2020 Volume 60 Issue 12 Pages 2659-2668
Experimental studies have been undertaken on the controlled solidification of iron oxide-rich melts in the system “Fe2O3”–CaO–SiO2–Al2O3 in air to determine the mechanisms of phase and microstructure formation during the cooling.
Selected bulk compositions, containing approximately 2 wt% Al2O3 and CaO/SiO2 = 3.5, were cooled at a fixed rate of 2K/s from fully liquid melts. The samples were rapidly quenched from selected temperatures, and the microstructures and phases present were examined using scanning electron microscopy (SEM) and electron probe X-ray microanalysis (EPMA).
It has been shown that, on non-equilibrium cooling in air, the magnetite and hematite phases are retained to sub-solidus temperatures despite the presence of the pseudo-ternary peritectic reaction H + L → SFCA + L that would occur under equilibrium cooling. The SFCA and SFC-I phases appear to nucleate preferentially at the interfaces between the magnetite and liquid phases; this phenomenon appears to be associated with common crystallographic features in the magnetite and the SFCA phases.
It has also been shown that rapid formation of secondary hematite can take place through the liquid phase assisted oxidation of the primary magnetite grains. The mechanism of this reaction has not been previously reported.
The Silico-Ferrite of Calcium and Aluminium (SFCA) group minerals are industrially important since they constitute a significant proportion of the material present in iron ore sinters, and their presence is believed to enhance the reducibility and mechanical strength of these sinter materials. Other phases typically observed in iron ore sinters are primary and secondary hematite (Fe2O3), magnetite (Fe3O4), dicalcium silicate (2CaO.SiO2, C2S), and glass.1)
During the iron ore sintering process the material in the sinter bed experiences a range of process conditions, in particular changes in temperature and oxygen partial pressure over time.2,3) Recent research4) has indicated that the phases and microstructures present in the final structures are formed principally during the cooling of the sample from the peak bed temperature. Systematic laboratory studies, undertaken under closely controlled temperatures and process conditions, have shown that the sinter product microstructure depends on the bulk composition of the samples, in particular, CaO/SiO2 ratio, alumina concentration, gas atmosphere and cooling rate.
Although it has been shown in previous studies that the SFC and SFCA phases can be formed under equilibrium conditions in the “Fe2O3”–CaO–SiO2 and “Fe2O3”–CaO–SiO2–Al2O3 systems, relatively little is known of the mechanisms and kinetics of formation of these phases, and the optimum process conditions for their formation.
The liquidus surface, and extent of the SFC primary phase field, in the iron-rich region of the “Fe2O3”–CaO–SiO2 system in equilibrium with air (nominally 21%O2) at 1 atm. total pressure has been established in previous studies by the authors.5) Systematic investigations of the controlled solidification of alumina-free “Fe2O3”–CaO–SiO2 melts in air6,7,8,9) have revealed unexpected observations. In particular, it has been shown that for a range of bulk compositions (Fe2O3 concentrations and CaO/SiO2 ratios) and cooling rates, the SFC phase does not form on cooling despite the fact that this phase is predicted to be thermodynamically stable under these conditions. Even for liquids within the SFC primary phase field SFC is not formed on cooling; rather intergrowths of the non-equilibrium SFC-I and Ca7.2Fe2+0.8Fe3+30O53 phases were observed dispersed throughout the melt but not associated with the pre-existing hematite.7) The observations demonstrated that the phases and microstructures formed under these conditions were strongly influenced by kinetic processes and not solely described by the chemical equilibria or Scheil-Gulliver solidification behaviour. This leads to a number of important questions about the mechanisms of formation of the SFC phase, and the process conditions necessary to maximise the production of the closely related SFCA phase during iron ore sintering.
The focus of the present study is to provide further detailed information on the sequence of reactions taking place during the cooling of alumina-containing melts, and the reaction mechanisms leading to the different product microstructures.
