2024 Volume 64 Issue 15 Pages 2167-2175
Foaming slag generated in the steelmaking process, especially in hot-metal pretreatment and electric arc furnaces, is a gas-liquid coexistent fluid with CO gas generated by the interfacial reaction between slag containing iron oxide and hot metal or carbonaceous materials. In addition, it is essential to understand the flow behavior of foaming slag during slag-tapping and the sedimentation behavior of iron particles, which affects iron yield, and to expand our knowledge of the viscosity of gas-liquid coexisting fluids for CFD modeling of these phenomena. In the present study, the apparent viscosity of a foaming slag was systematically investigated, which was generated by reacting CaO–SiO2–FexO slag with Fe–C alloy and varying the composition, gas phase ratio, and shear rate of the slag. By adding Fe–C alloy powder to the slag, bubbles were continuously generated in the molten slag, and foaming slag suitable for viscosity measurement could be prepared. It was found that the higher the amount of Fe–C alloy powder, the larger the gas phase ratio of the foaming slag due to an increase in the number of bubbles generated. The relative viscosity of the foaming slag was found to increase with the gas phase ratio. The higher the rotation speed, the smaller the relative viscosity of the foaming slag indicating shear-thinning characteristics. The relationship between shear rate and shear stress calculated from the viscosity of the foaming slag did not show general non-Newtonian fluid behavior.
In the converter-type dephosphorization process involving hot metal pretreatment and intermediate slag removal in the blast furnace-converter steelmaking process, the hot metal containing a saturated carbon concentration reacts with oxygen (O) in the slag or O2 gas injected for oxidation refining. This reaction of the hot metal with the oxygen generates a large number of fine CO gas bubbles in the molten slag; this two-phase gas–liquid coexistent fluid is called foaming slag. A sudden foaming phenomenon can cause problems such as slopping and waiting for the foam to settle during slag removal.1,2,3) In the steelmaking process that uses an electric furnace, a cold iron source such as scrap and direct-reduced iron has a significantly lower carbon content than that of hot metal, and therefore, foaming caused by the abovementioned reaction is less likely to occur. In this case, foaming slag is formed by injecting a carbon material to protect refractories inside the furnace. However, even in this case, it is difficult to control the foaming phenomenon that occurs because of the interface reaction in the high-temperature oxide melt.
Given this background, the foaming phenomenon of slag has been extensively researched at the laboratory scale in test and actual converters. The oldest studies in this field are those by Cooper and Kitchener,4) Hara et al.,5,6) and Kitamura and Okohira,7) while other studies have focused on bubble lifetime and foaming index,8,9,10,11,12,13,14,15,16,17,18,19) such as those by Ito and Fruehan.20,21,22) Understanding the flow behavior of foaming slag during slag removal3) and the settling behavior of iron particles that influence iron yield23,24,25,26) is essential to expand our knowledge regarding the viscosity of two-phase gas-liquid fluids for computational fluid dynamics (CFD) modeling.27,28,29) From a geoscience perspective, Uhira proposed that volcanic eruptions are caused by rapid foaming and a decrease in fluidity attributed to the decompression of gas components dissolved in magma. Based on this concept, Uhira prepared a two-phase gas–liquid fluid by suspending air in silicone oil and measuring the viscosity using a rotational method. Subsequently, some authors of this study30) prepared a simulated foaming slag in which bubbles were dispersed by introducing nitrogen gas into silicone oil through a porous plug and measured its apparent viscosity using the same rotational method. We demonstrated that the apparent viscosity of the simulated foaming slag was several times higher than that of only the liquid phase (silicone oil) in the liquid phase viscosity range of the steelmaking slag. Further, the apparent viscosity showed non-Newtonian properties that changed with shear rate. A follow-up study31) prepared simulated foaming slag with dispersed CO2 gas generated by the reaction of aqueous sodium hydrogen carbonate solution with oxalic acid, where the apparent viscosity was systematically measured. The results indicated that the apparent viscosity of the simulated foaming slag was at least several dozens of times, and in some cases, over 100 times, greater than the viscosity of the liquid phase. This result is significantly higher than that in previous reports where foaming was conducted via gas injection. Meanwhile, in a recent research study on high-temperature system, Martinsson et al.32) prepared a foaming slag by reacting molten 43CaO-32SiO2-25FeO (mass%) slag with carbon-saturated liquid iron at 1873 K and measured the apparent viscosity using the rotational method. This pioneering study confirmed the non-Newtonian property, wherein the apparent viscosity decreased with an increase in the rotational speed of the viscosity meter.
