The present invention relates to a method for determining the extent of electrochemical extraction of a metal (M) from a metal (M) oxide caused by a voltage applied between a cathode comprising (or consisting essentially of) or in contact with the metal (M) oxide and an inert metal alloy anode in an oxygen-dissolving molten electrolyte.

The extent to which electrochemical extraction of a metal (M) from a metal (M) oxide in an oxygen-dissolving molten electrolyte extraction has occurred is frequently measured after the completion of the reaction. This may be by (for example) microstructural analysis (eg X-ray diffraction) or a weight loss technique. Neither of these techniques is equipped to provide an instant assessment of the extent to which extraction has occurred whilst extraction is ongoing.

The conditions under which is carried out the electrochemical extraction of a metal (M) from a metal (M) oxide in an oxygen-dissolving molten electrolyte are harsh. Furthermore the extraction process may lead to the formation of aggressive or corrosive by-products typically under highly reducing conditions near to completion of the extraction process. These factors may preclude the use in situ of sensitive measuring equipment.

The present invention is based on the recognition that a current vs time plot usefully and reproducably characteristises the progress of an electrochemical extraction in an oxygen-dissolving molten electrolyte. In particular, the current vs time plot may be used to determine instantly and accurately the extent of extraction whilst extraction is ongoing.

Viewed from a first aspect the present invention provides a method for determining the extent of electrochemical extraction of a metal (M) from a metal (M) oxide caused by a voltage applied between a cathode comprising (or consisting essentially of) or in contact with the metal (M) oxide and an inert metal alloy anode in an oxygen-dissolving molten electrolyte, the method comprising: (a) measuring the current flow between the cathode and the inert metal alloy anode over a temporal range;

(b) relating a characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range to the extent of electrochemical extraction of the metal (M) from the metal (M) oxide.

By allowing the extent of electrochemical extraction to be determined instantly, the present invention makes it possible to exercise hitherto unattainable levels of control over electrochemical extraction in a molten oxygen-dissolving electrolyte.

Step (a) may be carried out discretely at intervals or continuously (eg to produce a current vs time plot). Preferably step (a) is carried out continuously.

Preferably the current flow between the cathode and the inert metal alloy anode over the temporal range includes a point of inflection. The point of inflection is followed by a steep rise in current promoted by the use of the inert alloy anode which is usefully exploited in quantitative measurements.

Preferably the current flow between the cathode and the inert metal alloy anode over the temporal range is substantially as illustrated in Figure 2 hereinafter.

In step (b), the characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be a qualitative characteristic.

The qualitative characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be a point of inflection. The qualitative characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be a current flow beyond a point of inflection (eg by a predetermined amount). The qualitative characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be that the current flow is increasing (eg by a predetermined amount). In step (b), the characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be a quantitative characteristic.

Preferably the quantitative characteristic is beyond a point of inflection of the current flow over the temporal range.

The use of a quantitative characteristic of the current flow beyond the point of inflection advantageously exploits a steep rise in current promoted by the use of the inert alloy anode. The quantitative characteristic is highly time-dependent beyond the point of inflection and accurate measurement of the current flow leads to precise determination of the extent of electrochemical extraction of the metal (M) from the metal (M) oxide. For example, a rate of extraction (%) of metal (M) from metal (M) oxide may be determined instantly and with high precision.

Preferably in step (b) the quantitative characteristic of the current flow is the measured current beyond a point of inflection of the current flow over the temporal range. Particularly preferably step (b) is: relating the measured current to the rate of extraction (%) of the metal (M) from the metal (M) oxide. Alternatively step (b) may be: relating the measured current to the oxygen content of metal (M).

Preferably the quantitative characteristic of the current flow is a threshold current beyond a point of inflection of the current flow over the temporal range.

Particularly preferably the method further comprises: (c) ceasing electrochemical reduction in response to the attainment of the threshold current.

Particularly preferably step (b) comprises: relating the threshold current to an extent of extraction beyond which is the onset of the formation of undesirable byproducts (eg gaseous by-products which may be noxious or environmentally undesirable such as chlorine). Particularly preferably step (b) comprises: relating the threshold current to an extent of extraction beyond which is the onset of corrosive conditions (eg anode corrosive conditions).

Particularly preferably step (b) comprises: relating the threshold current to a target rate of extraction (%) of metal (M) from metal (M) oxide.

The target rate of extraction is typically 99% or more, preferably 99.5% or more.

Particularly preferably step (b) comprises: relating the threshold current to a target oxygen content of metal (M). The target oxygen content is typically less than 2500ppm O ₂ by weight of metal (M), preferably less than 1500ppm O ₂ by weight of metal (M).

