2014 Volume 54 Issue 12 Pages 2713-2720
This study aims to utilize the high Al2O3 pisolitic ore in sintering process by designing the quasi-particle where the pisolitic ore is used as nuclei and ultra-fine hematite and magnetite ores are employed as adhering fines. The assimilation behavior between nuclei and adhering fines was investigated through microstructure analysis and it was correlated to sinter quality. When ultra-fine hematite ore was used as adhering fines, the low viscous melt of CaO·Fe2O3 was formed in the assimilation. Since this results in the low extent of Al2O3 localization and the porous structure, the detrimental effect of Al2O3 on the strength was not fully controlled. On the other hand, for the ultra-fine magnetite ore, 3CaO·Fe2O3·3SiO2 melt with high viscosity was predominantly participated in the assimilation. The assimilation was suppressed by the formation of ‘interfacial layer’. Due to the dense structure and high extent of Al2O3 localization, the detrimental effect of Al2O3 on strength was reasonably controlled. The quasi-particle comprising high Al2O3 pisolitic ore and ultra-fine magnetite ore showed the equivalent sinter quality to the quasi-particle sample consisting of nuclei of dense hematite and adhering fines of ultra-fine hematite resulting in the high sinter quality.
Outstanding change in ironmaking area of global steel industry is ascribed to the depletion of high quality iron ore resources and to the increasing proportions of gangue materials in raw materials. In particular, the loss on ignition (LOI) and the Al2O3 content of iron ores are expected to gradually increase.1,2) Under these circumstances, extensive use of low quality iron ores such as high Al2O3 pisolitic ore is unavoidable in ironmaking process.
There have been several researches concerning the adverse effects of pisolitic ore and Al2O3 content on sinter quality. It is well known that pisolitic ores have detrimental influence on sinter productivity due to its low bulk density and poor agglomeration behavior. In addition, the porous structure of sintered pisolitic ore has adverse effect on sinter strength, which is resulted from the thermal decomposition of high combined water in the ores during sintering.3,4,5) On the other hand, an increase of Al2O3 content in sinter mix results in the increase of the formation temperature and viscosity of melt. It deteriorates sinter quality such as strength and reduction degradation characteristics (RDI), which would cause some problems with permeability during the reduction in the blast furnace.1,6,7) To control the detrimental effect of Al2O3 on the sintering process, Lu et al.2) have suggested several solutions: (1) addition of magnetite, MgO and CaO for neutralization of Al2O3 effect, (2) dilution of Al2O3 by adding high reactive ores, (3) localization of Al2O3 by selective granulation, (4) optimization of raw materials and sintering process. Among the above solutions, the localization of Al2O3 might be feasible by employing quasi-particle concept.
With the intention of overcoming the declining quality of iron ore resources, the concept of quasi-particle has been introduced to utilize low quality iron ore in sintering process.1,8,9) Optimal design of quasi-particle structure would minimize the disadvantages of low quality ores, which can maintain and improve the sinter quality. For example, Kim et al.1) found that the productivity and quality of sinter were improved without increasing the Al2O3 content in sinter by segregating high Al2O3 pisolitic ore in the center of quasi-particle and by coating its surface with ultra-fine iron ore. The essence of quasi-particle design is to control the assimilation behavior because it has large influence on sinter quality such as strength and reducibility.10)
The present study aims to utilize the high Al2O3 pisolitic ore in sintering process by designing the quasi-particle where the pisolitic ore is used as nuclei and ultra-fine hematite and magnetite ores are employed as adhering fines. Figure 1 shows the concept of quasi- particle design employed in the present study. The assimilation behavior between nuclei and adhering fines was investigated through microstructure analysis. Depending on the assimilation characteristics of quasi-particle, the high content of Al2O3 may be localized in certain region and the adverse effects of Al2O3 can be suppressed. In addition, small scale sintering test was performed to correlate the assimilation behavior with sinter quality such as strength and reducibility.
Concept of quasi-particle employed in the present study.
