Electrochemical Behaviour of Dissolved Titanium Oxides during Aluminium Deposition from Molten Fluoride Electrolytes

Abstract

A small but significant amount of dissolved titanium is always present in the molten cryolite based electrolyte used during electrowinning of aluminium. The fact that titanium can appear in several oxidation states may be a cause for reduced current efficiency for aluminium, although codepositon of titanium should be thermodynamically favoured. Voltammetric studies were carried out in molten Na₃AlF₆ - Al₂O₃ (sat) containing TiO₂ at 1020℃. The results suggested that titanium was reduced in two steps from Ti (IV) to Ti (III) followed by reduction of Ti (III) to Ti metal.

Experiments were also carried out in industrial aluminium cells. Large quantities of TiO₂ were added to the electrolyte during electrolysis, and samples of electrolyte and produced metal were taken and analysed as a function of time after addition. The results suggested that titanium was codeposited at the aluminium cathode by a diffusion controlled reaction. The apparent current efficiency for titanium deposition was estimated to be higher than 90%.

The valency problem seems to be less challenging when depositing liquid titanium alloys from molten fluoride electrolytes. This approach may give rise to the development of production of a valuable titanium containing alloy TiAl₃.

1. Introduction

Experimental studies in controlled laboratory experiments and industrial cells were carried out to study the behaviour of important impurity elements in molten fluoride electrolytes with dissolved alumina used for producing aluminium. The results obtained for titanium may be of interest for the prospect of developing an electrolytic process for the production of titanium or titanium alloys.

The use of titanium is limited due to the high cost of the Kroll process, which is the only commercial production route for primary titanium. Electrolysis in a molten salt electrolyte is a possible way to produce titanium, and many attempts to develop an electrolysis process have been reported. Large cells and pilot plants have been run, but no commercial electrolysis production has been achieved. Worth mentioning is the work by Ginatta et al.¹⁾ which resulted in the successful development of a pilot cell for the electrowinning of titanium from TiCl₄ dissolved in a molten NaCl based electrolyte. The most famous approach is due to Fray et al.²⁾ who published a new electrolysis method, known as the FFC Cambridge method. Similar methods have been proposed by Suzuki et. al³⁾ and Okabe and Waseda⁴⁾.

The main obstacles for successful development of an electrochemical route for titanium production have been to fullfil the purity requirements of the titanium product, mainly due to the lack of an inert oxygen evolving anode and the problems associated with the existence of various valencies of dissolved titanium species.

In electrolytes having high oxide solubilities, such as CaCl₂, the failure to develop an inert anode for oxygen has led to carbon contamination of the cathode product. In general the oxygen level in electrochemically produced titanium has been too high. In addition the different valencies of dissolved titanium have caused inefficiencies in other electrolytes.

Various types of molten salt electrolytes have been tested and proposed. Titanium is known to form stable compounds and dissolved species of the oxidation states Ti (II), Ti (III), and Ti (IV). The solubility of Ti (IV) in molten salts is low, and most fundamental investigations have been done using Ti (II) or Ti (III) as the source of titanium. It is known that in molten chlorides titanium can be deposited from Ti (II), whereas in molten salts containing flouride Ti (II) is not stable⁵⁾. Also disproportionation reactions such as the following can occur   

\[ 3\ {\rm TiCl}_{2} = 2\ {\rm TiCl}_{3} + {\rm Ti} \](1)

In approaches using solid compounds, such as the FFC method, mainly TiO₂ or other Ti (IV) compounds have been used as the raw material.

In the context of the present studies, the behaviour of titanium species as an impurity in the electrolysis process for aluminium production is of relevance. Aluminium is produced by electrolysis in molten NaF-AlF₃-Al₂O₃ at ～955℃ Impurities enter the electrolyte mainly from the carbon anode and with the added alumina. Most impurities form soluble species in the electrolyte. Up to 90% of the titanium originates in the aluminium oxide, the rest coming from the anode carbon. Electrochemical investigations of Ti (IV) in fluoride electrolytes showed a two-step reduction mechanism for Ti (IV) with Ti (III) as a sole stable intermediate⁶⁾, according to   

\[ {\rm Ti}\ ({\rm IV}) \to {\rm Ti}\ ({\rm III}) \to {\rm Ti} \](2)

In chloride systems a three-step reduction of Ti (IV) was found, according to   

\[ {\rm Ti}\ ({\rm IV}) \to {\rm Ti}\ ({\rm III}) \to {\rm Ti}\ ({\rm II}) \to {\rm Ti} \](3)

Guang-Sen et al.⁷⁾ studied the electrochemical reduction of Ti (IV) in a molten equimolar KCl-NaCl mixture with additions of KF. At low concentrations of fluoride ion, the reduction was found to take place in accordance with reaction scheme (3), while the reduction was found to proceed according to eq. (2) at higher fluoride concentrations.

