Reoxidation of Al Ti Containing Steels by CaO Al 2 O 3 MgO SiO 2 Slag

Transcription

1 , pp Reoxidation of Al Ti Containing Steels by CaO Al 2 O 3 MgO SiO 2 Slag Dong-Chul PARK, In-Ho JUNG, 1) Peter C. H. RHEE and Hae-Geon LEE Department of Materials Science & Engineering, Pohang University of Science and Technology (POSTECH), San 31 Hyojadong, Pohang Korea. 1) Research Institute of Industrial Science and Technology, P.O. Box 135, Pohang Korea. (Received on May 24, 2004; accepted in final form on July 30, 2004 ) Reoxidation of liquid steel containing Al and Ti by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag was investigated at K with initial Al content of 820 mass ppm and Ti content varied from 100 to 500 mass ppm. It was observed that Al and Ti in the steel were simultaneously oxidized by SiO 2 in the slag, and the soluble oxygen was supersaturated during the course, particularly with respect to Al. Based on the experimental results, a new mechanism of the reoxidation reaction was proposed, which involves chemical reactions both at the metal/slag interface and in the bulk metal. Self-dissociation of SiO 2 into Si and O at the slag/metal interface was found to play an important role in both supersaturation of oxygen, and subsequent formation of complex oxide inclusions. Formation of inclusions having a two-layer structure where an Al 2 O 3 core was enclosed by complex Al Ti O oxide was explained in relation with supersaturation of oxygen in the steel. KEY WORDS: reoxidation; supersaturation; self-dissociation of SiO 2 ; inclusion with two-layer structure. 1. Introduction Reoxidation of steel after finishing refining causes a major problem for the production of clean steels. Al and Ti in particular are susceptible to oxidation by those reducible oxides such as FeO, MnO and SiO 2 in the slag, fluxes or refractories in the ladle and tundish. The oxidation reaction results in formation of non-metallic inclusions such as Al 2 O 3, TiO x and Al Ti O oxides which might cause process problems such as nozzle clogging in the continuous casting and product problems such as surface defects. The major oxygen source in tundish for reoxidation of steel includes reducible oxides like FeO, MnO and SiO 2 in the tundish slag and oxygen in the atmosphere. Since tundish is the final stage of the secondary steelmaking process, the reoxidation in this stage should be strictly controlled for production of a clean steel. Up until now, investigations on reoxidation in tundish have focused mostly on oxidation of Al with different slag compositions 1 5) and with oxygen from the atmosphere. 6,7) No systematic investigation on the combined oxidation of Al and Ti in steel with SiO 2 in a tundish slag has been reported yet. In the present study, oxidation of Al and Ti in liquid steel with a tundish slag containing SiO 2 was investigated at K. It has been attempted to elucidate mechanisms of the simultaneous oxidation of both Al and Ti in the steel by SiO 2 in the slag. 2. Experimental The master slag of 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) was prepared by melting a mixture of reagent grade CaCO 3, Al 2 O 3, MgO and SiO 2 in a graphite crucible under air using a high-frequency induction furnace (20 kw and 40 khz). The pre-melted slag was crushed and ground. Then any residual carbon was removed by heating the slag in air at K for 24 h in a box furnace. Iron alloys containing Al and Ti was prepared by melting about 300 g of high purity electrolytic iron with high purity Al (99.99 mass%) and Ti (99.99 mass%) in an alumina crucible using a high-frequency induction furnace. After maintaining the melt at K for about 30 min, it was allowed to cool to the room temperature in the furnace. The furnace was kept under inert atmosphere by blowing Ar gas purified by passing Mg(ClO 4 ) 2 and Mg chips at 450 C. The oxygen partial pressure in the furnace was measured to be about atm. The upper part of the cast alloy was removed to avoid a possible contamination by oxides or impurities which might have been accumulated. The chemical compositions of the samples were determined by employing various analytical methods described below. The starting compositions of the alloy for each experiment are listed in the zero time line in Table 1. Figure 1 shows the schematic diagram of experimental apparatus employed in the present study. Firstly, 30 g of surface-cleaned alloy was melted in an alumina crucible (25 mm I.D., 30 mm O.D. and 60 mm H) at K. After the alloy had been completely melted, the pre-melted slag ISIJ