Table 1 shows the chemical compositions of iron ores used in the present study. Four kinds of iron ores were used for the preparation of quasi-particles. High Al2O3 pisolitic ore (Ore A) and dense hematite ore (Ore C) were adopted as nuclei and ultra-fine hematite (Ore N) and magnetite (Ore J) ores were used as adhering fines of the quasi-particles. Each ore was sieved to 4–5 mm for nuclei and under 100 μm for adhering fines. The quasi-particles were granulated with 50% nuclei and 50% adhering fines on the weight basis of ores. CaCO3 was added to adhering fines to have the CaO/Ore value of 0.1 in the final sample composition. Table 2 shows the chemical compositions of prepared quasi-particle samples. Experiments and analyses were performed focusing on the quasi-particles comprising high Al2O3 pisolitic ore A (A-N and A-J). The quasi-particle samples with dense hematite ore C (C-N and C-J) were also prepared for the sake of comparison. Figure 2 shows the typical structure of high Al2O3 pisolitic ore (Ore A). It was comprised of hematite in the core and goethite surrounding it. The Si was concentrated in the core region of hematite, while Al was concentrated in the surrounding region of goethite.
Typical structure of high Al2O3 pisolitic ore (Ore A).
In order to determine the thermal characteristics of the ores used as nuclei, loss on ignition (LOI) was measured in accordance with ASTM-D 7348. Firstly, the ores were pulverized and sieved to under 250 μm and dried for 2 hrs at 110°C. They were then heated to 950°C at a heating rate of 8°C/min and held for 2 hrs in Ar atmosphere. As a result, high Al2O3 pisolitic ore A showed higher LOI value compared with dense hematite ore C, which is due to the high content of combined water contained in the ore. This is because the goethite region of the ore becomes porous after dehydration.
To investigate the assimilation behavior of quasi-particle samples, assimilation degree was evaluated by the following method.5) About 30 g of granulated quasi-particles were first charged into a nickel crucible (35 mm in diameter, 40 mm in height). A vertical electrical resistance furnace equipped with automated sample loading system was employed and the quasi-particle samples were sintered by controlling the sample position inside the furnace maintained at 1400°C. The samples were pre-heated at 1100°C for 3 min, and then shifted to the zone of 1400°C for 2 min in air stream. Subsequently, the sample was removed from the furnace for cooling. After cooling, the samples were mounted in resin and cut above 5 mm from the bottom of the crucible. Finally, it was polished to 1 μm level with diamond suspension. Due to the porous characteristics of the sample, the second mounting was done on the sample surface for clear observation.
The assimilation degree was evaluated by applying image analysis on the cross-section of the samples. As shown in Fig. 3, the areas of void and residual ores were distinguished and calculated by counting the number of pixels corresponding to each area. Based on the area measurement data, the area of sintered sample was calculated from the difference between total sample area and void fraction. Then the assimilation degree was determined by the following equation:
An example calculation of determining the assimilation degree (C-J sample).
Backscattered electron imaging and EPMA mapping were performed for microstructure observation and EPMA sweeping area analysis was conducted for further investigation on phase compositions focusing on a single quasi-particle to understand the assimilation characteristics of the samples.
Figure 4 describes the EPMA sweeping area analysis method. Since the microstructure of sinter has great inhomogeneity even in macroscopic observation, it is difficult to obtain the representative information of sinter sample. In this respect, EPMA sweeping area analysis might be an effective method because it targets wider area of sample.11,12) For the quantification of Fe2O3, Al2O3, SiO2 and CaO, the beam of 15 kV and 50 nA was employed and high purity standard materials of Fe2O3 (99.9%), CaSiO3 (99.9%) and CaAl2O5 (99.9%) were used for the calibration of each oxide. An imaginary 20×20 grid structure was constructed on a sintered quasi-particle including the areas of nuclei and adhering fines. Each cross line of grid structure represents a quantitative measuring point. Due to the porous structure of sintered sample, only the quantification data of higher than 95% were collected in terms of total mass% of measured oxides.
EPMA sweeping area analysis method.
Small scale sintering test for quasi-particle sample was performed employing an image gold furnace. Quasi-particle samples (300 g) containing 3 mass% of coke breeze were prepared and charged into a mullite crucible (70 mm in diameter, 95 mm in height). Then it was heated to 750°C for 10 min in N2 atmosphere. After reaching 750°C, pre-heated air of 1000°C was introduced into the sample layer. After introducing air for 1 min, the furnace was turned off and the samples were sintered by heat generated from the combustion of coke breeze. The sintering temperature was assumed to be approximately constant because each quasi-particle sample contains same amount of coke breeze. The oxidation effect of magnetite on sintering temperature was ignored because the melt formation at adhering fines may limit the oxidation of magnetite. The tumble strength and reducibility were evaluated for the sintered samples. In the present study, due to the limited quantity of tested sample, the standard test methods for tumble strength (ISO 2171) and reducibility (ISO 8176) were modified so that they could be applied to smaller amount of sintered sample. That is, 100 and 1.5 g of sintered samples were tested for tumble strength and reducibility, respectively.
