS did not form the other compounds such as CaSO4, but that it remained as CaS.
Photograph of samples extracted from CaCl2–3.0 mol%CaS melt at 1223 K. Holding time are (a) 1.8 ks, (b) 3.6 ks, (c) 10.8 ks and (d) 25.2 ks.
A small amount of FeO existed mainly at the gray area. The reason of FeO existence may be related with a higher partial pressure of CaCl2 at a higher temperature. CaCl2 melt hardly reacts with water or iron oxides, but its gas can react with the stainless steel vessel. The equilibrium vapor pressure of CaCl2 is evaluated as 0.25 Pa and 0.79 Pa at 1173 K and 1223 K, respectively.23) The higher vapor pressure of CaCl2 may cause the side reactions more significantly at 1223 K. For example, the evaporated CaCl2 may react on the stainless steel wall as,
| \begin{equation} \text{FeO}+\text{CaCl$_{2}$(g)}=\text{FeCl$_{2}$(g)}+\text{CaO} \end{equation} | (12) |
| \begin{equation} \text{Fe$_{2}$O$_{3}$}+\text{3 CaCl$_{2}$(g)}=\text{2 FeCl$_{3}$(g)}+\text{3 CaO} \end{equation} | (13) |
Equilibrium vapor pressure of CaCl2 gas under coexisting with liquid CaCl2. Vapor pressures of iron chlorides gas were evaluated under the same equilibrium, considering that iron oxides react CaCl2 gas and produce FeCl2 or FeCl3.
When the melt contained these impurities, the mass of solid sample for ICP analysis was measured heavier than the true value. This error increases Wsample in eq. (11), and the solubility is evaluated a little smaller than true value. Therefore, the analyzed concentration for 10.8 ks was taken as the saturation value at 1223 K, where the contribution of FeO did not affect the solubility of CaS significantly.
3.3 Temperature dependencyThe solubility of CaS increased slightly as the holding temperature increased. Figure 7 shows the temperature dependency of solubility. The solubility, C, was approximated by Arrhenius type dependency as14,15)
| \begin{equation} C=C_{0}\exp(-E/\mathit{RT}) \end{equation} | (14) |
| \begin{equation} \ln C=0.4670-5337/T \end{equation} | (15) |
Arrhenius-type plot of CaS concentration (C in mol%) with inversed temperature (1/T).
Off course, the solubility of CaS is related with the liquidus line in the binary phase diagram, where the balance between partial free energy of CaS and that of CaCl2 plays a main role. Therefore, eq. (15) shows only an experimental approximation in this system.
Using the solubility of CaS, the liquidus line of CaS and the simple phase diagram were estimated. When all CaS added initially could dissolve in the melt, its region should be in a single phase of liquid in the binary phase diagram. By increasing the CaS content in the bath, however, the saturation was achieved. When the excess amount of CaS precipitates or the undissolved CaS remains, such conditions should be called as two-phase region of solid CaS and liquid.
Figure 8 illustrates the phase diagram using the melting point of CaCl2 and the solubility limits of CaS, in addition to the experimental conditions. Because the eutectic solidification was observed and no other compounds were found, the simple eutectic melting is expected. The eutectic temperature was not studied here, but it should be lower than the melting point of CaCl2. Figure 8 was thus drawn.
Phase diagram of CaCl2–CaS system estimated by CaS solubility in CaCl2. Closed and open circles show the initial conditions and the analyzed liquidus position, respectively.
The liquidus of CaS in CaS–CaCl2 system locates closer to CaCl2 than that of CaO in CaO–CaCl2 system.9–15) Namely, the solubility of CaS is much smaller than that of CaO (nearly 20 mol%CaO at 1173 K9–15)). It is known that a compound CaOxCly exists in CaCl2–CaO binary system,9,11) and that it affects the liquidus of CaO. However, a stable compound in CaCl2–CaS system was not found in this work, and a simple eutectic reaction as shown in Fig. 8 does not contribute to expand CaS solubility to a wider range. It is clear from the physical view that the different solubility is due to the differences of ionic radius of oxygen and sulfur, and of interaction force among oxygen, sulfur and CaCl2. A detailed quantitative study is expected in future.
A relatively large solubility of CaS in CaCl2 melt supports the proposed mechanism in our previous reports that the dissolved CaS decomposes electrochemically to form Ca in CaCl2 melt. The equilibrium solubility of CaS can predict the maximum soluble amount of CaS in the reaction crucible, which is helpful to assess the applicability of OS process of the sulfides.
The solubility of CaS in CaCl2 melt was measured experimentally in order to confirm the solution mechanism of CaS as an important reaction in the cycle of proposed reduction. By soaking the melt into the silica tube and rapid cooling, the solubility was analyzed by ICP-AES method. When 3 mol%CaS was added to CaCl2, a constant solubility of 1.77 ± 0.1 mol%CaS was measured after 3.6 ks holding at 1173 K. Within a short period of 0.9 ks, the soluble CaS arrived to the saturation value. This confirms that the byproduct CaS in the sulfide reduction quickly dissolves in the CaCl2–CaS melt and diffuses to the bulk.
Using temperature dependency of the maximum solubility of CaS and structural analysis of solidified salt, a binary phase diagram of CaCl2–CaS system was sketched as shown in Fig. 8.
This work was financially supported in part by the Grant-in-Aid for Scientific Research (KAKENHI, #17H03434), and Japan Mining Industry Association (JMIA).