Experiments have been carried out in the laboratory to study reactions in the “Fe2O3”–CaO–SiO2–Al2O3 system in air. The samples were prepared from mixtures of high purity oxide powders of Al2O3 (>99.5 wt pct.), CaCO3 (>99.95 wt pct.), Fe2O3 (>99.5 wt pct.) and SiO2 (>99.9 wt pct.), all supplied by Alfa Aesar). Each of the powders was separately precalcined to produce pure oxides before preparing the mixtures. The nominal bulk compositions, A and B, investigated in the present study are given in Table 1. The compositions were selected based on the published experimental phase equilibria data for the liquidus of the “Fe2O3”–CaO–SiO25) system and subliquidus data for the system “Fe2O3”–CaO–SiO2–Al2O3 and SFCA phases,10) all in air. The SFC5,11,12,13) and SFCA10) solid solutions are stable over a relatively narrow range of CaO/SiO2 ratios.
| Bulk comp. | CaO | Fe2O3 | SiO2 | Al2O3 |
|---|---|---|---|---|
| A | 7.8 | 88.0 | 2.2 | 2.0 |
| B | 19.6 | 72.7 | 5.7 | 2.0 |
Samples A and B with approximately CaO/SiO2 = 3.5 by weight have been selected for the present study. The equilibrium phase assemblages in the pseudo-ternary “Fe2O3”–CaO–SiO2 system for these conditions are illustrated schematically in Fig. 1. It is assumed in this representation that Al2O3 behaves in an equivalent manner to Fe2O3. It can be seen that on cooling pure iron oxide liquid the magnetite phase, Fe3O4 is the first solid phase to form. At sub-liquidus temperatures, the magnetite phase becomes unstable relative to hematite, Fe2O3. For samples containing CaO and SiO2, it has been shown5) that, under equilibrium conditions, the incongruently melting compound, SFC, is formed. For samples high in Fe2O3, at temperatures below the ternary peritectic (1529 K), two possible reactions can take place, i) H + L → H + SFC, where H represents solid hematite and SFC the solid silicoferrite of calcium phase, and ii) the peritectic reaction, H + L → L + SFC. The SFC primary phase field is present over a narrow range of compositions located between the dicalcium silicate (C2S), CF2 and hematite primary phase fields. At lower temperatures there is a pseudo-binary eutectic reaction involving the formation of both the SFC and the C2S phases from the liquid, L + SFC → L + SFC + C2S.5) Other reactions take place at temperatures below this eutectic temperature, but these are deliberately not investigated or presented in the present paper for clarity. The peak temperatures used in the study were designed to produce, at the start of the experiments, either fully liquid or liquid + primary phase samples.
Schematic of the section through the “Fe2O3”–CaO–SiO2 system at CaO/SiO2 = 3.5 in equilibrium with air, illustrating the presence of the incongruently melting compound, SFC and the associated pseudo-ternary peritectic reaction (based on data from5)) and the compositions A and B investigated in the present study.
Note that the liquidus temperatures and phase equilibria in the “Fe2O3”–CaO–SiO2–Al2O3 system in air have yet to be completely characterised in this range of compositions, so the critical temperatures for the phase transformations for the system containing 2wt% Al2O3 will differ from those shown in Fig. 1.
Details of the experimental techniques used in the present study have been reported in a previous publication by the authors.6) The techniques enable individual samples to be cooled at a controlled rate, and each stage of the solidification process to be identified. The ability to rapidly quench the 0.1 g samples, and to accurately capture the microstructures and measure phase compositions present at a particular temperature, are critical to the experimental design. In the present study, controlled cooling rate of 2K/s has been used in all experiments. This cooling rate has been selected since this is typical of the rates observed on cooling iron ore sinters.3)
Following rapid quenching of the samples into water from selected temperatures, the resulting microstructures visible on polished cross-sections of the samples were examined using optical microscopy and electron microscope techniques. Scanning electron microscopy was used to obtain backscattered electron images, and electron probe X-ray microanalysis (EPMA) with wavelength dispersive spectrometers (JEOL8200L, trademark of Japan Electron Optics Ltd., Tokyo) was used to accurately measure the chemical compositions of the individual phases. EPMA was used at an acceleration voltage of 15 keV and a probe current of 15 nA. The Duncumb-Philibert ZAF correction procedure supplied with the JEOL JXA 8200L probe was applied. Appropriate reference materials (Al2O3, CaSiO3, Fe2O3, SiO2 from Charles M. Taylor, Stanford, CA) were used as standards. The iron concentrations in the phases were recalculated to Fe2O3 to unambiguously report the compositions of these phases. Phase identification has also been carried out using X-ray powder diffraction (XRD) performed with a Bruker D diffractometer with copper Kα radiation operated at 40 kV and 40 mA.
In the present study two series of experiments have been undertaken with samples of 72.7 and 88 wt% Fe2O3 respectively, each containing approximately 2 wt% Al2O3 and CaO/SiO2 = 3.5 by weight.