In step (b), the characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be related to the extent of electrochemical extraction of the metal (M) from the metal (M) oxide qualitatively.

The qualitative extent of electrochemical extraction of the metal (M) from the metal (M) oxide may be an extent of extraction beyond which is the onset of the formation of undesirable by-products or the onset of corrosive conditions.

In step (b), the characteristic of the current flow between the cathode and the inert metal alloy anode over the temporal range may be related to the extent of electrochemical extraction of the metal (M) from the metal (M) oxide quantitatively.

The quantitative extent of electrochemical extraction of the metal (M) from the metal (M) oxide may be a rate of extraction (%) of metal (M) from metal (M) oxide, the oxygen content of metal (M), a target rate of extraction (%) of metal (M) from metal (M) oxide or a target oxygen content of metal (M). In a preferred embodiment, the metal (M) oxide is TiO ₂ , the molten electrolyte contains CaCl ₂ and the electrochemical extraction is carried out in the presence of an alkali metal (M ^a ) oxide.

The method of the invention is carried out at an elevated temperature typically in the range 600-1000 ⁰ C, preferably 850-1000 ⁰ C (eg about 900 ⁰ C).

In the method of the invention, the temporal range may be less than 20 hours, preferably less than 10 hours (eg 8 hours), particularly preferably less than 4 hours.

The voltage is typically less than the discharge potential of metals in the molten electrolyte. For example, the voltage may be less than 3.5V (eg about 3.1V).

In a preferred embodiment, the method of the invention is carried out in an oxygen deficient atmosphere (eg an inert atmosphere such as argon).

The electrochemical cell of which the cathode and one or more inert metal alloy anodes are a part may be calibrated straightforwardly to obtain a relationship between a characteristic of the current flow between the cathode and the inert metal alloy anode and the extent of electrochemical extraction of the metal (M) from the metal (M) oxide. For calibration purposes, the extent of electrochemical extraction may be measured by conventional techniques such as microstructural analysis or by measuring weight loss.

In a preferred embodiment, the electrochemical extraction is carried out in the presence of an alkali metal (M ^a ) oxide. The alkali metal (M ^a ) oxide may be a caesium, rubidium, lithium, sodium or potassium oxide. Preferably the alkali metal (M ^a ) oxide is lithium, sodium or potassium oxide. Particularly preferably the alkali metal (M ^a ) oxide is potassium oxide.

The alkali metal (M ^a ) oxide may be an additive or may be formed in situ by decomposition of a decomposable alkali metal (M ^a ) salt into the alkali metal (M ^a ) oxide. Preferably the alkali metal (M ^a ) oxide forms the alkali metal (M ^a ) metallate (M) phase from a reaction of the alkali metal (M ^a ) oxide with a metal (M") metallate (M) phase. Particularly preferably the metal (M") metallate (M) phase is a solid phase. Particularly preferably the metal (M") metallate (M) phase is a perovskite (or perovskite-type) phase. Preferably M" is an alkaline earth metal, particularly preferably Ca, Sr or Ba, most preferably Ca.

Preferably the diffusivity of oxygen in the alkali metal (M ^a ) metallate (M) phase is higher than the diffusivity of oxygen in the metal (M") metallate (M) phase.

Preferably the alkali metal (M ^a ) metallate (M) phase is a liquid. Preferably the alkali metal (M ^a ) metallate (M) phase is a transitional phase.

In a preferred embodiment, the alkali metal (M ^a ) oxide is an additive. Preferably the alkali metal (M ^a ) oxide is in admixture with the metal (M) oxide in (or in contact with) the cathode.

The alkali metal (M ^a ) oxide and metal (M) oxide may form a self-supporting mixture (eg a pellet, slab, sheet, wire, foil, basket or tube). The self-supporting mixture may be the cathode or may be contactable with the cathode.

The alkali metal (M ^a ) oxide may be present in the self-supporting mixture in an amount in excess of a trace amount, preferably in excess of 5 wt%, particularly preferably in excess of 10wt%, more preferably in excess of 20wt%. Preferably the alkali metal (M ^a ) oxide is present in the self-supporting mixture in an amount in the range 10-70wt%, particularly preferably 20-50wt%.

In a preferred embodiment, the alkali metal (M ^a ) oxide is formed in situ by decomposition of a decomposable alkali metal (M ^a ) salt. The decomposable alkali metal (M ^a ) salt may be thermally decomposable.

