2022 年 62 巻 4 号 p. 652-657
The relative effects of dolomite and serpentine/olivine as MgO flux sources on the formation of SFCA and SFCA-I iron ore sinter bonding phases during heating was investigated using in situ synchrotron X-ray diffraction. Results showed that a sinter mix containing serpentine/olivine as the primary source of MgO is more likely to form high quality sinter due to the higher proportion of SFCA-I being formed compared to SFCA. SFCA-I is the phase thought to impart high strength and good reducibility characteristics to iron ore sinter. For example, in mixtures containing 2 mass% MgO, when dolomite was used the maximum concentrations of SFCA-I and SFCA attained were 19 and 9 mass%, respectively, and when serpentine/olivine was used the maximum concentrations were 20 and 3 mass%, respectively. In mixtures containing 3 mass% MgO, the maximum concentrations were 18 and 4 mass% (dolomite) and 28 and 2 mass% (serpentine/olivine).
Dolomite, CaMg(CO3)2, and other MgO-bearing materials such as serpentine, (Mg,Fe)3Si2O5(OH)4, and olivine, Mg2SiO4, are commonly used in addition to limestone, CaCO3, as flux components for the production of iron ore sinters.1) The MgO is added to the sinter blend to satisfy blast furnace chemistry requirements where the presence of Mg ensures good slag flowability and desulphurisation properties. The addition of up to 3 mass% MgO is typical. Previous studies have shown that the major effect of MgO is to stabilise magnesiospinel and suppress complex calcium ferrite bonding matrix formation.1,2,3)
Suppression of SFCA and SFCA-I bonding phases and replacement by higher amounts of spinel and glassy silicates has important implications for the properties of sinter.4,5,6) The bonding phases are usually the major mineral constituents of the sinter structure and these impart strength to the sintered mass. A decrease in their overall abundance at the expense of spinel and silicate melt phases with increasing MgO levels can result in a considerable drop in sinter strength. In particular, the formation of a vitreous glassy matrix and precipitation of dicalcium silicate produce a high amount of stress causing weakness.6,7) Re-oxidation of magnetite to secondary skeletal hematite is less likely during cooling, however, which will have a positive effect on sinter reduction degradation index (RDI).
SFCA and SFCA-I are key bonding matrix phases, and a review of their composition, structure and formation conditions has recently been undertaken.8) The review was focused on a high-Fe, low-Si form called SFCA-I (e.g., Ca3.2Fe2+0.8Fe3+14.7Al1.3O28), and a low-Fe form called SFCA (e.g., Ca2.3Mg0.8Si1.1Al1.5Fe8.3O20). The review also referenced the crystal structure work on Mg-rich SFCA (SFCAM) of Sugiyama et al.9) A review of the phase equilibria in iron ore sinters has also recently been published.10) We have also recently compared the effects of the use of different alumina sources (i.e., gibbsite, kaolinite, and aluminous goethite) on SFCA and SFCA-I formation during heating. Alumina source had a significant effect on SFCA and SFCA-I formation, with the use of kaolinite maximising the formation of SFCA-I (often described as being platy, needle-like or fibrous in cross section, compared to SFCA which tends to be columnar or blocky) which is the more desirable bonding matrix resulting in higher sinter strength and better reducibility.11,12,13,14) On the other hand, iron ore containing gibbsite as the primary source of alumina is less likely to form high quality sinter due to the low amounts of SFCA and SFCA-I which form.15) A similar study on the effect(s) of MgO source (i.e. dolomite vs serpentine/olivine) on SFCA and SFCA-I formation during heating using in situ XRD would be another significant contribution to the field of iron ore sinter bonding phase formation fundamentals, and this paper reports the outcomes of these analyses. This paper differs from our previous study which reported the effect of MgO on the crystallisation of SFCA phases from melt during cooling.16) Whilst the relative effects of dolomite and serpentine flux have been assessed by Loo et al.,2) that study compared the amounts of spinel and SFCA which were present in pilot scale sinters, and there was no reference to, or quantification of, SFCA-I. A key distinction, therefore, between that previous work and the novel work presented here is that the formation of SFCA and SFCA-I during heating has been characterised in intricate detail.