3.1. 72.7%Fe2O3, 2wt % Al2O3, CaO/SiO2 = 3.5Experiments were carried out using 72.7%Fe2O3 and a peak temperature of 1623 K (1350°C). Under these initial conditions the samples were completely molten. Separate samples were cooled at a controlled rate to selected temperatures, and then quenched to retain the phases and structures present at these temperatures. The microstructure formed at 1503 K (1230°C) is illustrated in Fig. 2. It can be seen that at the surface of the sample hematite crystals have heterogeneously nucleated at the gas/liquid interface and grown on cooling, forming facetted dendritic structures. Separately, in the interior of the sample, SFCA crystals have formed. It can be clearly seen that even when these SFCA crystals intersect the hematite there is no sign of the SFCA phase covering or encapsulating the hematite surface. Figure 2 also shows a portion of the sample that was not rapidly cooled; fine SFCA dendrites can be seen (Fig. 2(d)) to grow from the existing SFCA but no SFCA dendrites are formed on the hematite/liquid interface. Instead, fine crystals of C2S, which have formed on quenching, are present on the hematite surface.
Backscattered electron micrographs of sections of 72.7 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5. Samples heated to 1623 K (1350°C) in air, cooled in air at 2 K/s to 1503 K (1230°C). a) Sample surface, b) detail of a), c) interior of the sample, d) detail of c).
Samples containing 88%Fe2O3 were heated to 1683 K (1410°C) in air and cooled in air at a constant rate of 2K/s to the selected temperatures of 1623 K (1350°C), 1523 K (1250°C) and 1503 K (1230°C).
3.2.1. Fe3O4 Primary Phase FieldAt 1683 K (1410°C) the sample with bulk analysis of 88%Fe2O3 is in the magnetite primary phase field. From the micrographs (see Fig. 3) it is estimated that, at this temperature, this sample consists of approximately 80% solids. EPMA analysis of the magnetite solid solution shows that it contains 1.7 wt.%Al2O3 and 1.2 wt.%CaO, and the liquid phase contains 2.9 wt.% Al2O3, 25.5 wt.% CaO, 63.6 wt.%Fe2O3, 8.7 wt.% SiO2. Microanalysis indicates the magnetite is uniform in composition across all grains. Some fine dendritic secondary hematite crystals have formed on the surface of the magnetite and within the liquid during the quenching process. The liquid oxide phase wets the surface of the solid oxides and appears to hold the oxide grains together by capillary forces.
Backscattered electron micrographs of sections of 88 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5. Sample heated to 1683 K (1410°C) in air for 240 s.
(1) Secondary Hematite
Primary magnetite phase, secondary hematite and liquid phases are present in the sample cooled to 1623 K (1350°C) (see Fig. 4). The magnetite phase was found to be present predominantly in the centre of the sample in the form of anhedral grains surrounded by a liquid oxide. The fused polycrystalline magnetite grains contained residual isolated spherical pores 1–10 μm diameter. EPMA analysis shows that there is approximately 1.7 wt.% Al2O3 and 2.0 wt.%CaO in solid solution in this primary magnetite phase.
Backscattered electron micrographs of sections of 88 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5. Samples heated to 1683 K (1410°C) in air, cooled in air at 2 K/s to 1623 K (1350°C).
The secondary hematite grains are present on the outer portion of the sample and adjacent to pores connected to the gas phase. Microanalysis of the hematite phase shows that it contains less than 0.1 wt.% CaO but 1.5 wt.%Al2O3, a concentration only slightly less than in the magnetite at this temperature. The liquid contains 3.4 wt.%Al2O3, 24.0 wt.%CaO, 63.3 wt.%Fe2O3, 9.7 wt.%SiO2. In the presence of the liquid phase, this hematite appears to form by two mechanisms; i) primary phase hematite by crystallisation from the melt during cooling, and ii) secondary hematite by liquid assisted oxidation of the existing magnetite grains. The formation of the secondary hematite involves not only a change in crystal structure but the redistribution of the Al2O3 and CaO between the liquid, hematite and magnetite phases. Evidence supporting the presence of these different reaction mechanisms can be obtained by close inspection of the microstructures.
The secondary hematite grains differ from the original magnetite in that they are irregular in shape and new grain boundaries have been created; these new boundaries are not pores but consist of thin films of liquid surrounding the solids. Figures 4(c) and 4(d) show the growth interface between the original magnetite grains and the growing hematite. Note that the hematite does not completely cover the surface of the magnetite grains, as would be expected in a gas/solid system,14,15) rather it moves progressively into the magnetite at almost point contact. At present, the crystal orientations in the magnetite and hematite phases at this interface have not been determined but the interface could possibly be the close packed oxygen plane, which is present in both crystal structures. The transformation from magnetite to hematite at such an interface would involve only the change from ABCABCA packing of oxygen atoms in magnetite to ABABA in hematite. As pointed out in16,17) this can be achieved through the relatively rapid cooperative movement of partial dislocations along the common interface rather than the slow long range diffusion of ions and reconstructive processes.
