Selection About Molten Salts Used For The Electrochemical Reduction of Porous TiO₂ pellets

A commercial TiO₂ powder (98%, 10022728, Sinopharm) was used as the starting material. Before preforming, about 10 wt% polyvinyl alcohol (PVA) solution (the PVA concentration is 25 wt%) was added into the powder to improve its formability. The damp powder was weighed and pressed into pellets under the axial pressure of 60 MPa. Then these pellets were sintered at 1050 °C for 4 h. The sintered pellets were sanded with sandpaper. Properties of the pellets prepared were: mass 2.5 g, diameter 28.4 mm, thickness 2.0 mm, and porosity about 45% which was measured by Mercury Porosimeter. The chloride salts used, including anhydrous LiCl (99%, L116328, aladdin), anhydrous NaCl (99.5%, 10019318, Sinopharm), anhydrous MgCl₂ (99.9%, M140955, aladdin), anhydrous CaCl₂ (96%, 10005860, Sinopharm) and SrCl₂·6H₂O (99.5%, S817918, Macklin), were firstly dried at 200 °C for 10 h in flowing argon (2 L min⁻¹) and then at 300 °C for 4 h under vacuum before use. Electro-deoxidation experiments were carried out in a pressure-tight vertical tubular stainless steel reactor which was heated by silicon carbide rods. The upper end of the reactor was closed by a stainless steel cover, which was equipped with feedthroughs for the electrode leads and a thermocouple as well as gas inlet and outlet. A sintered pellet attached to a molybdenum wire (2 mm diameter, 700 mm long) was used as the cathode; a high-purity graphite rod (10 mm diameter, 80 mm long) was used as the anode which was connected with a stainless steel collector; and an alumina crucible (80 mm inner diameter and 120 mm high) severed as the container for the electrolyte. Before heated, the reactor was evacuated to 10 Pa and then was replenished with high-purity Ar gas (99.999%). This process was repeated twice. Then Ar gas passed continuously through the reactor vessel at 2 L/min until the end of each experiment. Because of the large differences in the melting points and the saturated vapor pressures of the chloride molten salts used, the electrolysis temperatures were different: 900 °C in NaCl and CaCl₂ molten salts, 920 °C in SrCl₂ molten salt and 830 °C in LiCl and MgCl₂ molten salts. Similarly, because of the difference in the decomposition voltages of the chloride molten salts used, the applied voltage was also different but equal to or slightly lower than the decomposition voltage of the chloride molten salts used.

In order to investigate the actual reduction pathway of TiO₂ in the molten salts, partially reduced samples were produced through interrupting the electrolysis after different reduction times ranging from 5 min to 48 h. After each experiment, the sample was moved out from molten salt and cooled down rapidly at argon atmosphere. The recovered samples were rinsed with flowing tap water to remove the solidified molten salt clinging to the surface of samples and then were placed in semi-concentrated acetic acid. After 24 h, acetic acid solution was replaced with distilled water. After another 24 h, the samples were removed out and dried in a vacuum oven at room temperature.

The phase compositions of the partially reduced samples were analysed by X-ray diffraction analyser (X'Pert Pro MPDDY2094, PANalytical B.V with CuKα radiation). A light optical microscopy (OLS4100, Olympus) was used to investigate the polished sections. Microstructure and chemical composition of the samples were investigated by means of scanning electron microscopy and energy-dispersive X-ray analysis (S-4800, Hitachi). The content of oxygen in final product was tested by oxygen/nitrogen analyser (TC500, LECO). The contents of alkali or alkaline earth metal oxides in the used salts were estimated by the acid-base titration method.

