The characteristics of nonmetallic inclusions are expected to significantly affect the production process and the product quality of steel. The problem of nozzle clogging, for instance, is caused by the deposition of large-sized inclusions and clusters with high melting point on the inner wall of submerged entry nozzle, which leads to unstable casting and causes crack defects on slab.[1,2,3,4]

Aluminum is one of the most popular deoxidizers because of its strong affinity to oxygen, thereby a large number of Al2O3 formed in molten steel. Mg was usually introduced to the Al-killed molten steel from refining slag or refractory containing MgO.[5,6] Many studies have proposed that trace amount of Mg was sufficient to modify Al2O3 to MgO–Al2O3 through theoretical calculations and experiments.[7,8] Although inclusions sizes can be reduced by means of this modification, both Al2O3 and MgO–Al2O3 are inclusions with the characteristic of high melting point and high hardness, which caused nozzle clogging problem and product quality defects.[9,10,11] In order to alleviate the harm of these inclusions, Ca treatment has been widely used to modify them to calcium aluminate in the liquid phase. Many studies were performed to discuss the thermodynamic,[12] kinetics,[13,14,15] and evolution mechanism[16] for the modification process in Al–Mg–Ca–O deoxidized steel under both laboratory scale experiments and industrial trials. However, these studies mainly investigated the inclusion information through postmortem analysis in quenched high-temperature samples.

The technology of confocal laser scanning microscopy (CLSM) originated in the 1980s is capable of observing in situ experimental phenomena at high temperature, so as to directly reflect the characteristics of inclusions. The precipitation behavior of inclusions during steel solidification[17] and agglomeration behavior of various inclusions caused by capillary interaction[18,19,20,21] were investigated by using CLSM. However, due to the lack of inclusions control during observation and the limitations of CLSM experimental methods, the transient evolution of inclusions characteristics during its modification, especially the transition from solid inclusions to liquid ones, has not been investigated by in situ observation yet.

In the present work, the alloy chosen is Fe–21Cr–11Ni stainless steel, in which Al2O3 and MgO–Al2O3 inclusions were expected to be modified to the liquid phase because of their harm to casting process and products quality. To avoid the interference between different kinds of inclusions in the in situ observation, the samples containing a single type of inclusions (Al2O3 or MgO–Al2O3) and the sample with Ca treatment were prepared according to thermodynamic calculation. Then, the novel CLSM experiments were carried out to document and quantify the evolution of inclusions from solid phase to liquid ones, with an emphasis on their sizes change.

Thermodynamic calculation was performed by FactSage 8.1 software and thereby the stability phase diagram of inclusions was evaluated in Fe–21Cr–11Ni–Al–Ca–(0.0008 Mg)–O melts at 1873 K (1600 °C). Figure 1(a) shows that Al2O3, CaO, CaO–Al2O3 solid and liquid oxides are in equilibrium with molten steel in the Ca and Al contents ranging from 5 × 10–6 to 0.005 pct, and 0.001 to 1 pct. When the melt contains 0.001 pct of Mg, Al2O3 and Al2O3–CaO solid oxide are replaced by MgO–Al2O3 and MgO, as shown in Figure 1(b). Approximately 0.0002 to 0.001 pct of Ca is sufficient to modify the inclusions to CaO–Al2O3 liquid oxide in both two melts.

Fig. 1
figure 1

Calculated inclusion stability diagram of the Fe–21Cr–11Ni–Al–Ca–Mg–O melts at 1873 K: (a) without Mg; (b) with 0.0008 pct Mg

Full size image

On the basis of thermodynamic calculation, the steel samples containing Al2O3 or MgO–Al2O3 inclusions were prepared at 1873 K (1600 °C) in a MoSi2 furnace under argon atmosphere. Different amounts of Al, Mg, and Ca were added into the melts containing a certain amount of electrolytic iron, chromium metal, and electrolytic nickel, and then sampled after 30 min to achieve sufficient homogenization of the melts. The details of the experimental operation were described in detail in other literature.[22] In the current work, sample A was deoxidized by Al, Mg treatment after Al deoxidation was used in sample B. Sample C was only treated with Ca and without Al and Mg addition.

