Method and means for metal production in chloride melts

Introduction

The present invention relates to a method and means for production of a metal or a metal alloy in an electrolysis cell.

Methods like the FFC-Cambridge and OS process have been suggested in the past for cheap production of metals and alloys from a CaCI ₂ -based electrolyte.

The anode material that has been mentioned specifically in these patents is carbon-based. However, the use of carbon anodes in the FFC-Cambridge process or similar processes based on CaCb-CaO electrolytes creates substantial problems that need to be overcome in order to produce cheap metal and alloys. Most important, formation of floating carbon sludge and carbon deposition on the cathode has been observed repeatedly. One to two percent carbon is enough to make the electrolyte sludgy, which might reduce the electrolyte circulation and the gas release from the electrolyte. In addition, the carbon will increase the risk of creep current and short-circuiting between the electrodes. The carbon can also dissolve in the electrolyte as carbonate, and a cyclic reaction is set up where oxide is cycled between the anode and the cathode, and where carbon is transported from the anode to the cathode. This cycle will lead to significant losses in current efficiency (CE) with respect to metal production.

The present invention offer some candidate inert anodes solving the carbon problems by removing carbon from the system. Another beneficial by-effect is that oxygen, instead of the greenhouse gas CO ₂ , is being produced at the anode. The anode will be used to produce a metal from a "non-metal species" (raw material) containing the metal ion.

Prior Art

Many people have tried to find a cheap and easy way of producing metals and alloys, and a replacement for the expensive and time consuming Kroll process for titanium manufacturing has been give special attention because of the remarkable properties of titanium making it the metal of the future limited basically only because of production cost.

FFC-Cambridge have filed several patents protecting the process they call "electro deoxidisation" (WO 99/64638, WO 02/40725 A2, WO 03/048399 A2, WO 2004/018735 A1 etc.) where they apply a potential below the decomposition of the electrolyte, but high enough to reduce a substrate, which is connected to the cathode by DC current. Another company, Qinetic, has also filed many patent applications dealing with this process (f. ex. WO 01/62994 A1 , WO 01/62995 A1 and WO 01/62996 A1 ).

Several metals has been produced by the FFC-Cambridge process, see Figure 1 :

The FFC-Cambridge Process can be summarised as follows, using titanium dioxide (rutile or anatase) and titanium as an example.

TiO ₂ (solid, cathode) => molten salt electrolysis (CaCI ₂ -CaO)

=> Ti (solid, cathode) + O ₂ /CO ₂ (anode) (1 )

The anode product is dependent on the anode material. Oxygen is evolved at an inert anode, and CO ₂ when a consumable carbon anode is used. ■

The process is a solid-state electrolysis process, where the raw material is reduced as a solid, unlike the traditional approach where the raw material. is dissolved in the electrolyte and then reduced and deposited at the cathode.

The in-situ processes start with a calciothermic reduction process, where calcium metal is being made at the cathode by reduction of the electrolyte:

CaCI ₂ + 2e ^" = Ca + 2Cl ^" (2)

The calcium then reacts with the raw material forming a metal, illustrated here with TiO ₂ :

TiO ₂ + 2Ca = Ti + 2CaO (3)

The raw material can also be reduced directly by adding calcium metal to the CaCI ₂ melt containing a raw material one wants to reduce, which the calcium metal subsequently does.

Several authors present similar processes for production of metals like titanium.

1. The Electronically Mediated Reaction (EMR) is described by Okabe and Sadoway in Materials Research, vol. 13, no. 12, 1998, pp. 3372-3377.

2. The Molten Salt Electrolysis (MSE) Process is described by Okabe, T.H. et al in J. Alloys and Compounds, 184, 1992, pp. 43-56.

3. The "OS Process" is described by Ono and Suzuki in JOM, Feb. 2002.

All of these processes are "metallothermic reduction processes". For processes 2 and 3 one produces the reducing agent (Ca metal) by deliberately applying a potential high enough to reduce the CaCl ₂ in the electrolyte to form Ca metal. This calcium metal then reduces the substrate "Calciothermically".

