2014 Volume 54 Issue 4 Pages 827-835
In the current investigation, the thin film method was employed to clarify the formation mechanism of the oxide layers at the interface between the Al2O3–SiO2–C refractory and liquid Fe. A reacted layer was formed in such a way that initially FeO-enriched liquid layers are widely distributed on the Fe surface and FeO and SiO2 in the liquid layer are reduced by Al in liquid Fe to develop solid Al2O3-enriched layer of inclusions. Some inclusions in liquid Fe might be produced by the remaining oxygen in Ar gas being supplied through the nozzle to prevent the adherence of inclusions onto it. The oxide layers at the interface estimated by thermodynamic calculation are in good accordance with the experimental results of the reaction between Al2O3–SiO2–C refractory and Al-killed/Al-Ti-killed steels previously reported.
Production of ultra clean steel has led a growing emphasis on the proper control of nonmetallic inclusions in steels. Refractory materials have been indicated as one of the major sources of non-metallic inclusions in the molten steel during ladle metallurgy.1,2,3,4,5) Al2O3 agglomeration onto a tundish nozzle has been a serious problem with the continuous casting process of steel. Nozzle clogging by agglomerated Al2O3 lead to a non-uniform flow of molten steel in the mold and to the defects in products from large Al2O3 inclusions separated from the clogged nozzle.6) Several investigations on the clogging mechanism and its prevention have been reported.6,7,8,9,10,11) In particular, the interfacial reactions between the nozzle and molten steel have been receiving extensive attention since the formation and growth of inclusions generated by the interfacial reactions are ascribed to the nozzle clogging. This indicates that the reaction products between the constituting oxides in nozzles and deoxidation elements in molten steel are accumulated at the interface. There have been numerous studies about nozzle clogging in Al-killed and Al/Ti-killed steels, which are supported by the consistent observation that the accumulated oxide layers on the nozzle surface are similar to the nozzle materials in terms of the constituting elements.
Sasai et al.12) clarified the formation mechanism of reacted layer in Al-killed and Ti-killed steels. They reported that SiO2 in refractory was reduced by carbon to produce SiO and CO gases and that SiO2 in refractory reacted with Al or Ti in molten Fe to generate Al2O3 or Ti3O5 inclusions. Kaneko et al.13) also reported that Al or Ti in molten Fe directly reduced SiO2 in refractory. In most cases, it is well known that a reticulate layer is formed on the surface of a refractory and another tight layer of alumina cluster is added onto it. In the case of Al-killed steel, it is difficult to make a difference between the reticulate and tight layers since Al2O3 is the major component in both layers. However, although the Al–Ti–O layer on the surface of a refractory and the layer of Al2O3 cluster were easily identified in the case of Al/Ti-killed ultralow carbon steel, it is not easy to separate both the layers from a viewpoint of nozzle clogging. Therefore, it is required to clarify the effect of interfacial reaction between actual nozzle and molten steel by employing new experimental technique.
It is required to investigate the formation and behavior of inclusions at the interface adjacent to the refractory since it is difficult to clearly identify the interface in the conventional experimental methods employing the reaction of refractory with molten steel. Therefore, in the present study, a thin film method was employed to clarify the effect of interfacial reaction between Al2O3–SiO2–C refractory and liquid Fe on the formation and evolution of inclusions at the interface.
According to Dekkers et al.,14,15) the formation of metastable secondary particles was unavoidable in the course of sampling regardless of the sampler used. It is believed that this is the most ambiguous factor affecting the analysis of the morphology and size of inclusions adjacent to the interface between the refractory and molten iron. Therefore, the thin film method developed in the current study would facilitate the clear explanation of the effects of supersaturation degree, reaction time and surface roughness of the substrate on the morphology and size distribution of alumina inclusions formed at the iron/refractory interface. The validity of the method suggested in the current research was evaluated by analyzing the morphology and composition of inclusions using SEM and EPMA. That is, it was confirmed that the current method could sufficiently be reasonable for simulating interfacial reactions between the substrate and liquid iron.