Preferably the decomposable alkali metal (M ^a ) salt is in admixture with the metal (M) oxide in (or in contact with) the cathode. Particularly preferably the mixture of decomposable alkali metal (M ^a ) salt and metal (M) oxide is a self- supporting mixture (eg a pellet, slab, sheet, wire, basket, foil or tube). The decomposable alkali metal (M ^a ) salt may be present in the self-supporting mixture in an amount in excess of a trace amount, preferably in excess of 5wt%, particularly preferably in excess of 10wt%, more preferably in excess of 20wt%. Preferably the decomposable alkali metal (M ^a ) salt is present in the self-supporting mixture in an amount in the range 10-70wt%, particularly preferably 20-50wt%.

Preferably the decomposable alkali metal (M ^a ) salt is decomposable into one or more gaseous species. The gaseous species may be selected from the group consisting of water and carbon dioxide. Decomposition of the alkali metal (M ^a ) salt into one or more gaseous species may advantageously promote electrochemical reduction by forming porosity within the cathode. Continuous formation of pores permits fast transport of molten electrolyte species (eg CaO and CaCl ₂ ) which accelerates chemical reduction.

The decomposable alkali metal (M ^a ) salt may be an alkali metal (M ^a ) halide, carbonate, bicarbonate, hydrogen sulphide, hydrogen sulphate, nitrate, chlorate or sulphate. Preferably the decomposable alkali metal (M ^a ) salt is an alkali metal (M ^a ) bicarbonate.

The decomposable alkali metal (M ^a ) salt may be a caesium, rubidium, lithium, sodium or potassium salt. Preferably the decomposable alkali metal (M ^a ) salt is a lithium, sodium or potassium salt. Particularly preferably the decomposable alkali metal (M ^a ) salt is a potassium salt.

In a preferred embodiment, the decomposable alkali metal (M ^a ) salt may be present with an amount of endogenous hydroxide ions.

In a preferred embodiment, the decomposable alkali metal (M ^a ) salt may be present with an amount of exogenous hydroxide ions. Preferably the exogenous hydroxide ions are provided by an alkaline additive. The alkaline additive may be an alkali metal hydroxide (such as lithium, sodium or potassium hydroxide), an alkali metal hydride (such as lithium, sodium or potassium hydride) or an alkaline earth metal hydroxide. The alkaline additive may be added to the oxygen-dissolving molten electrolyte. The metal (M) may be a reactive metal element, semi-metal element, metal alloy or metalloid element.

In a preferred embodiment, the metal (M) forms a solid perovskite (or perovskite-type) phase in the molten electrolyte. The solid perovskite phase may be an alkaline earth metal (eg Ca) metallate (M) phase.

The metal (M) may be one or more metals selected from the group consisting of group HA metals, group IIIA metals, group IVA metals, group B transition metals, rare earth metals and alloys thereof. Preferably the metal (M) is one or more metals selected from the group consisting of Mg, Al, Si, Ge, group IVB transition metals, group VB transition metals, group VIB transition metals, group VIIB transition metals, group VIIIB transition metals, lanthanides, actinides and alloys thereof. Particularly preferably the metal (M) is one or more metals selected from the group consisting of group IVB transition metals, group VB transition metals, group VIB transition metals, group VIIIB transition metals, actinides and alloys thereof. Especially preferably the metal (M) is one or more metals selected from the group consisting of Ti, Nb, Ta, U, Th, Cr, Fe, steel and Zr. More especially preferred is one or more metals selected from the group consisting of Ti, Nb, Ta and Zr. Most preferred is Ti.

During electrochemical extraction, Ti advantageously forms sub-oxides (eg Magneli phases, TiO and Ti metal) which contribute to a sharp rise in current beyond a point of inflection.

The alkali metal (M ^a ) metallate (M) phase may be M ^a ₂ MO ₃ or M ^a ₄ MO ₄ . Preferred is M ^a ₄ MO ₄ . For example, where M is titanium, the preferred phase is M ^a ₄ Ti0 ₄ .

The metal (M) oxide may be the cathode or the metal (M) oxide in admixture with either the alkali metal (M ^a ) oxide or the alkali metal (M ^a ) salt decomposable into the alkali metal (M ^a ) oxide may be the cathode. Preferably the metal (M) oxide in admixture with either the alkali metal (M ^a ) oxide or the alkali metal (M ^a ) salt decomposable into the alkali metal (M ^a ) oxide is the cathode.