BHP Billiton is another of the companies working on the MSE process for Ti-production (WO 02/066711 A1 , WO 02/083989 A1 etc.,), but their approach is very similar to the FFC- Cambridge approach.

All these process has CaCI ₂ -∞ntaining electrolyte in common. The only anode material that has been mentioned specifically in these patents is carbon-based. However, an "inert anode" has been mentioned (WO 03/048399 A2), without being specified.

In general the anodic reaction (see Figure 2) on carbon will be

C+ x "O ² " = CO _x + 2x e ^' (4)

In other words, this reaction implies a net consumption of carbon.

The cathodic reaction of a metal "Me" can simply be explained by the following equation, here explained by reduction of an oxide:

MeO _2x + 2 e ^4x" = Me + x "O ^2" " (5)

Carbonate formation

Carbon is a candidate material as anodes for the titanium production by all the electrolysis processes mentioned above. However, carbon cause some undesired effects in the cell. Most important, formation of floating carbon sludge and carbon deposition on the cathode has been observed repeatedly. It is believed that this is due to formation of dissolved carbonate from a reaction between CO ₂ formed at the anode and O ^2' in the electrolyte.

It is known that considerable amounts of CO ₂ can dissolve in CaCI ₂ -CaO melts (Maeda and McLean, Transactions of the ISS, September 1986, p 61-65), and the solubility of calcium carbonate increases with increasing CaO activity and reduced temperature. The reaction is simply:

CaO + CO ₂ = CaCO ₃ (6)

The equilibrium constant for this reaction is:

K = ^aCaCQ3 = 0.55 , at 900°C /2/ (7) ^a Ca0 ^' PcO ₂

It is reasonable to assume that the partial pressure of CO ₂ is close to 1 during electrolysis, unless CO production is favoured over CO ₂ production. The activity of CaO in the CaCI ₂ -CaO system at 900°C is shown in Figure 6.

Figure 6 shows that the activity coefficient in the CaO concentration range typically used during FFC production of Ti (0.5-5 mol%) is about 2.6. M. Maeda and A. McLean (Transactions of the ISS, September 1986, p 61-65) report that ^x cd ^ — = 0.6 at 900 ⁰ C (8) ^x O ² - Pco ₂

in the CaCI ₂ -CaO-CO ₂ system.

Assuming a partial pressure of CO ₂ equal 1 , the equilibrium CaCO ₃ concentration at 900°C then becomes:

XCaCO ₃ = 0- ⁶ - *CaO ( ⁹ )

The CaCO ₃ concentration in the electrolyte after some time of operation can therefore be expected to be comparable to that of CaO.

The carbonate ions can be reduced at the cathode according to the reaction

CO ₃ ² - + 4e = C + 3O ² - (10)

i.e. leading to carbon formation at the cathode, according to Ito, Shimada and Kawamura (Proc. Electrochem. Soc. 1992, p. 574). The oxide thus released can either react with the cathode (to e.g. calcium titanates) or be transported to the anode for further carbonate formation. In the latter case, a cyclic reaction is set up where oxide is cycled between the anode and the cathode, and where carbon is transported from the anode to the cathode (see Figure 3). This cycle will lead to significant losses in current efficiency (CE) with respect to Ti production. Current efficiency can drop to less than 5 percent. It is reasonable to assume that the higher the oxide concentration, the higher the carbon formation at the cathode and the higher the losses in CE. One should therefore strive to run the cell at the lowest possible CaO concentration without chlorine formation. Low CaO will also favour oxygen removal from the metal product.

