3.2.1 Morphology and Composition of Inclusions
Figure 1 shows the typical inclusion morphology and composition of both heats during RH vacuum treatment. As can be seen, the inclusions are globular and the composition is mainly CaO–Al2O3–MgO throughout the RH treatment.
Figure 1. Typical inclusion morphology and composition during RH treatment. a) 0 min, b) 6 min, c) 10 min, d) 16 min, e) 22 min, f) 28 min.
Download figure to PowerPoint
CaO–MgO–Al2O3 ternary diagrams are applied to demonstrate the composition change of inclusions in the present study, since the main components of inclusion are the three oxides CaO, MgO, and Al2O3, of which the sum is not less than 85% for all samples. The inclusion compositions of all samples for two heats are shown in Figure Fig. 2 and Fig. 3, respectively. In these two figures, each open circle represents an individual inclusion, and the filled square represents the average composition of all inclusions detected. The line is calculated by Thermo-Calc within which is the liquid region below 1873 K. The inclusions in this region are fully liquid and classified as C12A7 type inclusions, while those outside this region are non-liquid ones including fully solid or solid + liquid complex inclusions and can be divided into two types based on the composition. One is C3A type inclusions of which the CaO content is over 50%, the other is Al2O3 · MgO type inclusions of which the CaO content is less than 50%. From the two figures, it can be found that inclusions at different times mainly are C12A7, C3A and Al2O3 · MgO in both heats. The average compositions of inclusions are locating in the liquid region for all samples, but changes slightly with time. Al2O3 increases while CaO and MgO decreases, which are caused by the loss of [Ca], [Mg] and the reaction expressed by Equation (1).
Figure 2. Change of composition distribution of inclusions with time during RH vacuum treatment in Heat 1. a) 0 min, b) 6 min, c) 10 min, d) 16 min, e) 22 min, and f) 28 min.
Download figure to PowerPoint
Figure 3. Change of composition distribution of inclusions with time during RH vacuum treatment in Heat 2. a) 0 min, b) 6 min, c) 10 min, d) 16 min, e) 22 min, and (f) 28 min.
Download figure to PowerPoint
3.2.2 Number of Inclusions
Figure 4 shows the change of inclusion number with time during RH treatment. It can be observed that the total number of inclusion decreases with time. And that is the same for both fully liquid and non-liquid inclusions. And the removal ratio of inclusion for Heat 1 and 2 can reach up to 63% and 72% after 28 min treatment. As for the decrease of the inclusion, Miki et al. thought inclusion coagulation due to collision are mainly responsible for the decrease in inclusion population for the small inclusions less than 5 µm, and inclusion flotation by attachment to argon bubbles decreases mainly the number of large inclusions. They are removed to the slag layer above the ladle or the top free surface of the RH degasser. It can also be observed that liquid inclusions are far more than non-liquid ones. It means that the C12A7 type inclusions are much more than the total of C3A and Al2O3 · MgO type inclusions.
Figure 5 compares the variation of inclusion number with time for two heats. In Y axis, t is time (min), Nt and N0 are inclusion number at t and the start of the RH treatment. As seen in Figure 5, the inclusion number decreases but the decreasing rate slows down with time. In addition, the trends of Nt/N0 for both heats are almost the same. The result indicates that under the same operating conditions, the initial number of inclusion plays a vital role in the final inclusion number after RH treatment, the more the initial inclusions, and the more the final inclusions. So efforts must be made to control the inclusion number before RH treatment to get a highly clean steel in a short time. Also it can determine how much time is needed to get the aimed inclusion content based on Figure 5. Since inclusion is closely related to the defects of steel, and different steel grade has a different requirement on inclusion content. It made possible to select the optimum operating condition matching for the required quality, such as RH circulation time.
Figure 6 shows the variation of the inclusion number of different types with time in each heat. For both heats, both liquid and non-liquid inclusions decrease with time, but non-liquid inclusions decrease faster than liquid ones at the beginning of RH treatment. And the removal ratios, which can be calculated by Equation (2), are different, it only needs 6 min for non-liquid inclusions to get a removal ratio of 60%, while it costs 28 min to get the same removal ratio for liquid ones. And the removal ratio is 60% for liquid inclusions and 80% for non-liquid ones after the RH treatment. These results indicate that non-liquid inclusions can be removed more easily and thoroughly during the RH treatment.