2 Table 1. Analyzed compositions of metal and slag in the present study ISIJ 1670

3 Fig. 1. Schematic diagram of the experimental apparatus used in the present study for reoxidation experiments. Fig. 2. Variation of Al and Ti for the reoxidation of (Al Ti)-containing steel by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag in comparison with those of Al or Ti containing steels at K. of 10 g was dropped into the crucible through the quartz tube (17 in Fig. 1) which had been kept purged by blowing a deoxidized He gas. Ten grams of the slag was enough to cover the entire metal surface in the crucible. Then the system was kept at K for a predetermined length of time (3, 6, 9, 12, 15, 20, 25 or 30 min). When the predetermined time lapsed, the whole crucible assembly was rapidly pulled out of the furnace and quenched in a He gas stream, in liquid nitrogen and finally in water. Quenched slags were analyzed by the XRF technique. The quenched alloy samples were sectioned and the middle part of the samples was used for various analyses in the present study. Oxygen and nitrogen contents were analyzed by NO spectrometry and the concentrations of other elements by ICP-AES. Total Al, Ti and Si in metal were analyzed from the method described in JIS G Soluble and insoluble Al are taken as acid soluble and acid insoluble, respectively. However, because a soluble and insoluble Ti could not be separated from the dissolution in acid solution, electrolyte extraction method was used to measure the insoluble Ti concentration. This apparatus can extract the inclusions because the metallic composition is only dissolved by electric energy. Electrolytic solution used in the extraction method is 2 wt% Ba T.E.A (2 % Triethanol amine 1% Tetramethylammonium chloride Methanol). The dissolved solution with metallic ions and inclusions is separated by 0.1 mm membrane filter. After that, the filter was analyzed for the insoluble Ti concentration. A sample (Exp. 4 at 10 min) was used for examination of oxide inclusions. The sample was polished and etched for 3 to 5 s in an etchant composed of potassium meta-bisulfite and distilled water of the ratio 1 to 10. Then the sample was examined by an optical microscopy and SEM. For high-definition images of inclusions, the SEM examinations were carried out using a JEOL JSM-6330F with a field emission gun. Oxide inclusions were randomly selected and examined. The composition of inclusions was analyzed by the energy dispersive X-ray spectra (EDS) with the acceleration voltage of 20 kv and full ZAF corrections. Area mapping around each inclusion was also performed in order to determine the distribution of elements and average composition of an inclusion. The analysis was carried out for at least 5 min in order to obtain enough intensity to resolve elemental peaks. Inca system was used to examine the results of EDS analysis. (Inca is the trademark of Oxford Instruments, UK.) 3. Results The change of the composition of alloys and slags determined in the present study was listed in Table 1. The amount of soluble oxygen was recalculated from the difference of the total oxygen and insoluble oxygen Reoxidation of Al, Ti or (Al Ti)-containing Steel Experimental results on reoxidation of three different steels, namely Al-containing, Ti-containing and (Al Ti)- containing steels at K are compared in Fig. 2. The initial concentration of Al was the same at about 820 mass ppm for both Al-containing and (Al Ti)-containing steels, and the initial concentration of Ti was also the same about 310 mass ppm for both Ti-containing and (Al Ti)-containing steels. In all cases, the slag of the same compositions (14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 in mass%) was used. As can be seen in the figure, the oxidation of Al is ISIJ

