The present invention relates to a process for recovering a titanium dioxide product from a titanium oxide-containing roasted mass.

Iron oxides are present in many mineral ores such as mineral ores of chromium, titanium and copper. The removal of iron from these ores is essential for the extraction of the metal and its oxides (Sibum H (1997) Titanium Handbook of Extractive Metallurgy, Weinheim, Wiley- VCH). Various techniques for the removal of iron from the mineral ores have been exploited including reduction roasting, smelting and acid leaching in concentrated hydrochloric acid (see Gueguin M and Cardarelli F (2007) "Chemistry and mineralogy of titania-rich slags. Part 1 - Hemo- ilmenite, sulphate, and upgraded titania slags" Mineral Processing and Extractive Metallurgy Review 28(1): 1-58; Kahn J A (1984). "Non-Rutile Feedstocks for the Production of Titanium" Journal of Metals 36(7): 33-38; and Lasheen T A I (2005). "Chemical beneflciation ofRosetta ilmenite by direct reduction leaching" Hydrometallurgy 76: 123-129). However these techniques are not very efficient in removing iron compounds and generate toxic waste (Sibum [supra]; and Doan P (2003). Sustainable development in TiO2 industry: Challenges and Opportunities. TiO2 intertech Miami, Florida).

WO-A-2005/028369 discloses the recovery of titanium dioxide from a roasted mass using hot water to produce an insoluble residue in aqueous solution. Adding to the aqueous solution a source of alkalinity causes the selective separation of a fine precipitate from which may be recovered metal values including iron compounds. Metal values including iron compounds may also be recovered from the aqueous solution by acidification with a weak organic acid. The insoluble residue may be acid leached in an inorganic acid to remove zirconium, niobium, uranium and thorium compounds.

Iron compounds may be dissolved by treatment with organic acids such as oxalic acid. Studies have been performed on the dissolution of hematite in oxalic acid (see Panias D Taxiarchou M Douni I Paspaliaris I and Kontopoulos A (1996). "Dissolution of hematite in acidic oxalate solutions: The effect of ferrous ions addition" Hydrometallurgy 43(1-3): 219-230) and ascorbic acid (see Suter D, Banwart S and Stumm W (1991) "Dissolution of Hydrous Iron(Iii) Oxides by Reductive Mechanisms" Langmuir 7(4): 809-813; and Banwart S, Davies S and Stumm W (1989). "The Role of Oxalate in Accelerating the Reductive Dissolution of Hematite (Alpha-FejOi) by Ascorbate" Colloids and Surfaces 39(4): 303-309).

The present invention seeks to improve the recovery of a titanium dioxide product from a titanium oxide-containing roasted mass by exploiting an organic acid to effectively remove iron species (eg iron oxide) and alkali metal or alkaline earth metal species from the roasted titanium mineral by leaching.

Viewed from a first aspect the present invention provides a process for recovering a titanium dioxide product from a titanium oxide-containing roasted mass comprising:

(a) subjecting the titanium oxide-containing roasted mass or a substantially water-insoluble residue thereof to leaching in an organic acid solution to produce an acid leachate and the titanium dioxide product.

The titanium oxide-containing roasted mass is obtainable by roasting a titanium oxide-containing composition. The titanium oxide-containing composition is a mixture of metal oxide species in compound form or forms which include titania (TiO ₂ ). The titanium oxide-containing composition may be synthetic or (preferably) natural such as a powder, ore or mineral or a mixture thereof. The titanium oxide- containing composition may be a residue from a chlorination or sulphatation process.

A preferred titanium oxide-containing composition is a titanium rich material such as an ore (eg ilmenite, anatase, ilmenite beach sands, low grade titaniferrous slag, natural rutile or perovskite). The titanium oxide-containing composition may further include at least one iron species such as a ferrous or ferric species (preferably an iron oxide such as FeO, Fe ₂ O ₃ or Fe ₃ O ₄ ). The titaniferrous mixture may further comprise alumina or silica.

Preferably the titanium oxide-containing composition is an ore selected from the group consisting of ilmenite, anatase, perovskite and mixtures thereof. Step (a) may be preceded by: (aO) providing a titanium oxide-containing composition with one or more alkali salts to produce a charge; and

(aθl) oxidatively roasting the charge to produce the titanium oxide-containing roasted mass.

Step (aO) may further comprise providing the titanium oxide-containing composition with at least one of an alumina-containing additive and a calcium oxide-containing additive.

Preferably the charge is without an alumina-containing additive and a calcium oxide- containing additive.

Preferably the one or more alkali salts is one or more alkali metal or alkaline earth metal salts. Preferably the one or more alkali salts is one or more carbonates, hydroxides, bicarbonates or sulphates of a group IA or group HA metal or a mixture thereof. For example, the one or more alkali salts may be selected from the group consisting OfNa ₂ CO ₃ , K ₂ CO ₃ , Na ₂ SO ₄ , K ₂ SO ₄ , NaOH, NaHSO ₄ , KHSO ₄ , KHCO ₃ , NaHCO ₃ and KOH. The amount of alkali salt may be calculated based on the formation of alkali compounds of TiO ₂ , Fe ₂ O ₃ , Al ₂ O ₃ , SiO ₂ , and P ₂ O ₅ present in the composition {eg ore).

In a preferred embodiment, the one or more alkali salts is a potassium or sodium salt, particularly preferably a potassium salt. A potassium salt advantgeously promotes removal of iron.

