Accelerated Electro-Reduction of TiO2 to Metallic Ti in a CaCl2 Bath Using an Inert Intermetallic Anode

In the FFC-Cambridge process, the cathodic dissociation of oxide and CO/CO2 production on carbon anode is the basis for metal production in a CaCl2 bath. Using an inert intermetallic anode, the CO2 evolution can be eliminated altogether with acceleration in the electro-reduction kinetics. In the presence of a carbon anode, the cathodic dissociation of TiO2 suffers from slow reduction kinetics of TiO2 to Ti metal, which can be enhanced significantly by the incorporation of alkali species in the TiO2 pellet at the cathode and in the CaCl2 bath in the presence of an intermetallic inert anode. With inert anode and incorporation of K+-ion in the TiO2 matrix and in the salt bath, nearly full metallization with greater than 99% of Ti metal containing 1500 ppm of oxygen was possible to achieve in less than 16 h of electro-reduction. The microstructural and chemical analysis of the metallic phase and its morphology revealed the presence of a layer of titanium metal that forms in the fast reduction reaction step in less than 5 h, after which the reaction rate slows down significantly before terminating in 16 h. The investigation showed that two different types of microstructures of metallic titanium were evident—a thin sheet-like material on the outer periphery of the reduced pellet and the dendritic core which was found to be under the peripheral sheet of the metallic layer. The mechanism of morphological and microstructural changes in the reduced form of titanium metal is explained.

In the FFC process, the cathodic dissociation of TiO 2 in a CaCl 2 bath with carbon anode leads to the evolution of CO 2 , which is expected, as explained in the FFC-Cambridge process 4 . However, the cyclic voltammetric 5,6 and calcium thermic 5 have enabled the characterization of slow reaction kinetics due to the formation of interfacial CaTiO 3 phase as a chemical barrier for the metallization at cathode [4][5][6][7][8] . The complementary interfacial reactions are shown below arising from the dissociation of TiO 2 and due to the residual solubility of O 2− ions in the CaCl 2 liquid.
Note that the reactions (1a)-(1f) are interconnected for ionic transport and electronic charge balance in the overall electro-reduction of TiO 2 . The O 2− ions, when discharge at the carbon cathode, produce CO 2 via reaction (1f), and the electrons liberated reduce the Ti 4+ to Ti metal, as shown in reaction (1d). In this investigation the use of inert metallic anode is discussed below which eliminates the CO 2 evolution at anode and yields pure oxygen via reaction (1g). The production of commercial grade titanium metal powder and pure oxygen is, therefore, far more attractive than what has been demonstrated in the literature [4][5][6][7][8][9] using carbon anode.
The main motivation for this article is to demonstrate a laboratory-scale proof-of-principle of the production of metallic titanium and commercial purity oxygen generation for a range of applications, including further purification of evolved oxygen for hospital use. In this article, we have also analysed the microscopic reason for the formation of calcium titanate (CaTiO 3 ) perovskite structure, in which the O 2− ion lattice diffusion is known to be slow (1.323 × 10 -6 -4.033 × 10 -5 cm 2 s −1 ) at 900 °C 9 . By comparison, the value of the atomic diffusion of oxygen in β-titanium alloy is of the order of (~ 10 −6 cm 2 s −1 ) 10 which forms as an intermediate phase before the metallization with ultra-low oxygen in Ti metal is achieved at the cathode. The report on the formation of alkaline earth perovskite (CaTiO 3 ) is significant because the presence of alkaline earth perovskites on the surface of partially reduced pellets acts as an insulating barrier for the O 2− charge transport into the CaCl 2 bath, which is known to have a limited solubility of O 2− ions 6-8 at the partially reduced oxide and perovskite interface. Also, the perovskite phase is an insulator which means during the cathodic dissociation (reaction 1d), the electronic transport is impeded and leads to slowing down of the overall metal reduction process. In previous investigations 7 , the porosity of the pellets of TiO 2 was increased by incorporating fugitive agent polyethylene, which were burnt before commencing the reaction, so that a large surface area were possible to maintain for a uniform rapid electroreduction, when compared with the pellets of TiO 2 without polyethylene. The investigation 7 concluded that when the potential at 3.0 V (below the decomposition potential of CaCl 2 ) was applied the minimum level of oxygen achieved was approximately 3000 ppm after 48 h using a carbon anode. Furthermore, when the potential was raised from 3 to 3.15 V the time to achieve the residual oxygen concentrations at 3000 ppm in the reduced pellet was halved from 48 to 24 h. The experimental evidence, shown in the literature, confirm although the speed of reaction increases, a complete or near complete metallization is not possible by enhancing the insitu porosity alone, as the insulating perovskite was thermodynamically more stable under the cathodic dissociation condition, due to the difference in the Gibbs energies for the formation of (− 821 kJ per mole of CaTiO 3 versus − 733 kJ per mole of TiO 2 ). Table 1 below summarizes the Gibbs energy ((ΔG o , kJ mol −1 ) change 11 for the molecular form of the chemical reactions which may be relevant in explaining the perovskite stability in the absence and presence of K + -ions in the salt bath present at the interface between the pressed pellets of TiO 2 and CaCl 2 -KCl. In this article, the cathodic dissociation of TiO 2 in the presence of K + -ion in the salt mixture and in the pellet has been characterized. At the end of this article, we also compare and give a critical summary of other electro-reduction techniques used in the lab for reactive metal production.