At each side of the hematite growth tip, which is typically between 2–5 μm diameter, the hematite product is separated from the magnetite phase by a thin liquid film. The nucleation and growth of hematite on the original magnetite surfaces leads to the subdivision of the original magnetite crystals into smaller hematite subgrains. Behind the growth tip the hematite and magnetite phases are separated by planar interfaces and a thin liquid film. The arrow-like shape of the hematite crystals formed indicates that the rate of growth of the secondary hematite at these liquid/solid interfaces is constant.
The proposed reaction mechanisms and the reaction pathways for the liquid-assisted oxidation of the existing magnetite grains are shown schematically in Fig. 5. Since the concentrations of Al2O3 and CaO in the secondary hematite are lower than in the original magnetite these components are rejected into the magnetite ahead of the growing hematite/magnetite interface or into the liquid phase (step e3). These components may be removed by i) volume diffusion into the existing magnetite grains, and ii) diffusion along the hematite/magnetite interface. Volume diffusion coefficients in magnetite are relatively low even at the temperatures encountered in the present study.18) EPMA analysis does not indicate significant concentration profiles in the magnetite ahead of the growing interface. Also, no new Al2O3- or CaO-containing phases have been detected in the reaction product structures. The reactions and phase transformations are taking place over relatively short reaction times. All these observations indicate that the interfacial diffusion pathway is the dominant mass transfer mechanism in this case, and that the ultimate destination for the Al2O3 and CaO from the reaction interface is the liquid phase adjacent to the hematite/magnetite interface.
Schematic of the proposed reaction mechanisms taking place during the liquid phase oxidation of magnetite.
The liquid film, since it wets the surfaces of the solid oxides, forms a continuous pathway from the surface of the sample, the gas/liquid interface, to hematite/magnetite interface. The presence of the liquid close to the growing hematite crystal tip provides not only a sink for the Al2O3 and CaO but also a means of rapidly transferring oxygen from the gas phase to the magnetite/hematite interface. The present observations provide firm evidence to support the liquid phase mechanism of oxidation of the magnetite as proposed in.3) That is, the reaction at the gas/slag interface
| (1) |
(step e1) results in the consumption of electrons. These electrons are supplied from the reaction (step e2)
| (2) |
The overall reaction
| (3) |
The magnetite phase, which is unstable under these oxidising conditions, dissolves in the thin liquid film separating the magnetite and hematite phases. Hematite, which is now thermodynamically stable, is formed from the melt (step e4).
(2) Hematite Crystallisation
It can also be seen in Fig. 4(d) that the hematite is also growing into the melt, forming dendritic type crystals. The larger dendrites originating from the dense hematite are primary phase hematite formed during the controlled cooling of the melt. The fine scale dendrites, with multiple side arms, are formed during the quenching of the samples and are therefore not present at the reaction temperature.
3.2.3. Formation of SFCAThe upper limit of stability of the SFCA phase, has been shown to vary with bulk composition of the material.10) By analogy with the phase equilibria in the SFC system,5) this is equivalent to the temperature for the peritectic reaction, H + L → SFCA + L. The critical temperature for decomposition of hematite and SFCA formation in samples containing 2wt%Al2O3 in air have not been determined explicitly but SFCA has been observed in samples containing 1wt%Al2O3 at 1513 K (1240°C)10) and 2wt%Al2O3 at 1523 K (1250°C).4) It is therefore anticipated that, in the sample 88 wt% Fe2O3, 2 wt% Al2O3vand CaO/SiO2 = 3.5, SFCA will be present at 1523 K (1250°C).
Figure 6 shows the cross section of a sample cooled to 1523 K (1250°C). It can be seen that, near the outer surface of the samples in contact with air, Fig. 6(a), in addition to the secondary hematite, SFCA crystals and residual liquid are present. The SFCA present is in the form of needle-like, high aspect ratio crystals. In some cases, where there is a “rough” or irregular interface between the hematite and SFCA, the SFCA appears to originate at the hematite surfaces. In other cases, the SFCA appears go around the hematite crystals, Fig. 6(c) with apparently no interaction between the two phases.
