Phase Transformations and Deoxidation Kinetics during the Electrochemical Reduction of TiO₂ in Molten CaCl₂

Abstract

There are still some different understandings about deoxidization process from TiO₂ to Ti by FFC Cambridge process. Herein, an investigation was carried out on the phase transformations and deoxidation kinetics of the electrochemical reduction of TiO₂ to Ti in molten CaCl₂. The reaction goes through four different stages involving several phase transformations. The first stage is the deoxidation of TiO₂ with the formation of CaO and CaTiO₃. At this stage, Cl₂ is forcibly released from the anode because of the consumption of Ca²⁺ on the cathode, balanced with Cl⁻ at the anode. The second stage is the deoxidation of CaTiO₃ and titanium suboxides. These two stages take 45% of the entire deoxidation time from TiO₂ to Ti_[O]. The third stage is the deoxidation of titanium suboxides. This stage takes about 55% of the total time. The final stage is further deoxidation of Ti–O solid solution by electrochemical reduction or by calcium thermal reduction. There are no obvious kinetic barriers in the formation of CaTiO₃, but the rate of deoxidation of titanium suboxides and CaTiO₃ is relatively slow. Therefore, the limiting step from TiO₂ to Ti by the FFC process in molten CaCl₂ is deoxidation of titanium oxides and the interphase of CaTiO₃.

1. Introduction

In the past two decades, a novel molten salt electrochemical method, namely the FFC Cambridge process,¹⁾ has been developed, which makes it possible to produce metals and alloys by direct electrochemical reduction of their oxides in molten salt. However, there is ambiguity about the deoxidation pathway due to the formation of complicated intermediate perovskite compounds. In 2002, Chen et al.²⁾ first proposed that calcium titanate is formed by the reaction between CaO, formed by Ca²⁺ in molten salt and O²⁻ discharged from titanium oxides, and TiO₂ or the partially reduced titanium oxides.

The mechanism and kinetics of deoxidation at a constant potential have been the subject of many investigations.^3–5) In the early papers, it was suggested that the shape of the current–time curves depends on the conductivity of the oxide and the solubility of oxygen in the final metal.⁶⁾ It was then found, by taking niobium as an example, that the initial current value and the behavior of the current depend on the porosity and, perhaps, on the effective surface area of the pellets.⁷⁾ A 3PI (three-phase interline) metal/oxide/electrolyte model was also proposed to understand the deoxidation process of metal oxides by the FFC process.⁸⁾ Alexander et al.⁹⁾ have studied the electro-deoxidation pathway of TiO₂. By analyzing partially reduced samples at different times, it was concluded that the electro-deoxidation of TiO₂ takes place in four stages. At first, TiO₂ reacts with Ca²⁺ to produce titanium suboxides and CaTiO₃. These suboxides are then reduced to TiO by the action of Ca²⁺, with more CaTiO₃ produced as a by-product. A chemical reaction then occurs between TiO and CaTiO₃, and the lower-valent CaTi₂O₄ is produced. In the final stage, CaTi₂O₄ is electrochemically reduced to TiO, which is then further reduced to Ti by the formation of a Ti–O solid solution, Ti_[O]. A particularly well-resolved current–time curve for the reduction of TiO₂ samples in molten CaCl₂ at 900°C under an argon atmosphere and using a graphite anode was obtained by Schwandt et al.³⁾ A relationship between the current–time change and the different reaction products was proposed. Initially, the reduction is fast and the current is high. However, the formation of Ti₂O₃ and CaTiO₃, is marked by a significant decrease in current, because the subsequent reduction to TiO and CaTiO₃ proceeds at a much slower rate.

Other researchers have claimed that ex-situ perovskitization can significantly increase the speed and efficiency of the electro-reduction of solid TiO₂ in molten CaCl₂.¹⁰⁾ Therefore, some attempts have been made to react titanium oxide with calcium oxide prior to electrolysis to obtain CaTiO₃, which is then used for electro-deoxidation. This process has been used with other oxides as well. The results show that nucleation and growth are rate-controlling steps at certain stages in the reaction sequence. Reaction speeds cannot be determined by electrochemical reduction and transport considerations alone, but can also be affected by the method of forming the intermediate perovskite compounds, such as CaTiO₃ and CaTi₂O₄. Another voltammetric study to investigate the intermediate phases formed during the electro-deoxidation of TiO₂ in molten CaCl₂ has been carried out.³⁾ A mixture of Ti₂O₃, Ti₃O₅, and CaTiO₃ was observed first, where the titanate phase was inferred to be formed via the combination of Ca²⁺, O²⁻, and the remaining TiO₂. The second stage was the reduction of Ti₂O₃ or CaTiO₃ to TiO. The final stage was the conversion of TiO to Ti.