The role of alkali (Na, K) ions on the chemical breakdown of crystalline structures of ferruginous ilmenite (FeTiO 3 ), tantalite, and chromite minerals above 700 °C in oxidizing and reducing conditions have been extensively studied recently [12][13][14][15][16] in the context of the selective separation of metal oxides as water-soluble and insoluble products. In these investigations, the evidence for the formation of alkali-rich titanate liquid and lattice strain-induced fracture in situ of crystalline alkali complexes (e.g. titanate and chromite) of above 750 °C have also been explained and reported [15][16][17] . Both the evidence for latticeinduced cracking and formation of alkali-titanate based liquid 18 during chemical reactions of ferruginous mineral concentrate with alkali are relevant in the context of cathodic dissociation of TiO 2 and promoting decomposition of CaTiO 3 in situ. The thermodynamic basis for selecting the incorporation of LiCl in the CaCl 2 salt bath and K + -ions in the TiO 2 pellet at the outset of electroreduction process is demonstrated via the chemical reactions 2c--2d in Table 1, from which it is evident that at high temperatures (e.g. 900 °C), both the lime and calcium perovskite are unstable with respect to the thermodynamic stabilities of CaCl 2 and K 2 TiO 3 in the molten salt mixture, respectively. Also, note that in Table in reaction 2e that Li 2 O + KCl is more stable than LiCl + K 2 O.The K 2 O-TiO 2 phase diagram also confirms the evidence for the presence of alkalititanate liquid above 768 °C in the K 2 TiO 3 phase composition range 19 Based on the literature on alkali-titanate complexes [12][13][14][15][16][17] , the molecular representation of the analysis of thermodynamic equilibrium reactions in Table 1, which correspond to the cathodic and anodic reactions (1a-1f), may help in describing the overall electro-reduction of TiO 2 , from which the resulting microstructure and composition of titanium metal may be explained. Experimental evidences from previous studies on carbon anode show that the current decreases gradually, and finally stabilizes at a low value between 0.2 and 0.4 A, which has been attributed due to the presence of an insulating perovskite barrier. On the other hand the thermodynamic equilibrium conditions for reactions, shown in Table 1, show that the when alkali ions (Li + , K + ) are present, these ions enhance the electro-reduction the TiO 2 pellets at cathode both with and without the inert anodes. In this investigation the effect of mixed K + /Li + ions in breaking down the perovskite phase, which may be decomposed under the favourable thermodynamic alkali-cation exchange condition. The data for potassium ion exchange reactions are also presented for comparison with the lithium ions only in reactions 2c-2e in Table 1. We also show for the first time, a comparative study on the use of an intermetallic anode with carbon anode for Ti-metal production using the FFC-Cambridge process. Nearly full metallization was achieved with the use of inert anode and the reaction mechanism is explained with the help of phase stability analysis based on thermodynamic equilibrium and microstructural evidence.