Backscattered electron micrographs of sections of 88 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5. The samples were heated to 1683 K (1410°C) in air and cooled in air at a constant rate of 2 K/s to selected temperatures of 1523 K (1250°C).
Within the centre of the sample (shown in Fig. 6(b)) some of the original magnetite crystals are still present. The reason for this is that, as explained above, mass transfer of oxygen and ferric iron into the centre of the sample is required for complete oxidation. Where the transfer of oxygen from the surface has been impeded, such as adjacent to the platinum foil, or the diffusion path is long, magnetite grains are retained. These magnetite crystals are completely surrounded by thin SFCA crystals, which appear to be intimately connected to the magnetite host. It can be seen, in fact, that the SFCA phase appears to be growing into the magnetite from several different directions. The presence of the layer of SFCA crystals appears to completely separate the magnetite from the liquid phase. This solid layer once fully formed would appear to assist the retention of magnetite and prevent oxidation to hematite at this lower temperature. The SFCA crystals are faceted and in contact with the liquid form facetted dendrites; these microstructures indicate twinning has taken place within the crystals to form the side arms of the dendrites. The fine crystal size indicates these SFCA crystals formed at lower temperatures during the cooling process. In the same sample, adjacent to the magnetite are hematite crystals in direct contact with the liquid but also appearing to host SFCA growing from the surfaces; note the hematite crystals are not completely covered with SFCA. Since, as explained above, the sample is progressively oxidised from the gas/liquid interface to the interior of the samples it is not clear whether this SFCA was originally nucleated on magnetite grains that were subsequently oxidised to hematite, or the SFCA was formed directly on existing hematite crystals. The hematite grains are irregular in shape indicating that they are secondary rather than primary hematite grains.
These microstructures are further exemplified in the examples given in Fig. 7 for the sample cooled to 1503 K (1230°C). The SFCA crystals clearly penetrate the magnetite grains from multiple directions. Note that, at this temperature the dicalcium silicate phase, C2S has formed from the melt on the SFCA surfaces and as dendrites. EPMA analyses of the SFCA associated with the hematite and magnetite, are provided in Table 2. Figure 8 shows that the compositions of the SFCA lie on the C4S3–CF3 join. There are no systematic differences between the compositions of the SFCA in contact with the magnetite and the “hematite” surfaces. Close examination of the micrographs (Figs. 7(c) and 7(d)) and the compositional data indicates that the SFCA composition across the thickness of the laths is not uniform. From these cross-sections it appears that the Fe2O3 concentration in the crystals is highest at the centre (lighter shade in backscattered mode) and lowest adjacent to the liquid; this “coring effect” is consistent with a change in liquid composition during cooling and solidification of the melts. This compositional zoning within SFCA has been observed in samples of industrial iron ore sinter material (see Fig. 9).
Backscattered electron micrographs of sections of 88 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5. The sample heated to 1683 K (1410°C) in air, cooled in air at 2 K/s to 1503K (1230°C). a) Sample surface, b) interior of the sample, c) and d) details of the interior.
| Wt.% | Mol.% | ||||||
|---|---|---|---|---|---|---|---|
| Al2O3 | CaO | Fe2O3 | SiO2 | AlO1.5 | CaO | FeO1.5 | SiO2 |
| 3.77 | 17.71 | 72.13 | 6.45 | 5.28 | 22.55 | 64.51 | 7.67 |
| 4.01 | 16.67 | 72.28 | 6.04 | 5.69 | 21.52 | 65.51 | 7.28 |
| 4.29 | 15.96 | 73.83 | 5.94 | 6.05 | 20.44 | 66.40 | 7.10 |
| 5.01 | 15.65 | 73.78 | 4.67 | 7.12 | 20.23 | 67.00 | 5.64 |
| 4.20 | 15.61 | 74.53 | 5.54 | 5.95 | 20.08 | 67.32 | 6.65 |
| 5.12 | 15.34 | 74.56 | 3.98 | 7.31 | 19.91 | 67.96 | 4.82 |
| 4.69 | 15.49 | 74.55 | 4.30 | 6.69 | 20.11 | 67.98 | 5.21 |
| 4.28 | 14.16 | 75.75 | 3.65 | 6.24 | 18.76 | 70.49 | 4.51 |
| 4.49 | 14.32 | 76.21 | 3.19 | 6.51 | 18.90 | 70.66 | 3.93 |
| 3.14 | 14.72 | 77.22 | 3.67 | 4.55 | 19.41 | 71.52 | 4.51 |
| 4.54 | 13.45 | 77.93 | 3.21 | 6.55 | 17.65 | 71.86 | 3.93 |
| 4.05 | 13.63 | 78.52 | 3.74 | 5.81 | 17.76 | 71.88 | 4.55 |
| 4.44 | 13.84 | 77.48 | 2.32 | 6.48 | 18.38 | 72.27 | 2.87 |
| 4.62 | 12.84 | 79.94 | 2.58 | 6.64 | 16.78 | 73.42 | 3.15 |
| 4.19 | 13.05 | 78.39 | 2.19 | 6.17 | 17.46 | 73.64 | 2.73 |
| 4.49 | 11.84 | 80.61 | 1.96 | 6.56 | 15.74 | 75.26 | 2.43 |
Compositions of SFCA measured in the present study, recalculated assuming Al3+ – Fe3+ substitution behaviour and plotted against the C4S3–CF3 join. (bulk composition 88 wt% Fe2O3, 2 wt% Al2O3 CaO/SiO2 = 3.5).