The one or more alkali salts may be a carbonate (eg sodium or potassium carbonate).

In a preferred embodiment, the one or more alkali salts is an OH-containing salt. The presence of OH groups promotes the formation of ferrihydrite which facilitates leaching. Particularly preferably the one or more alkali salts is a hydroxide or bicarbonate, more preferably a hydroxide. Typically the organic acid solution is a solution of one or more weak organic acids. The weak organic acid is substantially non-dissociative in water. Typically the organic acid solution is without a mineral acid.

Preferably the organic acid solution is a reducing medium. A reducing medium advantageously promotes the removal of iron and alkali salt. Particularly preferably the organic acid solution is capable of reducing Fe(III) to Fe(II).

Preferably the organic acid solution is a solution of a carboxylic (eg mono or polycarboxylic), sulphonic or sugar acid (eg aldonic, uronic or aldaric acid) or a mixture thereof. Particularly preferably the organic acid solution is a solution of a carboxylic or sugar acid or a mixture thereof.

The carboxylic acid may be a saturated or unsaturated, acyclic (eg linear or branched) or cyclic (eg monocyclic or polycyclic) carboxylic acid which is optionally substituted (eg ring or chain substituted). The carboxylic acid may be aromatic or aliphatic. The carboxylic acid may be a monocarboxylic or polycarboxylic acid. Preferably the carboxylic acid is a dicarboxylic acid.

The sugar acid is preferably ascorbic acid.

The organic acid solution may be a solution of at least one of the group consisting of benzoic acid, ascorbic acid, formic acid, oxalic acid, citric acid and acetic acid.

In a preferred embodiment, the acid in the organic acid solution is a metal ligand or chelate. Particularly preferably the acid is a ligand or chelate of iron, an alkali metal or alkaline earth metal. More particularly preferably the acid is a ligand or chelate of iron.

In a preferred embodiment, the organic acid solution is a mixture of oxalic acid and ascorbic acid.

The concentration of oxalic acid may be in the range 0.1 to 0.3M, preferably in the range 0.15 to 0.25M, particularly preferably about 0.2M. The concentration of ascorbic acid may be in the range 7x10 ^"3 M to 8 xl0 ^"3 M, preferably in the range 7.4 xlO ^"3 M to 7.7 xlO ^"3 M, particularly preferably about 7.55xlO ^"3 M.

Preferably in step (a) the particles of the titanium oxide-containing roasted mass are in suspension in the organic acid solution. In this embodiment, the dissolution of iron is advantageously controlled by the chemical reaction occurring on the surface of the particles.

Preferably in step (a) the particles of the titanium oxide-containing roasted mass are maintained in suspension in the organic acid solution by stirring. Stirring is carried out by a stirrer operating typically at a rate in the range 100 to 2000rpm, preferably in the range 500 to 1750rpm, particularly preferably in the range 1000 to 1500 rpm. By way of example, the optimum condition for dissolution occurs typically when the solution is stirred by a stirrer operating at about 1300rpm.

Preferably in step (a) the organic acid solution is at a pH of 4 or less, preferably in the range 2.5 to 4, particularly preferably 3 to 3.5.

The process may further comprise:

(b) adding an alkaline solution to maintain the pH of the organic acid solution in the range 3 to 4.

In step (b), the alkaline solution may be sodium hydroxide.

Preferably in step (a) the organic acid solution is at an elevated temperature (eg 60- 9O ⁰ C), particularly preferably about 70 ⁰ C.

Step (a) may be carried out for a time in the range 5 to 300 minutes, preferably 100 to 250 minutes (eg about 4 hours). The titanium dioxide product may be separated from the acid leachate by a standard technique such as filtration. Preferably step (a) is carried out anaerobically. Typically step (a) is carried out in the presence of an inert gas. For example, argon gas may be passed through the organic acid solution. A convenient rate at which to pass through the argon gas is 300ml/min.

Step (a) may be preceded by:

(aO2) washing the roasted mass with an aqueous medium to produce an aqueous solution and the substantially insoluble residue of the roasted mass.

The aqueous medium may be water or an alkali solution (eg a dilute alkali solution). In step (aO2), water-soluble alkali compounds such as metal (eg sodium) aluminate, silicate, ferrite, chromate, vanadate and phosphate may be dissolved in the aqueous medium. Aqueous medium may be added repeatedly to wash the substantially insoluble residue (typically until the pH of the washings reaches about 7).

Preferably the aqueous solution supports a colloidal layer rich in rare-earth oxides.

Particularly preferably the process further comprises:

(c) isolating the colloidal layer from the aqueous solution; and

(d) recovering the rare-earth oxides from the colloidal layer.

The process may further comprise:

(c) washing the titanium dioxide product in an acid.

The titanium dioxide product may be washed in (for example) a mineral acid such as HCl, H ₂ SO ₄ , HNO ₃ or H ₃ PO ₄ (eg a mineral acid such as 5% HCl).

The titanium dioxide product is preferably in the form of synthetic rutile or synthetic anatase (or a mixture thereof). The process of the invention is capable of achieving TiO ₂ with a purity of 92wt% or more, preferably 94wt% or more, more preferably 97wt% or more.