Given this background, this study aimed to expand the fundamental knowledge about the flow behavior of two-phase gas-liquid fluids in the steel refining process. To that end, we prepared foaming slag suitable for viscosity measurements by reacting CaO–SiO2–FexO slag with Fe–C alloys and systematically investigated the apparent viscosity by changing the slag composition, gas phase fraction, and shear rate.
Table 1 presents the composition of the sample used in this study. Special grade reagents CaCO3, SiO2, and Fe2O3 (Sigma-Aldrich Japan Co., Ltd.) were used in the experiment. Each reagent was weighed to the specified composition and mixed in an alumina mortar. Next, the mixed powder sample was filled into a platinum crucible and melted in air at 1773 K, after which the slag was poured onto a copper plate and quenched to prepare the measurement sample.
| CaO | SiO2 | Fe2O3 |
|---|---|---|
| 40 | 40 | 20 |
| 30 | 30 | 40 |
| 20 | 20 | 60 |
The viscosity of the foaming slag is measured using the rotational method, and therefore, variations in the foaming height and gas phase fraction during measurement are undesirable. We first selected a carbon source (foaming agent) that can generate bubbles in the molten slag in a stable manner at 1673 K. Figure 1 shows a schematic of the equipment used in the experiment. A total of 50 g of the measurement sample was filled into a SUS310S crucible (55 mm ϕ outer diameter × 37 mm ϕ inner diameter × 70 mm height), melted in air at 1673 K, and held for 60 min. Subsequently, a foaming agent was placed onto the slag through an alumina tube. Then, CO and CO2 gas were generated by C+(O)=CO, C+2(O)=CO2 and the reaction with the generated CO (i.e., CO+(O)=CO2) to produce the foaming slag, following which the foaming lifetime was measured. We used five types of powders as foaming agents: Fe-1.443mass%C, Fe-0.185mass%C, and Fe-0.53mass%C (Japanese Iron and Steel Certified Reference Material) and carbides ZrC and NbC (Japan New Metals Co., Ltd.).
The foaming height was measured using Fe-0.53mass%C as the foaming agent. A total of 50 g of the measurement sample was filled into a SUS310S crucible (55 mm ϕ outer diameter × 37 mm ϕ inner diameter × 70 mm height), melted in air at 1673 K, and held for 60 min. Subsequently, the liquid height of the molten slag, which serves as a reference for the foaming height, was electrically measured by utilizing the electrical conductivity of the molten slag. One terminal of an LCR meter (Hioki E.E. Corporation, 3522-50) was connected to a platinum electrode (tip diameter = 10 mm), and the other terminal was connected to the stainless-steel crucible. The platinum electrode was lowered from the top of the molten slag, and the electrode contacted the molten slag to form a circuit and detect the current. The electrode height at this time was measured to determine the reference liquid height of the molten slag. Further, Fe-0.53mass%C powder was added into the molten slag through an alumina tube for generating a foaming slag, and the foaming height was measured in the same manner. For foaming height hfoam and initial liquid height hliquid, the gas phase fraction ϕ (vol%) of the foaming slag can be expressed as
| (1) |
Foaming slag was also prepared using a similar method independently of the above experiment, and bubbles that appeared on the foaming slag surface were photographed from above. A total of 100 bubbles were randomly selected from the photographed images, and the average value of the diameter equivalent to the projected area circle was considered as the bubble diameter.