Four sinter mixture samples were prepared: 1) SM4/5-2MgO-Dolomite, 2) SM4/5-2MgO-Serpentine/Olivine, 3) SM4/5-3MgO-Dolomite, and 4) SM4/5-3MgO-Serpentine/Olivine. Each mixture contained either 75.81 mass% Fe2O3, 13.80 mass% CaO, 3.49 mass% SiO2, 4.90 mass% Al2O3 and 2 mass% MgO (i.e., 2MgO), or 75.05 mass% Fe2O3, 13.66 mass% CaO, 3.45 mass% SiO2, 4.85 mass% Al2O3 and 3 mass% MgO (i.e., 3MgO). The base mixture, SM4/5, has been utilised in several of our previous investigations.16,17,18,19) The compositions were designed to model the reactive ultrafine (< 1 mm) component of a sinter mix rather than being representative of the bulk composition of a sinter blend which contains coarse nuclei (up to 6.3 mm in size), gangue and flux particles.
The mixtures were prepared from synthetic hematite, Fe2O3 (Acros Organics, 99.999%); calcite, CaCO3 (Thermo Fisher, 99.95%); quartz, SiO2 (Sigma Aldrich, 99.995%); and gibbsite, Al(OH)3 (Alcan OP25 Super White, 99.9%). Figure 1 shows XRD patterns collected ex situ for the dolomite and serpentine/olivine powders used. Rietveld refinement-based quantitative phase analysis of these XRD patterns revealed the dolomite powder contained 0.6 mass% quartz and 0.3 mass% calcite in addition to dolomite. The serpentine/olivine powder contained serpentine (lizardite and antigorite, 32 mass%) and olivine (forsterite, 18 mass%) along with smectite clay (nominally montmorillonite, (Na,Ca)0.3(Al,Mg)2Si4O10(OH)2.n(H2O), 22 mass%), chlorite (nominally clinochlore, Mg5Al(AlSi3O10)(OH)8, 16 mass%), pyroxene (nominally diopside, MgCaSi2O6, 7 mass%), magnesite (MgCO3, 4 mass%) and quartz (1 mass%). Table 1 shows the bulk compositions of the dolomite and serpentine/olivine samples as determined by X-ray fluorescence analysis. Weighed powders were micronized in ethanol for 4 min g−1, centrifuged and dried at 333 K. They were then remixed by hand in a mortar and pestle to ensure homogeneity.
XRD patterns (ex situ) collected for the (a) dolomite and (b) serpentine/olivine powders. The numbers in parentheses are the Miller indices of the major peaks of the constituent phases.
| Oxide (mass%) | Dolomite | Serpentine/Olivine |
|---|---|---|
| Fe2O3 | 1.0 | 7.9 |
| SiO2 | 1.1 | 40.4 |
| Al2O3 | 0.3 | 2.1 |
| CaO | 30.0 | 2.2 |
| MgO | 20.7 | 35.2 |
| Mn3O4 | 0.2 | 0.1 |
| Na2O | 0.05 | 0.1 |
| K2O | 0.05 | 0.01 |
| LOI1000 | 46.4 | 11.5 |
In situ synchrotron XRD (S-XRD) experiments were performed for the four sinter mixture samples on the powder diffraction beamline at the Australian Synchrotron.20) Throughout heating the Anton Paar chamber was fed by a continuous flow of a 0.5% O2 in N2 gas mixture (pO2 = 5 × 10−3 atm); Hsieh and Whiteman found that this pO2 maximised the formation of Ca-rich ferrites while still producing mineral assemblages similar to those found in industrial sinters.21) For the 2MgO mixtures, a heating rate of 20 K min−1 was used from 298–1623 K. For the 3MgO mixtures, a heating rate of 20 K min−1 was used from 298–873 K, then 10 K min−1 for 873–1623 K which corresponds to the region of Ca-rich ferrite phase formation and decomposition. Although the temperature-time profiles used were significantly slower than those encountered in industrial sintering, it allows for phase formation/decomposition events to be characterised with reasonable temperature resolution, and for XRD data to be of sufficient quality to establish trends in phase behaviour with increasing MgO concentration. Because different heating rates were used for the 2MgO and 3MgO experiments, comparisons are only made between results obtained in dolomite and serpentine/olivine experiments with the same MgO concentration, and not between experiments with different MgO concentrations.