As shown in Fig. 1(a), the lower supporting material of the sample assembly is the alumina single crystal substrate whose top side was coated with pure aluminum (>99.999%) of variable thickness to control the Al content in liquid Fe using the general sputtering method (70 W, Ar 5 mL/min). And a pure Fe foil of 100 μm thickness (>99.99%, 15×15 mm2) was placed on top of the coated side in a way that the aluminum on the substrate can diffuse through the layer of liquid Fe. Initial compositions of the Fe foil used in the current experiment are listed in Table 1. In case pure Al (>99.999%) of about 1 μm thickness was coated on the substrate, the Al content in liquid iron was estimated to be about 0.06 mass% under the assumption that all the Al dissolved into the pure Fe foil. As the upper substrate, Al2O3–SiO2–C refractory was used. Table 2 shows the chemical composition of Al2O3–SiO2–C refractory before and after decarburization. The Al2O3–SiO2–C refractory was sufficiently decarburized by the preheating. The surface roughness of the refractory was not available since the measured values were much beyond the upper limit of measuring capacity. Then the sample assembly was wrapped by ash-less paper, dipped into Al2O3 powder in a graphite crucible, and a cylindrical tungsten weight was placed on top of Al2O3 powder.
Schematic diagram of (a) sample assembly and (b) experimental setup.
A vertical electrical resistance furnace was employed as shown in Fig. 1(b). The temperature was automatically controlled within ±2 K using an R-type thermocouple (Pt-13%Rh/Pt) and a PID controller. The sample assembly was put into the furnace purged with Ar (6N purity) gas preliminarily purified by passing through Mg chips at 723 K to remove residual oxygen. After equilibrating for 3 hours, the graphite crucible containing the sample assembly was quickly withdrawn out of the furnace and quenched with water. The refractory/Fe interface containing inclusions was observed using FE-SEM (Hitachi, S-4200). The concentration of oxygen and nitrogen in the iron sample was analyzed by the combustion method using LECOTM TC-300.
Figure 2 shows the surface image of Fe sample obtained after raw Al2O3–SiO2–C refractory reacted with liquid Fe containing 0.06 mass% of Al for 3 hours at 1823 K. It was observed that the reacted layer of approximately 100 μm exists between the refractory and Fe. Due to the limitation in analyzing the compositional change of the constituents in the reacted layer by macroscopic analysis, the reacted layer was further magnified and analyzed by SEM/EDS. According to SEM/EDS analysis of the reacted layer, its averaged composition was analyzed to be 1.5%FeO-3.7%SiO2-94.8%Al2O3. When moving from the refractory side to Fe, the proportion of Al2O3 gradually decreased, which is similar to the results of the reacted layer formed between Al2O3–C refractory and liquid Fe.16) In order to observe the compositional change of the constituents, EPMA line mapping was carried out across the interface along the lines 1 and 2 in Fig. 2 and the results are shown in Fig. 3. It is clear that the concentrations of carbon and Si increased in Fe after the reaction. It is believed that some inclusions comprising FeO–SiO2–Al2O3–MgO exist in the reacted layer, which has been solidified after melting. It is expected that the increase of carbon in liquid Fe is ascribed to the dissolution of carbon in the refractory or to the introduction of CO gas formed by the reaction between SiO2 and carbon in the refractory. The increase of Si in liquid Fe is due to the reaction between Al in liquid Fe and SiO gas formed by the reaction between SiO2 exposed to the surface of the refractory and carbon as explained by Eqs. (1) and (2).
SEM image of the reacted layer formed by the reaction between raw Al2O3–SiO2–C refractory and liquid Fe.
Line mapping results of interfacial area after reaction between the refractory before preheating and Al-killed steel.