The present invention will now be described in a non-limitative sense with reference to the Examples and accompanying Figures in which: Figure 1 : Eh-pH diagram of K-Ti-Fe-O-H at 333K;

Figure 2: Schematic diagram of the reactor assembly;

Figure 3a: Backscattered image of ilmenite grain after roasting with potassium carbonate followed by reduction leaching;

Figure 3b: EDX of the particle confirming the presence of titanium and oxygen;

Figure 4a: Microstructure showing particles after roasting with sodium carbonate followed by reduction leaching;

Figure 4b: EDX of the particle confirming the presence of titanium and oxygen;

Figure 5: Synthetic rutile and anatase obtained after the benefϊciation of ilmenite ore;

Figure 6: Eh-pH diagram of Na-Fe-Ti-O-H constructed using FACTSAGE thermodynamic software (Bale, C, Pelton, A. D., Thompson, W. D., Melancon, J. and

Eriksson, G. FACTSAGE. Ecole Polytechnique CRCT, Montreal, Quebec, Canada);

Figure 7: Microstructure showing very fine particle precipitation (BSE 30Ox);

Figure 8: Comparison of the change in pH of the reaction mixture against time when leaching is carried out in HCl and acetic acid solution;

Figure 9: Percentage removal of iron with different stirring speeds;

Figure 10: Percentage removal of iron with time when roasted ilmenite and anatase is aeration leached;

Figure 11 : Percentage removal of sodium with time when roasted ilmenite and anatase is aeration leached;

Figure 12: Synthetic rutile and anatase obtained after the beneficiation of ilmenite ore;

Figure 13 a: Microstructure of ilmenite roasted with potassium carbonate and aeration leached;

Figure 13b: EDX corresponding to the micrograph of Figure 13 a;

Figure 14: Comparison of the roasted and leached product when ilmenite was roasted with lithium carbonate followed by reduction leaching;

Figure 15: FTIR of a ferrihydrite containing sample obtained after differential dissolution;

Figure 16: Rare earth oxides floating on the surface of water;

Figure 17a: Backscattered image of the colloidal layer obtained during water leaching;

Figure 17b: EDX of the bright phase in the backscattered image;

Figure 17c: EDX of the grey phase in the backscattered image; Figure 18: Cross-sectional SEM image of partially reacted ilmenite with a fragmented product layer of potassium titanate/ferrite (roasting temperature 1123K for 60 minutes in air);

Figure 19a: Partially reacted ilmenite grain after roasting with potassium carbonate for 60 minutes;

Figure 19b: EDX of product layer showing the formation of potassium titanate and ferrite;

Figure 19c: EDX of unreacted ilmenite grain;

Figure 20: Cross-sectional scanning electron micrograph cross-sectional images of ilmenite after a roasting reaction with sodium carbonates in air at 1123K. The peripheral product layer (sodium titanate/ferrite) grows in thickness with time;

Figure 21 : Cross section of weathered ilmenite grain (a) and final product (b) after limited leaching in an organic acid; and

Figure 22: An EDX spectrum (top left) showing iron only from the cubic structure (iron oxide) occasionally seen in a matrix of rutile particles in the background for which the EDX shows only traces of K. This material appears beige white in colour and is quite fine as indicated by the micron bar in the SEM image.

A reactor assembly 1 for performing the process of the invention in the following Examples is shown schematically in figure 2. The reactor assembly 1 comprises a beaker 2 fitted with a variable speed stirrer 3. A condenser 4 is fitted on the top of the beaker 2 to restrict the evaporation of water. The beaker 2 contains a nozzle 5 through which argon gas (Ar) may be passed into a leaching solution 6 to remove oxygen and maintain anaerobic leaching conditions. There are two openings in the beaker 2 for the measurement of temperature of the solution by a thermometer 6 and of pH of the solution by a pH probe 7.

Example 1

The experiments described in Example 1 compare leaching in oxidising and reducing media which lead to the production of synthetic rutile having a TiO ₂ concentration greater than 92wt%. Unless specified otherwise, the roasted ore upon which the leaching experiments were conducted was prepared in a similar manner to the preparation OfK ₂ CO ₃ and Na ₂ CO ₃ roasted and washed ores referred to specifically below.

Experimental

The reactor assembly of Figure 2 was mounted on a heating mantle and the temperature was varied between 323 and 353K. Aeration leaching was carried out in a solution OfNH ₄ Cl, methanol and acetic acid. Reduction leaching was carried out in a solution of ascorbic acid and oxalic acid.

Results

The principle of leaching is based on the Eh-pH diagram shown in figure 6. During leaching, no potential was applied and the value of hydrogen electrode potential E _H was zero. When the pH of the medium is altered, various phases are likely to form. At zero potential, Na ⁺ ions can be removed from sodium titanate below pH 12. The shaded region beyond pH 12 represents the region which is attained during washing with water. In this region sodium titanate and iron oxide are stable. However there is some recovery of sodium. To recover sodium and remove iron from a roasted ilmenite sample, the pH has to be maintained below 4 which is the shaded area in figure 6.

Initial leaching experiments were performed by adding 20% HCl to a beaker with the roasted sample. The solution was stirred at 300rpm for 4 hours which led undesirably to complete destruction of particle shape and size distribution. The microstructure of the leached sample shown in Figure 7 exhibited particle segregation into a range of smaller size fractions. The bright phases are zircon and monazite grains and their sodium compounds. The light grey colour phases are Fe-Ti-rich compounds and the dark grey regions are Ti-rich phases. There are impurities such as zircon and monazite phases which were locked inside the mineral matrix and liberated when the excessive dissolution of alkali occurs in a strong acid medium. The physical liberation of finer zircon and monazite phases is undesirable from the point of view of controlling gravity separation of such minerals. The acid leaching based separation of heavy mineral phases (namely zircon and monazite) is not practically suited against the observation of colloidal medium formed during the water washing of roasted mineral with KOH.