Experimental
The electrolysis experiments were performed in a molten salt mixture of CaCl 2 -LiCl (180gms CaCl 2 and 20gms LiCl) inside an alumina crucible. The chloride salts were initially dried for 24 h at 320 °C by removing moisture, followed by slowly heating it to 900 °C at a rate of 1 °C min −1 in a flowing atmosphere of argon, maintained at a rate of 500 ml min −1 . We used two different types of anodes: a carbon anode for comparison, followed by replacement of carbon anode with an intermetallic Al-Ti-Cu anode 20 for enhanced current carrying capacity. The oxide pellet (1 g) at the cathode was prepared by mixing 99.95% pure TiO 2 with analytical grade (99.9% pure) KHCO 3 in a molar ratio of 1:0.5 so that at the electroreduction temperature of 900 °C, so that the reaction between TiO 2 and the decomposed KHCO 3 , which yields K 2 TiO 3 and K 4 TiO 4 which then forms a liquid phase, as shown in the isothermal ternary section 19 . Both the cathode and the anode were fixed on to a steel holder and fastened using a molybdenum wire, ensuring a good electrical contact. Once the mixture of 90 wt%CaCl 2 -10 wt%LiCl was molten, the electrodes were lowered down into the salt bath. The cell was continuously maintained with a flow of argon gas atmosphere. A constant voltage of 3.1 V was applied through the electrodes and the current was continuously monitored using a computer-controlled software. After 15 h of electrolysis, the electrodes were extracted from the salt bath and the reduced pellet at the cathode was thoroughly washed using water. The reduced pellets were analysed for the presence of phases and the resulting microstructural changes were examined using a Cambridge Camscan 4 FEG SEM, which was operated at 20 kV. The SEM was equipped with an energy dispersive X-ray spectrometer (EDS Link Oxford). Wherever required, the phase identification was also carried out using the Philips X-ray diffractometer using CuK α radiation diffraction, which was necessary for determining the reaction mechanism. During electro-reduction of TiO 2 , the partially reduced pellets were also removed after different time intervals between 0.5 h and 4 h of reaction at 900 °C for the verification of phases formed during complete metallization. The results of X-ray powder diffraction, SEM, and EDX analyses are discussed for explaining the reaction mechanism of electro-reduction of TiO 2 to metallic titanium.

Resultsand Discussion
The current-time plots for the electro-reduction of TiO 2 with carbon and inert anodes are compared in Fig. 1a and b, respectively. It is evident that in case of each anode, there is a fast transient before 5 h after which the rate of either rise or fall in current slows down significantly. Note that the rate of change in current with time in the electro-reduction experiments with inert anode differs significantly from that observed for carbon anodes, when both the anodes were used for the electro-reduction of TiO 2 :K 2 O = 1:0.5 mixed pellet. Since the cell potential is fixed at 3.1 V, the trends in the rise and fall in current with time represent the decreasing and increasing cell resistance, respectively. The reason for the decreasing current in carbon anode, as reported previously, is due to the formation of the dominant perovskite (CaTiO 3 ) phase [6][7][8] . By comparison, the rising current in case of inert anode is due to increased metallization via the electro-reduction of TiO 2 which increases the overall conductivity with time. It was also observed that when the TiO 2 was not mixed with K + -ions in the TiO 2 pellet, the formation of perovskite was observed with both types of anodes, leading to almost cessation of the overall electro-reduction. For analysing the mechanism of the electroreduction reaction and its distinct characteristics of rising current with time, the phase changes with time were characterised using the X-ray powder diffraction technique. The partially reduced samples were extracted at 0.5, 1, and 4 h of reduction and washed and dried before X-ray diffraction. In Fig. 2, the diffraction pattern of the partially reduced pellets at 0.5, 1, and 4 h of time intervals are compared, from which the diminishing diffraction intensities of oxides of titanium (Ti 3 O 5 , TiO), perovskite, and CaTi 2 O 4 phases confirm that in the presence of K + -ion in the pellet the reducibility of TiO 2 increases significantly. Although the Ti-metal starts forming from a time interval of 0.5 h during the electroreduction, the phase intensities of perovskite and CaTi 2 O 4 diminish rapidly in 4 h of reduction. The metallisation of Ti metal in the CaCl 2 -LiCl bath follows the thermodynamic equilibrium condition which we have verified using the predominance area diagram in the Ti-Ca-O-Cl diagram in Fig. 3. No potassium complexes or salts were observed in the diffraction data, which may be explained because of high solubility of alkali and CaCl 2 in water.