The assimilation starts with the formation of melt at the adhering fines, and subsequently the melt reacts with the nuclei. The melt property plays an important role in determining the assimilative characteristics and final sinter quality.10,13) Figure 5 shows the measured assimilation degrees of the quasi-particle samples. The C-N and A-N samples showed higher assimilation degree than those of C-J and A-J samples. This result would be ascribed to the adhering fines comprising quasi-particle. It is commonly known that magnetite is less reactive and has poor characteristics in the melt formation compared with hematite.14,15) Therefore, lower assimilation degree was obtained for the samples comprising ultra-fine magnetite ore J (C-J and A-J) compared with the samples comprising ultra-fine hematite ore N (C-N and A-N). In order to explain the assimilation behavior of quasi-particles, the microstructure analyses were conducted for A-N and A-J samples particularly.
Measured assimilation degree of quasi-particle samples from image analysis.
The quasi-particle structures after assimilation are presented in Figs. 6 and 7 for the A-N and A-J samples, respectively. Large number of micro and macro pores was observed at the sintered region in A-N sample as shown in Fig. 6(a). Since the assimilation of quasi-particle takes place by the interaction between nuclei and the melt formed in the adhering fines, it is believed that the viscosity of melt is one of the key factors to determine the final pore structure. In case the ultra-fine hematite ore N was used as adhering fines of quasi-particle, the melt of calcium ferrite with low viscosity might initially be produced, which makes contribution to the formation of the porous structure in A-N sample. The irregular shape of pores might be ascribed to the increase in the melt viscosity, which resulted from the dissolution of Al2O3 and SiO2 in Ore A into the assimilated region in the progress of assimilation. Therefore, higher assimilation of nuclei and adhering fines was observed in A-N sample.
Backscattered electron images of A-N sample. (a) Microstructure of a sintered quasi-particle. (b) Interface between relict ore and sintered region. (c) Adhering fines region. H: hematite, CF: calcium ferrite, S: silicate.
Backscattered electron images of A-J sample. (a) Microstructure of a sintered quasi-particle. (b) Interface between relict ore and sintered region. (c) Adhering fines region. M: magnetite, CF: calcium ferrite, S: silicate.
On the other hand, in the case of A-J sample, limited number of pores and dense structure were observed in the region of adhering fines as shown in Fig. 7(a). Since magnetite is thought to be less reactive than hematite and calcium ferrite melt cannot be formed directly from the magnetite, it was expected that magnetite has poor characteristics of melt formation, which limits the assimilation.14,15) When the ultra-fine magnetite ore J was used as adhering fines of quasi-particle, limited formation of calcium ferrite melt with low viscosity was expected. Therefore, the assimilation of nuclei and adhering fines was observed only in the narrower region with the formation of ‘interfacial layer’ of silicate as shown in Fig. 7(b). Figures 6(c) and 7(c) clearly shows the different melt formation behaviors of ore N and J in the regions of adhering fines, respectively. The formation of large amount of calcium ferrite indicates that the calcium ferrite melt was easily produced for the case of hematite ore N, while formation of the network of fused magnetite-magnetite grain boundaries implies the limited generation of calcium ferrite melt for the magnetite ore J.14)
To further investigate the different assimilation behaviors of A-N and A-J samples, EPMA mapping analysis was carried out for the elements of Fe, Al, Si and Ca. Since most of Si was initially located in the nuclei ore, the initial shape of nuclei could be identified by the distribution of Si concentration. The relict ore which did not react with the melt in the progress of assimilation could be identified from the absence of Ca distribution. In the case of A-N sample, as shown in Fig. 8, high concentration of Al was observed in the goethite region of nuclei. In addition, little amount of Al was distributed throughout the entire region of quasi-particle, which indicates that some of Al2O3 diffused out from the nuclei to the region of adhering fines in the progress of assimilation. Consequently, the A-N sample showed low extent of Al2O3 localization due to the high assimilation behavior. On the other hand, in the case of A-J sample, it was observed that Al was localized only in the nuclei region as shown in Fig. 9. It is believed that the high extent of Al2O3 localization in A-J sample was resulted from the limited assimilation behavior. In addition, it appears that the localization of Al2O3 is affected by the formation of ‘interfacial layer’ between nuclei and adhering fines, which block the diffusion of Al2O3.