Figure 4(a) shows the interface formed by the reaction between the preheated Al2O3–SiO2–C refractory and liquid Fe containing 0.1 mass% of Al for 3 hours at 1823 K. As can be expected by the Reactions (1) to (3), the lower amount of Al2O3 inclusions can be formed in case less SiO(g) and CO(g) will be generated by the decreased activity of carbon in the preheated refractory. Thus, liquid Fe containing slightly larger amount of Al was used in a way that similar amount of Al2O3 inclusions would be formed in liquid Fe by driving the forward reactions of Eqs. (2) and (3). It was observed that the reacted layer of about 50 to 170 μm exists between the refractory and Fe. This reacted layer is separated from the coarse refractory and looks denser than the refractory. As previously performed, the reacted layer was further magnified as shown in Fig. 4(b) and analyzed by SEM/EDS. The average composition of zone A located adjacent to Fe was 5.4%FeO-50.8%SiO2-43.8%Al2O3. Zone B located closer to the refractory was analyzed to be 0.6%FeO-58.8%SiO2-40.6%Al2O3. That is, when moving from the refractory side to Fe, the proportion of SiO2 gradually decreased while the contents of FeO and Al2O3 increased. Although the compositions of A and B are in solid state according to FeO–SiO2–Al2O3 phase diagram at 1823 K,17) the growth of reacted layer cannot be explained by the transfer of oxygen in terms of SiO(g) and CO(g) produced by the reaction between SiO2 and carbon in the refractory since the refractory contains almost no carbon. Therefore, the current results are ascribed to the involvement of oxygen as a surface active element in the reaction, which is in good agreement with the previously reported ones.18,19) That is, FeO-enriched inclusions (FeO–SiO2–Al2O3) observed in Al-killed Fe are gradually reduced to Al2O3 by Al dross initially injected. It is believed that the reacted layer can be formed in such a way that initially FeO-enriched liquid layer of inclusions are widely distributed on Fe surface. Then, FeO and SiO2 in the liquid layer are reduced by Al in liquid Fe, which ends up with the development of solid Al2O3-enriched layer of inclusions with increasing the number of inclusions. Then the inclusions might be pushed back toward the refractory as Al2O3 content in the inclusion increases.
SEM image of the reacted layer formed by the reaction between preheated Al2O3–SiO2–C refractory and liquid Fe.
As shown in Fig. 5, EPMA line mapping was carried out across the interface along the lines 1 and 2 in Fig. 4(a) to observe the compositional change of the constituents. The increase of carbon content in Fe layer was remarkably not observed compared with the case of the raw refractory. In general, the carbon in liquid Fe is normally identified due to the reaction between carbon-containing refractory and Fe. This is ascribed to the low carbon content remaining in the preheated refractory with other oxide constituents unchanged. Therefore, it is believed that the supply of oxygen toward the surface of liquid Fe is not available by the reaction between SiO2 and carbon in the refractory as previously mentioned. On the other hand, it is believed that the increase of Si in Fe is ascribed to the reduction of SiO2 exposed to the refractory surface after decarburization directly by Al in liquid Fe as explained by Eq. (4). This can be supported by the EPMA mapping result that the reacted layer shows very high Al2O3 content and slight amount of FeO, SiO2 and MgO. After all, it is believed that the increase of Si in Fe is ascribed to the non-uniformity of the preheated refractory.
Line mapping results of interfacial area developed by the interfacial reaction between the preheated refractory and liquid Fe.
In case the raw refractory contains carbon, the reaction between the SiO(g) or CO(g) supplied from the refractory and Al in liquid Fe could take place to produce Al2O3 inclusions as represented by Eqs. (2) and (3). However, in the case of decarburized refractory, the reaction between SiO2 exposed to the refractory surface and Al in liquid Fe could occur to generate Al2O3 inclusions and to increase Si level in Fe as indicated by Eq. (4). From the remaining FeO in the reacted layer, the reaction between Al in Fe and FeO–SiO2–Al2O3 inclusion resulted from the interaction between FeO and refractory could proceed.
Before the reaction between SiO2 exposed in the refractory and liquid Fe is examined, the reaction between pure quartz and liquid Fe containing no Al was considered. Figure 6 shows the surface of Fe reacted with quartz for 3 hours at 1823 K. It is clear that no inclusions were observed on Fe surface. According to EDS analysis, only small amount of FeO and Fe was detected. In order to thermodynamically examine the decomposition of pure SiO2, [%Si] content in liquid Fe was estimated based on the initial [%O] in Fe listed in Table 1.
SEM image on Fe surface between quartz and liquid Fe.