One of the aims of the beneficiation process was to maintain the average particle size greater than 100 micron to permit the product to be used in a fluidised bed chlorination reactor for the production of pigment grade TiO ₂ . If the particle size is lower than 100 microns, the loss due to excessive fluidisation during chlorination will increase in the reactor. During leaching with HCl, the pH of the solution increased as the sodium titanate decomposed to sodium ions and titanium dioxide. As the process of leaching in an HCl medium yields TiO ₂ of nearly 82% purity with varying particle size, a further leaching experiment was carried out with organic solvents in order to control pH of the leaching medium.

Optimisation ofpH

It was necessary to establish the basic conditions under which the pH must be maintained during the leaching process for maximum retention of the shape and size of rutile. Initial assessment of pH control involved using a strong acid (HCl) and a weak acid (CH ₃ COOH).

Figure 8 compares the changes in pH with time when leaching was carried out in 30% HCl and acetic acid solutions. It is evident from figure 8 that pH varies between 3.75 and 4.25 in an acetic acid medium. The pH increases to 6 when a mineral acid such as HCl is used. The increase in pH due to the addition of HCl can be attributed to the low concentration of HCl which when consumed during the leaching stage results in an increase in pH. After 5 hours leaching with HCl solution, the total iron remaining in the residue was estimated to be 10.56wt%. The low removal of iron is attributed to the fact that the pH of the medium increased above 5 within 2 hours. This was not the case when acetic acid was used. It is for this reason that the removal of iron was better as pH varied between 3.75 and 4.25. After 5 hours, the total iron concentration remaining in the residue was 7.5wt%. The analysis of iron after 4 hours was found to be marginally higher than after 5 hours and this time was considered therefore to be optimum. The main inference from this experiment is that control of pH is essential for maintaining the removal of iron and alkali salt. Aeration leaching

30 grams of roasted material were suspended in a 500 ml beaker containing 1.5% (w/v) NH ₄ Cl and 0.5% (v/v) methanol solution. Acetic acid was added to the solution and the pH was adjusted to 3. To keep the roasted particles suspended, the solution was stirred at constant speed. During leaching, air flow was maintained at a constant rate of 0.5 lit/min. It is evident from figure 9 that better removal of iron was achieved with increased stirring speed which reached a maximum when the solution was stirred at 1300rpm. This increase is attributed to a higher suspension volume when the stirring speed is increased from 900 rpm to 1300 rpm. Under these conditions the surface area of the chemical reaction increases and leads to high removal of iron. However, it can be seen that at 1400 rpm, removal of iron has decreased slightly. This was because some particles accumulated on top of the beaker and had to be constantly reintroduced into the medium.

Removal of iron and sodium

During leaching, samples were collected for analysis of the total iron and sodium present in the solution. Figures 10 and 11 are plots of percentage removal of iron and sodium respectively when roasted ilmenite and anatase minerals were treated via aeration leaching.

It is evident from figure 10 that removal of iron from roasted ilmenite is higher than from anatase. This is because anatase has comparatively less iron oxide present in the mineral lattice. The kinetics of leaching for ilmenite for the first hour are fast but slow down with time. The total removal of iron from ilmenite after 5 hours was about 57wt%. The removal of iron from roasted anatase ore is almost linear and at the end of 5 hours only 20wt% of iron was removed.

Similar observations were made in the removal of sodium ions from the roasted material (see figure 11). In the first 3 hours, the rate of removal of Na ⁺ ions was much faster than between 4 and 5 hours where the reaction tends to plateau. However for both ilmenite and anatase the removal of sodium ions was around 60wt%. Table 1 compares the chemical composition of ilmenite and anatase minerals after the benefϊciation steps of roasting and leaching.

Table 1 : Chemical composition of ilmenite and anatase ores after roasting followed by aeration leaching at T=25°C, atm= O ₂ at 500ml miή ¹

It is evident from Table 1 that the TiO ₂ concentration for anatase has increased from 56wt% to 69wt%. There is little difference in the TiO ₂ concentration between untreated ilmenite and the roasted and leached sample. Although significant amounts of sodium were recovered in the water washing and leaching stages, it was confirmed by chemical analysis that the ores contained a little over 15wt% of sodium oxide. To remove the remaining sodium oxide, an acid treatment was performed with 4M HCl solution. Figure 12 compares the diffraction patterns after roasting, aeration leaching and acid leaching from which it is evident that beneficiated material contains both rutile and anatase. After aeration leaching, anatase and rutile peaks start growing. However peaks of sodium titanate and ferrite were present and therefore acid leaching was needed.

Given that the concentration of constituents below 5wt% cannot be detected from XRD analysis, verification was necessary using chemical analysis. Table 2 compares the chemical composition of the ores after acid treatment. It is evident that the TiO ₂ concentration increased to 91wt% for ilmenite and 88wt% for anatase. Furthermore the concentration of rare earth oxides and phosphorus pentoxide decreased from the concentration present in the mineral. Similar experiments were performed when ilmenite and anatase were roasted with potassium and lithium carbonate. As lithium carbonate forms more stable titanates and ferrites than those formed with Na ⁺ and K ⁺ ions, leaching of lithium titanates resulted in no removal of iron or lithium. However, when leaching was performed with a roasted sample treated with potassium carbonate, the results were similar to those for sodium.