Bhagat et al.¹¹⁾ used white beam synchrotron XRD to monitor phase evolution in situ during the electrochemical reduction of TiO₂, and demonstrated that the reduction starts by converting TiO₂ to suboxides that improve the ionic conductivity of the oxide pellets. TiO₂ quickly forms substoichiometric phases, which improve the ionic conductivity compared to that obtained with stoichiometric TiO₂. Rogge et al.¹²⁾ studied the combination of CaTiO₃ and titanium metal powder with CaCl₂ flux at 1000°C to form CaTi₂O₄, but CaTi₂O₄ was not formed when the titanium powder was oxidized to TiO, indicating that the comproportionation reaction probably has a small positive ΔG. But the Gibbs free energy of formation of CaTi₂O₄ is unknown,⁹⁾ there is a certain controversy in paper by Rogge et al.

Based on the above researches, it can be obtained that the interphases appeared and their deoxidization processes are complicated. There is still ambiguity in understanding the deoxidation mechanism from TiO₂ to Ti by the FFC Cambridge process. The objective of the present study was to gain insight into the details of the reduction pathway that occur during the electrochemical reduction of TiO₂ to Ti in molten CaCl₂. Partially reduced specimens taken at various times, including after short-time electrolysis, were used to establish the reaction pathway by applying XRD, scanning electron microscopy, and energy-dispersive X-ray techniques in order to gain a greater insight into the phase transformations and deoxidation kinetics of the process. By doing so, we aim to gain a better understanding of the deoxidation mechanism of TiO₂ based on the reported studies.

2. Experimental

2.1 Preparation of the oxide cathode

Commercial TiO₂ powder of 98% mass specified purity was used as the oxide precursor. The TiO₂ powder, screened at 140 mesh, was uniaxially pressed at 2.5–5 MPa with a 12 mm diameter into a 1.3 g preformed cylinder. The cylinders were subsequently sintered at 1173 K for 2 h, in order to obtain a cathode with sufficient mechanical strength for later handling and processing. The sintered cathode is a cylinder with a diameter of 12 mm and a height of 3.75 mm, and its porosity is about 39.9%.

2.2 Experimental procedure

A dense alumina crucible of 100 mm internal diameter and 100 mm height containing about 500 g CaCl₂ was then placed inside the vertical tubular reactor. The water-cooled upper end of the reactor was sealed with a stainless steel cover and a silicone-sealed feed. The sealed reactor was then flushed three times by evacuation followed by filling with argon gas. The reactor was then heated at a constant rate of 5 K/min until the target temperature was reached, with an argon flow of 120 mL/min.

The sintered TiO₂ pellet tied to a steel rod served as the cathode while a graphite rod 5 mm in diameter and 50 mm in height was used as the anode. The system was left to stabilize for 1 h after the target temperature 1173 K was reached. Then the cathode was slowly immersed in the electrolyte. The graphite rod was also immersed into the molten salt. A constant voltage of 3.1 V was applied between the two electrodes. The outlet was connected to a gas bottle containing starch and potassium iodide solution in order to detect chlorine evolution (0.2 g/L soluble starch and 0.08 g/L potassium iodide).

Electrolysis was terminated at set times and both electrodes were removed from the molten salt and placed in the upper part of the reactor. The electrolyzed samples were taken out from the furnace after the whole reactor was cooled to room temperature under an argon atmosphere. The recovered samples were washed in water and then dried. The dried samples were embedded with epoxy resin and ethylenediamine. The inlaid samples were cut and partly polished.