The in situ observation of inclusions in the above samples was carried out in CLSM (Lasertec VL2000DX), which is equipped with a He–Ne laser and an infrared image furnace. Prior to heating, the chamber was filled with high-purity Ar gas (purity > 99.999 pct) at a constant flow of 50 to 60 mL/min to evacuate the air. The cycle of vacuum and purging with Ar was performed three times to clean the chamber thoroughly. The temperature profile of experiments is shown in Figure 2(a). First, samples A and B were heated to 1873 K (1600 °C) with a heating rate of 300 °C/min. During the 10 min holding time, the inclusions behavior in molten steel was recorded by the CRT monitor at a rate of 30 frames per second. After observation, the heating power was directly turned off so as to quench the samples immediately and retain the characteristics of inclusions at high temperature. Subsequently, another sample A or B was heated in the same way and quenched immediately after observing for 5 min (corresponding to the red-dotted line in Figure 2(a)), which was then processed to prepare for the following CLSM experiment of Ca treatment, and the processing method is shown in detail in Figure 2(b). A large number of inclusions and agglomerated clusters were floated up to the sample surface during observation. Thus the steel disks containing inclusions and clusters can be obtained by cutting the upper third of the quenched samples. Then the steel disks containing the inclusions or clusters from sample A and B were placed on the surface of Ca-treated sample C to form sample A + C and B + C, respectively, and the inclusion behavior was observed in situ in the next 5 min (corresponding to the green-dotted line in Figure 2(a)). After observation, the quenched samples were cross sectioned and polished, for analyzing the morphology and chemical composition of the inclusions by SEM–EDS. The compositions of samples before and after in situ observation were determined by ICP-OES with an accuracy of ± 0.5 ppm, as shown in Table I, which are marked as red star in Figure 1, indicating that the experimental results were in accord with the theoretical design, that is, the main inclusions in samples A and B should be Al2O3 and MgO–Al2O3, respectively, and the compositions of inclusions after Ca treatment were entered into the region of liquid oxide.

Table I Chemical Analyses of Experimental Samples (Mass Percent, Pct)
Full size table
Fig. 2
figure 2

Schematic of CLSM experiments: (a) temperature profile; (b) method of samples processing

Full size image

The behaviors of various inclusions in molten steel were observed in situ. Figure 3(a) shows the agglomeration process of Al2O3 inclusions in sample A. Two solid particles with cluster shapes attracted each other to form clusters 1 + 2, then particle 3 moved rapidly towards clusters 1 + 2, and a larger cluster of inclusions consisting of 1 + 2 + 3 was formed in 0.4 s. MgO–Al2O3 inclusions in sample B were very small and tended to be dispersed, and non-agglomeration behavior among them within 34 s is exampled in Figure 3(b). Seven inclusions are randomly distributed in molten steel over time. Even though the distance between inclusions 6 and 7 was shortened to less than 20 μm shown in Figure 3(b3), they did not further agglomerate, but moved away from each other to a distance of about 60 μm shown in Figure 3(b4).

Fig. 3
figure 3

Images from the in situ observation of pure inclusions (a) Al2O3 particles in Sample A at 860.18 s (a1), at 860.98 s (a2) at 861.78 s (a3), at 862.18 s (a4); (b) MgO-Al2O3 particles in Sample B at 942.18 s (b1), at 948.68 s (b2), at 972.28 s (b3), and at 975.42 s (b4)

Full size image

In Ca-treated sample A, two typical variations of inclusion characteristics were captured, which were determined by the reaction between Al2O3 inclusions and Ca shortly after the sample melting. It is obvious from Figures 4(a1) to (a4) that a solid inclusion with irregular shape gradually transformed into typical spherical liquid inclusions in about 10 s. In addition, some of the large-sized clusters first disaggregated into smaller size inclusions and then gradually transformed to spherical ones, as shown in Figures 4(b1) to (b4). The agglomeration force derived from cavity bridge force is known to the main mechanism of inclusions agglomeration, which is significantly affected by the wettability between inclusions and molten steel.[23,24] The disaggregation phenomena of inclusions may be mainly due to the wettability increasing of inclusions during their transformation caused by Ca treatment. The evolution behavior of inclusions in Ca-treated sample B is represented by Figures 4(c1) to (c4), in which three irregularly shaped MgO–Al2O3 inclusions with smaller size were successively liquefied into spherical inclusions within 30 s. It is worth mentioning that the spherical liquid inclusions formed after Ca treatment in both sample A and B did not agglomerate over time.