Thermodynamic analysis

Thermodynamic analysis is very helpful in estimating reaction pathways and conditions in some systems. Usually electro-deoxidation of metal oxides takes place in an oxygen-containing chloride molten salt system. The factors influencing the elecro-deoxidation of TiO₂ involve the applied potential on cathode, temperature, O²⁻ ions activity and properties of the used molten salt. In order to represent the systems well, the thermodynamic data was converted to diagrams of the equilibrium potential E (relative to the standard chlorine electrode) against the activity of oxide ion (expressed as its logarithm, pO²⁻) in the chloride molten salts, which has obvious advantages in predicting and understanding the electrochemical reduction process of TiO₂ in different chloride molten salts.¹⁵ The equilibrium potentials (E) of the electrode reactions are calculated from the free energies as Eq. 1 (Nernst equation) and the potential zero is taken as the potential of a chloride electrode (1.013 × 10⁵ Pa) in contact with a chloride molten salts at unit chloride activity. Pure component was used as the standard state of solid and liquid phases; and in case of gas phase, the gas of 1.013 × 10⁵ Pa is the standard stare (P^θ). For O²⁻ ions, the pure oxide of the cation of the chloride molten salt is chosen as unit activity of oxide ions. According to the above, the thermodynamic diagrams of E-pO²⁻ in the systems of Ti-O-Cl-M (M=Li, Na, Mg, Ca, Sr and Ba) were plotted, as shown in Fig. 1.

$$\begin{eqnarray}&&E=-{\rm{\Delta }}G/nF\end{eqnarray} \tag{ 1 }$$

Zoom In Zoom Out Reset image size

From the E-pO²⁻ diagrams, the deposition potentials of Li, Na, Mg, Ca, Sr and Ba metals of unit activity are −3.36 V, −3.17 V, −2.46 V, −3.23 V, −3.39 V and −3.49 V (vs Cl₂/Cl⁻) respectively in corresponding alkali or alkaline earth metal chloride molten salts. At a lower O²⁻ ion activity, TiO₂ always can be reduced into Ti theoretically in all of the studied molten salts, which proceeds in a sequence from high-to-low order titanium oxides with the negative shift of the potential. However, so low O²⁻ ion activity could not last for a long time. During the electro-deoxidation of titanium oxides, O²⁻ ions will be released ceaselessly, which inevitably lead to the increase of the O²⁻ ion activity around the titanium oxides particles. At a higher O²⁻ ion activity, titanium oxides, including TiO₂, Ti₄O₇, Ti₃O₅, Ti₂O₃ and even TiO, could react with alkali and alkaline earth metal ions (Li⁺, Na⁺, Mg²⁺, Ca²⁺, Sr²⁺ and Ba²⁺) and O²⁻ ions to form the corresponding titanates before the formation potential of Ti metal except the cases of Ti₂O₃ and TiO in Ti-O-Cl-Mg system and TiO in Ti-O-Cl-Ca system. From Fig. 1c, when the O²⁻ ion activity reaches unit activity in MgCl₂ molten salt, Ti₂O₃ is stable at the potential of around −1.73 V ∼ −1.63 V and TiO is stable at around −2.22 V ∼ −1.73 V. In CaCl₂ molten salt saturated with CaO, TiO is stable at the potential of around −2.81 V ∼ −2.59 V, as shown in Fig. 1d. With the decrease of the O²⁻ ion activities, the above-mentioned potential ranges will shift negatively.

From the above, the electrochemical reduction of titanium oxides is inevitably accompanied by the formation of titanates. Titanates are stable at higher O²⁻ ion activity and the reduction usually requires a more negative potential. During the reduction of titanates, amounts of O²⁻ ions would be released, which could not diffuse promptly out of the pellets but deposit inside the pellets by the form of alkali or alkaline earth metal oxides. Thus the O²⁻ ion activity inside the pellets would be very high during electro-deoxidation, even reaching unit activity. From the E-pO²⁻ diagrams, at unit O²⁻ ion activity, Li₂TiO₃, SrTiO₃ and BaTiO₃ can be directly reduced into Ti metal in corresponding chlorides molten salts and the equilibrium potentials are −3.28 V Cl₂/Cl⁻ (corresponding to +0.08 V vs Li⁺/Li), −3.17 V Cl₂/Cl⁻ (corresponding to +0.22 V vs Sr²⁺/Sr) and −3.47 V vs Cl₂/Cl⁻ (corresponding to +0.02 V vs Ba²⁺/Ba) respectively; MgTiO₃ is firstly reduced into Ti₂O₃ at −1.63 V vs Cl₂/Cl⁻ (corresponding to +0.83 V vs Mg²⁺/Mg), and then the Ti₂O₃ is reduced into TiO and Ti metal successively at −1.73 V vs Cl₂/Cl⁻ (corresponding to +0.73 V vs Mg²⁺/Mg) and −2.22 V vs Cl₂/Cl⁻ (corresponding to +0.24 V vs Mg²⁺/Mg); CaTiO₃ is firstly reduced into TiO at −2.59 V vs Cl₂/Cl⁻ (corresponding to +0.94 V vs Ca²⁺/Ca), and then the TiO is reduced into Ti metal at −2.81 V Cl₂/Cl⁻ (corresponding to +0.42 V vs Ca²⁺/Ca); with regard to Na₂TiO₃, it cannot be reduced even when the electrode potential reaches the deposition potential of Na in NaCl molten salt. Obviously at unit O²⁻ ion activity, the formation potentials of Ti metal in the chloride molten salts are very close to the deposition potentials of the corresponding alkali or alkaline earth metal, especially in LiCl and BaCl₂ molten salts where the formation potentials of Ti metal almost reach the deposition potentials of Li metal and Ba metal.