When the conditions at the cathode are sufficiently reducing to form dissolved Ca, it is reasonable to assume that this may react with dissolved carbonate:

2Ca + CO ₃ ^2" = 2CaO + C + O ^2" (11)

Such a reaction will lead to carbon formation in the bulk of the electrolyte, and the carbon will float to the surface of the melt since its density is lower than the electrolyte. One to two percent carbon is enough to make the electrolyte sludgy, which might reduce the electrolyte circulation and the gas release from the electrolyte. In addition, the carbon will increase the risk of creep current and short-circuiting between the electrodes. Towards the end of a run, when the cathodic potential is close to the Ca potential, significant carbon formation has been observed, even at high anodic overpotentials.

Deposition of an insulating layer of CaTiO ₃ has been observed on carbon anodes when TiU ₂ is being reduced in a CaCI ₂ -CaO melt at 900 ⁰ C. The mechanism is caused by the CaO dependent solubility of CaTiO ₃ , or some solubility of reduced Ti-species, e.g. TiO, and since "CaO" is being consumed at the anode interphase, one might pass the saturation limit of CaTiO ₃ in this layer as the CaO activity decline towards the anode surface.

Chlorine formation is observed at high anodic potentials, but only when there is little CaO in the electrolyte. At CaO concentration of approximately 1 w%, no or little chlorine is detected, even when very high anodic potentials are applied. Is appears that oxygen is produced instead. Two mechanisms are proposed. Either chlorine is produced on the anode, and a subsequent reaction between Cl ₂ and CaO gives CaCI ₂ and O ₂ , or oxygen is produced directly on the anode deposit. In either case, a CaTiO ₃ deposit is assumed. The CaTiO ₃

hinders direct contact between the carbon and CaO in the electrolyte, preventing CO ₂ formation. The CaTiO ₃ anode is consequently a candidate for an inert, oxygen-producing anode.

BHB Billiton discusses how to reduce the problems related to carbon in the process in patent application WO 03/076692 A1 , using an oxygen ion conducting membrane (yttria stabilised zirconia) to separate the anolyte and catholyte, but it is currently believed that the chemical stability of this membrane in CaCI ₂ -based electrolytes is a problem. In other words, problems related to the carbon cannot be totally avoided without the finding s of a totally inert ion conducting membrane.

From the above, it should be clear that the use of carbon anodes in the FFC-Cambridge process or similar processes based on CaCI ₂ -CaO electrolytes creates substantial problems that need to be overcome in order to produce cheap metal and alloys like titanium and ferrotitanium.

The best solution is to remove carbon completely from the system, and replace the carbon anode with an inert anode material producing oxygen instead. This process will also be a much cleaner process, because production of the greenhouse gas CO ₂ is avoided. No process disturbance is caused by periodic replacements of worn carbon anodes, and controllable process conditions are possible to maintain since the anodes are dimensionally stable.

For the FFC-Cambridge and the OS processes, no proper inert anode material have been found in the open literature nor patented.

The aluminium industry has searched for an inert anode material since Hall invented the process in 1886. So far, the aluminium industry has not found the potential inert anode candidate materials attractive enough to replace their carbon anodes. The electrolyte used (cryolite), seems to be too corrosive for the anode candidates tested, but the work continues because of the enormous economical and environmental potentials the findings of a good inert anode material have.

For the FFC-Cambridge and the OS processes, no proper inert anode material has been found proposed in the open literature.

Further prior art literature:

1. Bak, T.; Nowotny, J.; Sorrell, C. C; Zhou, M. F.; Vance, E. R. "Charge transport in CaTiO ₃ : II. Thermoelectric power." Journal of Materials Science: Materials in Electronics, vol. 15, no. 10, 2004, pp. 645-650.

2. Bak, T.; Nowotny, J.; Sorrell, C. C; Zhou, M. F.; Vance, E. R. "Charge transport in CaTiO ₃ : 1. Electrical conductivity." Journal of Materials Science: Materials in Electronics, vol. 15, no. 10, 2004, pp. 635-644.

3. Bak, T.; Nowotny, J.; Rekas, M.; Sorrell, C. C; Vance, E. R. "Effect of Gd on electrical conductivity of CaTiO ₃ " Journal of the Australasian Ceramic Society, vol.34, no.1 , 1998, pp. 182-186.