4 slower in the (Al Ti)-containing steel than in the Al-containing steel. The same is true for the oxidation of Ti, but the difference of the rate of oxidation between the Ti-containing and (Al Ti)-containing steels is much more significant. This implies that the interaction of Ti with oxygen is strongly influenced by co-existence of Al. Figure 3 clearly shows the influence of Al on the oxidation rate of Ti in the steel: the oxidation rate of Ti decreases with increasing the Al content. Fig. 3. Observed dependence of the oxidation rates of Ti, d[ti] sol. /dt, on the Al contents in the initial stage of the reoxidation of (Al Ti)-containing steel. An important point to be noted in this section is that both Al and Ti oxidize simultaneously and the concentration of both Al and Ti decreases as time lapses Change of Al, Ti and Si Concentrations Figure 4 shows the change of Al, Ti and Si concentrations in the steel during the reoxidation process at K. Full details of experimental results are given in Table 1. The initial content of Al was the same at about 820 mass ppm in all experiments, but the initial content of Ti was varied from 100 to 500 mass ppm. The initial composition of slag was the same in all experiments at 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%). The general trend of the concentrational change of Al, Ti and Si in steel was found similar in all experiments: the concentrations of Al, Ti and Si exhibit a linear change in the initial stage of the reoxidation, followed by a gradual approach to a certain value for each of these Supersaturation of Oxygen Figure 5 shows the change of the oxygen concentration with time during reoxidation of the steels. It is seen that the total oxygen (the sum of both the dissolved oxygen and the oxygen tied up as oxides) gradually increases with time until it reaches a steady value for each initial condition. The concentration of the dissolved oxygen shows a rather pecu- Fig. 4. Variation of Al and Ti for the reoxidation of (Al Ti)-containing steel with different initial Ti content by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag at K. (a) 100, (b) 200, (c) 300, (d) 400 and (e) 500 mass ppm Ti. Filled square, open square and filled circle symbols represent Al, Ti and Si contents, respectively ISIJ 1672

5 ISIJ International, Vol. 44 (2004), No. 10 Fig. 5. Variations of oxygen contents during reoxidation of (Al Ti)-containing steel. Half-filled and open symbols represent total oxygen and soluble oxygen contents respectively. The oxygen contents in bulk alloys in the equilibrium with slag are presented as filled symbols. liar trend in its change with time: it stays low until about 10 min after the initiation of reoxidation, and then sharply increases with time until it has reached around 50 mass ppm. The difference between the total and dissolved oxygen concentrations indicates the amount of insoluble oxygen, i.e., oxide inclusions and hence the inclusions in the bulk metal tend to increase in the initial stage due to reoxidation, and then decrease due to flotation up to the surface. The dissolved oxygen concentrations in equilibrium with the bulk metal and slag were computed using the FactSage8) thermodynamic software, and the results are also included in Fig. 5. It is seen that the dissolved oxygen content tends to increase beyond the values for equilibrium with the rest of the system: the difference between the measured and computed values of the dissolved oxygen is negligibly small in the initial stage of reoxidation, but becomes significant from about 10 min after the initiation of reoxidation. The difference eventually diminishes toward the completion of the reoxidation. The above phenomenon, which is termed as supersaturation of oxygen, was also reported by Lee et al.9) through their study of reoxidation of Fe Al alloy by CaO Al2O3 FetO at K. Through oxidation study of Fe Al alloy by CaO Al2O3 SiO2 ( MnO and/or FeO) slag at K, Sun and Mori5) also reported that the supersaturation occurred with slags of CaO Al2O3 MnO (and/or FeO), and that the supersaturation could occur with slag of CaO Al2O3 SiO2, provided that SiO2 content be high. Fig. 6. SEM images of typical complex oxide inclusions observed in the bulk metal (Exp. 4 at 10 min). The inclusions showed two-layer structure; Al2O3 core enclosed by Al Ti O oxide. 