Table 2: Chemical composition of beneficiated ilmenite and anatase ores (T= 25° C) via oxidative leaching and acid washing with 4MHCl

The microstructure of the beneficiated mineral after roasting with potassium carbonate and aeration leaching is shown in Figure 13 a. The micrograph shows the segregated particles which occur due to breakage of the lattice in the presence of potassium ions. The EDX in Figure 13b confirms the presence of titanium dioxide with smaller peaks of potassium, iron, magnesium and silicon. Chlorine is present due to the acid leaching with HCl.

Aeration leaching is a successful method for recovering alkali ions as well as removing the iron remaining after roasting.

Leaching in a Reducing Medium The main aim of leaching is to recover most of the alkali salts and to remove impurities such as iron oxide. The Eπ-pH diagram in figure 6 indicates that below pH 4, Fe ²⁺ and Na ⁺ ions are produced whereas TiO ₂ is stable. Therefore leaching experiments were performed in reducing medium to convert the Fe ³⁺ ions which were generated during oxidation roasting to Fe ²⁺ ions. This was achieved using the setup shown in figure 2 by the addition of ascorbic acid which acted as a reducing agent. The reaction mechanism for the reduction leaching was studied in detail by spectroscopy.

(I) K ₂ CO ₃ Roasted Ore

From figure 1, it is evident that below a pH of 4 iron and potassium exists in the ionic form whereas TiO ₂ is stable. Therefore by controlling the pH of the medium it is possible to recover titanium dioxide from the roasted mass.

Ilmenite ore was roasted with a stoichiometric proportion Of K ₂ CO ₃ at 1148K for 4 hours. The roasted mass was then washed in water to remove all soluble potassium ferrites, silicates and aluminates. The substantially insoluble residue was then subjected to leaching in a solution of oxalic and ascorbic acid in the reactor assembly 1 illustrated in figure 2. As the stirring rate and the pH were already optimised in the aeration leaching process, these parameters were unchanged. The pH of the solution was maintained by oxalic and ascorbic acid which being weak acids dissociate slowly and therefore maintain the pH below 4 throughout the leaching process. The pH of the solution was adjusted to 3 by the addition of NaOH. The temperature was varied between 333 and 363 K. The best result was obtained when leaching was performed at 343K beyond which the particles broke into very small fragments having particle size of less than 75μm. The solution was stirred at a constant rate of 1300rpm and argon gas was passed into the solution at 300ml min ^"1 . Leaching was carried out for 4 hours after which the residue was filtered and washed with 5% HCl solution.

The micrograph in Figure 3a shows a synthetic rutile particle and lots of veins which can be attributed to the removal of iron and potassium in the leaching process. The corresponding EDX in Figure 3b has dominant peaks of Ti and O which suggest the formation of synthetic rutile and peaks attributable to the presence of potassium, calcium, magnesium and iron. Chemical analysis proved that the concentration of TiO ₂ was 97wt%. The remaining 3wt% was made up of potassium, calcium and magnesium oxides.

(2) Na ₂ CO ₃ Roasted Ore

Ilmenite ore was roasted with a stoichiometric proportion Of Na ₂ CO ₃ at 1123K for 4 hours. The roasted mass was then washed in water to remove soluble potassium ferrites, silicates and aluminates. The substantially insoluble residue was then subjected to leaching in a solution of oxalic and ascorbic acid in the reactor assembly 1 illustrated in figure 2. The pH of the solution was adjusted to 3 by the addition of NaOH. The temperature of the leaching solution was maintained at 7O ⁰ C. The solution was stirred at a constant rate of 1300rpm and argon gas was passed into the solution at 300ml min ^"1 . Leaching was carried out for 4 hours after which the residue was filtered and washed with 5% HCl solution.

Reduction leaching led to the formation of synthetic rutile having 94% TiO ₂ with the remainder containing sodium oxide, aluminium oxide, silicon dioxide and iron oxide. The backscattered image in Figure 4a illustrates the beneficiated ore. The corresponding EDX in Figure 4b shows dominant peaks of titanium and oxygen and small peaks of aluminium, silicon and iron. Furthermore it was observed that by controlling the roasting and leaching conditions both synthetic anatase and rutile can be obtained. The diffraction pattern in Figure 5 compares the roasted, leached and acid washed sample. The diffraction pattern of the acid leached sample exhibits both rutile and anatase peaks.

(3) Li ₂ CO ₃ Roasted Ore

When reduction leaching was performed on ilmenite roasted with lithium carbonate, synthetic rutile was not produced due to the presence of stable lithium titanate.

Figure 14 compares diffraction patterns for the roasted and leached products for the lithium system. It is evident that after reduction leaching there is no difference in the lithium titanate peaks. However goethite peaks marked as 'x' in the roasted product are missing in the leached sample which confirms that there was some removal of iron.

Conclusion

Roasting of ilmenite with different alkali salts resulted in the formation of alkali titanate which can be leached with organic acid to produce synthetic rutile.

Although two different types of leaching were performed, reductive leaching resulted in better removal of iron and alkali salt. The primary condition of leaching was to maintain the pH below 4 and therefore the choice of the organic compounds was crucial. Ascorbic acid converts Fe ³⁺ to Fe ²⁺ which then reacts with oxalic acid to form iron oxalate and dissolves in the solution. Oxalic acid is a weak acid and therefore does not dissociate readily in water. This helps to maintain the pH of the solution between 3 and 4. Oxalic acid also reduces Fe ³⁺ to Fe ²⁺ and forms iron oxalate directly. In the presence of ascorbic acid, the reduction of iron starts as early as 15 minutes into the leaching process. The acid solution also removes alkali to produce synthetic rutile at a concentration greater than 97% TiO ₂ .

