ISIJ International, Vol. 43 (2003), No. 9, pp. 1316–1325 Reduction of Titania–Ferrous Ore by Carbon Monoxide Eungyeul PARK and Oleg OSTROVSKI1) Formerly Ph.D. Student, School of Materials Science and Engineering, University of New South Wales, Sydney, Australia, now at Max-Planck-Institut für Eisenforschung, Max-Planck-Str. 1, 40237 Düsseldorf, Germany. 1) School of Materials Science and Engineering, University of New South Wale, Sydney 2052, Australia. E-mail: [email protected] (Received on December 9, 2002; accepted in final form on March 17, 2003 ) The reduction of titania–ferrous ore (ironsand) containing 57.2 wt% of iron and 7–8 wt% of TiO2 was in- vestigated in non-isothermal and isothermal reduction experiments using CO–CO2–Ar gas mixtures in a lab- oratory fixed bed reactor. Samples in the course of reduction were characterised using XRD, EPMA and SEM. Two types of particles were identified in ironsand: 1) homogeneous particles of titano-magnetite with cubic spinel structure (a major type); and 2) non-homogeneous particles, characterised by lamellar structure of rhombohedral titano-hematite, exsoluted from the titano-magnetite. ϭ Ϯ Titano-magnetite, which is a magnetite–ulvospinel solid solution (Fe3O3)1Ϫx(Fe2TiO4)x, with x 0.27 0.02, was reduced to metallic iron and titanium sub-oxides. Titanium had a strong effect on the mechanism and rate of reduction of iron oxide in ironsand. KEY WORDS: titano-magnetite (Fe3ϪxTixO4); magnetite (Fe3O4); ulvospinel (Fe2TiO4); titano-hematite; reduc- tion; carbon monoxide. tion in the course of reduction. This paper presents results 1. Introduction of reduction of titania–ferrous ore (New Zealand sand) by Titanium-containing iron ores, which are found through- carbon monoxide in comparison with hematite iron ore and out the world in large deposits, are becoming alternative magnetite, and examination of ores phase composition and sources of iron.1–4) Conventional processing of this ore morphology. (ironsand) includes carbothermic reduction of iron oxides in a rotary-kiln. However, solid state carbothermic reduc- 2. Experimental tion is relatively slow, has high-energy consumption, and a narrow range of operating temperatures.5,6) This stimulates Chemical compositions of ironsand and hematite iron ore research into gaseous reduction of titania–ferrous ores. examined in this paper are presented in Table 1. The oxy- Reduction of ironsand by carbon monoxide was studied gen associated with iron in the form of FeO and Fe3O4 was from early 1970’s.4,7,8) It was established that titano-mag- calculated using the composition of ironsand given in Table 2ϩ 3ϩ netite (TTM) is a solid solution Fe3ϪxTixO4 of magnetite 1 (the ore contained 24.2 % Fe and 33.0 % Fe ). The cal- (Fe3O4) and ulvospinel (Fe2TiO4), with cubic spinel struc- culated oxygen content was 21.1 wt%. The ores particle ture. The cation (Fe2ϩ, Fe3ϩ and Ti4ϩ) distribution in a TTM size was in the range of 125–212 mm. The surface area was lattice depends on the composition of titanium and temper- 1.2 m2/g for the ironsand, and 1.6 m2/g for the hematite iron 9–14) ature. Fe3O4–Fe2TiO4 system is characterised by the ore. The mass of an ore sample in each experiment was 2 miscibility gap between magnetite and ulvospinel.15–17) grams. Partial oxidation of TTM can form rhombohedral titano- Ore reduction using CO–CO2–Ar gas mixture was stud- 9,10,18–21) hematite (TTH, Fe2ϪyTiyO3), which has a low solu- ied in a laboratory fixed bed reactor in a vertical tube elec- bility in TTM.10) Formation of this phase by the oxidation tric furnace. Experimental set-up and reactor schematic are of TTM is called exsolution. It occurs via migration of shown in Fig. 1. After a sample was put into the reactor, the cations, resulting in enrichment of titanium.20,21) It was reactor was purged with argon. Then the reactor was moved established that the gaseous reduction of TTM ironsand to the hot zone of the furnace. A sample was heated up to is much slower than that of hematite and magnetite required temperature in argon atmosphere. Then the reduc- iron ores.3,4,7,8) The complete reduction of iron oxides in ing gas mixture was introduced to the reactor. After certain ironsand by carbon monoxide requires temperature above reaction time, the reactor was pulled off from the hot zone 1 273 K and high reducing potential.4,7,8) Understanding of of the furnace and quenched. limitations in the reduction of titania–ferrous ore requires In non-isothermal experiments, a sample was heated in detail investigation of TTM structure and phase transforma- the reactor from 200 to 1 373 K with the rate of 100 K/h, © 2003 ISIJ 1316 ISIJ International, Vol. 43 (2003), No. 9 Fig. 1. (a) Schematics of experimental set-up: 