Sample of material from industrial iron ore sinter exhibiting compositional zoning within the platy-SFCA phase formed on magnetite.31)
A necessary precondition for SFC and SFCA formation from the liquid is that, on cooling, the liquid composition passes through the SFC and SFCA primary phase fields; this is achieved primarily through control of the CaO/SiO2 ratio of the bulk material. It has been shown,6,7,8,9) however, that this on its own is not sufficient to guarantee the formation of the SFC phase. For example, in the “Fe2O3”–CaO–SiO2 system in air, hematite, magnetite, dicalcium silicate, calcium diferrite and calcium ferrite phases, all having primary phase fields in this system, have been observed to form on cooling. Analysis of the microstructures and phase assemblages formed at different temperatures during these various cooling sequences indicate that prior to the nucleation of C2S, CS, CF2 and CF phases, the melts become noticeably undercooled with respect to the new phase. This indicates difficulties in heterogeneous nucleation of these phases. The SFC-I phase is formed by nucleation directly from the melt and by heterogeneous nucleation on magnetite, but no SFC is observed in these samples.6,7,8,9) An example of SFC-I formation on undercooled magnetite in the alumina-free “Fe2O3”–CaO–SiO2 system in air is given in Fig. 10. The SFC-I phase was identified from EPMA analysis using the approach previously identified by the authors.7)
Backscattered electron micrograph showing the nucleation of SFC-I on magnetite. Sample 88 wt% Fe2O3, CaO/SiO2 = 3.5, heated to 1683 K (1410°C) in air, cooled in air at 2 K/s to 1483 K (1210°C).
Although there are limited data available on the phase equilibria and liquidus of the “Fe2O3”–CaO–SiO2–Al2O3 system in air10) in the compositional range of interest to the present investigation, the present study has shown that the presence of alumina results in the formation of SFCA during cooling of these melts and that nucleation appears to take place preferentially on the surfaces of the magnetite crystals.
X-ray diffraction studies have demonstrated that the crystal structures and site occupancies of cations in Ca2.1Mg1.2Fe5.55Si1.50Al3.65O20 (SFCA)19), Ca2Mg2Fe4.45Si2.15Al3.4O20 (SFCA),20) Ca2.90Mg0.95Fe10.11Al5.99O28 (SFCA-I)21,22) and SFCA-II23,24,25) are crystallographically closely related. It has also been shown that the alumina-free silico ferrite of calcium (SFC) phase has a crystal structure identical to the SFCA phase.26,27)
The SFCA group of minerals contain common polysomes, groups of structural units that are repeated in these crystal structures in selected sequences.26,28) The spinel polysome, <Sp> is a module of the spinel structure, oriented in the <111> directions, i.e. normal to the close packed oxygen plane, consisting of one tetrahedral and one octahedral unit. The pyroxene polysome, <Py> is a module, equivalent to one silicate chain of the pyroxene structure with both tetrahedral and octahedral units. The crystal structure of the SFCA phase, illustrated in Fig. 11,20) consists of alternating <Sp> and <Py> polysomes, i.e. <SpPy>,19,20,24,29) while the SFCA-I structure comprises of a repeat pattern of two <Sp> polysomes and one <Py> polysome, i.e. <SpSpPy>.21,22) In these minerals the cations have been shown to be distributed across both spinel <Sp> and pyroxene <Py> polysomes. Within these structures, Fe2+ and Fe3+ have been shown to preferentially occupy sites in the spinel polysome. Within the spinel polysome both tetrahedral and octahedral sites exist with Fe2+ preferentially occupying the octahedral sites and the tetrahedral sites fully occupied by Fe3+.29) In the pyroxene polysome, Mg2+ and Ca2+ occupy the octahedral sites and Si4+ and Al3+ the tetrahedral sites.19,20)
The crystal structure of SFCA, detailing the arrangement of the spinel (Sp) and pyroxene (Py) polysomes.20)
It has been demonstrated10) that the SFC and SFCA phases form solid solutions with notional end members CF3 and C4S3, and CF3, CA3 and C4S3 respectively, although these endmembers are not themselves thermodynamically stable at 1 atm pressure.[13] Based on the compositional range of the SFC and SFCA phases, the compositional range is explained by the substitution reaction; 2(Fe3+, Al3+) = (Ca2+, Fe2+) + Si4+.