As is known to all, Ti metal has a large solvent capacity and a strong affinity for oxygen, and the solubility of oxygen in Ti metal is up to 13.7 wt% at 900 °C.¹⁶ Thus the Ti metal obtained at the above-mentioned equilibrium potentials of necessity contains amounts of oxygen. Usually the deoxidation of Ti-O solid solution is one of the main control step during the electro-deoxidation. The deoxidation limit is impacted by the applied potential, temperature, O²⁻ ions activity and properties of the used molten salt. In order to estimate the deoxidation limits in different molten salts, the thermodynamic properties of Ti-O solid solution were required. Toru H. Okabe et al. measured the deoxidation limit of Ti-O solid solution at different temperatures by Ca metal, and gave the thermodynamic properties of Ti-O solid solution in Beta phase and the Gibbs free energy of reaction 2, as shown by Eq. 3.^17,18

$$\begin{eqnarray}&&1/2\,{{\rm{O}}}^{2}\left({\rm{g}}\right)={\rm{O}}\left[{\rm{Ti}}\right]\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{\rm{\Delta }}G=-583000+88.5T\left[1173-1373\,{\rm{K}}\right]\end{eqnarray} \tag{ 3 }$$

According to their experiment conditions, Eq. 3 is only valid for dilute Ti-O solid solution which contains oxygen below 0.2 wt%. Assuming that the applicable scope of Eq. 3 was expanded to Ti-O solid solution containing 1.0 wt% oxygen, the deoxidation limits of Ti-O solid solution and the corresponding conditions in different systems could be represented as the red lines in the E-pO²⁻ diagrams. From these, for obtaining low-oxygen Ti, a low O²⁻ ion activity and a very negative potential even reaching the deposition potentials of alkali or alkaline earth metals would be required. For example, in the system of Ti-O-Cl-Li at 900 °C, only when the O²⁻ ion activity around the cathode is lower than $1.8\times {10}^{-2},$ the titanium metal containing as low as 1 wt% oxygen can be obtained before the deposition potential of Li metal; and for obtaining the Ti metal containing as low as 0.2% and 0.01% oxygen, it is required that the O²⁻ ion activities are lower than $4.2\times {10}^{-3}$ and $2.1\times {10}^{-4}$ respectively. In other systems, the calculation results are similar with these in Ti-O-Cl-Li system, as shown in Table I. Here it should be noticed that for obtaining low-oxygen Ti (the oxygen content is below 1.0 wt%) before the deposition potentials of alkali or alkaline earth metals of unit activity, the O²⁻ ion activities in corresponding systems must be lower than unit activity, except in Ti-O-Cl-Ca system. In Ti-O-Cl-Ca system, Ti metal containing 1.0% and 0.2% oxygen could be obtained at −3.06 V vs Cl₂/Cl⁻ (corresponding to +0.17 V vs Ca²⁺/Ca) and 3.14 V vs Cl₂/Cl⁻ (corresponding to +0.09 V vs Ca²⁺/Ca) at unit O²⁻ ion activity; while for obtaining the Ti metal containing 0.01% oxygen before the deposition of Ca metal, the O²⁻ ion activity must be below 0.26.

Table I.  Oxygen contents of the reduced samples in different molten salts and the corresponding experimental conditions, and the thermodynamic conditions for obtaining low-oxygen and ultra-low-oxygen Ti.

