The scope of the study entailed (1) the phase equilibria studies to re-draw Jochens’ phase diagram in air for validation purposes and at pO2 applicable in titaniferous smelting for the establishment of a reliable phase diagram applicable to titaniferous slags, (2) production and beneficiation of selected titaniferous slags for the production of titania products, and (3) evaluation of an integrated flowsheet for the processing of titanomagnetite to produce Fe, V and Ti products.
1.4.1 Phase equilibria studies in the Ca-Mg-Al-Si-Ti-O system
The phase relations in the multicomponent Ca-Mg-Al-Si-Ti-O system are in fact very complex (Pistorius, 2011). Following the typical approach to studying phase equilibria in multicomponent and multiphase systems, the phase equilibria study in the current testwork included (a) survey of the available thermodynamic and phase equilibria data in the Ca-Mg- Al-Si-Ti-O system, (b) calculation and redrawing of suitable equilibrium phase diagram using a thermochemical software, and (c) strategic experimentation to validate calculated phase diagrams.
Literature survey in the Ca-Mg-Al-Si-Ti-O system
To understand the phase equilibria in multicomponent or higher order systems, subsystems are generally reviewed (Jung, et al., 2009; Wulandari, et al., 2009; Jak & Hayes, 2004). For the literature review of thermochemical and phase equilibria data in the Ca-Mg-Al-Si-Ti-O system, numerous subsystems were reviewed. Due to numerous possible lower order combinations in the high order Ca-Mg-Al-Si-Ti-O system and the fact that the titaniferous slags are characterised by the presence of TiO2, the reviewed binary, ternary and quarternary subsystems were subjectively limited to those containing TiO2.
Calculation of phase equilibria
It is essential to take advantage of available computational thermochemical and thermodynamic software for simulating and depicting complex phase equilibria. In the present work, the thermodynamic predictions of phase equilibria in the Ca-Mg-Al-Si-Ti-O system were conducted using FactSage software, an established commercial computer package used worldwide for the calculation of multi-component phase equilibria and thermodynamic properties (Bale, et al., 2002).
In his study of the Ca-Mg-Al-Si-Ti-O system applicable to titaniferous slags in air, Jochens (1967) reported, amongst other things, that 4(MgO.2TiO2).Al2O3.TiO2 or stoichiometrically 4MgTi2O5-Al2TiO5 crystallises as the primary phase at relatively high MgO concentrations in the reviewed compositional range. The latest FactSage 7.2 version does not have Al2TiO5 modelled as an endmember in the pseudobrookite (M3O5) solid solution database (Centre for Research in Computational Thermochemistry, 2018). In addition, Berezhnoi and Bulko (1955) and Krajewski (1992) reported in their studies of the Al-Mg-Ti-O system that under
reducing conditions, Ti3O5 would crystallise in the M3O5 solution. Hence, the latest FactSage 7.2 model would not be able to correctly predict the phase relations in the Ca-Mg- Al-Si-Ti-O system in air and at low pO2 atmospheres in which case the M3O5 solid solution is anticipated to be composed of MgTi2O5, Al2TiO5, and Ti3O5 endmembers.
Pseudobrookite solid solution thermodynamic models for the MgTi2O5-Ti3O5, MgTi2O5- Al2TiO5, and Al2TiO5-Ti3O5 subsystems were developed and integrated to ternary MgTi2O5- Al2TiO5-Ti3O5 system, and subsequently incorporated into FactSage as a private pseudobrookite solid solution database. The principles of the CALculation of PHAse Diagram (CALPHAD) method were adopted during the development of the private pseudobrookite solid solution database for the correct calculation of phase equilibria using the evaluated and optimized thermochemical data in the literature. The application of the CALPHAD method in the current study is shown in Figure 2. It should be noted that the experimentation for the optimisation of the private pseudobrookite database fell outside the scope of the current work and was not conducted; instead optimized thermodynamic data from literature was used (Bale, et al., 2002; Pelton, et al., 1998). Following the established thermodynamic approach, the pseudobrookite solid solution was modelled using the sublattice model coupled with Compound Energy Formalism (CEF) and Redlich-Kister polynomial. In real multicomponent and multi-phase systems, the Gibbs energy of a solid solution can be expressed as shown in Equation [1.1] (Pelton, 2006; Hillert, 2001; Shi, et al., 1992).