Based on Eq. (6), about 0.026 mass% of Si would be in equilibrium with 0.017 mass% of O for the pure SiO2. Thermodynamically it is likely that pure SiO2 directly decomposes into Si and O in liquid Fe, which results in the dissolution of some Si in Fe for the initial O content of 0.017 mass%. Therefore, it was confirmed that pure quartz might decompose into Si and O in liquid Fe without forming any kind of inclusion. Figure 7 shows the surface of quartz reacted with liquid Fe containing 0.2 mass% of Al. Spherical particles and coral-shaped cluster of inclusions were observed. According to EDS analysis, the composition of spherical particle A in Fig. 7 was 31.6%FeO-62.4%SiO2-6.0%Al2O3 and that of part B was 1.38%FeO-51.5%SiO2-47.2%Al2O3. On the other hand, Fig. 8 shows the surface of Fe reacted with quartz. According to EDS analysis, the composition of coral-shaped cluster of inclusions was 4.0%FeO-29.5%SiO2-66.5%Al2O3. Therefore, it can be understood that FeO and SiO2 can be reduced by Al in liquid Fe.
SEM image on quartz surface between quartz and liquid Fe.
SEM image on Fe surface between quartz and liquid Fe.
Figure 9 shows the surface of Al2O3 single crystal facing liquid Fe. Spherical particles and coral-shaped cluster of inclusions were observed. According to EDS analysis, the composition of spherical particle A in Fig. 9 was 24.1%FeO-67.2%SiO2-8.7%Al2O3 and that of part B was 20.3%SiO2-79.7%Al2O3. It is understood that the Si supplied by the interfacial reaction between non-uniform SiO2 in the refractory and liquid Fe would have the opposite diffusion path to Al and that Al and Si have counter-diffusion each other. Therefore, the SiO2 exposed by non-uniformity of refractory can be reacted with Al in liquid Fe, which results in forming solid Al2O3-enriched inclusions as represented by Eq. (4). After all, it is believed that this process might cause the serious nozzle clogging.
SEM image on the surface of Al2O3 single crystal substrate between Al2O3 and liquid Fe.
The surface activity of oxygen, JO is defined as the slope of the surface tension, σ vs XO at infinite dilution:
Al–O equilibrium in liquid Fe.20)
It can be considered that the inclusions in liquid Fe can be formed due to the oxygen included in Ar gas which is supplied through nozzle to prevent the adherence of inclusions onto it. Figure 11 shows the analyzed content of oxygen in liquid Fe through which Ar was blown for 3 hours when the initial content of oxygen was 0.05 mass%. Two kinds of Ar gases were used. The first Ar gas (99.999% purity) was sufficiently deoxidized by passing through Mg turnings preliminarily and the second Ar gas was directly supplied without deoxidation. In the chemical analysis, the specimen surface was slightly dissolved into HCl(1 + 10) for 30 min in such a way that FeO on the surface could not affect the analysis of oxygen in the bulk of Fe. The activity of FetO in Fig. 11 was evaluated according to the following equilibrium between liquid Fe and dissolved oxygen:
Effect of oxygen partial pressure on the oxygen in liquid Fe.
Based on the formation of inclusions at the interface between the refractory and liquid Fe, thermodynamic calculation was carried out to predict the characteristics of reacted layer and the thickness of inclusion layer. The calculation was performed using thermodynamic calculation software, FactSage5.2. Fundamental assumptions and procedure for the calculation are as follows: (1) The composition in the inner part of the refractory remains unchanged since only the surface is decarburized in actual process. (2) The initial reaction occurred among the liquid Fe, the produced gases from refractory materials and the remaining constituents in the refractory of 1 mole. (3) Solid inclusions resulted from the reaction stick to the refractory and that they are not involved in the subsequent reactions. And the liquid phase and gases continue to react with 1 mole of Fe.