EPMA area and line mapping of Fe, Si, Ca and Al for A-N sample.
EPMA area and line mapping of Fe, Si, Ca and Al for A-J sample.
In particular, the distribution of Ca provides some remarkable differences in the assimilation behaviors between A-N and A-J. In the case of A-N, the concentration of Ca was gradually decreased from the adhering fines region to relict ore. It suggests that the calcium ferrite melt which formed at the adhering fines has large reactivity with nuclei (especially at the dehydrated goethite region), resulting in the high assimilation in A-N. On the other hand, in the case of A-J, the rapid decrease in Ca concentration was observed across the interfacial layer into the nucleus. This might indicate that the interaction between melt and nuclei was suppressed, which results in the limited assimilation of A-J.
For in-depth investigation on phase compositions by focusing on a single quasi-particle, EPMA sweeping area analysis was conducted, which will facilitate the understanding of the assimilation behavior of the samples. The analysis results were plotted on the CaO–SiO2–Fe2O3 ternary diagram as shown in Fig. 10. Figure 10(c) provides two domains of melting point below 1250°C indicated as domains α and β.16) Since the primary fields of 3CaO·Fe2O3·3SiO2 and CaO·Fe2O3 were located in domain α and β, respectively as indicated in Fig. 10(c), the primary phases producing initial melts were assumed to be 3CaO·Fe2O3·3SiO2 domain α and CaO·Fe2O3 for domain β. In the case of A-N in Fig. 10(a), it was found that the distribution of chemical compositions was closely related to the domain β. This result might indicate that the calcium ferrite melt was initiated at the adhering fines and mainly participated in the assimilation of A-N. On the other hand, in the case of A-J in Fig. 10(b), high concentration of chemical compositions was found along the line connecting the composition of 3CaO·Fe2O3·3SiO2 and Fe2O3, which represents the compositions of ‘interfacial layer’ between nuclei and adhering fines of A-J. This result suggests that the initial melt probably formed with the composition of 3CaO·Fe2O3·3SiO2 and the composition of melt moved to the Fe2O3 side in the progress of assimilation.
Compositional distribution of (a) A-N and (b) A-J on CaO–SiO2–Fe2O3 system with (c) melt regions below 1250°C.
The assimilation behaviors of quasi-particles can be explained on the basis of penetration phenomena derived by capillary pressure. The Washburn equation expresses the kinetics of penetration as follows:17,18)
h: the length of the liquid column (i.e. penetration length),
t: penetration time,
r: radius of capillary,
γ: surface tension,
θ: contact angle,
η: viscosity of the liquid.
Equation (3) shows that the penetration length of melt depends on the both the nuclei property such as porosity (r) and melt properties such as surface tension (γ), contact angle (θ) and viscosity (η). In the present study, it was assumed that the nuclei property was constant since the identical nuclei ores were used for A-N and A-J. However, melt properties varies significantly depending on the composition and temperature. It indicates that the penetration behavior is essentially determined by the melt properties under some assumption that the nuclei property, penetration time and temperature are constant. Some available information was adopted from the literature to assess the impact of melt properties on the penetration behavior. The surface tensions were reported at 1410°C with assumption of θ=0 for Fe-saturated iron silicate melts with 0–34.98 wt% SiO2 and Fe-saturated calcium ferrite melts with 1.76–26.90 wt% CaO in the range of 419–577 and 560–575 N·m–1, respectively.19) On the other hand, in the present study, the melt viscosities of initial compositions were estimated by using the thermodynamic calculation16) as shown in Fig. 11. This result showed that the viscosity of 3CaO·Fe2O3·3SiO2 has about 5 times higher than that of CaO·Fe2O3 at 1300°C. Since the penetration behavior is essentially determined by melt properties in terms of the ratio of surface tension to viscous force (i.e. γcosθ/η), the above discussion indicates that the assimilation behavior of primary melt is believed to dominantly be affected by the melt viscosity. Therefore, the lower assimilation degree of A-J can be explained by the higher viscosity of 3CaO·Fe2O3·3SiO2 melt.
Viscosity of 3CaO·Fe2O3·3SiO2 and CaO·Fe2O3.