𝐺𝑚 = ∑ ∑ 𝑦𝑖 𝑗 𝑖′𝑦𝑗′′𝐺𝑖𝑗0 − 𝑇𝑆𝑐𝑜𝑛𝑓𝑖𝑔+ 𝐺𝑒𝑥𝑐𝑒𝑠𝑠 [1.1] Where 𝑦𝑖′ and 𝑦𝑗′′ are site fractions on first and second sublattices, 𝐺𝑖𝑗0 is the Gibbs energy of formation of the compound 𝑖𝑗, also called ‘endmember’, 𝑆𝑐𝑜𝑛𝑓𝑖𝑔 is the ideal entropy of mixing. The main parameters of the CEF stage of the model are the endmember Gibbs energies. The 𝐺𝑒𝑥𝑐𝑒𝑠𝑠 is the excess Gibbs energy of mixing of the real solution described by the Redlich-Kister polynomial.
FactSage 7.2 equipped with the private pseudobrookite solid solution database was used to calculate and reproduce Jochens’ phase diagram (Figure 1) in the same compositional range in air. The scope of FactSage calculations was increased to include the predictions of phase equilibria in the same and other compositional ranges at low pO2 applicable to titanomagnetite smelting.
Figure 2: CALPHAD method adopted for database development and calculations of phase diagrams
Experimentation
Systematic experimental equilibrium phase stability studies in air and at pO2 of 10-16 atm using the established integrated high temperature equilibration, quench and Electron Probe Micro Analysis (EPMA) approach (Jak & Hayes, 2010) were undertaken to cross-validate the FactSage predictions of phase relations in the Ca-Mg-Al-Si-Ti-O system.
1.4.2 Beneficiation of titaniferous slag
A slag chemistry with a minimum chance of the crystallisation of the detrimental spinel solid solution [(Mg)(Al,Ti)2O4] was identified on the updated phase equilibria at low pO2 applicable to titanomagnetite smelting. The targeted titaniferous slag was produced by smelting using the conventional and cold crucible induction furnaces. The slag was subjected to beneficiation testwork to demonstrate the production of a saleable titania product from a titaniferous slag.
Some processes dedicated to the upgrading of dilute titania resource, like titaniferous slags, to saleable TiO2 products are reported in the open literature (Hassell, et al., 2016; Van Vuuren & Tshilombo, 2011; Zhang, et al., 2007; Fouad, 2005; Becker & Dutton, 2002). These processes have several disadvantages including the production of materials of still
insufficient titania grades to qualify as high titania materials, and poor titania recoveries. There is a market for the titania product with a minimum of 70wt% TiO2 (Shijiazhuang Leveling import and export Co., Ltd., 2014); the first prize is however to produce a suitable feedstock for the preferred chloride pigment production process.
Titaniferous slags typically contain high levels of alkaline earth impurities like MgO and CaO from the gangue, and from the flux in the case of dolomite/ limestone fluxed smelting. An oxidative-reductive roast and leach process, typically known as the Upgraded Slag (UGS) process, is established for upgrading titania slags with high levels of alkaline earth impurities. In the UGS process, the phase composition of the slag is modified through roasting to increase the leachability of the impurities. The UGS process is used commercially at Rio Tinto Fer et Titane (RTFT) in Canada for upgrading the SORELSLAG® slag generated from the Allard Lake ilmenite that contains relatively high levels of the alkaline earth impurities (Borowiec, et al., 1998; Doan, 1996). The scope of the current work involved a preliminary investigation of the best combination of UGS process conditions, in particular the leaching conditions, for the beneficiation of the selected titaniferous slag with little or no spinel to produce a saleable titania product.
An integrated flowsheet for the production of steel, vanadium and titanium products from a titanomagnetite feed was developed and subsequently subjected to economic evaluation using the discounted cash flow (DCF) model.