Laboratory in situ XRD experiments were also performed, in the range 298 to 1623 K and at the same pO2, for the dolomite and serpentine/olivine starting materials. Details of the synchrotron and laboratory experimental setups and conditions have been described in detail previously.18) It is important to note that the synchrotron and laboratory experiments were performed using different X-ray wavelengths (λ = 1.1055 Å (0.11055 nm) for synchrotron; λ = 1.789 Å (0.1789 nm) for laboratory). Rietveld refinement-based quantitative phase analysis (QPA) of the individual S-XRD datasets was performed using TOPAS (Version 6). Details of the QPA procedure have been described previously.22)
Figure 2 shows plots of accumulated in situ XRD data, viewed down the intensity axis and with temperature plotted vs 2θ, collected for the laboratory experiments designed to examine the thermal decomposition of the dolomite (Fig. 2(a)) and serpentine/olivine (Fig. 2(b)) starting materials. As is apparent from Fig. 2(a), the decomposition of CaMg(CO3)2 to CaO and periclase (MgO) commenced at ~900 K and was complete by ~1080 K. As temperature increased further the CaO and MgO peaks narrowed due to increasing crystallite size and/or decreasing lattice microstrain in the two phases. As is apparent from Fig. 2(b), decomposition of serpentine [(Mg,Fe)3Si2O5(OH)4] commenced above 900 K, initially associated with loss of chemically bound water, before the major decomposition/breakdown reaction was complete by ~1130 K which resulted in the formation of additional olivine [forsterite, Mg2SiO4].23) In addition to Mg2SiO4, peaks corresponding to enstatite (MgSiO3) were also observed in the in situ XRD data collected above ~1150 K. The retarded development of enstatite until temperatures above ~1150 K is attributed to the nature of the reaction by which it is formed (i.e., reaction between olivine and silica).24)
In situ laboratory XRD data collected for the (a) dolomite and (b) serpentine/olivine starting materials, viewed down the intensity axis, during heating in the range 298–1623 K and at pO2 = 5 × 10−3 atm.
Figure 3 shows the plot of accumulated in situ S-XRD data collected during the experiment performed for the SM4/5-3MgO-Dolomite mixture. During heating the sequence of phase decomposition/formation events was: decomposition of Al(OH)3; decomposition of CaCO3 and CaMg(CO3)2; formation of CaO and MgO; formation of dicalcium ferrite (C2(F1-xAx)); formation of monocalcium ferrite (CF) and CFA (composition determined in Reference 17); formation of spinel; formation of SFCA-I; formation of SFCA; and, formation of gehlenite (Ca2Al2SiO7). At 1623 K the phase assemblage was spinel in a FeO–Fe2O3–CaO–MgO–Al2O3–SiO2 melt. Due to the low abundance of MgO and the broad profile of the MgO peaks, the characteristic (200) MgO peak–which is evident in Fig. 2(a), at ~50° 2θ–is not apparent in Fig. 3 (this peak would be expected at ~30.4° 2θ in this experiment, owing to the different X-ray wavelength used).
In situ S-XRD data collected for the SM4/5-3MgO-Dolomite mixture, viewed down the intensity axis, during heating in the range 298–1623 K and at pO2 = 5 × 10−3 atm.
Figures 4(a)–4(d) show the accumulated S-XRD data over narrower 2θ and temperature ranges for the SM4/5-2MgO-Dolomite, SM4/5-2MgO-Serpentine/Olivine, SM4/5-3MgO-Dolomite and SM4/5-3MgO-Serpentine/Olivine mixtures, respectively. The narrower 2θ range of ~12.5–20° 2θ was selected as it incorporates the (111) spinel peak at ~13.7° 2θ and includes the angular range between 18–20° 2θ where the identification of SFCA-I and SFCA is clearest. It is evident from these plots that the spinel phase begins to form at a lower temperature in the 2MgO-Dolomite (~1260 K) and 3MgO-Dolomite (~1180 K) mixtures, compared to the Serpentine/Olivine mixtures (~1300 and ~1240 K for 2MgO and 3MgO, respectively). This indicates that the Mg present in periclase (i.e., after the decomposition of dolomite in the 2MgO-Dolomite and 3MgO-Dolomite mixtures) is more reactive–and therefore available to stabilise magnesiospinel at lower temperatures–than the Mg present in Mg2SiO4 and MgSiO3 in the 2MgO-Serpentine/Olivine and 3MgO-Serpentine/Olivine mixtures. Whilst it is not possible using the S-XRD data to categorically determine whether all the periclase, Mg2SiO4 and MgSiO3 was consumed by solid-state reaction at high temperatures, we do consider it likely given the fine-grained nature of the materials and, therefore, their high reactivity.
In situ S-XRD data collected for the (a) SM4/5-2MgO-Dolomite, (b) SM4/5-2MgO-Serpentine/Olivine (c) SM4/5-3MgO-Dolomite and (d) SM4/5-2MgO-Serpentine/Olivine mixtures over the range ~12.5 to 20° 2θ and 1057 to 1623 K.