Delimarskii et al.⁸⁾ investigated the electrodeposition of titanium from molten cryolite varying content of TiO₂ at 1050℃. A three-electrode configuration was used for the linear voltammetry mesurements. The cathode was a rod of spectral pure graphite (A = 0.2 cm²). A graphite rod was used as reference electrode, while the graphite crucible served as the anode. Argon was purged through the cell during the experiments. Their results showed that two plateaus were observed in cryolite melts with varying TiO₂ content, which indicated that Ti(IV) was reduced in a two-step process,

Qui et al.⁹⁾ studied the deposition of titanium with cyclic voltammetry. The working electrode was made of graphite and the graphite crucible served as the counter electrode. The reference electrode was aluminium contained in a corundum tube with a molybdenum wire as the current lead. In pure molten cryolite with alumina a cathodic current peak and a corresponding anodic current peak were due to aluminium deposition and oxidation. When TiO₂ was added to cryolite instead of Al₂O₃ two new peaks appeared. These peaks were assigned to the reduction of Ti (IV) and the subsequent oxidation of Ti metal., according to   

\[ {\rm Ti}\ ({\rm IV}) + 4\ {\rm e}^{-} \to {\rm Ti} \](4)

and   

\[ {\rm Ti} \to {\rm Ti}\ ({\rm IV}) + 4\ {\rm e}^{-} \](5)

The potential difference between the aluminium deposition peak current and the titanium deposition peak current was found to be ～0.32 V, which is very close to the theoretical difference in the standard potentials for the two reactions.

Raj and Skyllas-Kazacos¹⁰⁾ carried out electrochemical studies in cryolite-alumina melts at 1050℃. The working electrode was made of graphite, a tungsten electrode was used as a quasi-reference electrode, while the carbon crucible containing the electrolyte served as the counter electrode. Potential sweep measurements showed that two new current peaks appeared after TiO₂ was added to the melt, i.e. one reduction peak and one oxidation peak. These peaks were similar to the results of Qui et al.⁹⁾ due to reactions (4) and (5). The results suggested that alloy formation of Al-Ti was taking place.

Devyatkin et al.¹¹⁾ applied cyclic voltammetry to study the electrochemical behaviour of Ti (IV) in cryolite-alumina melts with additions of TiO₂, MgTiO₂ and CaTiO₂. The experiments were performed at 1300 K in air. The working electrode was a platinum wire (0.15–0.2 cm²), while O²|O²⁻ (Pt) was used as reference electrode. The counter electrode and the crucible were also made of platinum. Voltammograms recorded in the system Na₃AlF₆- Al₂O₃ showed several peaks that were assigned to the formation of intermetallic compounds between aluminium and platinum. In the system Na₃AlF₆- Al₂O₃-TiO₂ two new processes were observed. Based on diagnostic criteria the first process was found to be a reversible one-electron charge transfer with a soluble product, i.e. reaction (2). The second process was believed to be a three-electron transferw ith formation of metallic titanium, i.e. reaction (2). The deposition of titanium was found to be followed by an irreversible chemical reaction according to   

\[ {\rm Ti} + 3\ {\rm Ti} {\rm O}_{2} \to 2\ {\rm Ti}_{2} {\rm O}_{3} \](6)

When investigating the system Na₃AlF₆- Al₂O₃-MgTiO₃ a new peak appeared. This process was an irreversible four-electron reaction in accordance with reaction (4), followed by an irreversible chemical reaction accodring to   

\[ {\rm Ti} + 3\ {\rm MgTiO}_{3} \to 2\ {\rm Ti}_{2}{\rm O}_{3} + 3\ {\rm MgO} \](7)

The system Na₃AlF₆- Al₂O₃-CaTiO₃ gave rise to one new peak. It was concluded that this was in accordance with a reversible four-electron charge transfer reaction with the formation of metallic titanium according to reaction (4).