Alternatively the metal (M) oxide may be in contact with a cathode. In this embodiment, the metal (M) oxide may be in admixture with the alkali metal (M ^a ) oxide or the alkali metal (M ^a ) salt decomposable into the alkali metal (M ^a ) oxide. Alternatively the metal (M) oxide may be in the electrolyte in contact with the cathode. The alkali metal (M ^a ) oxide or the alkali metal (M ^a ) salt decomposable into the alkali metal (M ^a ) oxide may be in the electrolyte in contact with the cathode. The cathode may be a metal substrate such as steel which may be in the form of a cathodic bath, crucible or basket.

The oxygen-dissolving molten electrolyte may be (or contain) a compound of an alkali metal (eg Li, K or Cs), alkaline earth metal (eg Mg, Ca, Sr or Ba), Zn, Al or Y (or a mixture thereof). Preferably the oxygen-dissolving molten electrolyte contains a compound of Ca.

The oxygen-dissolving molten electrolyte may be (or contain) a hydrogen phosphate, dihydrogen phosphate or halide. Preferred is a halide (eg a chloride or fluoride), particularly preferably a chloride. The oxygen-dissolving molten electrolyte may be CaCl ₂ -containing or cryolite.

Preferably the molten electrolyte contains (eg consists essentially of) CaCl ₂ . Particularly preferably the molten electrolyte contains CaCl ₂ and an alkali metal halide (preferably a chloride). Preferred is a mixture of CaCl ₂ and KCl or of CaCl ₂ and LiCl.

Preferred is an inert metal alloy anode which is substantially unreactive with oxygen. Preferred is an inert metal alloy anode which is substantially insoluble in the molten electrolyte.

Preferably the anode is composed of an Al-E-Cu based alloy comprising an intermetallic phase of formula: Al _x EyCu ₂

wherein:

E denotes one or more metallic elements; x is an integer in the range 1 to 5; y is an integer being 1 or 2; and z is an integer being 1 or 2.

The Al-E-Cu based alloy may be substantially monophasic or multiphasic. Preferably the intermetallic phase is present in the Al-E-Cu based alloy in an amount of 50wt% or more (eg in the range 50 to 99wt%). Preferably the Al-E-Cu based alloy further comprises an ordered high-temperature intermetallic phase of E with aluminium, particularly preferably Al ₃ E. Other intermetallic phases may be present.

In a preferred embodiment, the Al-E-Cu based alloy is substantially free of CuAl ₂ . This is advantageous because CuAl ₂ has a tendency to melt at the elevated temperatures which are deployed typically in the method of the invention. Preferably CuAl ₂ is complexed.

In a preferred embodiment, the Al-E-Cu based alloy falls other than on the E poor side of the tie line joining Al ₃ E and ECu ₄ (eg on the E rich side of the tie line joining Al ₃ E and ECu ₄ ).

In a preferred embodiment, the Al-E-Cu based alloy comprises an intermetallic phase falling on or near to the tie line joining Al ₃ E and ECu ₄ .

In a preferred embodiment, the Al-E-Cu based alloy falls other than on the E poor side of the tie line joining Al ₃ E and AlECu ₂ (eg on the E rich side of the tie line joining Al ₃ E and AlECu ₂ ).

In a preferred embodiment, the Al-E-Cu based alloy comprises an intermetallic phase falling on or near to the tie line joining Al ₃ E and AlECu ₂ . In a preferred embodiment, the Al-E-Cu based alloy falls other than on the E poor side of the ξ, Al ₅ E ₂ Cu, EAlCu ₂ and β-ECu ₄ phase tie line (wherein ξ is a phase falling between Al ₃ Ti and Al ₂ Ti with 3 at% or less of Cu (eg 2-3 at% Cu)).

In a preferred embodiment, the Al-E-Cu based alloy comprises an intermetallic phase falling on or near to the ξ, Al ₅ E ₂ Cu, EAlCu ₂ and β-ECu ₄ phase tie line.

Preferably the intermetallic phase is Al ₅ E ₂ Cu. Particularly preferably the Al-E- Cu based alloy further comprises Al ₃ E.

Preferably the intermetallic phase is EAlCu ₂ . Particularly preferably the Al-E- Cu based alloy further comprises P-ECu ₄

The anode may be composed of a homogenous, partially homogenous or non- homogeneous Al-E-Cu based alloy.

Typically E has a potential in the anode which is lower than it would be in the molten electrode.

In a preferred embodiment, the anode develops a passivating layer. Preferably the passivating layer withstands oxidation in anodic conditions.

In a preferred embodiment, E is a single metallic element. The single metallic element is preferably Ti.