2.3. Foaming Slag Viscosity MeasurementFigure 1 shows a schematic of the viscosity measurement device. A total of 50 g of the measurement sample was filled into a SUS310S crucible (55 mm ϕ outer diameter × 37 mm ϕ inner diameter × 70 mm height), melted in air at 1673 K, and held for 60 min. Next, Fe-0.53mass%C powder was placed onto the slag through an alumina tube to prepare the foaming slag. Subsequently, the spindle was immersed 20 mm into the foaming slag, and the torque generated on the spindle was measured using a rheometer head (Anton Paar, DSR502). The detailed shape of the spindle has been published elsewhere.33) The experimental conditions for this measurement are listed in Table 2. The gas phase fraction of the foaming slag (40–60 vol%) and spindle rotational speed (20, 30, 40, and 50 rpm) were changed for each slag composition listed in Table 1 to measure the torque. A total of six measurements were performed for each condition, and the average value was used as the measured viscosity. The measured values were substituted into the equation to obtain a calibration line at each spindle rotational speed using silicone oil with a known viscosity at room temperature to determine the apparent viscosity. Figure 2 shows an example of a calibration line indicating the relationship between the torque obtained at a spindle rotational speed of 20 rpm and the viscosity of silicone oil. This line shows a very good linear relationship (R2 = 0.9998) between the torque and viscosity.
| Gas-phase fraction | vol% | 40–60 |
| Rotational speed | rpm | 20, 30, 40, 50 |
| Shear rate | s−1 | 4.50, 6.75, 9.00, 11.3 |
| Temperature | K | 1673 |
We investigated the foaming behavior of the foaming slag when Fe-1.443mass%C powder was added as a foaming agent. Adding Fe-1.443 mass%C to the slag resulted in a vigorous foaming reaction and the formation of foaming slag. However, the SUS310S crucible melted during the experiment, which caused the slag sample to flow out, interrupting the experiment. The melting of the SUS310S crucible can be attributed to Fe-1.443mass%C, which exists as a liquid phase at the experimental temperature because of its low liquidus temperature of 1290°C. Thus, we investigated the foaming behavior of foaming slag after adding foaming agents such as Fe-0.185mass%C (liquidus temperature: 1490°C), Fe-0.53mass%C (liquidus temperature: 1450°C), ZrC (melting point: 3540°C), and NbC (melting point: 3900°C), which had liquidus temperatures or melting points higher than the experimental temperature (1400°C) in the present study. A gas was generated after adding any of these foaming agents because of the reaction, forming the foaming slag. Figure 3 shows the bubble lifetime when each foaming agent was added. These results indicated that the values were ~5, 20, 40, and 15 min for Fe-0.185mass%C, Fe-0.53mass%C, ZrC, and NbC, respectively. When considering the time required to perform viscosity measurements using the rotational method described in Section 2.3, a stable foaming height needs to be maintained when rotating the spindle and detecting the torque to ensure that a long foaming lifetime is desirable. We first considered using ZrC, which had the longest foaming lifetime; however, we found that it was lifted up by bubbles and floated to the top of the foaming slag. If a solid floats on the foaming slag in this manner, then the spindle and solid will interact during viscosity measurement, affecting the measured value. Thus, it is desirable for the foaming agent to sink, and therefore, we adopted Fe-0.53mass%C, which has the next longest foaming lifetime, as the foaming agent.
Figure 4 shows the relationship between the additive content of Fe-0.53mass%C powder and the gas phase fraction in the foaming slag. This showed that the gas phase fraction in the foaming slag increased with an increasing additive content of Fe-0.53mass%C powder in all slag compositions. This can be attributed to CO and CO2 gas being generated in the molten slag via the reactions C+(O)=CO and C+2(O)=CO2 with the Fe-0.53mass%C powder and the reaction CO+(O)=CO2 on the surface of the generated CO bubbles. In addition, a comparison with the same additive content of Fe-0.53mass%C powder showed that the gas phase fraction in the foaming slag was larger in slags with a lower Fe2O3 content. The gas phase fraction is expected to decrease because the CO and CO2 gas generation reaction becomes milder when considering that the activity of (O) decreases with a decrease in the FexO content in CaO–SiO2–FexO slag;34,35) however, in reality, the opposite result is obtained.