Example 2 - Role of Ferrihydrite in the leaching and separation of rare-earth oxide impurities

Table 3: Chemical composition of magnetic and non-magnetic fractions of ilmenite ore in wt%

A differential dissolution technique (Schwertm.U 1973 "Use of Oxalate for Fe Extraction from Soils" Canadian Journal of Soil Science 53(2): 244-246; Jambor, J. L. and Dutrizac, J. E. (1998). "Occurrence and constitution of natural and synthetic ferrihydrite, a widespread iron oxyhydroxide" Chemical Reviews 98(7): 2549-2585) was carried out in a dark room by shaking the solution for 2 hours at room temperature after which the iron content in the solution was ascertained from atomic absorption analysis. Oxalic acid was added during the differential dissolution process to maintain the pH at 3. It was shown that ferrihydrite or poorly crystalline iron oxides are readily soluble in ammonium oxalate solution at pH 3 whereas goethite and hematite are much more resistant. It was found that the solution of ammonium oxalate contained 54 ppm of iron thus proving the presence of ferrihydrite. Further presence of ferrihydrite was confirmed by performing an FTIR on the dried sample obtained after the differential dissolution process. The FTIR of the sample is presented in figure 15 and shows distinctive peaks for oxalic acid and ammonium oxalate. Two OH stretching bands observed at 3000 and 3200cm ^"1 are attributable to the carboxylic group and hydrated water in ammonium oxalate. The C=O and C-O stretching bands at 1700cm ^"1 and 1230cm ^"1 respectively are attributable to the carboxylic acid group. The N-H and C=O bond stretching bands at 1409cm ^"1 and 1595cm ^"1 are due to the presence of ammonium oxalate. However the Fe-O stretching band is present at 490cm ^"1 which confirms the formation of iron oxalate. Iron oxalate cannot form at room temperature by dissolving hematite in oxalic acid (see Taxiarchou, M., Panias, D., Douni, I., Paspaliaris, I. and Kontopoulos, A. (1997). "Dissolution of hematite in acidic oxalate solutions" Hydrometallurgy 44(3): 287-299). Thus the above analysis points to the presence of ferrihydrite in ilmenite ore which is responsible for the leaching of iron from the ilmenite lattice in the natural environment and the formation of pseudobrookite. It was important to characterise the iron hydroxyl-oxide phase as the formation of such phases in laboratory scale would assist in easy removal of iron in the leaching stage. The other important characteristic of ferrihydrite is that it has a large surface area and exists as nanoparticles with high adsorption capacity. Ferrihydrite is also stable at pH 4 to 7 making it ideal for processing minerals.

Adsorption of impurities on ferrihydrite

When ilmenite was roasted with a potassium salt (eg KOH at 1023K for 4 hours), a major structural stress developed within the ilmenite lattice leading to macroscopic fracture in ilmenite grain. Such a macroscopic change in chemical reaction led to the liberation of dispersed rare earth oxides in the matrix of ilmenite. Once liberated, these particles formed a colloidal layer and floated on the surface of water when the roasted material was treated with cold distilled water. Figure 16 is the colloidal residue which was collected from the filter paper and is predominantly rich in rare- earth oxides.

The colloidal layer was filtered and analysed using XRF. Due to unavailability of rare earth oxide standards, a semi quantitative XRF analysis was performed. Table 4 lists the chemical composition of the colloidal layer.

Table 4: Semi-quantitative chemical composition of the colloidal layer floating on the surface of water after roasting ilmenite ore with KOH. FeO.Tiθ ₂ 'KOH=l:2.2

It is evident from table 4 that there is a high concentration of calcium, lanthanum and cerium and a small concentration of phosphorus and iron. The presence of phosphorous and rare earth oxides is indicative of the mineral co-existence of monazite which is often associated with rare-earth bearing ilmenite.

The colloidal layer (Figure 16) was analysed using SEM. The backscattered image in Figure 17a shows two distinct phases. The EDX of the bright phase in Figure 17b confirms the presence of iron, calcium, titanium, phosphorus and cerium phases. The EDX for the dark grey phase shown in Figure 17c is predominantly titanium, potassium and a whole range of rare-earth oxides. Smaller peaks of strontium suggest that the perovskites of strontium co-exist with Ca.

Comparing figures 17b and 17c, it is evident that phosphorus and rare earth oxide phases co-exist with iron-rich phases. This might be due to the prevalence of ferrihydrite or hydro-oxy oxides of iron which are abundant under neutral to strong pH under oxidising condition. Thus from the various analyses it could be said that ferrihydrite present in ilmenite ore is responsible for the leaching of iron from the ilmenite lattice thereby increasing the concentration of TiO ₂ in the ore.

Influence ofbicarbonates and hydroxides

Table 5 presents the results of a comparison of the removal of iron from magnetic and non magnetic fractions of ilmenite after roasting with Na ₂ CO ₃ , NaHCO ₃ and NaOH followed by water washing. It is evident that the reaction with sodium hydroxide improves the removal of iron much more than the other sodium compounds. The increase in iron removal is due to the numerous cracks grown in the ilmenite grain during roasting (see figure 18). These cracks increase the conversion from ilmenite to corresponding sodium titanate and sodium ferrite.