4. Chinarro, E., Jurado, J. R., Figueireido, F. M., Frade, J. R., "Bulk and grain boundary conductivity of CaO.97Ti1 -xFexO3-d materials," Solid State Ionics, vol. 160, 2003, pp.161 -169.

5. Dunyushkina, L A.; Gorbunov, V. A.; Babkina, A. A.; Esina, N. O. "High- temperature electrical transport in Al-doped calcium and strontium titanates." Ionics, vol. 9, no.1 & 2, 2003, pp. 67-70.

6. FactSage 5.3.1, Thermfact Ltd., Montreal, Canada, available from http://www.factsaqe.com/.

7. Hwang, Soon Cheol; Choi, Gyeong Man. : "Nonstoichiometry and mixed conductivity of calcium zirconate." Solid State Ionics: The Science and Technology of Ions in Motion, Proceedings of the Asian Conference, 9th, Jeju Island, Republic of Korea, June 6.-11., 2004, pp. 381-388. ,

8. Longo, V.; Ricciardiello, F. Electrical conduction mechanism in solid solution calcium zirconate(IV)-strontium zirconate(IV). Fac. Ing., Univ. Degli Studi Trieste, Trieste, Italy. Universita degli Studi di Trieste, lstituto di Chimica Applicata e Industriale, [Pubblicazioni], vol. 89, 1979, p. 26.

9. Vashook, V.; Vasylechko, L; Knapp, M.; Ullmann, H.; Guth, U. "Lanthanum doped calcium titanates: synthesis, crystal structure, thermal expansion and transport properties." Journal of Alloys and Compounds, vol. 354, no.1 & 2, 2003, pp.13-23.

10. Iwahara, H.; Esaka, T.; Mangahara, T. "Mixed conduction and oxygen permeation in the substituted oxides for calcium titanate (CaTiO ₃ )." Journal of Applied Electrochemistry, vol. 18, no 2, 1988, pp. 173-177.

11. Xie, S.; Liu, W.; Wu, K.; Yang, P. H.; Meng, G. Y.; Chen, C. S. Oxygen permeation and electric conductivity study in CaFe0.2Ti0.8O3-d" Proceedings of the China International Conference on High-Performance Ceramics, 1st, Beijing, China, Oct. 31- Nov. 3, 1998, 1999, pp. 519-522.

12. Yang, Wei; Li, Guangqiang; Sui, Zhitong. "Coprecipitating synthesis and impedance study of CaZM -xl nxO3-D (x = 0 0.03 0.07 0.09 0.1)." Journal of Materials Science Letters, vol. 17, no. 3, 1998, pp. 241-243.

Present invention:

Possible candidate materials for an inert anode for use in CaCI ₂ -CaO based electrolytes, such as the FFC-process and the OS process, are identified. Chemical stability and cost of the anodes are emphasised.

General criteria for an inert anode in these processes can be summarised as follows:

1. Chemically inert to oxygen and chlorine

2. No (or at least low) dissolution rate in the electrolyte used

3. Sufficient electrical conductivity

4. Produce no species with high vapour pressure

5. Minor dissolution should not introduce detrimental effects to the cathode product

6. High melting point to sustain operation at elevated temperature without creep

7. High thermal and mechanical shock resistance . 8. Easy to connect to the electrical circuit

9. Acceptably low cost

A fundamental approach further described is based on the knowledge that oxides more basic than CaO are converted to chloride by the melt. More acidic oxides will form solid compounds with CaO or to some degree dissolve in the electrolyte by an oxide ion accepting mechanism. The solubility of candidate materials can be measured, or estimated based on compound melting points and assumptions about melt structure.

With the present invention it is possible to produce high-quality metals in a carbon-free electrolyte enabling a low-cost process which is both competitive with alternative process and for some metals and alloys represents a totally new possibility. Any person skilled in the art can optimise the process conditions and results described here further.