Murai et al. found the removal of total oxygen content (T.O) during RH treatment is in the form of Equation (3), of which C is T.O and K0 is the removal rate of T.O. Similar to their result, the removal of inclusion during RH is also in such form. But in this study, C is Nt/N0 and the K0 is the inclusion removal rate. The calculation result is shown in Figure 7, the ln(Nt/N0) is the mean value of two heats, and the absolute value of the slope is the inclusion removal rate K0. It can be observed that ln(Nt/N0) is proportional to time and decreases steadily throughout the time for liquid inclusions, but for non-liquid inclusions, it decreases quickly at 6 min, then slows down. The difference could be explained as follows: for non-liquid inclusions, they are abundant in the first 6 min, and they are easy to collide into bigger ones, so they are removed faster. After 6 min, most of the inclusions have been removed and the number of remaining inclusions is so small that the collision and coagulation hardly happen even though a big stirring energy exists, so the remaining inclusions can only be removed in the form of small mono-particles, rather than big coagulated ones. But the removal rate of smaller inclusions particles is slower than that of bigger ones, so the removal rate slows down after 6 min. As for liquid inclusions, however, the inclusions are abundant throughout the time, they can collide and coagulate into bigger ones and get removed, so the removal rate does not change. Furthermore, it could be concluded that in the first 6 min both non-liquid and liquid inclusions are removed in coagulated forms, and the non-liquid inclusions are removed faster than liquid ones, which are 0.150/min and 0.035/min. After 6 min, non-liquid inclusions are removed as mono-particles and liquid inclusions are still removed in coagulated forms, non-liquid inclusions are smaller than liquid ones, but non-liquid inclusions are decreased in almost the same speed as liquid ones, both are 0.035/min. All this could prove that non-liquid inclusions are removed more easily than liquid ones
As mentioned above, liquid inclusions are C12A7 type, while non-liquid inclusions can be divided into C3A and Al2O3 · MgO type inclusions. Figure 8 shows the removal behavior of C3A and Al2O3 · MgO type inclusions. It can be observed that the removal ratio of C3A type inclusions are nearly the same in both heats, and that's also the case for Al2O3 · MgO type inclusions. Just like the removal behavior of total non-liquid inclusions in Figure 7, both C3A and Al2O3 · MgO type inclusions decrease quickly in the first 6 min, then slow down. And they are removed easily than C12A7 type inclusions. These results obtained from Figure Figs. 7 and 8 indicate that there must be some property that can determine the similarity of removal behavior of either liquid (C12A7) or non-liquid inclusions (C3A and Al2O3 · MgO) in both heats, and might explain why non-liquid inclusions can be removed more easily than liquid ones.
The contact angle, work of adhesion and interfacial energy between steel melt and inclusion are employed to explain the different removal behaviors of liquid and non-liquid inclusions.
Since it's impossible and unnecessary to study the removal behavior of every single particle of thousands of inclusions, average compositions of C12A7, C3A and Al2O3 · MgO in this study are applied. Due to the similar inclusion behavior of both heats and nearly consistent inclusion composition in each heat with time during RH treatment, the first sample at RH 0 min in Heat 1 is selected, and the composition is shown in Table 3. As can be seen, the C3A type inclusions have a higher CaO content, and C12A7 type inclusions have a medium CaO content, while Al2O3 · MgO type inclusions have higher Al2O3 and MgO.
Table 3. Average composition of inclusions at RH 0min in heat 1
|Non-Liquid||Al2O3 · MgO||64||29||7|
The surface energy γSG is 1844 cm−2 calculated by Equation (7). The [O] and [S] used in this calculation is 0.00026% and 0.0012%. The [O] is calculated by Equation (8) assuming the activity of Al2O3, the activity coefficient of [O] and [Al]s are unity and [Al]s at RH 0 min in Heat 1 is 0.042%
Table 4 shows the interfacial energy between inclusion and gas (γIG), contact angle (θC), the work of adhesion (Wad), the interfacial energy (γSI) between inclusion and steel melt at 1873 K. 54.3° and 670 erg cm−2 are chosen as θC and γIG of C12A7 since its average composition is very close to the data reported by Bretonnet et al. The visible contact angle α for C12A7 is needed to calculate γSI since it's liquid at 1873 K, and the data is 35.7° in their study. The γSI and Wad for C12A7 are calculated by Equation Eq. (5) and Eq. (6). Shinozaki et al. have measured the contact angle of Al2O3 · MgO, they found it nearly did not change with the increase of MgO for molten steel with small oxygen content. The measured contact angle of Al2O3 · MgO with 20% MgO is 134.1°. So 134.1° is applied here to represent the contact angle of Al2O3 · MgO in which MgO is 29% in this study. The γSI and Wad for Al2O3 · MgO are calculated by Equation Eq. (4) and Eq. (6). Because no θC and γIG are available for C3A to date, only Al2O3 · MgO is compared with C12A7 to elucidate the different behaviors between liquid and non-liquid inclusions.
Table 4. Interfacial energy between inclusion and gas (γIG), contact angle (θC), work of adhesion (Wad), the interfacial energy (γSI) between inclusion and steel melt at 1873 K
|Particle||Oxide||γIG/erg cm−2||θC/deg||Wad/erg cm−2||γSI/erg cm−2|
|Al2O3 · MgO||512||134.1||561||1795|
As can be seen in Table 4, the θC of Al2O3 · MgO is 134.1°, much larger than C12A7, 54.3°. Besides, Al2O3 · MgO has a higher γSI than liquid one C12A7, which means the Al2O3 · MgO inclusions are more likely to collide and detach from steel melt. Both behaviors will reduce the total interfacial energy of the inclusion-steel system, which is the product of interfacial energy multiplying the contact area, through reducing the contact area, and these processes are spontaneous. Furthermore, Al2O3 · MgO has a much lower Wad, which indicates it will be much easier to be removed from the steel bulk since it needs less energy to detach form the steel. Table 4 also lists relevant date for Al2O3 and MgO. It can be inferred that for a given particle size, Al2O3 can be removed faster than Al2O3 · MgO, MgO, and C12A7, since it has the largest γSI and lowest Wad. And it's supported by other authors'[2, 3, 5] experiment result, they all found that Al2O3 can be removed quickly from the steel melt. Furthermore, Yin et al. confirmed that the collision and agglomeration of liquid CaO–Al2O3 particles are more difficult compared with Al2O3 at inert gas/molten steel interface due to the different interfacial parameters. Also it can be deduced form Table 4 that MgO can be removed faster than C12A7, because the γSI of MgO is larger than C12A7 but the Wad of MgO is smaller than C12A7.