7 (see the dotted lines). The diagram shows that the mixture of Al2TiO and Ti3O5 forms a pseudobrookite solid solution. Although an accurate phase determination was not attempted in the present study, it might be able to conclude that the outer layer of the complex inclusions is in the form of pseudobrookite solid solutions with the mean composition of 71mass%Al2TiO5 29mass%Ti3O5 (Al/Ti molar ratio 1/1) in average Oxide Inclusions: Insoluble Oxygen Oxide inclusions in the bulk metal, which are the origin of insoluble oxygen in chemical analysis, were selected randomly from a sample (Exp. 4 after 10 min), and subjected to chemical analysis and phase determination. Over 80 % of observed inclusions were Al2O3 of the size near 10 m m in diameter. The rest were in the form of the two-layer structure: an Al2O3 core surrounded by Al Ti O complex oxide. Figure 6 shows typical examples of inclusions found in the present study. The composition of the outer Al Ti O oxides was found to vary: the Al/Ti molar ratio was in the range of 2/3 to 2/1 with the mean value of around 1/1. This information of the molar ratios is graphically represented in the Al2O3 Ti2O3 TiO2 ternary phase diagram10) as seen in Fig. 4. Discussion 4.1. Chemical Reactions It will be reasonable to assume that the major reoxidation of steel take place at the metal/slag interface. However, it can also be expected that the reoxidation may occur in the bulk metal to some extent, provided that oxygen necessary for reoxidation be available in the bulk metal. (1) Reactions at the metal/slag interface: 4Al 3(SiO2) 2(Al2O3) 3Si...(1) 4Ti 3(SiO2) 2(Ti2O3) 3Si...(2a) Ti (SiO2) (TiO2) Si...(2b) ISIJ

6 Fig. 7. The composition of the outer Al Ti O oxides in Fig. 6 plotted in the Al 2 O 3 Ti 2 O 3 TiO 2 phase diagram. 10) The analyzed Al : Ti ratio are plotted as dotted lines. The pseudobrookite solid solution exists between AlTi 2 O 5 and Ti 3 O 5 end-members. Fig. 8. Schematic sketch for the possible reactions involved in reoxidation process. (1) to (6) correspond to the reactions (1) to (6) in the text, respectively. Fig. 9. Variation of Ti and Al in reoxidation of (Al Ti)-containing steels by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag at K. Observed values are represented by symbols. The calculated values for the chemical equilibrium state between metal and slag are shown as curved lines. (SiO 2 ) Si 2O...(3) (2) Reactions in the bulk metal: 2Al 3O Al 2 O 3 (s)...(4) 3Ti 5O Ti 3 O 5 (s)...(5) xal yti zo Al x Ti y O z (s)...(6) The above reactions are schematically represented in Fig. 8. Ti can be dissolved into slag in the form of Ti 2 O 3 or TiO 2. Ti 3 /Ti 4 ratio of dissolved Ti oxide in slag is varied with slag composition as well as oxygen partial pressure. According to the previous studies, 11,12) the Ti 3 /Ti 4 ratio increases with decrease of slag basicity and decrease of oxygen partial pressure. Unfortunately, since there has been no experimental data for the same slag composition as in the present study, the Ti 3 /Ti 4 ratio for the present experiments was roughly estimated to be about 0.5 based on the previous studies 11,12) and this value was used in the calculations relevant to Ti oxidation by SiO 2 at metal/slag interface. In the case of the Ti oxidation reaction in the bulk metal, Ti 3 O 5 is formed as oxidation product at the concentration of Ti less than several thousands mass ppm. In previous studies, 13,14) only the interfacial reactions represented by Eqs. (1) and (2a) (or (2b)) were taken into consideration to explain the reoxidation process. Moreover, most of the previous studies have not taken the phenomenon of oxygen supersaturation in the metal into account in interpretation of the reoxidation process. However, as observed in the present experiments, there is a considerable extent of supersaturation of oxygen occurring in the reoxidation of (Al Ti)-containing steels (see Fig. 5). Considering that SiO 2 in the slag is the only source of oxygen for reoxidation of the steel, the oxygen responsible for the supersaturation should be originated from the self-dissociation of SiO 2 according to the reaction represented by Eq. (3). The supersaturation of oxygen in the bulk metal renders possibility of reoxidation in the bulk metal according to the reactions represented by Eqs. (4) (6) Oxidation of Ti in the Initial Stage The reoxidation process of steel by reaction with slag, particularly the change of Al and Ti concentrations in the steel, may be simulated in the following manner, provided that the reoxidation proceed in such a way that the metal/ slag interface is always in chemical