Metallic impurities that are more noble than aluminium tend to deposit at the liquid aluminium cathode in the industrial process¹²⁾. It has been shown^13,14) that cations of such impurities are reduced at the cathode at the limiting current density (i_lim) which is given by the following equation:   

\[ {\rm i}_{\rm lim} = {\rm nFkc}^\circ \](8)

where k is the mass transfer coefficient and c° is the concentration of the dissolved impurity element species in the bulk of the electrolyte. Studies of impurities in industrial cells have been carried out by analyses of samples of electrolyte and metal as a function of time after additions of known amounts of compounds containing impurities. The concentration of the impurity species under investigation versus time after addition can be expressed as follows¹⁴⁾   

\[ c = c_{o} \exp \left( - \frac{A}{V}kt \right) \](9)

where A is the area of the active cathode and V is the volume of the electrolyte. By knowing the latter quantities, the mass transfer coefficient can be determined. Typical values have been reported to be in the range from 10⁻⁵–10⁻⁶ m/s^13,14), iron giving slightly higher values than silicon and titanium. However, such mass transfer coefficients will depend more on design, technology and operation of industrial cells. High convection of electrolyte and metal pad or unstable cells will cause higher values.

2. Experimental

The experiments were carried out in a standard furnace under argon atmosphere. The electrolyte temperature was mesured by a thermocouple inside a platinum tube placed in the melt.

Great care was used to avoid using oxide materials in the cell exposed to the electrolyte. A tungsten wire (2.0 mmØ) was used as working electrode, while the counter electrode was made of platinum. Wires of platinum and tungsten served as quasi reference electrodes. In experiments with dissolved alumina a liquid aluminium reference electrode was used, where liquid aluminium was placed inside an alumina tube with a slot for electrolyte contact and a tungsten wire.

Natural hand-picked cryolite (Kryolittselskabet) was the main electrolyte constituent. Al₂O₃ (Fluka, pro analysis) and TiO₂ (Alfa Johnson Matthey, 99%+ rutile) were dried at 200℃ before use. The electrolyte was contained in a platinum or a sintered alumina crucible.

Experiments in industrial cells were carried out by adding known amounts of TiO₂ in the electrolyte and analysing electrolyte samples taken as a function of time after additions. Also samples of produced aluminium were taken and analysed.

Industrial measurements to study the behaviour dissolved titanium species during aluminium electrolysis were carried out in a cell with prebaked anodes. The solid crust on top of the cells was removed near one end of the cell above the tapping hole, which is where aluminium is tapped regularly, and and a known amount of TiO₂ was added to the electrolyte. Samples of electrolyte and liquid aluminium were taken before the addition and at certain intervals after the addition. Bath samples were analysed by ICP or XRF and metal samples were analysed by optical emission spectrography. The bath analysis techniques could not distinguish between different oxidation states of titanium, so only the total content of dissolved titanium was determined.

3. Results and Discussion

3.1 Electrochemical studies in molten fluoride electrolytes

Figures 1 and 2 show results from cyclic voltammetry in molten Na₃AlF₆- Al₂O₃ (sat) and pure molten Na₃AlF₆, both with additions of TiO₂ at 1020℃. Clear corresponding cathodic and anodic current peaks were recorded at potentials less cathodic than deposition of aluminium. These reactions were assumed to be due to deposition and oxidation of aluminium. The peak potential separation indicated that the n value was close to 3, which suggests that titanium is deposited according to   

\[ {\rm Ti}\ ({\rm III})\ ({\rm diss}) + 3\ {\rm e}^{-} \to {\rm Ti}\ ({\rm s}) \](10)

The cathodic current increase at more anodic potentials was then assumed to be due to the following reaction   

\[ {\rm Ti}\ ({\rm IV})\ ({\rm diss}) + {\rm e}^{-} \to {\rm Ti}\ ({\rm III})\ ({\rm diss}) \](11)

A more careful look at the voltammograms suggested that a small alloying of titanium with tungsten also took place. The results also indicated that the titanium deposition was diffusion controlled and also that aluminium was codepositing at slightly more cathodic potentials.

Fig. 1

Cyclic voltammograms on a tungsten electrode at 200 mV/s in molten Na₃AlF₆- Al₂O₃ (sat) containing 1 mass% and 2 mass% TiO₂ at 1020℃. Tungsten reference electrode.

Fig. 2

Cyclic voltammograms on a tungsten electrode at 50, 100, and 200 mV/s in molten Na₃AlF₆ containing 1 mass% TiO₂ at 1020℃. Platinum reference electrode.

These results are in agreement with reports from other studies in fluoride melts where titanium is deposited from Ti (III) because Ti (II) species are unstable⁷⁾.

3.2 Industrial studies in aluminium electrolysis cells

Figure 3 shows the analysed content of titanium in electrolyte samples as a function of time after addition of TiO₂ to the electrolyte of an industrial aluminium producing cell. After about 24 hours the titanium content was found to stabilise at values close to the starting level.

Fig. 3

The content of titanium in the electrolyte as a function of time after addition of TiO₂ to an industrial aluminium electrowinning cell.