The phases identified during the course of electro-reduction were found to be in good agreement with the predicted equilibrium phase composition, as shown in Fig. 3. It is evident from the equilibrium analysis that the electro-reduction bath after 1 h of reaction must be reaching oxygen potential below log 10    Note that during electrolysis, the potassium bicarbonate (KHCO 3 ) mixed with TiO 2 pellet decomposes to K 2 O via 2KHCO 3 = K 2 O + H 2 O + 2CO 2 . The decomposition reaction occurs rapidly and combines with the TiO 2 by forming potassium titanate complexes, as shown in Eqs. (3a) and (3b). Equation (3c) shows the anion-cation exchange reactions. The Gibbs energy change for these reactions are large and negative at 900 °C.
The potassium titanate, formed in situ, melts at 900 °C and breaks down the perovskite barrier and allows a greater surface area for electro-reduction. This means that the liquid rich in K 4 TiO 4 becomes the main source of the O 2− anion transport which is mediated by the presence of reciprocal salt mixture, shown in reaction 3c. This is possible because under the applied potential of 3.1 V, the K 4 TiO 4 liquid continues to dissociate and generate K 2 O which then follows equilibrium conditions in Eqs. (3a) and (3b). Under these thermodynamic conditions, the perovskite phase does remain stable, and decomposes readily which is why the electro-reduction As shown in Fig. 4a, the dense metallic microstructure formed as a result of the initial stage of electro-reduction is a consequence of the presence of complex titanate liquid. Once the dense metallic layer forms, there is a volumetric change resulting in the shrinkage which appears to form discreet pores on the peripheral surface resulting. Since the weight ratio of K 2 O:TiO 2 is 1:2, the overall reaction suffers from the paucity of K + -rich liquid inside the pores of unreduced TiO 2 and its oxides. The resulting reaction slows down after 5 h with each type of anode and then progresses at much reduced rate, as shown in Fig. 1a and b.
In Fig. 5a the high magnification image of the dendritic region in Fig. 4a is shown and compared. The dendritic structure of titanium metal is evident and also dominated by metallic titanium. There is little evidence for the presence of any perovskite phase in Figs. 4a, b and 5a, b. The effect of increased K 2 O:TiO 2 = 1:1 ratio on the reduction is shown by presenting the microstructural evidence in Fig. 6 in which the formation of dense outer layer is evident and this layer seems more continuous around the peripheral surface than that in Fig. 4. Since the Gibbs energy changes for (3a)-(3c) reactions at 900 °C are large and negative, which suggest that in the presence of freshly formed K 2 O from the decomposition of KHCO 3 , the perovskite surrounding the unreacted TiO 2 is removed by the formation of potassium titanate (K 4 TiO 4 )-rich liquid via reaction (3a). The solid K 2 TiO 3 may form via reaction (3b) only below 500 °C 19 , as it decomposes to form K 4 TiO 4 and K 8 Ti 5 O 14 phase mixture. It should be noted that since the Gibbs energy change for the formation of K 2 TiO 3 is much smaller than that for K 4 TiO 4 , which is why it is the liquid rich in K 4 TiO 4 is likely to be more prevalent than K 2 TiO 3 .
The presence of porosity, which may arise as a result of the solidified microstructure during the phase equilibrium condition of electroreduction with simultaneous evolution of CO 2 from the decomposition of KHCO 3 , may be responsible for maintaining the porosity on the surface of the pellet. It is these microscopic pathways through which the CaCl 2 liquid may reach the inner core of unreduced pellet for removing O 2− ion for maintaining the steady state of ion transport during the electro-reduction process. On the other hand, where the CaCl 2 liquid remains in contact with more impervious metallic Ti, the interstitial diffusion of atomic oxygen might be the viable transport for the O 2− ions transport: O 2− ↔ [O] atomic + 2e either across the metallic layer or through the interdendritic pores of titanium metal. This is likely because the values of the atomic diffusivities of oxygen in α and β forms of Ti metal are of the order of ∼ 1 × 10 -6 cm 2 s −1 , when compared with the 1.323 × 10 -6 -4.033 × 10 -5 cm 2 s −1 for O 2− anion diffusion in CaTiO 3 9,10 under high and low partial pressures of O 2 . Since at a later stage of electroreduction, the overall oxygen potential of the cell reduces, when compared at the start of the electro-reduction, the apparent diffusivity O 2− anion  also increases by three times at 900 °C 9 . The slow atomic and anionic diffusion in metallic Ti and perovskite phase, respectively, may become