equilibrium: (1) A small portion of the slag is added to the steel, and the system is then allowed to be equilibrated. The change of steel compositions, particularly of the Al and Ti contents, is computed. (2) Another small portion of the slag is added, and the same is done as above. (3) The above is repeated until the steel has been substantially reoxidized. Figure 9 shows the change of Ti along with the change of Al due to the reoxidation by the slag, which was computed according to the above-mentioned sequence by using the FactSage 8) thermodynamic software. It is seen that, in the initial stage of reoxidation where the Al content in bulk metal is still high, the rate of decrease in the Ti content is sluggish. Actual changes of Al and Ti contents observed in the present study are superimposed in the figure and show vastly different from the computed predictions: The rate of the Ti oxidation is much higher in the actual case than in 2004 ISIJ 1674

7 Fig. 10. The relations of d[ti]/d[al] and [Al] in the steel during reoxidation. The symbols and lines are the same as in Fig. 9. the prediction which is based on the interfacial equilibrium. This discrepancy is much more clearly seen in Fig. 10 where d[ti]/d[al] is contrasted against Al content in bulk metal (in mass ppm). In general, the trend of the observed variation is similar to the calculated ones. The d[ti]/d[al] increases with increase of initial Ti content and decrease of Al content in the steel. However, there is significant difference between the observed d[ti]/d[al] values and calculated ones. In the initial stage where Al content is high enough, the observed values are several times higher than the calculations, which means that Ti in the steel is oxidized much faster than the thermodynamic equilibrium state. The disagreement between the actual experimental results and thermodynamic predictions suggests that the oxidation of Al and Ti does not proceed in a manner that the chemical equilibrium is maintained at the metal/slag interface. A possible approach to resolve this apparent disagreement is to look at the interfacial reactions in a microscopic (atomic) kinetic point of view. Suppose that a Ti atom is oxidized by reacting with SiO 2 in slag to form a titanium oxide. Since this oxide is unstable in existence of high Al, it is highly susceptible to the attack of neighboring Al atoms, particularly when the Al concentration is high. However, some titanium oxides may be able to survive from the Al attack owing partly to a large spatial distance between Al and Ti atoms at the interface and subsequently to dissolution into the bulk slag. If this is the case, the faster oxidation rate of Ti in existence of high Al can be explained. However, this view is tentative and warrants a further study in the future Supersaturation of Oxygen In the present study, the source of oxygen for oxidation of Al or Ti in the steel is SiO 2 in the slag. Since the oxygen partial pressure of the Ar gas used to purge the reaction chamber was measured to be about atm, any noticeable oxidation by the ambient gas can be ruled out. SiO 2 in the slag may act as the source of oxygen supply in two different ways; namely, 1) by directly reacting with either Al or Ti (represented by Eqs. (1) and (2)), and 2) by self-dissociation into Si and O (represented by Eq. (3)). If the first case is the major mechanism, the increase of Si in bulk metal should agree with the decrease of Al and Ti in bulk metal in mass balance according to the stoichiometric Fig. 11. Fig. 12. The comparison of Si content calculated from the mass balance for the Al and Ti reactions at interfaces (reactions (1) and (2)) and actual measured Si content. The difference represents Si content enriched by SiO 2 selfdissociation (reaction (3)). Variation of Si contents in metal with time. Si(1) is the Si content calculated by relation of [Si] [Si] ini. (28/32) D[O] sol. D[Si] (based on the changes of Al and Ti contents); Si(2) is Si content measured in the experiment of reoxidation of (Al Ti)-containing steel; Si(3) is Si content measured from the separate SiO 2 dissociation experiment. relationship given in Eqs. (1) and (2). Figure 11 shows an example of the change of the Si content in the steel observed in the present study, and the change calculated from the stoichiometric relationship