The model explained in the introduction was tested by plotting log concentration versus time after addition as shown in Fig. 4. Good fits to the model were obtained, and the mass transfer coefficient of titanium was found to be ～5∙10⁻⁶ m/s from measurements in a 250 kA prebake cell. Similar mass transfer coeffcients were determined for other metallic elements such as iron, silicon, and manganese; all being more noble than aluminium¹⁴⁾.

Fig. 4

Plot of log concentration for dissolved titanium versus time after addition of TiO₂.

Analyses of the produced metal showed that the titanium content increased during the period after the addition. By analysing metal samples taken as a function of time after adding TiO₂ it was also possible to estimate the faction of added titanium that was codepositing with aluminium at the cathode during the experiment. The average fraction of codepositing titanium was found to be ～0.9. This number suggests that the current efficiency for titanium codeposition with aluminium should be at least 90% if a relatively high content of oxide titanium oxide could be maintained. The valency issues seem to be less of a challenge when depositing a liquid alloy such as Al-Ti. According to the Al-Ti phase diagram a mixture of TiAl₃ and pure Al will be formed upon cooling of the liquid Al-Ti alloy produced by electrolysis.

4. Conclusions

Voltammetric studies were carried out in molten Na₃AlF₆- Al₂O₃ (sat) containing TiO₂ at 1020℃. The results suggested that titanium was reduced in two steps from Ti (IV) to Ti (III) followed by reduction of Ti (III) to Ti metal.

These results are in agreement with reports from other studies in fluoride melts where titanium is deposited from Ti (III) because Ti (II) species are unstable.

Experiments were also carried out in industrial aluminium cells. Large quantities of TiO₂ were added to the electrolyte during electrolysis, and samples of electrolyte and produced metal were taken and analysed as a function of time after addition. The results suggested that titanium was codeposited at the aluminium cathode by a diffusion controlled reaction. The apparent current efficiency for titanium deposition was estimated to be higher than 90%.

The valency problem seems to be less challenging when depositing liquid titanium alloys from molten fluoride electrolytes. This approach may give rise to the development of production of a valuable titanium alloy TiAl₃.

Acknowledgements

Dr. Trond Eirik Jentoftsen is acknowledged for carrying out many of the experiments. Hydro Aluminium is thanked for financial support.

REFERENCES

• 1)  M.V. Ginatta: Industrial Plant for the Production of Electrolytic Titanium, Proceedings 6th World Conference on Titanium, Cannes, France, 6-9 June, 1988.
• 2)  D.J. Fray, G.Z. Chen and T.W. Farthing: Nature 407 (2000) 361. 10.1038/35030069
• 3)  R.O. Suzuki, K. Teranuma and K. Ono: Metall. Mater. Trans., B 34 (2003) 287. 10.1007/s11663-003-0074-1
• 4)  T.H. Okabe and Y. Waseda: JOM 49 (1997) 28. 10.1007/BF02914710
• 5)  E. Chassaing, F. Basile and G. Lorthiair: J. Appl. Electrochem. 11 (1981) 187. 10.1007/BF00610979
• 6)  M. Makyta, K. Matiasovsky and V.I. Taranenko: Electrochim. Acta 34 (1989) 861. 10.1016/0013-4686(89)87120-8
• 7)  C. Guang-Sen, M. Okido and T. Oki: J. Appl. Electrochem. 18 (1988) 80. 10.1007/BF01016208
• 8)  Y.K. Delimarski, A.P. Krymov, V.F. Makogon, V.I. Shapoval and V.V. Nerubashchenko: Ukr. Khim. Zh. 45 (1979) 803.
• 9)  Z. Qiu, M. Zhang, Y. Yu, Z. Che, K. Grjotheim and H. Kvande: Aluminium 64 (1988) 606.
• 10)  S.C. Raj and M. Skyllas-Kazacos: Electrochim. Acta 37 (1992) 1787. 10.1016/0013-4686(92)85082-V
• 11)  S.V. Devyatkin, G. Kaptay, J.-C. Poignet and J. Bouteillon: High Temp. Mater. Process. 2 (1998) 497. 10.1615/HighTempMatProc.v2.i4.50
• 12)  K. Grjotheim and K. Matiasovsky: Aluminium 59 (1983) 687.
• 13)  H. Grini Johansen, J. Thonstad and Å. Sterten, Light Metals 1977, p.253.
• 14)  T.E. Jentoftsen, G.M. Haarberg, B.P. Moxnes, A. Buen and J. Thonstad: Mass Transfer of Iron, Silicon and Titanium in Hall-Heroult Cells, Proceedings, 11th Int. Aluminium Symposium, ed. A. Solheim and G.M. Haarberg, September 2001, Trondheim-Bergen-Trondheim, Norway, p.217.