2.3 Characterization

The ground samples were analyzed by XRD (Philips D/max 2500PC generator over a scan range of 10°∼80°). Fracture surfaces were investigated using SEM (TESCAN VEGA II) equipped with energy-dispersive X-ray spectroscopy (EDS, Oxford INCA Energy 350) working at an accelerating voltage of 15 keV, secondary electrons were detected, and images were taken at magnifications of 1000X to 5000X.

3. Results and Discussions

In this work, the obtained shape of the current-time curve and the main phases appearing during electrolysis are similar to previous reports, as shown in Fig. 1. However, an interphase of CaO (in the form of CaCO₃ due to washing in water) appeared in the initial stage during electrolysis, which is quite different from other reports, as shown in Figs. 2 and 3. The main phases of titanium oxide include TiO₂, Ti₄O₇, Ti₂O₃, TiO, Ti₂O, and Ti (Ti–O solid solution), and the main interphases are CaTiO₃, CaO, and TiC. The results of EDS analysis at the spots indicated in Fig. 3 are shown in Table 1.

Fig. 1

SEM images of the partially reduced samples and the current-time curve of TiO₂ electrochemically reduced in molten CaCl₂ for 24 h.

Fig. 2

XRD patterns of the samples electrolyzed for various times.

Fig. 3

SEM images and elemental distribution of the partially electrolyzed sample for (a) 5 min, (b) 40 min and (c) 60 min.

Table 1 EDS analysis at the spots indicated in Fig. 3. (Atomic percent)

A few titanium suboxides were found and some CaO appeared in the initial stage of electrolysis. At this stage, chlorine gas was forcibly released on the anode, although the applying voltage was lower than the decomposition voltage of CaCl₂. A turning point was observed on the current–time curve (shown in Fig. 1), when no more chlorine gas released. We believe that this point indicates the end of the formation of CaTiO₃ on the cathode. After this time, Ca²⁺ in the molten salt no longer takes part in the reaction on the cathode. In addition, after this turning point, deoxidation of CaTiO₃ and titanium suboxides occurred. To understand the entire deoxidation process from TiO₂ to Ti metal by the FFC process in molten CaCl₂, the following questions need to be further discussed:

3.1 Stage I: Deoxidation of TiO₂ and formation of CaO on the surface and CaTiO₃ in the interior of the cathode

Analysis of the partially electrolyzed samples showed that CaTiO₃ was formed after a short electrolysis time. TiO₂ can still found in the sample as well as a small amount of CaTiO₃ after 5 min of electrolysis, as shown in Fig. 3(a). The EDS results show that the Ca:Ti:O atomic ratio at point I is 15.03:16.11:68.85(≈ 1:1:3) and the Ti:O atomic ratio at point II is 25.62:71.62(≈ 1:2) in Table 1. CaTiO₃ is found to be distributed near TiO₂. This means that TiO₂ is easily deoxidized under electrolysis, although it is a semiconductor. However, when the electrolysis time was increased to 40 min, in addition to CaTiO₃, some CaO appeared on the surface of the cathode as balls with poor conductivity, as shown in Fig. 3(b). The EDS results show that the Ca:O atomic ratio at point III is 19.03:66.19(≈ 1:1) and the Ca:Ti:O atomic ratio at point IV is 25.62:71.62(≈ 1:1:3) in Table 1. The polished section of the sample electrolyzed for 60 min shows a dense layer on the surface, as shown in Fig. 3(c). The EDS results show that the Ca:O atomic ratio at point V is 22.49:76.51(≈ 1:1) and the Ti:O atomic ratio at point VI is 39.72:55.23(≈ 1:1) in Table 1. The appearance of CaO in the initial time period differs from Bhagat’s results,¹¹⁾ which inferred that CaO only appears at the final stage of deoxidation of TiO₂ due to limited dissolution of CaO in molten salt. From this study, it can be concluded that CaO also appeared in the initial stage of deoxidation of TiO₂. However, in the interior of the sample, the main phases are CaTiO₃ and titanium suboxides.