with Al and Ti given in Eqs. (1) and (2). It should be noted that the observed Si content is consistently higher than the calculated one. The existence of excess Si to a substantial extent implies that there is an additional source of Si increase other than the Al and Ti oxidations represented by Eqs. (1) and (2). This additional source must be the self-dissociation of SiO 2 mentioned above and represented by Eq. (3). The Si content was recalculated by considering change of dissolved oxygen by relation of [Si] [Si] ini. (28/32) D[O] sol. D[Si] (based on the changes of Al and Ti contents). The results are plotted in Fig. 12. As can be seen in the figure, the recalculated Si is in a good agreement with the measured Si, indicating that the excess Si in metal is caused by the self-dissociation of SiO 2. In order to confirm that the observed excess Si in metal is truly from the self-dissociation of SiO 2 in the slag, a separate experiment (Exp. 8 in Table 1) was carried out to determine the rate of the self-dissociation of SiO 2 into steel ISIJ

8 by keeping the slag composition being the same as that used in the present main study and with the steel containing no Al or Ti, but some dissolved oxygen about 80 mass ppm. The results are also given in Fig. 12. Although the rate of self-dissociation of SiO 2 in existence of Al or Ti in the steel could be different from the one without Al and Ti, this figure strongly supports the view that SiO 2 in the slag does dissolve into the metal by dissociation into Si and O. The origin of the oxygen for the supersaturation shown in Fig. 5 can then be attributed to the self-dissociation of SiO 2. Figure 13 shows a typical example of the relationship among changes with time of Al and O contents, and the rate of the Al content change. It is of interest to note that the supersaturation of oxygen becomes significant when the rate of the Al content change begins to decrease sharply. This appears to imply that when the rate of Al oxidation is high, SiO 2 at the interface participates mostly in the Al oxidation, and hence the self-dissociation of SiO 2 is negligible. When the rate of Al oxidation becomes sluggish, however, SiO 2 available for self-dissociation increases, and hence the selfdissociation of SiO 2, in other words, the supersaturation of oxygen, becomes significant. The extent of supersaturation in the system of M O M x O y (s) in steel (activity of M x O y 1.0) may be defined as x y am ao S...(7) MO ( ) x y 1/ KM where K M is the equilibrium constant for the reaction of xm yo M x O y (s)...(8) Fig. 13. The relationship among changes with time of Al and O content, and the rate of the Al content change. Fig. 14. Variation of soluble oxygen contents during the reoxidation of (Al Ti)-containing steels by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag at K plotted in Al (a), Ti (b) and Si (c) deoxidation curves. Noted that the supersaturation against Al 2 O 3 is much higher than Ti 3 O 5. No supersaturation against SiO 2 was observed. Filled symbols represent the final compositions of each experiment ISIJ 1676

9 Figures 14(a) 14(c) show the experimental data of the change of the soluble oxygen content against the Al, Ti and Si contents, respectively. The curves in the figures were calculated using FactSage 8) thermochemical software with a new FACT database for liquid steel based on a new Associate Model. 15) While the classical Wagner formalism 16) for dilute liquid solutions Fe M O assumes that all dissolved elements (M and O) exist as separate atoms and are distributed randomly, the Associate Model assumes that dilute elements which have a strong affinity for oxygen dissolve as associated molecules (M*O), as well as separate atoms (M and O), and that all species are distributed randomly. The dissolved species are at equilibrium: M O M*O with Gibbs energy of this reaction Dg M*O which is the only model parameter. (In certain cases, M 2 *O associated molecules are also assumed to describe deoxidation phenomena in liquid steel more accurately.) Furthermore, it has been found 15) that Dg M*O is independent of temperature in all cases. In this way, no interaction parameters between M and O are required. The new Associate Model reflects the actual structure of the solution more closely than the classical Wagner formalism 