From the above results, the assimilation mechanisms of A-N and A-J were summarized in Fig. 12. For the A-N sample which was comprised of high Al2O3 pisolitic ore A as nuclei and ultra-fine hematite ore N as adhering fines:
Summary of assimilation mechanisms of (a) A-N and (b) A-J.
a) Calcium ferrite melt initially forms in the region of the adhering fines by the reaction of ultra-fine hematite N with CaO.
b) Assimilation starts with the penetration of calcium ferrite melt into the nuclei and then the melt with low viscosity reacts with dehydrated goethite of nuclei.
c) In the progress of assimilation, the viscosity of melt gradually increases as Al2O3 and SiO2 of the nuclei dissolved into the melt.
d) Finally, granular hematite and silicate phases were precipitated upon cooling and Al2O3 were mainly localized in the assimilated region. However, due to the high fluidity of the melt, some amount of Al2O3 diffused out to the region of adhering fines.
On the other hand, for the A-J sample which was comprised of high Al2O3 pisolitic ore A as nuclei and ultra-fine magnetite ore J as adhering fines:
a) Since the calcium ferrite cannot be formed directly from magnetite,14) it is expected that the melt of fayalite (2FeO·SiO2) preferentially forms at the interface between nuclei and adhering fines.
b) Subsequently, the fayalite melt reacts with CaO to generate the melt mainly comprising the composition of 3CaO·Fe2O3·3SiO2. This transformation could be interpreted based on the dependence Fe3+/Fe2+ ratio on CaO content.22) That is, the Fe3+/Fe2+ ratio increased with increasing CaO content in the melt:
c) As a result of the high viscosity of 3CaO·Fe2O3·3SiO2 melt, the assimilation in A-J sample takes place only at the interface between nuclei and adhering fines by forming the interfacial layer with the compositions along the line from 3CaO·Fe2O3·3SiO2 to Fe2O3. The formation of interfacial layer could block the Al2O3 diffusion so that Al2O3 localized only in the nuclei region.
Small scale sintering test was carried out to evaluate the sinter quality of quasi-particle samples. Figure 13 shows the effects of Al2O3 and SiO2 content on the tumble strength and reducibility of the sample. A decrease of tumble strength and an increase of reducibility were found in the samples with nuclei of high Al2O3 pisolitic ore A compared with those of the samples with nuclei of dense hematite ore C. This result is in good agreement with the previous researches2,6) about the detrimental effect of high Al2O3 on tumble strength. Furthermore, regardless of the nuclei ores, lower tumble strength and higher reducibility were observed in the samples with adhering fines of ultra-fine hematite ore compared with that of the samples with adhering fines of ultra-fine magnetite ore.
Effect of Al2O3 and SiO2 content in the ores on the (a) tumble strength and (b) reducibility of quasi-particle samples.
In particular, the A-J sample showed reasonable sinter quality from viewpoints of the compromise between the tumble strength and reducibility. The sinter quality of A-J sample was similar to that of C-N sample which is considered a suitable sinter sample in the present work. However, in the case of A-N sample, it is probably not suitable for the sinter sample due to the low tumble strength even though it has the highest reducibility. Consequently, it can be concluded that the ultra-fine magnetite ore J is more suitable adhering fines than ultra-fine hematite ore N in terms of sinter quality.
The assimilation behavior of quasi-particle was investigated to utilize the high Al2O3 pisolitic ore as nuclei of quasi-particle. From the findings, the following conclusions were obtained.
(1) In case the ultra-fine hematite ore was used as adhering fines of quasi-particle, CaO·Fe2O3 melt with low viscosity was predominantly participated in the assimilation. Due to the porous structure and low extent of Al2O3 localization, the detrimental effect of Al2O3 on the strength of quasi-particle was not fully controlled.
(2) When the ultra-fine magnetite ore was employed as adhering fines of quasi-particle, 3CaO·Fe2O3·3SiO2 melt with high viscosity was predominantly participated in assimilation. As a result, the assimilation was suppressed by the formation of ‘interfacial layer’ along the line connecting the compositions of 3CaO·Fe2O3·3SiO2 and Fe2O3 in CaO–SiO2–Fe2O3 ternary system. Due to the dense structure and high extent of Al2O3 localization, the detrimental effect of Al2O3 on the strength was reasonably controlled.
(3) The quasi-particle comprising high Al2O3 pisolitic ore and ultra-fine magnetite ore showed competitive sinter quality similar to the quasi-particle sample consisting nuclei of dense hematite and adhering fines of ultra-fine hematite, resulting in the high sinter quality.