We discuss the results shown in Fig. 4 using the foaming index, which is an equation that expresses the bubble lifetime in the foaming slag using the liquid properties of the slag, such as viscosity, density, and surface tension, as parameters.20,21,22) We use the foaming index Σ formulated by Ito et al. from an experiment using CaO–SiO2–FeO slag that is expressed as
| (2) |
where Σ, η, ρ, and σ represent the foaming lifetime (s), slag viscosity (Pa∙s), slag density (kg/m3), and surface tension (N/m), respectively.
Table 3 lists the molten physical properties of the slag for each composition.36,37) Figure 5 shows the calculated foaming index for each slag, which indicates that the foaming index increased with decreasing Fe2O3 content in the slag (i.e., bubble lifetime in the foaming slag is longer). As shown in Table 3, the surface tension and density of the CaO–SiO2–Fe2O3 slag decreased with decreasing Fe2O3 content, and the foaming index increased according to Eq. (2); however, the change in surface tension and density with respect to Fe2O3 content is relatively small. Meanwhile, the CaO–SiO2–Fe2O3 slag viscosity increased with decreasing Fe2O3 content, increasing the foaming index; however, the change was over an order of magnitude greater than those of the surface tension and density, and it is believed to have a large effect on the foaming index. Thus, the molten physical properties of the slag, especially the viscosity, is suggested to be the dominant influence over the increase in the generated CO and CO2 gas with the abovementioned increase in the Fe2O3 content. The slags with high Fe2O3 content and low viscosity struggle to maintain foaming and require large amounts of CO and CO2 gas to maintain the foaming height (gas phase fraction), and therefore, more Fe-0.53mass%C powder is required to achieve the same gas phase fraction compared to that for slags with low Fe2O3 content.
| 20mass%Fe2O3 | 40mass%Fe2O3 | 60mass%Fe2O3 | |
|---|---|---|---|
| Viscosity, η /Pa·s | 0.18 | 0.12 | 0.062 |
| Surface tension, σ /Nm−1 | 0.41 | 0.47 | 0.50 |
| Density, ρ /kgm−3 | 3400 | 3600 | 4000 |
Figure 6 shows an observational photograph of the top of the 40CaO-40SiO2-20Fe2O3 (mass%) foaming slag as an example. Figure 7 shows the relationship between the gas phase fraction in the foaming slag and the bubble diameter observed on the top surface of the foaming slag. These results indicate that the average bubble diameter ranged from 0.4 to 0.6 mm and increased with an increasing gas phase fraction in the foaming slag. Further, the average bubble diameter was found to increase with increasing Fe2O3 content in the slag because of an increase in the reaction volume caused by the increase in the activity of FexO, which increased the bubble coalescence frequency caused by the increase in the gas phase fraction of the foaming slag and the effect of the physical properties of the melt. For the latter, as shown in Table 3, a higher Fe2O3 content in the slag is known to result in a lower viscosity and higher surface tension.36,37) Therefore, the promotion of drainage caused by the increase in fluidity in the bubble liquid film38) and the aggregation and coalescence of fine bubbles attributed to the increase in surface tension can explain the increase in the measured bubble diameter. Figure 7 shows that the error bars for the obtained bubble diameter are very large, and therefore, we limit our discussion to qualitative aspects.