Both the hydroxide and bicarbonate have OH ^" groups within the structure. Their presence promotes the formation of ferrihydrite which otherwise is impossible to form. This is why the kinetics of reaction are much slower with Na ₂ CO ₃ compared with NaOH and NaHCO ₃ .

Table 5: Percentage removal of iron when two fractions of the ore were treated with different alkali salts

The reason for this catastrophic change in the structure can be explained on the basis of the following decomposition reactions. The protons from the electrolytic (galvanic) dissociation OfH ₂ O diffuse much more rapidly than the alkali ions. The resultant change in the volumes of titanate and ferrite crystallites with protons is much larger than in the absence of protons.

The decomposition of alkali salts takes place via the following reactions:

2NaOH(S) = Na ₂ O(S) + H ₂ O (g) [1]

4NaHCO ₃ =2Na ₂ CO ₃ + 2H ₂ O (g) + 2CO ₂ (g) [2]

Further the steam generated from the above reactions dissociates:

H ₂ O (g) = H ⁺ + 0H ^' [3]. The presence of hydroxyl groups with alkali accelerates the attack by providing protons (H ⁺ ions) in the lattice to compensate for the charge deficiency due to the diffusion of Na ⁺ ions which replaces the Fe ²⁺ sites. Since a proton is much smaller in size than an alkali ion, the diffusion is much more rapid which allows reaction to commence at lower temperature.

Effect of potassium salts

When ilmenite was roasted with potassium carbonate, bicarbonate and hydroxide, there was breaking of the product layer to numerous fragments as seen in figure 19. As K ⁺ ions diffuse into the ilmenite lattice, a strain is created proportional to the concentrations of H ⁺ and K ⁺ which is too large to be accommodated within the lattice structure. Thus the product layer of potassium titanate and ferrite cracks. Samples collected at intermediate stages of roasting showed the same phenomenon. Figure 19 illustrates that the product layer of potassium titanate and ferrite had already fragmented after one hour of the roasting process. The EDX in figure 19b represents the reacted ilmenite grain showing potassium, titanium and oxygen peaks whereas Figure 19c confirms the unreacted core of ilmenite showing iron, titanium and oxygen peaks. However due to the extensive fragmentation of the ilmenite grain, the reaction between ilmenite and potassium carbonate led to an increase in removal of iron which was established by analysing the iron content after roasting and water leaching. Table 6 compares the removal of iron when both fractions of ilmenite were roasted with lithium, sodium and potassium carbonates.

Table 6: Removal of iron for both magnetic and non magnetic fractions of ilmenite after roasting with different alkali salts followed by water leaching

From Table 6 it is evident that the removal of iron is maximised when roasting is carried out with potassium carbonate. There is no removal of iron in the case of lithium carbonate due to the formation of stable lithium titanate and ferrite. The removal of iron in the case of roasting with sodium carbonate is higher than that with lithium carbonate but lower than that with potassium carbonate. Iron removal also depends on the solubility of the different ferrites in water. Lithium ferrite is not water soluble and no removal of iron was apparent during the water washing stage.

Due to the unavailability of rare earth oxide standards such as Nd, U and Th, suitable oxide markers phosphorus pentoxide, alumina, cerium and lanthanum oxides were chosen. Table 3 indicates that the non magnetic fraction of the ilmenite ore had higher concentrations of rare earth oxides and was used for investigating the maximum removal when treated with potassium salts. Table 7 compares the list of markers used for quantifying the rare earths oxides in the magnetic and non magnetic fractions of the ore. Comparing the concentrations of rare earth oxides in Tables 3 and 7, it is evident that the concentrations are not the same. This can arise because the concentrations of rare earth oxides are not uniformly spread in the ilmenite mineral. However, Table 7 shows the same trend that the concentrations of rare earth oxides are higher in the non magnetic fraction of the ilmenite ore compared to the magnetic fraction.

Table 7: Rare earth oxides present in magnetic and non magnetic fractions of ilmenite ore

After roasting the non magnetic fraction of ilmenite ore with potassium hydroxide followed by a water leaching step, the material obtained was analysed using XRF. Table 8 compares the concentrations of rare earths oxides in treated non magnetic fraction with the untreated fraction. It is evident from Table 8 that 80wt% of rare earth oxides was removed by roasting the non magnetic fraction of ilmenite ore with potassium hydroxide followed by water leaching.

Table 8: Comparison of concentrations of rare earth oxides in the initial and treated samples of non magnetic fractions of ilmenite ore.

Example 3 - Particle Size and Purity of Synthetic Rutile

Table 9 - Chemical compositions of various ores in wt%

Particle size is affected by most of the operational parameters within the alkaline roasting and reductive leaching steps. Unreacted Bomar ilmenite (see Table 9) has a particle size ranging from 100 - 450μm which when mixed with the alkali salt at elevated temperatures reacts and binds within the grains.

The alkali salt selected for the reaction influences greatly the resulting size of the reacted ilmenite grains. Lithium, sodium and potassium salts behave in a similar manner by forming a product layer through which the alkali ions tend to diffuse toward the unreacted core (see Figures 20 and 18).