The distribution of cations between the pyroxene and spinel polysomes suggests that the SFCA and SFCA-I phases can be stabilised by the addition of specific cations. In particular, it is suggested that the addition of Al3+, shown by Hamiliton et al.9) to distribute preferentially to the tetrahedral lattice sites, stabilises the pyroxene polysome, and in turn the SFCA <SpPy> and SFCA-I <SpSpPy> phases, as the result of the higher bond strength of Al–O relative to Fe–O.
The SFC-I, SFC and SFCA crystals have a common feature, they all contain almost complete layers of close packed oxygen atoms. This was pointed out by Hamilton et al.19) in relation to SFCA, who indicated that this provided a common structural relationship with hematite, which has a close packed plane, (100). This common plane represents a potential nucleation site for SFCA minerals during the solidification process. It should be pointed out, however that magnetite has four (4) sets of close packed planes of oxygen within its crystal structure, represented by the <111> family of planes, (111), (1-11), (11-1) and (-111).30) This being the case it would be expected that the probability of nucleation of SFCA on magnetite crystal surfaces would be greater than on hematite.
There is, however an additional, perhaps more important factor, to be taken into account. SFCA minerals actually contain spinel polysomes that are identical to those present in magnetite. The presence of these polysomes may explain why the SFCA phase appears to grow into and consume the magnetite crystals, as shown in Figs. 6 and 7 from multiple but specific directions. In contrast, the SFCA formed on hematite appears to grow predominantly into the melt not into the hematite phase (also shown in Figs. 6 and 7).
There are other examples that illustrate the influence of crystal structures on nucleation and association of phases on solidification. In the “Fe2O3”–CaO–SiO2 system in air the CF2 phase has been found to nucleate on the surfaces of hematite crystals.6,7,8,9) Both CF2 and hematite have fully occupied close packed oxygen planes; unlike the “almost” close packed plane of the SFCA phase, which has systematic vacancies. The CF2 crystal is also a polysomatic mineral, containing a ‘spinel’ polysome and a ‘Motif-F’ polysome.31) Both the CF2 and SFCA phases have spinel polysomes, however in SFCA, the spinel polysome is oriented normal to the oxygen close packed planes whilst in the CF2 crystal it is parallel to the oxygen plane. These differences in crystal structures appear to significantly influence the heterogeneous nucleation of new phases from the melt.
The preferential nucleation of the SFC-I and SFCA phases on magnetite, observed in the present study, in the controlled cooling experiments suggests that the SFC-I, SFC and SFCA crystals share a particular structural relationship with magnetite but not with hematite. The SFCA group of phases contains both spinel and pyroxene modules. It is argued that these spinel modules, identical to the magnetite crystal structure, enable the SFCA group of minerals to form at the magnetite/liquid interface.
In summary, whilst both hematite and magnetite have close packed planes on which SFCA nucleation can take place, the presence of the spinel polysomes, common to magnetite and SFCA phases, appears to significantly increase the probability of SFCA and SFCA-I nucleation and growth on the magnetite phase.
The sintering of iron ore is a complex process involving changes to the physical states of the phases present in the feed and products, and heterogeneous chemical reactions taking place in constantly changing temperatures. This complexity has made it difficult to identify and understand the key reactions taking place. Fundamental experimentally-determined information on phase equilibria in the “Fe2O3”–CaO–SiO25,11,12,13)and “Fe2O3”–CaO–SiO2–Al2O310) systems have provided the essential first steps for understanding and interpreting these changes. The authors have argued in recent research3,4,5,6,7,8,9) that the phases and microstructures observed in sinter products are primarily formed as a result of solidification of the melts, and that the complexity of the system can be best understood by systematic experimental studies of these solidification processes and reactions taking place on cooling.