The current calculation was carried out for 1 mole of raw refractory of 47%Al2O3-24%SiO2-28%C and 1 mole of Fe containing 600 ppm of Al. It was assumed that the refractory was not decarburized. When the change of reaction were infinitesimal compared with that in the previous stage, the reaction was forced to stop. In order to calculate the thickness of reacted layer, it was assumed that 1 mole of Fe was contained in a cube. The contact area between the refractory and liquid Fe was calculated to be 1.9×1.9 cm2. The densities of Mullite, Al2O3, FeO–TiO2 FeO–SiO2–Al2O3(slag) were 3.52, 3.05, 4.44 and 3.0 g/cm3, respectively. The calculation result is shown in Fig. 12(a). Based on Fig. 12(a), the calculation was discussed by comparing it with the experimental results. It is experimentally evident that FeO formed on Fe surface reacts with Al2O3–SiO2–C refractory. FeO and SiO2 are reduced by sufficient carbon in the refractory, resulting in the formation of Al2O3-enriched layer. Thus, it is likely that FeO–SiO2–Al2O3 are reduced by Al in liquid Fe to generate the layer of Al2O3–SiO2, which is in good agreement with some experimental results.24,25,26) Another Al2O3 layer of 1.8 μm could originate from the reduction of SiO2 in mullite (3Al2O3⋅2SiO2) layer by Al in liquid Fe. However, the mullite layer could be destroyed due to the unstability during cooling in actual experiments. And it is likely that Al2O3 inclusions were agglomerated and attached to the reacted layer, which resulted in the formation of Al2O3-enriched layer on the refractory observed in the current study. Therefore, it can be considered that the thermodynamic calculation show the stepwise progress of forming the first and second reacted layer.
Calculation results for the reaction between Al2O3–SiO2–C refractory and Al-killed Fe and for that between Al2O3–SiO2–C and Al/Ti-killed Fe.
Reaction between Al2O3–SiO2–C refractory and Al/Ti-killed Fe. [A: Al–Ti–O, T: Ti–Al–(Fe)–O, U: unaltered layer, D: decarburized layer, P: Product layer, M: metal droplet, BM: bulk metal]27)
Another calculation was carried out for 1 mole of the raw refractory of 47%Al2O3-24%SiO2-28%C and 1 mole of liquid Fe containing 0.06 mass% of Al and 0.06 mass% of Ti. The other conditions are similar to those previously mentioned. The calculation results are shown in Fig. 12(b) and are compared to the experimental results by Kwon et al.27) As previously explained, the Al2O3 layer was formed in the first reaction, and simultaneously FeO on Fe surface could combine with TiO2 in liquid Fe to generate FeO–TiO2 layer. However, Al2O3- and TiO2-enriched inclusions are mixed together in the experimental results reported by Kwon et al.27) This indicates that Al2O3 particles and FeO–TiO2 inclusions mix together in the first reaction since FeO and TiO2 exist in the solid-liquid coexisting region of the phase diagram as shown in Fig. 12(b).
The thin film method was carried out to investigate the formation mechanism of interfacial inclusions between refractory and molten steel because it has the only interfacial properties in boundary layers except for the inclusions and reactions in bulk metal. The new experimental methodology was effective in clarifying the effect of refractory materials on the interfacial reaction between Al2O3–SiO2–C refractory and Al-killed/Al-Ti-killed steels. Based on the findings, the following conclusions were obtained.
(1) According to SEM/EDS analysis of the interface, when moving from the refractory side to Fe, the proportion of SiO2 gradually decreased while the contents of FeO and Al2O3 increased.
(2) The reacted layer is formed in such a way that initially FeO-enriched liquid layers are widely distributed on the Fe surface and then FeO and SiO2 in the liquid layer are reduced by Al in liquid Fe to develop solid Al2O3-enriched layer of inclusions.
(3) SiO2 exposed by non-uniformity of refractory can be reacted with Al in liquid Fe, which results in forming solid Al2O3-enriched inclusions. This might cause the nozzle clogging.
(4) The inclusions in liquid Fe can be increased due to the oxygen remaining in Ar gas being supplied through nozzle to prevent the adherence of inclusions onto it.
(5) According to the thermodynamic calculation, the oxide layers formed at the interface are in good agreement with the experimental results previously reported for that between Al2O3–SiO2–C refractory and Al-killed/Al-Ti-killed steels.