In an alternative preferred embodiment, E is a plurality (eg two, three, four, five, six or seven) of metallic elements. In this embodiment, a first metallic element is preferably Ti. Typically the first metallic element of the plurality of metallic elements is present in a substantially higher ambunt than the other metallic elements of the plurality of metallic elements. Each of the other metallic elements may be present in a trace amount. Each of the other metallic elements may be a dopant. Each of the other metallic elements may substitute Al, Cu or the first metallic element. The presence of the other metallic elements may improve the high-temperature stability of the alloy (eg from 1200 ⁰ C to 1400 ⁰ C).

In a preferred embodiment, E is a pair of metallic elements. In this embodiment, a first metallic element is preferably Ti. Typically the first metallic element of the pair of metallic elements is present in a substantially higher amount than a second metallic element of the pair of metallic elements (eg in a weight ratio of about 9:1). The second metallic element may be present in a trace amount. The second metallic element may be a dopant. The second metallic element may substitute Al, Cu or the first metallic element. The presence of a second metallic element may improve the high-temperature stability of the alloy (eg from 1200 ⁰ C to 1400 ⁰ C).

Preferably the pair of metallic elements has similar atomic radii. Preferably the atomic radius of the second metallic element is similar to the atomic radius of Cu. Preferably the atomic radius of the second metallic element is similar to the atomic radius of Al.

In a preferred embodiment, E is one or more of the group consisting of group B transition metal elements (eg first row group B transition metal elements) and lanthanide elements. Preferably E is one or more group IVB, VB, VIB, VIIB or VIIIB transition metal elements, particularly preferably one or more group IVB, VIIB or VIIIB transition metal elements.

In a preferred embodiment, E is one or more metallic elements of valency II, III, IV or V, preferably II, III or IV.

In a preferred embodiment, E is one or more metallic elements selected from the group consisting of Ru, Ti, Zr, Cr, Nb, V, Co, Ta, Fe, Ni, La and Mn. In a particularly preferred embodiment, E is one or more metallic elements selected from the group consisting of Ti, Fe, Cr and Ni.

Preferably E is or includes a metallic element capable of reducing the tendency of CuAl ₂ towards grain boundary segregation at an elevated temperature. In this embodiment, the metallic element capable of reducing the tendency of CuAl ₂ towards grain boundary segregation at an elevated temperature may be the second metallic element of a plurality (eg a pair) of metallic elements. Particularly preferably E is or includes a metallic element capable of forming a complex with CuAl ₂ . Preferred metallic elements for this purpose are selected from the group consisting of Fe, Ni and Cr, particularly preferably Ni and Fe, especially preferably Ni.

Preferably E is or includes a metallic element capable of reducing the tendency of the first metallic element or Cu to dissolve in molten extractant. In this embodiment, the metallic element may be the second metallic element of a plurality (eg a pair) of metallic elements. Preferred metallic elements for this purpose are selected from the group consisting of Fe, Ni, Co, Mn and Cr, particularly preferably the group consisting of Fe and Ni (optionally together with Cr).

Preferably E is or includes a metallic element capable of promoting the passivation of the surface of the anode in the presence of a molten electrolyte. For this purpose, the metallic element may form or stabilise an oxide film. In this embodiment, the metallic element may be the second metallic element of a plurality (eg a pair) of metallic elements. Preferred metallic elements for this purpose are selected from the group consisting of Ru, Fe, Ni and Cr. Particularly preferably E is Ti, Fe, Ni and Cr in which the formation of a combination of oxides such as iron oxides, chromium oxides, nickel oxides and alumina advantageously promotes passivation.

Preferably E is or includes a metallic element selected from the group consisting of Zr, Nb and V. Particularly preferred is V or Nb. These second metallic elements are advantageously strong intermetallic formers. In this embodiment, the metallic element is the second metallic element of a plurality (eg a pair) of metallic elements.

Preferably E is or includes a metallic element capable of forming an ordered high-temperature intermetallic phase with aluminium metal. Particularly preferably E is or includes a metallic element capable of forming Al ₃ E. Preferably E is or includes Ti. A titanium containing alloy typically has electrical resistivity in the range 3 to 15 μohm cm at room temperature.

Preferably the intermetallic phase is Al ₅ Ti ₂ Cu. Particularly preferably the Al- Ti-Cu based alloy further comprises Al ₃ Ti.

Preferably the intermetallic phase is TiAlCu ₂ . Particularly preferably the Al- Ti-Cu based alloy further comprises P-TiCu _4.

In a preferred embodiment, E is or includes Ti and a second metallic element selected from the group consisting of Fe, Cr, Ni, V, La, Nb and Zr, preferably the group consisting of Fe, Cr and Ni. The second metallic element advantageously serves to enhance high-temperature stability of the Al-Ti-Cu phases.