						Thermodynamic conditions for Ti
						1 wt% oxygen		0.2 wt% oxygen
Salt	T	t	P	CMO	CO	Potential	Activity	Potential	Activity
LiCl	830	48	3.35	0.9	6.8	−3.36	1.8 × 10−2	−3.36	4.2 × 10−3
NaCl	900	10	3.2	—	—	−3.17	1.0 × 10−10	−3.17	1.9 × 10−11
MgCl2	830	48	2.45	1.4	4.6	−2.46	0.51	−2.46	0.11
CaCl2	900	48	3.15	1.1	0.3	−3.06	1	−3.14	1
SrCl2	920	48	3.4	0.6	5.4	−3.39	0.26	−3.39	5.4 × 10−2
BaCl2	—	—	—	—	—	−3.49	1.7 × 10−2	−3.49	3.0 × 10−3

Salt—Molten salt used; T—Electrolysis temperature, °C; t—Electrolysis time, h; P—the applied voltage, V; C_MO—Concentration of corresponding alkali or alkaline earth metal oxides in the used molten, mol%; C_O—Oxygen concentration in the reduced sample, wt%; Potential—Equilibrium potential required for obtaining the Ti metal, vs Cl₂/Cl⁻; Activity—the value of the highest O²⁻ ion activity for obtaining the Ti metal.

From the above, the electro-deoxidation of titanium oxides and titanates is greatly influenced by O²⁻ ion activity in systems. Here the pure oxide of the cation of the chloride molten salt is chosen as unit activity of oxide ion, which means that the chloride molten salt saturated with the corresponding alkali or alkaline earth metal oxide is the standard state for O²⁻ ions. Table II lists the solubility data of some alkali and alkaline earth metal oxides in corresponding chloride molten salts.^19–22 Li₂O, CaO, SrO and BaO have great solubility respectively in the corresponding chloride molten salts; the solubility of MgO in molten MgCl₂ is very little; as for the solubility of Na₂O in molten NaCl, no relevant data was found. According to the definition of the standard state for O²⁻ ions, the activity coefficients of O²⁻ ions in the chloride molten salts saturated with the corresponding alkali or alkaline earth metal oxides were calculated, as shown in Table II. Assuming that the activity coefficients of O²⁻ ions in the chloride molten salts were independent on the O²⁻ ions concentration at a certain temperature, the O²⁻ ions activity in unsaturated chloride molten salts at the temperature could be estimated. However, activity coefficient is usually a function of the composition and temperature. For getting the accurate activity coefficient data, there is still a tremendous amount of work to be done.

Table II.  Solubility data and calculated activity coefficients of alkali and alkaline earth metal oxides in corresponding chloride molten salts.

Oxides	Li2Oat 650 °C	Na2O	MgOat 900 °C	CaOat 900 °C	SrOat 920 °C	BaOat 900 °C
Solutility/mol %	11.5	—	1.4	19.5	10.5	12.5
Activity coefficient	8.67		71.4	5.13	9.53	8

From the thermodynamic analysis, TiO₂ can be reduced into Ti metal in any of the studied molten salts above as long as the reduction conditions were satisfied. However during the practical electro-deoxidation experiments, some conditions are uncontrollable or hardly achieved. For example, in Ti-O-Cl-Na system, Ti metal can be obtained only when the O²⁻ ion activity is below 10⁻⁸, which is almost impossible to be achieved and also uncontrollable during the electro-deoxidation. In addition, considering the kinetics factors and some unexpected mesophase, the actual reduction pathways and deoxidation limits of porous TiO₂ pellets might differ greatly with the thermodynamic analysis. Thus electro-deoxidation experiments of porous TiO₂ pellets were performed in the molten salts to determine the actual reduction pathways and the possible deoxidation limits of the porous TiO₂ pellets. The applied voltage was equal to the decomposition voltage of the chloride molten salt used. Because of the strong toxicity of BaCl₂, we have no condition to make the electrolysis in it. So the electro-deoxidation of TiO₂ in molten BaCl₂ was not discussed here.