Based on the results of short time electrolysis, it can be concluded that TiO₂ can trap electrons immediately from the electrode and discharge O²⁻ when the power is switched on. The O²⁻ first reacts with Ca²⁺ from the molten salt to form CaO, which then quickly reacts with undeoxygenated TiO₂ to form CaTiO₃. However, CaO cannot react with titanium suboxides. There is a large amount of TiO₂ in the interior of the sample, which does not have enough time to deoxidize before reacting with CaO to form CaTiO₃. Therefore, at this stage, three different layers co-exist on the cathode: a CaO layer, a titanium suboxide layer, and a CaTiO₃ layer. It is unnecessary for all TiO₂ to transform into CaTiO₃ during the electrochemical reduction by the FFC process from TiO₂ to Ti. The microtopography and species present at the cathode strongly depend on the electrolysis time, as shown in Figs. 1 and 2. At this stage, the following processes may be occurring:

① The TiO₂ cathode is surrounded by calcium ions which can freely enter into the cathode. ② The TiO₂ cathode obtains electrons and O²⁻ is released from the cathode. ③ Ca²⁺ traps O²⁻, forming CaO immediately. Due to the bonding energy of Ca²⁺ and O²⁻ being higher than Ti^x+ and O²⁻, CaO is formed easily. Then, a new phase of CaO is formed on the surface of titanium suboxides, as shown in Fig. 3(b). Ca²⁺ located in the inner part of the cathode can also trap O²⁻ forming CaO, which can then react with the remaining TiO₂ to form CaTiO₃. ④ When deoxidation of TiO₂ on the surface sample and the reaction between CaO and TiO₂ in the interior of sample are complete, Ca²⁺ will no longer be consumed. This procedure takes about 5.6% of the total electrolysis time from TiO₂ to Ti–O solid solution, i.e., the formation of CaTiO₃ in molten CaCl₂ salt is relatively fast. The overall cathodic reactions leading to the formation of the experimentally observed mixtures may be written as follows:   

\begin{align} \text{Cathode:}\ &\text{TiO$_{2}$} + \text{2ae$^{-}$}\rightarrow \text{TiO$_{\text{z}}$} + \text{aO$^{2-}$}\\ &\ (\mathrm{z} = 2-\mathrm{a},\ 0<\mathrm{a}<0.5) \end{align} (1)

  

\begin{equation} \text{Surface of the sample:}\ \text{Ca$^{2+}$}+ \text{O$^{2-}$} \rightarrow \text{CaO} \end{equation} (2)

  

\begin{equation} \text{Interior of the sample:}\ \text{CaO} + \text{TiO$_{2}$}\rightarrow \text{CaTiO$_{3}$} \end{equation} (3)

The anodic reaction may involve the discharge of chloride ions, although the formation of carbon monoxide and carbon dioxide as the anodic off-gases is thermodynamically more favorable than the formation of chlorine. The details of the anodic behavior are discussed later.   

\begin{equation} \text{Anode:}\ \text{2Cl$^{-}$}\rightarrow \text{Cl$_{2}$} + 2\mathrm{e}^{-} \end{equation} (4)

The current–time curve in Fig. 1 shows a sharp decrease during this stage.

3.2 Stage II: Deoxidation and decalcification of CaTiO₃

In the interior of the sample, CaTiO₃, CaO, and TiO_z exist at the same time. It was observed in the first stage that the remaining TiO₂ in the interior of the cathode changed to CaTiO₃, which is different from the results of Schwandt et al.³⁾ The second stage of the electrochemical reduction of CaTiO₃ is characterized by the dominant presence of CaO and titanium suboxides. This means that CaO only reacts with TiO₂ to form CaTiO₃ without reacting with titanium suboxides. Meanwhile, the outer layer of titanium suboxides obtained in the first stage continues to be deoxidized.   

\begin{align} \text{CaTiO$_{3}$} + 2\beta\mathrm{e}^{-}&\rightarrow \text{CaO} + \text{TiO$_{\text{z}}$} + \beta\text{O$^{2-}$}\\ &\quad \ (\beta = 3-\mathrm{z}-1) \end{align} (5)

  

\begin{equation} \text{TiO$_{\text{z}}$} + 2\mathrm{e}^{-}\rightarrow \text{TiO$_{(\text{z}-1)}$} + \text{O$^{2-}$} \end{equation} (6)

It can be assumed that the electrochemical deoxidation of CaTiO₃ and titanium suboxides is the dominant reaction in the second stage. In addition, in the adjacent layer (the core), only CaO and Ti₂O₃ can be observed. No CaTi₂O₄ phase appeared during the electrochemical process and titanium suboxides did not react with CaO to form CaTiO₃, which is quite different from previous reports. When CaO and TiOz (0 < z < 2) coexist with CaTiO₃, the layer seems to be much denser than the adjacent layer containing only CaO and titanium suboxides, and its color is also different from that of the other layers. There is no CaO on the surface. One possible reason is that CaO produced in the first stage is dissolved in the molten CaCl₂.