The anode may be composed of an Al-E-Cu based alloy obtainable by processing a mixture of 35 atomic % Al or more (preferably 50 atomic % Al or more), 35 atomic % E or more (wherein E is a first metallic element as hereinbefore defined) and a balance of Cu and optionally E' (wherein E' is one or more of the additional metallic elements hereinbefore defined).

In a preferred embodiment, the anode is composed of an Al-E-Cu based alloy obtainable by processing a mixture of (65+x) atomic % Al, (20+y) atomic % E (wherein E is a first metallic element as hereinbefore defined) and (15-x-y) atomic % Cu, optionally together with z atomic % of E' (wherein E' is one or more of the additional metallic elements hereinbefore defined) wherein E' substitutes Cu, Al or E.

In this embodiment, the alloy may be obtainable by casting, preferably in an oxygen deficient atmosphere (eg an inert atmosphere). For example, a mixture may be melted in an argon-arc furnace under an atmosphere of argon gas and then solidified in an argon atmosphere. Alternatively in this embodiment, the alloy may be obtainable by flux-assisted melting, vacuum arc or vacuum melting using a resistance furnace. Contamination by O, C, N, S or P should be minimised. In a preferred embodiment, the anode is at least as conducting at elevated temperature (eg at 900 ⁰ C) as a carbon electrode. Preferably the anode is more conducting at elevated temperature (eg at 90O ⁰ C) than a carbon electrode.

It has been recognised that the dissociation of molten CaCl ₂ electrolyte into chlorine substantially coincides with the attainment of a desirable level of metal (M) extraction from a metal (M) oxide.

Viewed from a further aspect the present invention provides a method for determining the extent of electrochemical extraction of a metal (M) from a metal (M) oxide caused by a voltage applied between a cathode comprising (or consisting essentially of) or in contact with the metal (M) oxide and an inert metal alloy anode in molten CaCl ₂ , the method comprising:

(A) measuring the evolution of chlorine over a temporal range;

(B) relating the onset or level of chlorine evolution over the temporal range to the extent of electrochemical extraction of the metal (M) from the metal (M) oxide.

By allowing the extent of electrochemical extraction to be determined instantly and whilst the reaction is ongoing, the present invention makes it possible to exercise hitherto unattainable levels of control over electrochemical extraction in molten CaCl ₂ .

Step (A) may be carried out discretely at intervals or continuously (eg to produce a chlorine evolution vs time plot). Preferably step (A) is carried out continuously.

Particularly preferably step (B) is: relating the onset or level of chlorine evolution to the rate of extraction (%) of the metal (M) from the metal (M) oxide. Alternatively step (B) may be: relating the onset or level of chlorine evolution to the oxygen content of metal (M).

Preferably the method further comprises:

(C) ceasing electrochemical reduction in response to the attainment of a threshold level of chlorine evolution. Particularly preferably step (B) comprises: relating the threshold level of chlorine evolution to a target rate of extraction (%) of metal (M) from metal (M) oxide.

The target rate of extraction is typically 99% or more, preferably 99.5% or more.

Particularly preferably step (B) comprises: relating the threshold level of chlorine evolution to a target oxygen content of metal (M). The target oxygen content is typically less than 2500ppm O ₂ by weight of metal (M), preferably less than 1500ppm O ₂ by weight of metal (M).

The present invention will now be described in a non-limitative sense with reference to the Examples and accompanying Figures in which:

Figs Ia and IB: XRD of a TiO ₂ + KHCO ₃ pellet roasted for 1 hour and electrolysed for 0.5 hours (see Fig Ia) and 1 hour (see Fig Ib) in a molten bath Of CaCl ₂ -LiCl showing phases of Ti (ICDD 5-682), CaTiO ₃ (ICDD 42-423), CaTi ₂ O ₄ (ICDD 11-29) and TiO (ICDD 8-117); and

Fig 2: Current vs time graph measured according to the method of the invention at an applied voltage of 3. IV.