5.1. Secondary Hematite FormationFrom detailed examination of the microstructural changes in the present and previous fundamental studies of magnetite oxidation, three mechanisms have been identified,
i) Liquid phase assisted oxidation, as observed in the current study and in,3)
ii) Bulk diffusion of cations through the solid hematite product oxide forming shrinking core microstructures, as in the case of gas/solid reactions,3,14,15) and
iii) Hematite lath growth, also in the case of gas/solid reactions.16,17)
Microstructures characteristic of mechanisms ii) and iii) are not commonly observed in most industrial sinters.
5.2. SFCA FormationPhase equilibrium studies have shown that SFC and SFCA phases are only formed from material having high CaO/SiO2 ratio, typically greater than 2, although this value will depend on the concentrations of the other chemical components present in the sinter.5,10,11,12,13) Typically, local CaO/SiO2 values lower than this critical value will result in the formation of isolated iron oxide grains in a matrix of glassy material; the glassy continuous phase leads to sinter with low mechanical strength and low softening temperatures.
It has been established10) that SFCA formation is favoured by the presence of Al2O3 in the bulk material and the SFC phase by high oxygen partial pressures.13) In the present study and previous studies3) it has been shown that the nucleation of SFCA is favoured by the presence of magnetite during the cooling of the melts and slow cooling rates from peak bed temperatures in above-solidus conditions. These findings appear to be consistent with the structures observed in industrial sinters.
In industrial iron ore sinter practice, the presence of magnetite can be achieved by the use of this mineral in the ore feed, and by generating the high peak bed temperatures and lower oxygen partial pressures at peak bed temperature.
Long residence times at above solidus temperatures (approximately 1200°C) will promote the nucleation and growth of the SFCA phase from the melts. These conditions can be achieved by, for example, decreasing the thermal load in the bed, adjusting process parameters to match velocity of heat and combustion fronts, and increasing fuel rates. In addition, in a previous study,3) it was demonstrated that increasing the proportion of liquid phase present by increasing the peak bed temperature, leads to high sinter density and low connected pore volume; this structure means the effective oxygen pressure within the lump sinter is lower than in the air that is being passed through the sinter bed. The result is the retention of magnetite formed at the peak bed temperature, a low proportion of secondary hematite and high mechanical strength.
Low peak bed temperatures and short times at above-solidus temperatures appear to result in limited densification and sintering, high connected pore volume, rapid oxidation of any original or primary magnetite that may be present to secondary hematite and low mechanical strength.
Experimental studies of the controlled cooling of selected “Fe2O3”–CaO–SiO2 and “Fe2O3”–CaO–SiO2–Al2O3 melts in air have shown that the reactions do not follow equilibrium or Scheil-Gulliver solidification pathways. In both of these systems hematite, formed as the primary phase, or as a result of oxidation of the primary magnetite, is, under the modest cooling rates encountered in industrial sintering processes, retained on cooling rather than undergoing decomposition as part of the peritectic reaction, L + H → SFC + L.
The SFC phase is not readily formed on cooling of “Fe2O3”–CaO–SiO2 melts, however the SFC-I phase is observed to form on metastable magnetite. In “Fe2O3”–CaO–SiO2–Al2O3 melts in air containing 2wt% Al2O3, the SFCA phase is found to nucleate preferentially at the interface between the magnetite and liquid phases. It is argued that this preferential nucleation is related to common crystallographic features in the magnetite and SFCA phases.
Rapid secondary hematite formation is shown to take place through the liquid phase assisted oxidation of the primary magnetite grains. This mechanism also results in the release, into the liquid phase, of calcium initially dissolved in the primary phase magnetite and a reduction in hematite grain size.
The findings obtained under these controlled laboratory conditions are consistent with microstructures observed in industrial iron ore sinters, and may assist in the interpretation of the origins of these complex microstructures and the further optimisation of the sintering process.
The authors would like to thank the Australian Research Council Linkage program and BHP for financial support to enable this research to be carried out, and the Centre of Microstructure and Microanalysis (CMM), the University of Queensland for providing electron microscope facilities that enabled the microanalytical measurements to be undertaken. This research was also supported by an Education Endowment Fund (EEF) scholarship from the Australasian Institute of Mining and Metallurgy (AusIMM) and an Australian Government Research Training Program (RTP) Scholarship.