From stage I, it is known that CaO and CaTiO₃ form quickly during electrolysis. Some of the Ca²⁺ from the electrolyte participates in the cathode reaction forming CaTiO₃ and CaO during stage I and finally these calcium species leave the cathode again. About 20 mole% CaO can be dissolved in molten CaCl₂.⁹⁾ This dissolved CaO can also be electrolyzed and Ca metal can deposit on the cathode at the experimental potential. In this study, no obvious hydrogen released when the electrolyzed sample is washed with water when the electrolysis is incomplete. In contrast, the release of hydrogen is very intense when the completely electrolyzed sample is immersed in water. A possible reason is that deposited Ca metal is used to reduce titanium suboxides or CaTiO₃ when the cathodic deoxygenation is incomplete. However, when the cathode is completely deoxidized, Ca metal deposits in the cathode. Ca in the molten CaCl₂ can reduce titanium oxides to titanium metal, which has been proven by the OS process.¹³⁾

Therefore, the main process in stage II is the disappearance of the calcium species (CaTiO₃ and CaO) formed in stage I. New phases appearing on the cathode are CaO from CaTiO₃ and Ca metal from the dissolved CaO. It takes much more time for decalcification of the cathode. The cathodic reaction during this stage may be accompanied by the deposition of calcium metal.

3.3 Stage III: Electrolysis of TiO_z or calcium thermal reduction of TiO_z (0 < z < 2)

From Fig. 4, it can be seen that the decomposition potential of CaO is more negative than that of TiO_z. Theoretically, TiO_z is easier to electrolyze than CaO when both are present. When calcium metal deposits on the cathode, there are two routes toward deoxidation of titanium suboxides. One possible route is the calcium thermal reduction of TiO_z.   

\begin{equation} \text{TiO$_{\text{z}}$} + z\text{Ca}\rightarrow \text{Ti$_{\text{[O]}}$} + z\text{Ca$^{2+}$} + \text{O$^{2-}$} \end{equation} (7)

Fig. 4

Decomposition potential of titanium oxides and CaCl₂.

This reaction is chemically driven because there is no flow of electrons. In the OS process, Suzuki et al.¹³⁾ have proved that pure Ca is not necessary to reduce TiO₂ into metallic Ti, and dissolved Ca can be used for TiO₂ reduction. No proof of calcium thermal reduction of TiO_z has been reported, however, probably because of the solubility of Ca and the rapid reaction between TiO_z and Ca. In contrast, the final Ti metal electrolyzed for 24 h shows a compacted lump with high strength and dense structure, as shown in Fig. 1. Why does the final titanium metal deoxidized from the TiO₂ powder sinter together instead of becoming titanium powder? One of the possible reasons is that the exothermic reaction of the calcium thermal reduction provides enough heat for solid sintering of the final titanium metal. When TiO_z is reduced to Ti_[O] completely, Ca metal is deposited on the cathode. Residual Ca then reacts with water to release H₂ during the salt removal process at room temperature.

A second possibility is the electrochemical reduction of TiO_z.   

\begin{equation} \text{TiO$_{\text{z}}$} + \mathrm{e}^{-}\rightarrow \text{Ti$_{\text{[O]}}$} + z\text{O$^{2-}$} \end{equation} (8)

This is because the electrochemical decomposition potential of the titanium suboxides is lower than the applied voltage. In this stage, titanium suboxides can also receive electrons to be further deoxidized. The sample electrolyzed for 8 h contained only TiO_z and Ti_[O]. Therefore, both calcium thermal reduction and electrochemical reduction may contribute to the reduction of TiO₂ to Ti–O solid solution. Herein, the difficulty is how to confirm which reduction mechanism is the dominant one. Further study is needed to resolve this question.