EXAMPLE 1

Method

Pellets were prepared by mixing l-2g of TiO ₂ with 0.2-0.5g of KHCO ₃ at different weight ratios. In each case, the mixture was heat treated for 1 hour at 1073K and pressed in a die at a pressure of 3643atm. A hole was drilled in the pellet with a 2mm drill bit. The pellet was suspended in a steel electrode which acted as a cathode with a molybdenum wire. An Al-Ti-Cu intermetallic anode was suspended on a steel electrode with a molybdenum wire. The two electrodes were connected to a power supply which was set to a constant voltage of 3.1V. Molten electrolytic mixtures of KCl-CaCl ₂ and LiCl-CaCl ₂ were prepared by taking 180gms of CaCl ₂ with 20gms of KCl and LiCl respectively. In each case, the mixture was transferred into a zircon crucible which was lowered into a furnace maintained at 32O ⁰ C. The mixture was heat treated for 24 hours and then transferred into an alumina crucible and heated to 800 ⁰ C at 0.5 ⁰ C per minute after which the temperature was raised to 92O ⁰ C at a rate of 2 ⁰ C per minute. During heating, argon gas was passed into the furnace at 500ml min ^"1 . Once the electrolyte was fully molten, the temperature of the furnace was lowered to 900 ⁰ C. The two electrodes were lowered into the furnace and a potential of 3.1V was applied using an Agilent 665 IA DC power supply. The experiments were carried out for a period of 8-24 hours.

Pellets were removed at intervals of 30 and 60 minutes of electrolysis and washed in water for 24 hours. The pellets were finely ground using a mortar and pestle for X-ray powder diffraction analysis. The diffraction was carried out using Cu- Ka as target at a scanning rate of 0.02° sec ^'1 .

Results

An increase in internal porosity was achieved readily in situ by the presence of KHCO ₃ in the TiO ₂ pellet. As KHCO ₃ decomposes, it produces potassium oxide, carbon dioxide and water. The liberated gaseous mixture of CO ₂ and H ₂ O increases the porosity in the pellet which enhances the contact surface area between CaCl ₂ and TiO ₂ and facilitates rapid cathodic dissociation of TiO ₂ .

Besides pore formation, a much more significant reaction takes place between K ₂ O and CaTiO ₃ . K ⁺ ions diffuse into the perovskite lattice which breaks the structure by forming more stable liquid potassium titanates as shown in equation [1] (ascertained from an equilibrium calculation performed using FACTSAGE see C. Bale et ai, FACTSAGE (Ecole Polytechnique CRCT, Montreal, Quebec Canada)). The calcium oxide formed in this reaction is dissolved in the molten salt bath until it reaches saturation:

CaTiO, + 2K ₂ O = K ₄ TiO ₄ + CaO ΔG = -334349.6 J mole ^"1 at T=900°C [1] As the diffusivity of O ^2" ions in the liquid phase is faster than in solid CaTiO ₃ , the reduction of K ₄ TiO ₄ to the Magneli phases through to Ti metal occurs rapidly as no major reorganisation of crystalline TiO ₂ is required. The Magneli phases (Ti ₄ O ₇ , Ti ₃ O ₅ ) all have a distorted rutile structure with a larger number of oxygen vacant sites. From a phase equilibrium analysis, it was established that the K ₄ TiO ₄ liquid phase can be in equilibrium with the Magneli phase and continue to shift the equilibrium with the progression of reduction to the metallic phase. The formation of liquid phase increases the reaction kinetics which is evident as Ti metal was observed within the first half an hour of electrolysis. The K ⁺ ions produced from the decomposition of potassium titanate reacts with the molten electrolyte and forms KCaCl _3.

By controlling the volume of the liquid phase of potassium titanate, the loss of Ti in the molten salt can be prevented. If the liquid phase drains out from the solid pellet into the CaCl ₂ bath, TiO ₂ is then irreversibly lost into the CaCl ₂ bath.

Figs Ia and IB are the XRD pattern of the pellet at 0.5 hours (see Fig Ia) and 1 hour (see Fig Ib) of electrolysis. Phases of Ti (ICDD 5-682), CaTiO ₃ (ICDD 42- 423), CaTi ₂ O ₄ (ICDD 11-29) and TiO (ICDD 8-117) are present. A comparison of Figs Ia and Ib shows that the perovskite peak is suppressed as perovskite is decomposed. After 20 hours of electrolysis, the XRD pattern (Fig 6) shows that titanium metal is present.

EXAMPLE 2

A number of experiments were carried out to change the ratio of potassium bicarbonate in the pellet in the range 10-50wt%. Experiments were also conducted on the two different types of molten salt containing CaCl ₂ -KCl and CaCl ₂ -LiCl mixtures at 90O ⁰ C with a constant voltage of 3.1V. Both processes yielded complete reduction Of TiO ₂ pellet to Ti metal. The residual concentration of oxygen dissolved in the Ti metal was determined by X-ray diffraction analysis (see M. Dechamps et ai, Scripta Metallurgica 11 (11), 941 (1977)) and was found to be 1350ppm by weight. A first experiment was carried out with a pellet containing 20wt% potassium bicarbonate in a CaCl ₂ -KCl bath for 8 hours. A Ti metal layer with a thickness of 500μm was formed beyond which there is a high concentration of calcium, titanium, potassium and chlorine.