16) and gives a much better description of the configurational entropy. For examples, in the case of Al deoxidation, the reactions Al O Al*O and 2Al O Al 2 *O were assumed and their reaction energies are Dg Al*O J/mol and Dg Al2 *O J/mol. The details of the Associate Model, and model parameters for Al, Ti and Si deoxidations are found in the recent study by Jung et al. 15) In Fig. 14, the solid lines represent the curves for supersaturation (defined in Eq. (7)) plotted with the activity of oxides being unity. Included also are the relationship between dissolved oxygen and metallic elements (Al, Ti and Si) in equilibrium with the slag used in this study (14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 in mass%). In the slag, the activities of Al 2 O 3 and SiO 2 were calculated to be 0.62 and 0.27, respectively, at K. 8) In the case of Ti deoxidation, it is very difficult to find the activity of titanium oxide in the slag due to the amount of the oxide being negligibly small. Therefore, a plot with the oxide activity being 0.1 is included just as a reference. It is of interest to note from Fig. 14 that although the oxidation process appears to overshoot the equilibrium condition, particularly for the case of Al, during the active reaction, the whole system gradually approaches the true equilibrium between the metal and slag. It is seen from Fig. 14(a) that, as the Al content in the steel decreases by oxidation, the dissolved oxygen content in the steel increases rapidly and becomes eventually supersaturated with respect to the Al O Al 2 O 3 equilibrium. The ultimate extent of supersaturation of oxygen is about S Al2 O The dissolved oxygen content tends to stay constant after this, even though the Al content decreases further. The dissolved O and Al contents eventually approach the true equilibrium condition when the Al content becomes low to the range of single mass ppm. As can be seen in Fig. 14(b), oxygen is also supersaturated with respect to the Ti O Ti 3 O 5 equilibrium in the course of Ti oxidation, but the extent of supersaturation is not as significant as that with respect to the Al O Al 2 O 3 equilibrium. For the case of system, Fig. 14(c) reveals that the dissolved oxygen content is below the value for the Si O SiO 2 equilibrium. The oxygen content appears to approach the equilibrium value only when the Al and Ti contents in the steel become substantially low (single mass ppm level). Based on the experimental results and discussions given above, the course of oxygen supersaturation in the steel can be summarized in the following manner: (1) During the initial period of reoxidation, where the Al and Ti contents in the steel are high, SiO 2 at the metal/slag interface is mostly used up for Al and Ti oxidation and hence not available for self-dissociation, which results in a very low concentration of dissolved oxygen in the steel. (2) As the Al and Ti contents decrease, more SiO 2 becomes available for self-dissociation and hence the dissolved oxygen content increases. (3) The dissolved oxygen content eventually goes beyond the value for equilibrium with the Al O Al 2 O 3 system, and hence becomes supersaturated with respect to the Al O Al 2 O 3 equilibrium. The reason for the supersaturation is not yet clearly known, but may be attributed to difficulty associated with homogeneous nucleation of oxides. (4) Toward the end of the reoxidation process, the entire system including the metal and slag approaches the global equilibrium and the dissolved oxygen content arrives at the value which is in equilibrium with both the metal and slag Formation of Complex Inclusions In the previous section (Sec. 3.4) it is reported that portion of inclusions found in the steel is in the form of two-layer complex oxides (Al Ti O complex oxide with an Al 2 O 3 core inside). Formation of this somewhat peculiar type of inclusions may be explained with the help of supersaturation discussed in the above section. As long as the supersaturation condition prevails during the course of reoxidation, it must always be possible to form an oxide in the bulk of the steel, provided that all barriers against nucleation of the oxide be overcome. As seen in Fig. 14, for a given dissolved oxygen content, Al shows much higher extent of supersaturation than Ti does. This implies that Al has much