3.4 Stage IV: Deoxidation of Ti–O solid solution (Ti_[O])

Due to the high bonding energy between oxygen and titanium, further deoxidation of Ti–O solid solution is very difficult. To obtain titanium with a lower oxygen content, a further refining process can be used. Lowering the oxygen partial pressure can also be tried.¹⁴⁾

4. Conclusions

Phase transformations and deoxidation kinetics of the electrochemical reduction of TiO₂ in molten CaCl₂ were studied based on partially reduced samples taken at different times during the experiments. The following conclusions were obtained:

An interphase of CaTiO₃ is formed by Ca²⁺ coming from the molten CaCl₂, O²⁻ released from TiO₂ and the remaining TiO₂. Titanium suboxides do not participate in the formation of CaTiO₃.

Ca²⁺ and Cl⁻ in the molten CaCl₂ are consumed as CaTiO₃ or CaO on the cathode and Cl₂ on the anode respectively during the stage of CaTiO₃ formation.

CaTiO₃ can be reduced to titanium suboxides. Ca²⁺ in the molten salt first forms interphases on the cathode and then the Ca²⁺ is released from the cathode again.

The main limiting step from TiO₂ to Ti by the FFC process in molten salt is the deoxidation of CaTiO₃ and titanium suboxides. The rates of deoxidation of CaTiO₃ and TiO are much slower.

The process from TiO₂ to metal includes four main stages: deoxidation of part of the TiO₂ and formation of CaTiO₃, deoxidation and decalcification of CaTiO₃, deoxidation of titanium suboxides, and deoxidation of Ti–O solid solution.

Ca metal may be produced on the cathode and it may take part in the reduction of CaTiO₃ and titanium suboxides as the reductant. This hypothesis needs to be further studied.

Acknowledgments

The authors acknowledge gratefully the financial support from the National Natural Science Foundation of China (Grant No. 51674054).

REFERENCES

• 1)   G.Z.  Chen,  D.J.  Fray and  T.W.  Farthing: Nature 407 (2000) 361–364.
• 2)   G.Z.  Chen and  D.J.  Fray: J. Electrochem. Soc. 149 (2002) E455–E467.
• 3)   C.  Schwandt and  D.J.  Fray: Electrochim. Acta 51 (2005) 66–76.
• 4)   S.  Wang and  Y.  Li: J. Electroanal. Chem. 571 (2004) 37–42.
• 5)   K.  Dring,  R.  Dashwood and  D.  Inman: J. Electrochem. Soc. 152 (2005) E104–E113.
• 6)  G.Z. Chen and D.J. Fray: Light Metals 2004, ed. by A.T. Tabereaux, (The Minerals, Metals & Materials Society, 2004) pp. 881–886.
• 7)   X.Y.  Yan and  D.J.  Fray: Metall. Mater. Trans. B 33 (2002) 685–693.
• 8)   G.Z.  Chen,  E.  Gordo and  D.J.  Fray: Metall. Mater. Trans. B 35 (2004) 223–233.
• 9)   D.T.L.  Alexander,  C.  Schwandt and  D.J.  Fray: Acta Mater. 54 (2006) 2933–2944.
• 10)   K.  Jiang,  X.  Hu,  M.  Ma,  D.  Wang,  G.  Qiu,  X.  Jin and  G.Z.  Chen: Angew. Chem. Int. Ed. 45 (2006) 428–432.
• 11)   R.  Bhagat,  D.  Dye,  S.L.  Raghunathan,  R.J.  Talling,  D.  Inman,  B.K.  Jackson,  K.K.  Rao and  R.J.  Dashwood: Acta Mater. 58 (2010) 5057–5062.
• 12)   M.P.  Rogge,  J.H.  Caldwell,  D.R.  Ingram,  C.E.  Green,  M.J.  Geselbracht and  T.  Siegrist: J. Solid State Electrochem. 141 (1998) 338–342.
• 13)   R.O.  Suzuki and  S.  Inoue: Metall. Mater. Trans. B 34 (2003) 277–285.
• 14)   Y.  Zhang,  Z.Z.  Fang,  P.  Sun,  T.  Zhang,  Y.  Xia,  C.  Zhou and  Z.  Huang: J. Am. Chem. Soc. 138 (2016) 6916–6919.