Figure 2 shows the I-t (current vs time) and E-t (voltage vs time) curves during the electro-deoxidation of TiO₂ pellets in the molten salts. In order to avoid a too large initial current, a constant-current electrolysis was conducted at the initial stage of electrolysis. With the electrolysis time went on, the corresponding voltage increased progressively. Until the voltage reached the default value, the constant-voltage electrolysis began. From Fig. 2, the durations of the constant-current electrolysis were usually several minutes. Then the current began to decrease. After about 2 h of electrolysis, the current kept a low level until the electrolysis terminated. The oxygen content of some reduced samples in different molten salts and the corresponding experimental conditions were listed in Table I.

Zoom In Zoom Out Reset image size

The detail discussion about the actual reduction pathways of porous TiO₂ pellets in the molten salts was placed in the Supplementary Materials. Figures. S1–S8 (available online at stacks.iop.org/JES/167/082501/mmedia) show the XRD patterns and the optical micrographs of the polished sections of the samples the electrolysis treatment of which was terminated after different durations in the molten salts. From that, the actual reduction pathways of porous TiO₂ pellets in different molten salts were determined and were listed in Table III. They are roughly similar and can be divided into three stages as shown in Fig. 3. In the first stage, alkali or alkaline earth metal ion (Mⁿ⁺, n = 1 or 2) was incorporated into porous TiO₂ pellets to form titanates and/or titanium sub-oxides (Ti_nO_2n−1, n = 1, 2, 3, 4 etc.) under the driving of the voltage. During this stage, no or very little oxygen was removed. In the second stage, the titanates and/or titanium sub-oxides were reduced into Ti-O solid solution. Meanwhile, amounts of O²⁻ ions would be released intensively. In third step, Ti-O solid solution was further deoxidized into low-oxygen Ti (the oxygen content is lower than 1%). Still, some differences existed among the reduction pathways of porous TiO₂ pellets in the molten salts, which mainly reflected in two aspects. The first is the reduction of the titanates: in LiCl, MgCl₂ and CaCl₂ molten salts, the corresponding titanates were firstly reduced into titanium sub-oxides (mainly TiO and Ti₂O₃) and then they were further reduced into Ti-O solid solution; while in SrCl₂ molten salt, the SrTiO₃ was directly reduced into Ti-O solid solution, and in NaCl molten salt, the sodium titanate could not further be reduced. The second is the possible deoxidation limits of TiO₂ pellets in the molten salts. Figure 4 shows the changes of the residual O content in the samples obtained after various length of time of polarization in different molten salts. After 48 h of polarization, the changes of the residual O contents of the reduced samples were very small, which was considered to reach quasi-deoxidation limits. From the figure, only in CaCl₂ molten salt, low-oxygen Ti can be prepared; while in other molten salts, just high-oxygen Ti (the oxygen content is more than 1 wt%) or Ti_xO (x = 2, 3 or 6) was obtained. Figure 5 shows the SEM images of the samples obtained after 10 h of polarization in NaCl and after 48 h of polarization in LiCl, MgCl₂, CaCl₂ and SrCl₂. Ti metal or Ti_xO (x = 2, 3 or 6) obtained in molten LiCl, MgCl₂, CaCl₂ and SrCl₂ showed varying degrees of sintering. The sample obtained in CaCl₂ molten salt was the most compact.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

From the above, low-oxygen Ti was got only in molten CaCl₂. While in LiCl, MgCl₂ and SrCl₂ molten salts, only high-oxygen Ti (the oxygen content is about 5 wt%) or Ti_xO (x = 2, 3 or 6) were prepared, the oxygen contents in which were about 4–10 wt%. From the E-pO²⁻ diagram, the main factors influencing the deoxidation limit of Ti metal are the applied potential and O²⁻ ion activity. Obviously the applied voltages in all experiments satisfied the requirement for preparing low-oxygen Ti. The alkali metal or alkaline earth oxides concentrations in the used electrolytes after 48 h of electrolysis were tested by acid–base titration. The results were listed in Table I. Here the MgCl₂ molten salt was saturated with MgO. Thus the O²⁻ ion activity in MgCl₂ molten salt was determined to be 1. But those in LiCl, CaCl₂ and SrCl₂ molten salts could not be determined because of the unknown activity coefficients. From the lower concentration of oxides in LiCl, CaCl₂ and SrCl₂ molten salts, it was surmised that the O²⁻ ion activities might be far less than unit avtivity. Still, the product obtained in MgCl₂ molten salt has a lower oxygen content than those in LiCl and SrCl₂ molten salts. Thus O²⁻ ion activity might be not the main factor influencing the deoxidation limit of Ti-O solid solution. The most predominant and fundamental factor might be the reducing ability of alkali or alkaline earth metals to titanium oxides, which might be the main reason lead to the great differences of electro-deoxidation conditions and limits for preparing Ti metal in different chloride molten salts.