These and further advantages can be achieved in accordance with the invention as defined in the accompanying claims.

Table of figures:

The invention shall be further explained by examples and Figures where:

• Figure 1 is showing a periodic table with the elements being reduced by the FFC- Cambridge process,

• Figure 2 shows the simplified chemical reaction between the dissolved oxygen ions in the electrolyte and the anode carbon material,

• Figure 3 shows the undesired shuttle reaction with carbonate formation,

• Figure 4 shows the solubility of Ti in CaCI ₂ as a function of CaO concentration at 90O ⁰ C,

• Figure 5 shows the solubility of Ni in CaCI ₂ as a function of CaO concentration at 900 ⁰ C,

• Figure 6 shows activities and activity coefficients in the CaCI ₂ -CaO system at 900 ⁰ C,

• Figure 7 is showing a SEM picture of titanium metal being produced by the FFC process with carbon anodes,

• ^' Figure 8 shows solubility of Ni from a NiO tablet in a CaCI ₂ -NaCI-I wt% CaO electrolyte as a function of NaCI content at 900 ⁰ C.

Inert anode material 1: NiO

Predictions based on the melting temperature of the compound indicate that NiO should have a moderately high liquid activity at 900 ⁰ C. Since nickel oxide is more acidic than calcium oxide, formation of oxychloride species with a resulting increase in solubility with CaO-concentration is expected and shown in Figure 5, which shows solubility of NiO in CaCI ₂ as a function of CaO at 900°. Complexing between Ni-O-Cl will give a negative deviation from ideality. Being nobler than Ti the dissolved nickel species will continuously discharge at the cathode. The actual anode wear will depend on the mass transfer conditions, which has to be optimised by design.

Materials based on pure NiO can readily be prepared by oxidising Ni plates Ni/NiO cermets

Test with pre-oxidised NiO anodes and Ni-NiO anodes at 1000 ⁰ C in 10 hours in CaCI ₂ -based electrolytes at 900 ⁰ C show very little sign of corrosion after 20 hours of operation.

The prospects of a highly conductive ceramic NiO-anode are low, as only monovalent compounds increase conductivity significantly.

Inert anode material 2: CaTiOsJn the CaO-TiOg system

CaTiO ₃ is deposited at the surface of the anode material during reduction of TiO ₂ in a CaCI ₂ -

CaO electrolyte, and is therefore an interesting inert anode candidate.

Three compounds exist in the CaO-TiO ₂ system, Ca ₃ Ti ₂ O ₇ , Ca ₅ Ti ₄ O ₁₃ and CaTiO ₃ (thermodynamic data exists for all three.) In contact with the electrolyte, CaTiO ₃ is the stable phase in the range 6-10 ^'4 < X _Ca o < 0.046, Ca ₅ Ti ₄ Oi ₃ exists in the range 0.046 < X _CaO < 0.106, and Ca3Ti2O7 is stable at X _Ca o > 0.106 (1173 K). This means that only CaTiO ₃ needs to be considered in relation with the FFC-Cambridge and the EMR/OS process.

FactSage reports the following fusion data for CaTiO ₃ :

δH° _fus = 106.64 kJ δS° _fus = 47.76 J/K T _fus = 2233 K

This gives a liquid activity of CaTiO ₃ at 1173 K of 6-10 ³ . Initial measurements show that the solubility of TiO ₂ increase with CaO-content and at 1 wt% CaO a solubility of approximately 1200 ppm Ti is measured (see Figure 4). Whether this solubility is acceptable from an economic standpoint is not considered here. However, as this compound does not introduce any new elements to the system it is attractive as a non-contaminating material and is an interesting inert anode material. By introducing for example NaCI to the electrolyte, the solubility of CaTiO ₃ is found to decrease, so the electrolyte composition can be optimised further to reduce the solubility even more by someone skilled in the art.