the rate governing step in achieving near completion of electro-reduction for near 100% metallization. A clear transition in the morphological distinction between the sheet-like and dendritic titanium metal may be the result of K 2 O-rich liquid assisted anion transport between time, t = 0 and t = 5 h on the surface of the pellet at the start of the electro-reduction. The surface reaction is then taken over by the anion transport in combination with interstitial diffusion of atomic oxygen in oxygenated β-Ti at later stage beyond 5 h of electroreduction. Direct evidence for the presence of oxygenated β-Ti may be confirmed by the presence oxygen peak in the EDX spectrum in Fig. 5b. Based on the anionic and atomic diffusivity data, given above, the estimated thickness of metallic layer growth may be of the order of 60-70 μm h −1 of the reduced pellet when using the inert anode. For comparing the effect of current on the reaction mechanism, a carbon anode was also used during electrolysis in the presence of KHCO 3 in the pellet, as shown in Fig. 1b. However, only partial metallization was achieved in this case, which is consistent with the experimental observations made by Schwandt et al. 20 and Alexander et al. J. Indian Inst. Sci. | VOL 102:1 | 127-137 January 2022 | journal.iisc.ernet.in 21 . In this study, we show that the low current is a manifestation of much slower metallization of TiO 2 in spite of the use of TiO 2 mixed with KHCO 3 . To support this point of view, in Fig. 7 we show the elemental maps of a cross-section of partially reduced TiO 2 sample which contained TiO 2 :KHCO 3 = 1:0.2. Although high current was observed at the start, as shown in Fig. 1b, the formation metallic layer was only confined to 500 μm. We also confirmed the formation of KCl via reaction (3c), which also supports other accompanying reactions in 3a and 3b, for example. It should be mentioned that no K 2 TiO 3 was found in the partially reduced sample.
Using the data in the current-time diagrams in Fig. 1a and b, the rate of mass change (Δm,g) has been plotted against time duration for electro-reduction. The change in mass was estimated using the Faraday's law. Based on the computed mass from the current-time plots, the apparent rates of reduction for the inert and carbon anodes were estimated and the derived data from the linear regression analysis (R 2 ≥ 0.998) in Fig. 8 clearly show that the overall kinetic constant for electroreduction is much slower for carbon than that for the inert anode. The comparison of kinetic data derived from Fig. 1a and b and the evidence from the X-ray diffraction and SEM analyses help in deucing that the inert anode with potassium salt in electrolyte and pellet may be better choice for molten salt based electro-reduction of TiO 2 . Before we conclude this article, we compare the above-mentioned process of Ti production with

Conclusions
When the inert anode was used for electroreduction of pressed pellets of TiO 2 :K 2 O = 1:0.5 mixture at 900 °C, nearly 100% metallization of the TiO 2 to metallic Ti was achieved in less than 16 h of electrolysis. The experimental evidence shows that more than 75% reduction was achieved within first 5 h when the decomposed KHCO 3 yielded K 2 O by aiding the formation of K 4 TiO 4 -rich liquid at 900 °C. The results of phase and microstructural examination also demonstrate that the intermediate CaTiO 3 phase decomposes under electro-reduction condition by exchanging anions via reactions (3a) and (3c), which then sustain the cathodic dissociation of TiO 2 present. The presence of K 4 TiO 4 -rich liquid appears to be responsible for the formation of thick continuous layer of Ti-metal on the peripheral surface pressed pellets. Once the K 4 TiO 4 -rich liquid has exhausted, the remaining oxide proceeds via much slower combination of O 2− ion diffusion in perovskite and atomic diffusion of oxygen in the β-Ti structure. There is also microanalytical evidence for the presence of larger concentrations of residual oxygen in the dendritic Ti-metal structure when compared with dense Ti-metal layer formed on the surface of the electro-reduced pellet. From the isothermal electro-reduction data, the rate constant was determined from the mass change versus time curves for the inert and carbon anode reduced pellets. It was found that the rate constant for carbon anode (k carbon = 0.003 g h −1 ) was nearly an order of magnitude smaller in magnitude than that observed for inert anode (k inert = 0.04 g h −1 ). With carbon anode, the presence of K 4 TiO 4 does not lead to complete reduction of TiO 2 as the perovskite does not decompose spontaneously.

Publisher'sNote
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.