Other observations from Fig. 4 are: i) the CF peaks in the data collected for the 2MgO-Serpentine/Olivine and 3MgO-Serpentine/Olivine mixtures (Figs. 4(b) and 4(d), respectively) are of noticeably lower intensity than those in the data collected for the Dolomite mixtures (Figs. 4(a) and 4(c)); ii) there are no β-SiO2 peaks evident in Fig. 4(d) for the 3MgO-Serpentine/Olivine mixture. This is because the quartz present as part of the serpentine/olivine powder (at 1 mass% level, see Section 2.1) was below the detection limit of the synchrotron measurement. The serpentine/olivine powder was the only Si-containing material added to make the starting 3MgO-Serpentine/Olivine mixture (i.e., no additional quartz was added to achieve the required Si content of 3.49 mass% SiO2); and iii) an additional olivine phase formed in the experiments performed for the 3MgO-Dolomite and 3MgO-Serpentine/Olivine mixtures. It is distinct from the serpentine decomposition product observed in Fig. 2(b), which was identified as forsterite. The peaks evident in Figs. 4(c) and 4(d) for this phase were matched with International Centre for Diffraction Data (ICDD) database entry 04-017-7265, which has the nominal formula Mg0.04Fe1.56SiO4.
3.3. QPA and Effects of Dolomite and Serpentine/Olivine on SFCA and SFCA-I FormationFigure 5 shows the results of the Rietveld refinement based QPA performed on the in situ XRD data collected during heating of the SM4/5-2MgO-Dolomite mixture, with the concentration of each phase shown as a function of temperature. Figure 6(a) shows the same QPA results but over a narrower temperature range (1023–1573 K), and Figs. 6(b), 6(c) and 6(d) show the results over the same narrower temperature range for the SM4/5-2MgO-Serpentine/Olivine, SM4/5-3MgO-Dolomite and SM4/5-3MgO-Serpentine/Olivine mixtures, respectively. Because of the very low abundance of CF in the experiments performed for the Serpentine/Olivine mixtures, curves for CF have been omitted from Figs. 6(b) and 6(d) for clarity.
Results of the Rietveld refinement-based quantitative phase analysis of the in situ XRD data collected in the range 298–1573 K for the SM4/5-2MgO-Dolomite mixture.
Results of the Rietveld refinement-based quantitative phase analysis of the in situ XRD data collected in the range 1023–1573 K for the (a) SM4/5-2MgO-Dolomite, (b) SM4/5-2MgO-Serpentine/Olivine, (c) SM4/5-3MgO-Dolomite and (d) SM4/5-3MgO-Serpentine/Olivine mixtures. The dashed vertical lines at 1273 and 1373 K in (a) and (b) coincide with the peak C2(F1-xAx) and CFA concentrations attained in the experiment performed for 2MgO-Dolomite. The dashed vertical lines at 1223 and 1323 K in (c) and (d) coincide with the peak C2(F1-xAx) and CFA concentrations attained in the experiment performed for 3MgO-Dolomite.
In Figs. 6(a) to 6(d), the initial and more gradual stages of spinel formation (e.g., between ~1260 and 1460 K for the 2MgO-Dolomite sample) are due to solid state reaction between hematite and the Mg-bearing material (e.g., MgO from the decomposition of dolomite). The more rapid increase in spinel concentration (e.g., in the range ~1460 to 1520 K) coincides with incongruent melting of SFCA-I and/or SFCA. This latter effect has been observed consistently in our previous in-situ XRD work,17,18,22) and the magnitude of the increase in spinel concentration depends on the total amount of SFCA phases which formed. The results in Fig. 6 indicate that for a given MgO concentration the use of dolomite results in a higher maximum concentration of spinel when compared with the use of serpentine/olivine. The maximum spinel concentrations for the 2MgO-Dolomite and 2MgO-Serpentine/Olivine mixtures were 20 and 12 mass%, respectively, and for the 3MgO-Dolomite and 3MgO-Serpentine/Olivine mixtures they were 23 and 20 mass%, respectively.
The results shown in Fig. 6, importantly, indicate that the use of serpentine/olivine results in significantly higher SFCA-I-to-SFCA ratio compared to the use of dolomite. For the 2MgO mixtures, the maximum concentrations of SFCA-I and SFCA are: 20 and 3 mass% for Serpentine/Olivine; 19 and 9 mass% for Dolomite. And, for the 3MgO mixtures: 28 and 2 mass% for Serpentine/Olivine; 18 and 4 mass% for Dolomite. Given that SFCA contains more Si than SFCA-I, this points to the reactivity of the Si-containing material in the Serpentine/Olivine mixtures (i.e., in the case of the 3MgO mixture, solely Mg2SiO4 and MgSiO3 which are the decomposition products of serpentine (see Section 3.1); in the case of the 2MgO mixture, Mg2SiO4 and MgSiO3 and a small amount of added quartz). The Mg2SiO4 and MgSiO3 appear to have lower reactivity than the quartz. Given that SFCA-I is the matrix phase thought to impart high strength and good reducibility characteristics to iron ore sinter, the significance of these results is that they suggest that for a given MgO concentration the use of serpentine/olivine as MgO source in the flux would produce sinter of superior quality compared to the use of dolomite.