Pure CaTiO ₃ has low electrical conductivity, and a doping agent is required in order to make this oxide sufficiently conducting. In oxidizing atmosphere, CaTiO ₃ is an extrinsic conductor (conductivity determined by amount and type of impurities). Table 1 shows reported conductivities in doped and undoped CaTiO ₃ materials.

W

11

Table 1 : Reported conductivities in doped and undoped CaTiO ₃ materials in air/O ₂ at 900 ⁰ C

Doping Amount of Conductivity Reference element M substitution (S/cm)

[cation fractionl none 0 0.003 [Balachandran et al, 1982], none 0 1.710 ^'5 [Bak, T, 1998]

Gd 0.15 0.0001 [Bak, T, 1998]

La 0.1 0.002 [Vashook, 2003]

Al 0.25 0.012 Dunyushkina, L.A, 2003

Fe 0.01 0.0034 [Chinarro et al, 2003]

Fe 0.15 0.07 [Chinarro et al, 2003]

Fe 0.2 0.08 [Xie,S, 1999]

Fe 0.2-0.5 0.1- 0.2 |ϊwahara1988l

The electrical conductivity of a nominally pure sample in oxygen at 900°C was reported to be 0.003 S/cm [Balachandran et al, 1982]. Substitution of 15-50% Fe on the B-site is reported to increase the conductivity to maximum 0.2 S/cm. Other doping agents listed in the table have not shown to give higher electrical conductivities. The low electrical conductivity makes it necessary to have a short conducting path with an additional conducting phase.

CaTiO ₃ is also a good oxygen ion conductor [Bak. T], which has probably influenced some of the reported values on electronic conductivity as the contribution from ionic and electronic conductivity are not separable in a standard measurement. For instance the effect of Al- doping is hard to explain as other than an effect on the ionic conductivity.

The ionic conductivity also means that the CaTiO ₃ could operate as an extension of the electrolyte so that oxidation took place not only at the surface but also in the bulk of the material. This will limit the choice of conducting phase to probably only noble metals. Though noble metals would withstand the oxygen there is a second requirement of absolutely no electrolyte penetration due to a potential reaction with chloride.

Because of the low electrical conductivity of CaTiO ₃ , a doping agent has to be introduced in order to increase the conductivity. CaTiO ₃ is a perovskite structure. Electronic conductivity of CaTiO ₃ might be increased by additions of transitions metal oxides having a d-electron configuration different from d° or d ¹⁰ . The following doping candidates will add d-electrons to the structure: addition of transition metal fraction up to 30% Mn ⁴⁺ (d ³ ), Fe ³⁺ (d ⁵ ), Co ²⁺ (d ⁷ ) and Ni ²⁺ (d ⁸ ) as oxides. The dopant will substitute Ti in the structure giving increased electronic conductivity of CaTiO ₃ .

Inert anode material 3: CaZrO ₃ in the CaO-ZrO? system

The system CaO-ZrO ₂ is quite similar to the CaO-TiO ₂ system. CaZrO ₃ has a higher melting point and a lower liquid activity of CaZrO ₃ (2-10 ^'4 at 1173 K), more than a decade lower than for CaTiO ₃ . No data on solubility is available. However, as CaZrO ₃ is even more temperature stable than CaTiO ₃ it can be expected to have a low solubility, too. Even though the zirconate could contaminate the product, it is considered a possible candidate due to its stability.

Table 2 shows reported conductivities in doped and undoped CaZrO ₃ materials. CaZrO ₃ have an intrinsic lower conductivity than CaTiO ₃ , and typically conductivity in pure CaZrO ₃ is reported in the range 10 ^'5 to 10 ^"4 S/cm. Doping with In have shown to increase the conductivity by nearly 2 decades. Nevertheless, the conductivity is probably too low for practical use. CaZrO ₃ is also an oxygen ion conductor and the same limitations apply as for CaTiO ₃ .