Fig. 4
figure 4

Images from the in situ observation of inclusions modification after Ca treatment (a) liquefying process of Al2O3 inclusions in Sample A + C at 1159.18 s (a1), at 1176.38 s (a2), at 1187.38 s (a3), at 1198.58 s (a4); (b) disaggregating and liquefying process of inclusions in Sample A + C at 1051.78 s (b1), at 1068.18 s (b2), at 1092.38 s (b3), at 1105.18 s (b4); (c) liquefying process of MgO-Al2O3 inclusions in Sample B + C at 1167.18 s (c1), at 1177.69 s (c2), at 1193.29 s (c3), at 1201.58 s (c4)

Full size image

The morphology and composition of the inclusions in quenched samples after observation were analyzed by SEM–EDS. As shown in Figure 5(a), the large-sized cluster of Al2O3 in sample A was composed of numerous small-sized ones. The small-sized MgO–Al2O3 inclusions shown in Figure 5(b) were distributed in sample B, indicating that even though the distance between them is < 4 μm, the two clusters still did not agglomerate further. After Ca treatment, the complex inclusion with MgO–Al2O3 core wrapped by CaO–Al2O3 shell was detected in sample B + C as shown in Figure 5(c). Yang et al.[25] proposed that due to the limitation of kinetic conditions, MgO–Al2O3 with large size was difficult to be completely modified to liquid CaO–Al2O3 inclusions, instead of forming the complex inclusions similar to those in this work. In addition, Figure 5(d) shows the typical characteristic of liquid CaO–Al2O3–(MgO) inclusions dispersed in both samples A + C and B + C, these inclusions were modified from Al2O3 and MgO–Al2O3.

Fig. 5
figure 5

The composition and morphology of typical inclusions after in situ observation: (a) pure Al2O3 cluster in sample A; (b) pure MgO-Al2O3 in sample B; (c) complex inclusions with MgO-Al2O3 core wrapped by CaO-Al2O3 shell in sample B + C; (d) liquid CaO-Al2O3-(MgO) in samples A + C and B + C

Full size image

Figure 6 presents the variation of average size (calculated by the equivalent diameter) of inclusion particles with respect to time. The average size of Al2O3 inclusion increased gradually with the increase of time, and its maximum value was 26.57 μm, which agglomerate to form large-sized clusters due to the strong attractive force caused by capillary interaction between them presented in the previous study.[20,21,26] On the contrary, the attractive force between the dispersed MgO–Al2O3 inclusions was weak,[26] which leads to no significant increase of their size in this study. After Ca treatment, the size of inclusions particles decreased significantly from 18.79 to 12.71 μm caused by the modification from Al2O3 to CaO–Al2O3 liquid inclusions. Ferreira et al.[27] and Verma et al.[28] proposed that the size of Al2O3 clusters can be reduced after Ca treatment, and two possible explanations for which were that new inclusions with smaller size precipitated, or new calcium aluminates formed from the disaggregation of the original clusters. The phenomenon of in situ observation shown in Figure 4(a) combined with statistics of inclusions size changes here conformed to the latter explanation. The transformation from MgO–Al2O3 to CaO–Al2O3–(MgO) liquid inclusions led to size increase, which probably due to the formation of the complex inclusions shown in Figure 5(c), and whose size was still smaller than that from modification of Al2O3. Moreover, the size of liquid inclusions was slightly increased because of the weak attractive force among them that was reported in other literatures.[29]

Fig. 6
figure 6

The variation of inclusion average size with respect to time

Full size image

To sum up, the various inclusions (Al2O3 and MgO–Al2O3) were controlled and obtained in samples by thermodynamic calculation, and the sample containing less than 0.001 pct of Ca that was capable of modifying these inclusions to CaO–Al2O3–MgO liquid inclusions. On this basis, a novel CLSM experiment was designed to simulate the modification process from high-melting point inclusions to liquid ones by Ca treatment, in which the transient evolution of the inclusions especially their sizes change, can be observed in situ in molten steel. The pure Al2O3 inclusions particles agglomerated rapidly to form large-sized clusters. On the contrary, the MgO–Al2O3 inclusions were easy to be dispersed in molten steel. After Ca treatment, the irregularly shaped Al2O3 were transformed to spherical liquid CaO–Al2O3, and the average size of them decreased from 18.79 to 12.71 μm owing to the disaggregation of the former Al2O3 clusters. For some of MgO–Al2O3 inclusions with large size, Ca treatment was unable to completely modify them, resulting in the formation of the complex inclusions with MgO–Al2O3 core wrapped by CaO–Al2O3 shell. The inclusions size after modification of MgO–Al2O3 was much smaller than that of modifying Al2O3. Once the liquid inclusions were formed, their size no longer increased significantly due to non-agglomeration among them. The CLSM experimental method proposed in this study is expected to be performed to investigate the transient evolution of other inclusions in the future.