1-reactor, 2-pressure gauge, 3-six way valve, 4-rotameter, 5-gas flow con- troller and 6-gas purifier. (b) Schematic experimental reactor. Table 1. Chemical compositions (wt%) of ironsand and hematite iron ore. and then kept at 1 373 K until the completion of the reduc- tion. CO–CO2–Ar gas mixture was prepared from highly purified argon, carbon monoxide and chemically pure car- bon dioxide using mass flow controllers. The gases were purified before mixing by passing through traps filled with Drierite and 4A molecular sieve to remove moisture. The inlet gas flow rate was maintained at 800 ml/min. The re- ducing gas was introduced into the reactor from the top and left the reactor through a porous plug at the bottom. The outlet gas was analysed on-line by mass spectrometer. Fig. 2. XRD patterns of ironsand, hematite and magnetite iron Samples reduced to different extent were analysed by ores. XRD and microprobe spectrometer (EPMA, Cameca SX- 50 Probe). The morphology of samples was observed by Field Emission Scanning Electron Microscope (FESEM, proportion of ironsand exhibited exsolution caused by par- HITACHI S-4500). The extent of reduction was calculated tial oxidation (weathering) of TTM to TTH, which has the as a mass fraction of oxygen in iron oxides removed in the XRD pattern identical to hematite. course of reduction. 3.1.1. Homogeneous Titanomagnetite Particles Particles with homogeneous TTM phase constituted the 3. Results and Discussion bulk of the ironsand. Quantitative elemental analysis across of such particles with a step of 1 micron is shown in Fig. 3. 3.1. Phase Characterisation of Ironsand The analysed elements were Fe, Ti, Al, Mg, Si and O. Iron The XRD patterns of ironsand, hematite iron ore and and titanium in the particle were distributed uniformly magnetite obtained by partial reduction of the hematite iron along the line of analysis with the average compositions of ore are shown in Fig. 2. The XRD pattern of ironsand in- 32.8 at% Fe and 2.98 at% Ti. However, the concentrations cluded peaks for magnetite, hematite and traces of of aluminium and magnesium fluctuated in the range of maghemite (g-Fe2O3). The peaks of Ti-rich phases such as 0.88–8.13 at% for Al, and 0.64–4.26 at% for Mg. The fluc- ulvospinel (Fe2TiO4), ilmenite (FeTiO3) and rutile (TiO2) tuations of the Al and Mg concentrations through the parti- were not detected. cle were independent of each other, and of the Fe and Ti The EPMA analysis showed that the main phase of the concentrations. The amount of silicon in the TTM ironsand New Zealand ironsand was homogeneous TTM. A small particle was at most 0.26 at% (0.16 wt%) along the 1317 © 2003 ISIJ ISIJ International, Vol. 43 (2003), No. 9 analysed line, although the concentration of SiO2 in the persed in the TTM matrix. ironsand measured by chemical analysis was 2.17 wt%. The average Fe/Ti atomic ratio of the analysed particle Quantitative point analysis showed that silicon was present was 10.0 with a standard deviation less than 2.6%. The predominantly in silicate inclusions. ratio of oxygen (after subtracting oxygen in aluminium, The mapping images of Fe, Ti, Al and O in a representa- magnesium and silicon oxides) to iron and titanium tive homogeneous particle of the ironsand are presented in ([O]/[FeϩTi]) was calculated to be 1.48, which is higher Fig. 4. The ulvospinel fraction (x) in the homogeneous than the expected value of 1.33 for the stoichiometric TTM. TTM solid solution, (Fe3O4)1Ϫx(Fe2TiO4)x, was found to be This discrepancy can be attributed to the residual minor ox- 0.27Ϯ0.02. Aluminium and magnesium oxides were dis- ides and low sensitivity of the EPMA for the light elements. 3.1.2. Non-homogeneous Particles While the most particles of ironsand were homogeneous TTM, some particles showed a phase separation with a tita- nium-rich phase having a lamella structure with an angle between lamellas close to 60° (Fig. 5). In literature, the phase separation in ironsand is assigned either to the misci- bility gap between magnetite and ulvospinel15–17) or to the exsolution between TTM and TTH.9,10,18–21) Fig. 3. Elemental distribution in a homogeneous ironsand parti- Fig. 4. Mapping of Ti, Fe, Al and O in the homogeneous particle cle. of ironsand. Fig. 5. Elemental analysis of a non-homogeneous ironsand particle. © 2003 ISIJ 1318 ISIJ International, Vol. 43 (2003), No. 9 The origin of the phase separation was examined by experiments and had no effect on experimental results. quantitative point analyses, which results are shown in Fig. Reduction of ironsand started when the temperature 5. The major grey phase was identified as TTM, what reached 1 073 K. The maximum rate of the reduction ap- agrees well with XRD analysis (Fig.