Table 2: Reported conductivities in doped and undoped CaZrO ₃ materials in air/O ₂ at 900 ⁰ C

Doping Amount of Conductivity Reference element M substitution (S/cm)

[cation fraction]

None. O 2E-5; Longo, F, 1979

None O 8E-4 Hwang, 2004

In 0.1 0.001 ^' WEI, 1998

For the same reasons given for CaTiO ₃ , addition of Mn ⁴⁺ (d ³ ), Fe ³⁺ (d ⁵ ), Co ²⁺ (d ⁷ ) and Ni :2+ (d ⁸ ), added as oxides, will increase the electrical conductivity of CaZrO ₃ .

Inert anode material 4: CaSnOq in the CaO-SnO ₉ system

SnO ₂ anodes are extensively used in the glass industry, and SnO ₂ was a promising candidate for an inert anode in the aluminium industry in the 80-ies and 90-ies. SnO ₂ anodes works OK in short-time experiments in CaCI ₂ -CaO and CaCI ₂ -NaCI-CaO melts, too. However, when CaO is present, as is the case in the described processes, CaSnO ₃ is formed at the SnO ₂ anode.

No literature data on the conductivity of CaSnO ₃ was found. Our own measurement shows that the electrical conductivity of the pure CaSnO ₃ compound is in the order 10 ^'4 S/cm at 900 ⁰ C. As comparison, commercial Sb-doped tin oxide anodes have typically a conductivity of 100 S/cm, six decades higher. CaSnO ₃ anodes need to be doped in order to work as an inert anode material.

However, the effect of the dopant on the electrical conductivity illustrates the problem that could arise from a solution/precipitation process at the surface. A precipitated layer of the stable compound would not get the designed level of dopant and must be expected to be non-conducting. Therefore, even a thin layer of pure CaSnO ₃ (or other CaO-based compounds) formed at surface of the SnO ₂ anode would lead to anode passiviation and has to be avoided by process optimisation.

No phase diagram has been found for a system containing CaSnO ₃ . However, CaSnO ₃ is reported as a stable phase along with Ca ₂ SnO ₄ . The free energy of formation of the perovskite CaSnO ₃ phase from its single oxides is reported to be -63 kJ at 900°C [Jacob and Chan, 1974]. This stabilisation is in between CaTiO ₃ (-89kJ) and CaZrO ₃ (-48 kJ). No data on solubility of Sn in CaCI ₂ -containing electrolyte is available.

For the same reasons given for CaTiO ₃ , addition of Mn ⁴⁺ (d ³ ), Fe ³⁺ (d ⁵ ), Co ²⁺ (d ⁷ ) and Ni ²⁺ (d ⁸ ), added as oxides, will increase the electrical conductivity of CaSnO ₃ .

Inert anode material 5: Ca ₂ Fe ₂ Os in the CaO-Fe ₂ O ₃ system

In this binary system there are two stable compounds, _: CaFe ₂ O ₄ and Ca ₂ Fe ₂ O ₅ . Initial measurements of Fe ₂ O ₃ solubility at 900 ⁰ C shows that the solubility increases with CaO- content and at 1 wt% CaO a solubility of approximately 1700 ppm Fe is measured. Again, the solubility could be too high for economical reasons, due to wear and need for metal cleaning of the product. But as Fe might not be a severe contaminant, especially when FeTi is to be produced, it is a very interesting material. As explained before, a better electrolyte composition with for example CaCI ₂ and NaCI may reduce solubility significantly. Reduced temperature is also expected to lower the solubility of these oxides.

For the same reasons given for CaTiO ₃ , addition of Mn ⁴⁺ (d ³ ), Fe ³⁺ (d ⁵ ), Co ²⁺ (d ⁷ ) and Ni ;2+ (d ⁸ ) as oxides will increase the electrical conductivity of Ca ₂ Fe ₂ O ₅ as oxides.