Lithium salts have an ionic radius of 0.68A which falls between the ionic radius of Fe ²⁺ and Fe ³⁺ (0.64 - 0.74A). Being comparable in size with Fe-ions, lithium ions diffuse fastest but the resulting reaction product is not soluble in water and does not facilitate preferential separation between alkali-rich titanate and alkali-complexed iron oxides. By comparison, sodium ion (Na ⁺ ) has an ionic radius of 1.02A which is greater than that of Ve ²⁺ [Fe ²⁺ and Ti ⁴⁺ (0.68A). The ionic size difference between alkali and parent cations of the ilmenite lattice creates much greater strain between the product layer (see Figures 20a-d and 18) and the mineral lattice. Consequently cracks form as the reaction continues with time. The cracks become more prevalent over shorter periods at higher roasting temperatures due to a larger diffusion speed of cations in the mineral and product layer lattices. The effect of time on the formation of product layer of alkali titanate is explained with reference to Figure 20 in which the thickness of the peripheral layer grows with increasing time. The unreacted core is ilmenite. Upon reaching a critical thickness the peripheral layer cracks and renews a fresh layer for roasting reaction between alkali and ilmenite. Therefore the alkali ion diffusion into the ilmenite structure is an important step in the mineral beneficiation process.

Potassium ion (K ⁺ ) has the largest ionic radius of 1.38 A which is why the reaction- induced strain between the ilmenite lattice and the product layer is the largest amongst the three alkali ions and contributes to rapid cracking of the product layer (see Figure 18). This exposes the unreacted ilmenite grain below the surface which further supports the conclusion that surface renewal via cracking of the product layer is the prevalent mechanism for product layer formation and its disintegration during alkali roasting of ilmenite in air. The effect with potassium salts is more vivid than with sodium salts as can be seen by comparing figures 20a-d with Figure 18. In Figure 18, the size of the fragments of product layer appear much smaller than in Figure 2Od.

Elevated temperature and longer reaction times for roasting also break the ilmentite grains further. Reaction becomes more efficient as the temperature increases until the initiation of the melting point. Once the solid outer layer begins to melt the reaction is deprived of oxygen and consequently the formation of alkali titanates/ferrites is severely limited.

Example 4 - Controlled leaching after roasting with K/Na bicarbonate or hydroxide

The removal of impurities such as oxides of iron, aluminium, silicon and rare-earth can further contribute to the reduction in particle size of ilmenite. This is mostly affected by the kinetics of oxidative and reduction leaching. The kinetics of oxidative leaching were studied at room temperature whereas the reduction leaching was studied between 313 K and 353K. The reaction rate has been shown to nearly double every 293K increase above room temperature to a maximum at 373K. Unreacted ilmenite grains can take days to break under organic acid dissolutions. Consequently by limiting the reaction during roasting, it may be possible to preserve particle size by compromising the overall residual concentration of impurity oxides. This is evidenced by demonstrating the seams of weathered ilmenite which are preferentially beneficiated after four hours with sodium hydroxide at 500°C. Leaching was performed at pH 2.6 under anaerobic conditions at 65°C whilst the solution was stirred at 500rpm for four hours.

The oxide seams shown in Figure 21a are normally made up of a mix of ferrihydrite and hematite which are highly defective and react preferentially during roasting. Areas around the seams and outer layers are generally the richest in rare earth materials and impurities. In these areas leucoxene forms in amorphous to crypto- masses which are unstable due to the charge difference of the iron leaving the lattice. The balance requirement of the lattice means it must be stabilized with impurities such as Al ₂ O ₃ , Cr ₂ O ₃ , SiO ₂ , V ₂ O ₅ and Na ₂ O.

Figure 21a shows a cross section of a weathered ilmenite grain. The iron has leached out of the ilmenite grain through weathering and formed hematite layers across the grain. Figure 21b is an image of the final product after both reaction stages. The acid has attacked only the already activated iron seams within the grain. By only performing a partial reaction, a large proportion of the impurities can be removed whilst maintaining the particle size around 100 - 200μm. However during these partial reactions rare earths and other impurities are often left behind on the surface. If silica, alumina or cerium is not removed during initial roasting, the reductive leach will have little effect on this material. In the case of cerium, crystallization occurs on the surface of ilmenite grains during leaching. The material not removed during roasting is mobilized during the leaching stage and nucleates forming crystals of impurities across the surface.

Partial roasts also produce a large quantity of residual alumina and silica grains in the product. Leaching has next to no effect on these materials achieving 0.2% dissolution over one hour.

Example 5 - Formation of High Purity (>97%) Rutile and Particle Size Nearly pure titanium dioxide can be achieved through a two-step process but at the expense of the original particle size. This is achieved by simply using an efficient roasting process which results in particle size reduction (see Figures 18 and 20) in which significant particle liberation is expected. By utilizing an elevated roasting temperature (eg above 823K), potassium bicarbonate salt reacts rapidly with ilmenite and releases rare-earth, silicate and zircon impurities.

During leaching in anaerobic conditions with oxalic and ascorbic acids, the titanium dioxide falls from suspension while the iron nucleates and forms cubic crystal structures (Figure 22). Although thermo-chemistry does not predict the presence of a solid Fe ²⁺ bearing solid phase which is rather unexpected, it does entail that in the vessel pH may be rising above 4 locally to cause the nucleation of hydroxy-oxide of iron as shown in Figure 6. The speed and type of crystal formation depends on concentration, pH, temperature, speed of oxidation and impurities present in solution. Under the leaching conditions there are four main compounds that are commonly formed. These are hematite, goethite, lepidocrocite and ferrihydrite. The titanium dioxide is flaky compared with the cube-like structures of iron oxides in Figure 22.