greater potential for forming an oxide than Ti does. When a certain local fluctuation in the bulk metal is large enough to remove the barriers against nucleation, an oxide will nucleate, and the oxide forming first should be Al 2 O 3 (Eq. (4)). When an Al 2 O 3 nucleus forms, the Al and O content in the vicinity of the oxide will become lower due to consumption for Al 2 O 3 formation. As the oxide grows, both the Al and O contents in the steel adjacent to the oxide will continue to decrease. The local condition will eventually arrive at a stage where the driving force for oxidation becomes the same for both Al and Ti in the vicinity of the oxide. Then simultaneous oxidation of Al and Ti will occur and form a complex Al Ti O oxide phase outside the Al 2 O 3 oxide (Eq. (6)). As the above reactions occur under the highly supersaturated condition, the rate of the reactions should be high so that the mass transfer in the steel may not be fast enough to remove the local concentration gradient in the vicinity of the oxide ISIJ

10 5. Conclusions Reoxidation of liquid steel containing Al and Ti by 14%CaO 35%Al 2 O 3 10%MgO 41%SiO 2 (in mass%) slag was studied at K in Al 2 O 3 crucible. The initial Al content was fixed at about 820 mass ppm and Ti content was varied from 100 to 500 mass ppm. The following results were obtained from the present study: (1) When steel contains Al and Ti, both of them oxidize simultaneously by SiO 2 in the slag, but at different rates. This is contrary to thermodynamic prediction; it is Al that oxidizes, keeping Ti mostly intact, particularly when the Al content is high. (2) The oxidation of Al is faster in the Al only-containing steel than in the (Al Ti)-containing steel. The same is true for Ti, but Ti shows much larger difference in the rate. (3) Soluble oxygen is supersaturated in the course of reoxidation to a substantial extent with respect to Al O Al 2 O 3 equilibrium, and the supersaturation becomes significant when the rate of oxidation of Al begins to decrease. (4) Oxide inclusions with two-layer complex structure, i.e., an Al 2 O 3 core surrounded by complex Al Ti O oxide, form in the bulk metal, although the vast majority of inclusions are in the form of Al 2 O 3 single phase. From the above findings, a mechanism of reoxidation of (Al Ti)-containing steel by SiO 2 in the slag is proposed: (1) Oxidation of Al and Ti at the metal/slag interface is governed more by kinetic driving force rather than by thermodynamic one, which leads to simultaneous oxidation of Al and Ti. (2) Apart from reacting with Al or Ti at the metal/slag interface, SiO 2 dissociates into Si and O, and dissolve in the bulk metal. This is responsible for the supersaturation of oxygen. (3) Oxidation reactions also occur in the bulk metal between the dissolved Al and supersaturated O to form Al 2 O 3, followed by oxidation of Al, Ti and O to form complex Al Ti O oxide. This is responsible for the formation of complex oxides with two-layer structure abovementioned. Acknowledgements The authors wish to express their appreciation to POSCO for providing financial support which enabled this study to be carried out. REFERENCES 1) Y. Tago, Y. Higuchi and S. Fukagawa: CAMP-ISIJ, 9 (1996), 60. 2) Y. Higuchi, Y. Tago, K. Takatani and S. Fukagawa: Tetsu-to-Hagané, 84 (1998), ) H. Goto and K. Miyazawa: ISIJ Int., 38 (1998), ) W. W. Huh and W. G. Jung: ISIJ Int., 36 (1996), S136. 5) H. Sun and K. Mori: ISIJ Int., 36 (1996), S34. 6) K. Sasai and Y. Mizukami: ISIJ Int., 36 (1996), ) K. Sasai and Y. Mizukami: ISIJ Int., 38 (1998), ) C. W. Bale, P. Chartrand, S. A. Degterov, G. Eriksson, K. Hack, R. Ben Mahfoud, J. Melancon, A. D. Pelton and S. Petersen: Calphad, 26 (2002), ) K. R. Lee and H. Suito: ISIJ Int., 35 (1995), ) S. C. Park, I. H. Jung, K. S. Oh and H. G. Lee: ISIJ Int., 44 (2004), ) G. Tranell, O. Ostrovski and S. Jahanshahi: Metall. Mater. Trans. B, 33B (2002), ) H. D. Schreiber, T. Thanyasiri, J. J. Lach and R. A. Legere: Phys. Chem. Glasses, 19 (1978), ) W. Pluschkell, B. Redenz and E. Schürmann: Arch. Eisenhüttenwes., 52 (1981), ) K. Okohira, N. Sato and H. Mori: Trans. Iron Steel Inst. Jpn., 14 (1974), ) I. H. Jung, S. A. Decterov and A. D. Pelton: Metall. Mater. Trans. B, 35B (2004), ) C. Wagner: Thermodynamic of Alloys, Addison-Wesley, Reading, MA, (1962), ISIJ 1678