Summarised, the selected candidate anode materials are based on NiO, CaTiO ₃ , CaZrO ₃ , CaSnO ₃ and Ca ₂ Fe ₂ O ₅ or a mixture thereof. The anode can be a pure oxide, a mixture of the said oxide(s) and a metal (composite/cermet), or a layer of one of the said inert anode materials on an electron conductive substrate. For a person skilled in the art, it is obvious that the exact composition of the inert anode might deviate from the stoichiometric formulas given.

Examples:

Example 1 : Solubility measurements:

• A tablet of the oxide of interest, approx. 050 x 7mm, is rotated in the electrolyte

• Approx. 1 kg pre-dried CaCI ₂ is heated to 900 ⁰ C in a crucible (alumina or Pt)

• The tablet is weighed and added in the afternoon and several melt samples are taken

• The tablet is removed the next morning and new melt samples taken

• The tablet is washed (boiled in acid) and weighed

• CaO is added to the melt

• The tablet is re-immersed in the melt

• Procedure is repeated until CaO saturation

• Melt samples are analysed for CaO and metals of interest (pH titration, ICP...)

As can be seen from Figures 4 and 5, the solubility of the Ti oxides is low at low CaO activity, but increases with increasing activity of CaO in the electrolyte. The solubility of Ni has a minimum at about 2 - 3 weight % CaO. This means that the process has to be operated at low CaO activity to protect the anode material. The current/voltage applied have to be optimised to prevent chlorine formation on the anode.

Example 2: CaCIp-NaCI-CaO electrolyte

Experiments performed in an electrolyte comprising CaCI ₂ -NaCI-CaO gives a lower solubility (see Figure 8) of Ni than experiments in a pure CaCI-CaO electrolyte. Less formation of CaTiOs at the carbon anode was also observed in CaCI ₂ -NaCI-CaO electrolytes than in CaCI ₂ -CaO electrolytes, indicating lower solubility of Ti, too. The electrolyte composition can be optimised further to give low solubility and thereby long-lived anodes and high purity products. The effect of NaCI in CaCI ₂ is also expected for the other inert anode candidates.

Example 3: NiO anode

Plates of oxidised Ni performed well as an inert anode for 24 hours in a CaCI ₂ -40% NaCI- CaO electrolyte at 900°C. After the experiment the nickel metal was covered with a layer of green NiO acting as an inert anode producing mainly oxygen. A cermet made of a mixture of Ni metal and NiO also showed promising behaviour as an inert anode with only some signs of corrosion after 24 hours.

Example 4: CaTiO ₃ anode

Experiments with carbon anodes in an CaCI ₂ -CaO electrolyte at 900 ⁰ C containing dissolved "Ti species" from the raw material TiO ₂ , which formed a thick layer of CaTiO ₃ at the surface of the carbon anode after a couple of hours, acting as an active anode.

An anode made of CaZrOa is expected to behave similar to CaTiO ₃ because of mutual chemical properties.

Example 5: CaSnOg anode

An anode made of SnO ₂ was tested in CaCI ₂ -40% NaCI with about 1.4 wt% CaO at 900 ⁰ C. As expected a thin layer at the surface of the SnO ₂ anode was converted to CaSnO ₃ , which acted as the inert anode. The experiment was terminated after 24 hours.

Example 6: LiCI electrolyte, titanium metal production

TiO ₂ powder was pressed into a cavity electrode and made the cathode in a three-electrode configuration with a Ni/NiO reference electrode and an anode. An electrolyte of LiCI was used at 850 ⁰ C and the TiO ₂ was held for 10 minutes at -2 V versus the reference electrode. Scanning electron microscopy confirmed the production of Ti metal.

Example 7: LiCI electrolyte, titanium alloy production

A blended oxide precursor of TiO ₂ and IVIoO ₂ was produced by conventional ceramic processing to give a porous pellet that was made the cathode in an electrolysis cell containing LiCI electrolyte at 850 ⁰ C. The cell voltage was increased to 3.3 V and held for 24 hours. Scanning electron microscopy confirmed the production of a Ti-Mo alloy.