For better comparison, the reducing ability needs to be quantised. The reducing ability behaves as follows: the activity of alkali or alkaline earth metals in corresponding chloride molten salt impacts on the deoxidation limit of Ti metal. Thus the activity of alkali or alkaline earth metals in corresponding chloride molten salt, balancing with a certain Ti-O solid solution (the thermodynamic property has been established), can be used to quantise the reducing ability. According to Nernst equation, the logarithm of the activity of alkali or alkaline earth metals (loga_m) in corresponding chloride molten salts is inversely proportional to the potential difference between the electrode potential and the standard electrode potential of reaction 4, as Eq. 5.

$$\begin{eqnarray}&&{{\rm{M}}}^{{\rm{n}}+}+{\rm{ne}}\to {\rm{M}}\end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&\mathrm{log}\,{a}_{M}=-\displaystyle \frac{nF\left(E-{E}^{\theta }\right)}{2.303RT}\end{eqnarray} \tag{ 5 }$$

Where a_M is the activity of alkali or alkaline earth metals, n is electron transfer number, F is Faraday constant, E^θ is the standard electrode potential of reaction 4, E is the electrode potential, R is gas constant and T is the temperature.

Assuming the thermodynamic properties of the Ti-O solid solution obtained at the equilibrium potential of Ti/TiO or titanates in all the chloride molten salts are identical, as shown in the E-pO²⁻ diagrams, the reducing ability of alkali or alkaline earth metals can be expressed by the potential difference between the formation potential (E) of Ti metal at unit O²⁻ ion activity and the deposition potential (E^θ) of alkali or alkaline earth metals of unit activity. In addition, considering the influence of electron transfer number n, the reducing ability of alkali or alkaline earth metals is expressed by n(E−E^θ). Here it should be noticed that Na metal cannot reduce titanium oxides into Ti at unit O²⁻ ion activity in theory. Thus its reducing ability was expressed to be zero. Figure 6 shows the reducing ability of Li, Na, Mg, Ca, Sr and Ba metals. From that, the reduced ability of Ca metal is strongest, followed by Mg, Sr, Li and Ba. Meanwhile, experiments indicated that in molten CaCl₂, the deoxidation limit of TiO₂ pellets was lowest, followed by in molten MgCl₂, SrCl₂ and LiCl, which agreed well with the sequence of the reduced ability of the corresponding alkali or alkaline earth metals. Thus it can be concluded that the reducing ability of alkali or alkaline earth metals impacts greatly on the electro-deoxidation of TiO₂ pellets in corresponding chloride molten salts. From the above, it can be surmised low-oxygen Ti was hardly obtained in BaCl₂ molten salts.

Zoom In Zoom Out Reset image size

From the above, E-pO²⁻ diagrams are helpful in estimating and understanding the electro-deoxidation pathways and conditions of TiO₂ pellets in different chloride molten salt though there are some differences between the experimental results and the thermodynamic analysis. Low-oxygen Ti was got only in molten CaCl₂ after about 48 h of electrolysis. In LiCl, MgCl₂ and SrCl₂ molten salt, only Ti₆O or high-oxygen Ti was obtained after 48 h of electrolysis, the oxygen content of which was about 4–7 wt%. While in molten NaCl, no Ti metal or Ti–O solid solution was obtained. The main factors influencing the electro-deoxidation of TiO₂ pellets include the applied potential, temperature, O²⁻ ion activity, polatization time etc But the most predominant and fundamental factor is the reducing ability of the metallic state of cation of the chloride molten salt to titanium oxides, which decides the electro-deoxidation limit of TiO₂ in the chloride molten salts.

The authors sincerely acknowledge financial support from the National Natural Science Foundation of China (51674076) and the Fundamental Research Funds for the Central Universities (N162502002).