In terms of the intermediate calcium ferrite phases C2(F1-xAx) and CFA, the maximum concentrations attained were: for 2MgO-Dolomite, 18 and 15 mass%; for 2MgO-Serpentine/Olivine, 19 and 17 mass%; for 3MgO-Dolomite, 17 and 17 mass%; and for 3MgO-Serpentine/Olivine, 19 and 15 mass%. So, no significant variation between experiments is apparent. What is apparent, however, is a shift of the peaks in the C2(F1-xAx) and CFA curves. In Fig. 6(a) which is for the 2MgO-Dolomite mixture, the dashed lines at 1273 and 1373 K coincide with the peaks in the C2(F1-xAx) and CFA curves. In Fig. 6(b) for the 2MgO-Serpentine/Olivine mixture, the peaks in the C2(F1-xAx) and CFA curves are shifted to higher temperature relative to the curves in Fig. 6(a). Similarly, In Fig. 6(d) for the 3MgO-Serpentine/Olivine mixture, the peaks in the C2(F1-xAx) and CFA curves are shifted to higher temperature relative to the curves in Fig. 6(c). These results suggest the incorporation of Mg into the C2(F1-xAx) structure, with the C2(F1-xAx) formation mechanism being CaO + Fe2O3 + Al-oxide + MgO, or CaO + Fe2O3 + Al-oxide + Mg2SiO4/MgSiO3. In addition, because C2(F1-xAx) is at least partially consumed by reaction with Fe2O3 to form CFA, this suggests the incorporation of Mg into the CFA structure. The temperature of the peaks is dependent on the reactivity of the MgO source.
Finally, all the results presented here only apply during heating up to a point of achieving spinel in a FeO–Fe2O3–CaO–MgO–Al2O3–SiO2 melt. We hypothesise that there would be no significant effects of using dolomite versus serpentine/olivine during cooling in the reactive ultrafine component of a sinter mix, with crystallisation of SFCA from the melt (note that there is no evidence in the literature of SFCA-I crystallisation occurring during cooling) being influenced solely by the MgO concentration and not its mineral form of introduction to the sintering process. Recent in situ XRD work confirmed an overall suppression, with increased MgO concentration, of the amount of Ca-rich ferrite phase formation during cooling.16) Of course, if there is undetected unreacted MgO or Mg2SiO4/MgSiO3 at the maximum temperature then our hypothesis may prove incorrect. In addition, variations of the amount of MgO contained in the spinel phase between the four systems and, therefore, variations in the melt composition may also mean our hypothesis is incorrect. This could be tested as future work.
The effect of MgO added in the form of dolomite and serpentine/olivine on the formation of SFCA and SFCA-I iron ore sinter bonding phases during heating in synthetic mixtures was investigated using in situ X-ray diffraction. Results showed that if dolomite is added as the fluxing agent to a sinter mixture, the MgO generated from dolomite decomposition is more reactive than the MgO-bearing phases generated from serpentine decomposition (olivine, (Mg,Fe)SiO4 and enstatite, MgSiO3) and magnesiospinel is stabilised to lower temperatures. The results also showed that if serpentine/olivine is added as the fluxing agent then the proportion of SFCA-I to SFCA is greater. This has important implications for the physical properties of sinter, since SFCA-I is thought to impart high strength and good reducibility characteristics to iron ore sinter, and therefore suggests that use of serpentine/olivine fluxing agent would be beneficial to achieve the favourable influence of MgO in the blast furnace.
This research was undertaken on the powder diffraction beamline (10BM1) at the Australian Synchrotron (part of ANSTO), Victoria, Australia, under beamtime award AS193/PD/15014. Rachel Pattel (formerly CSIRO Mineral Resources, Clayton, VIC, Australia) is thanked for sample preparation. James Manuel (CSIRO Mineral Resources, Pullenvale, QLD, Australia) is thanked for providing the dolomite and serpentine/olivine materials. Steve Peacock (CSIRO Mineral Resources, Clayton, VIC, Australia) is thanked for XRF analysis.