When the concentration of potassium bicarbonate was increased from 20wt% to 50wt% and electrolysis was performed for 20 hours, it was found that a uniform microstructure of Ti metal was formed across the cross section of the pellet with the majority of the area being metallised. Furthermore when the salt bath was replaced by LiCl-CaCl ₂ and 50wt% of potassium bicarbonate was mixed with TiO ₂ and electrolysed for 20hours, it also led to full metallisation. The reduction in the two molten salts proves that the formation OfK ₄ TiO ₄ liquid phase is important for increasing reaction kinetics and is independent of the molten salt used. During electrolysis, all the experiments showed an increase in the current with the inert metallic anode which is in sharp contrast with previous observations (see C. Schwandt and D. J. Fray, Electrochimica Acta 51 (1), 66 (2005); M. Ma et al., Journal of Alloys and Compounds 420 (1-2), 37 (2006); and R. O. Suzuki et al, Metallurgical and Materials Transactions B-Process Metallurgy and Materials Processing Science 34 (3), 287 (2003)).

Current-Time Analysis

Figure 2 displays the current-time plot for the reaction of Example 2. Although a smooth curve was observed, there was oscillation in the current with a variation of ± 0. lamps during electrolysis. It can be seen from figure 2 that there was a decrease in the current for the first half hour to a point of inflection after which the current increased rapidly. Beyond two hours, there was a slow increase in current which plateaus at around 4.0amps.

The large initial current is due to the use of the inert anode which has high conductivity and decreases cell resistance. The initial decrease in current in figure 2 is due to the formation of a perovskite phase (verified from the X-ray diffraction analysis). The XRD data for half hour electrolysis showed the presence Of CaTiO ₃ , CaTi ₂ O ₄ , Ti ₃ O ₅ , TiO and Ti metal phases. After 4 hours, almost 95% of TiO ₂ was reduced to Ti. In previous experiments (Alexander et al, Acta Mater ialia 54 (11), 2933 (2006) and Schwandt [supra]), no titanium metal had been observed in the first 30 minutes of the process.

From the Ti-O phase diagram, it is known that Ti ₃ O ₅ can never be in equilibrium with Ti from which it is concluded that (at an early stage) two simultaneous reactions occur. The first reaction is the formation of CaTiO ₃ , CaTi ₂ O ₄ and Ti ₃ O ₅ which dominates the phase constitution. The second reaction is the decomposition Of K ₄ TiO ₄ to form TiO and Ti metal. Since the Magneli phases are more electrochemically conducting and the Ti metal is formed in the first hour of electrolysis, an increase in current is eminent which is what is seen in figure 2. The diffraction pattern after one hour of electrolysis showed small peaks of CaTiO ₃ and predominant peaks of Ti, CaTi ₂ O ₄ and Ti ₃ O ₅ . XRD does not show the presence of the potassium titanate phase because it is a transitional liquid phase during electrolysis.

The amount of Ti metal produced is verified by microstructural analysis and by measuring the weight loss after electrolysis (as previously demonstrated by G. Z. Chen et al, Metallurgical and Materials Transactions B-Process Metallurgy and Materials Processing Science 35 (2), 223 (2004) in the case of electro-reduction of Cr ₂ O ₃ in molten CaCl ₂ ). After electrolysis of Ig OfTiO ₂ pellet for 20 hours, the pellet was washed in water for 24 hours and the weight of the pellet was measured again and was found to be 0.605g. The theoretical amount of Ti produced from Ig of TiO ₂ is 0.6g which is within the error of experimental observation thus verifying complete metallisation.

These measurements may be used to calibrate the electrochemical cell to obtain a relationship between current measured temporally and the rate of conversion OfTiO ₂ to Ti metal. In this way, the current vs time plot of a new sample of TiO ₂ can be exploited to determine the rate of conversion (whilst the extraction is ongoing) or a desired end-point of the electrochemical reduction (typically Ti metal at about 99.5%) to a high degree of accuracy.

Similarly the electrochemical cell may be calibrated to obtain a relationship between current measured temporally and the onset of corrosive conditions or conditions under which chlorine is evolved from the molten electrolyte. In this way, the current vs time plot of a new sample of TiO ₂ can be exploited to determine when electrochemical reduction should be ceased to prevent corrosion and/or production of chlorine.