Figure 8 shows the effect of the gas phase fraction on the viscosity (top row) and relative viscosity (bottom row) of the CaO–SiO2–Fe2O3 foaming slags. Here, the relative viscosities shown in the bottom row of Fig. 8 are relative values obtained by dividing the viscosity of the foaming slag by the viscosity of the liquid phase. These values indicate how many times higher the viscosity of the foaming slag is than that of the homogeneous liquid phase. Figure 8 (top row) shows that the CaO–SiO2–Fe2O3 foaming slag viscosity increased with increasing gas phase fraction for all compositions. This trend was similar to that for the simulated foaming slags in which nitrogen gas was dispersed in silicone oil30) and for the simulated foaming slags in which carbon dioxide gas was dispersed by mixing in an aqueous sodium bicarbonate solution with an aqueous oxalic acid solution.31) This phenomenon is attributed to the dispersion of the second phase, such as the gas or solid phase, in the liquid phase generating a shear rate and stress at the interface with the second phase that is different from the main flow direction (i.e., rotational direction in the case of viscosity measurement performed by the rotational method).39) Further, a higher rotational speed (shear rate) during viscosity measurement resulted in a lower viscosity for all compositions. Similar measurement results were obtained for the above-mentioned room temperature simulated foaming slag30,31) and suspension systems with a dispersed solid phase;40,41,42) however, a higher shear rate resulted in a further rearrangement of the dispersed second phase, resulting in the so-called “shear thinning” phenomenon where the flow can be propagated with lower stress. Recently, Martinsson et al.32) reacted molten 43CaO-32SiO2-25FeO (mass%) slag with carbon-saturated liquid iron at 1873 K to prepare foaming slag and measured the apparent viscosity using the rotational method. This study confirmed a non-Newtonian tendency in which the apparent viscosity decreased with the increasing rotational speed of the viscosity meter; a similar tendency was obtained in the present study.
Figure 8 (top row) shows that a decrease in the Fe2O3 content significantly increased the viscosity of the CaO–SiO2–Fe2O3 foaming slag with an increasing gas phase fraction. This indicates that the increase in the viscosity caused by the dispersion of the second phase is more prominent because the viscosity of the liquid phase increases with decreasing Fe2O3 content (Table 3). Figure 8 (bottom row) shows that the relative viscosities of the CaO–SiO2–Fe2O3 foaming slags are relative values obtained by dividing the viscosities of the foaming slags shown in Fig. 8 (top row) by the liquid phase viscosity, and therefore, changes with respect to the gas phase fraction and rotational speed are similar. The results clarified the existence of conditions under which the relative viscosity value was several hundred times the liquid phase viscosity, making this a very large value when considering that the relative viscosities of the foaming slags simulated at room temperature30,31) and the suspension system with the dispersed solid phase40,41,42) were at most ~120 and 60, respectively.
Figure 9 shows the relationship between the shear rate obtained from viscosity measurement conditions such as rotational speed, and the shear stress obtained by multiplying the viscosity and shear rate of the CaO–SiO2–Fe2O3 foaming slag. The results confirmed that the shear thinning phenomenon, in which the shear stress decreased with increasing shear rate, was observed in all compositions. However, the shear stress showed a significantly different shear rate dependency when compared to room temperature simulated foaming slags.30,31) In other words, the relationship between the shear stress and shear rate of the foaming slag at room temperature was classified as the so-called general non-Newtonian fluid,43) such as a pseudoplastic fluid, represented by an upward convex curve that passes through the origin, or a Herschel–Bulkley fluid, which is represented by an upward convex curve that does not pass through the origin and has a positive intercept. As shown in Fig. 9, the results of the present study showed that the relationship between the shear stress and shear rate cannot be classified as any of the non-Newtonian fluids, which indicates that the rheological properties of CaO–SiO2–Fe2O3 foaming slags change significantly with changes in shear rate. However, it seems necessary to conduct viscoelastic measurements that can evaluate the rheological properties at various angular velocities and frequencies, rather than at a constant shear rate.
We propose an empirical equation that can express the relative viscosity of CaO–SiO2–Fe2O3 foaming slags based on the Einstein–Roscoe equation44,45,46)
| (3) |
where η/ηL represents the relative viscosity, ϕ represents the gas phase fraction, and a and n are coefficients. In the present study, a = 1.0 when there is a liquid phase around the bubbles, and they can move freely. Some of the authors previously used the Einstein–Roscoe equation to propose another equation for predicting the apparent viscosity of the gas–liquid coexistent fluid. To this end, they considered the coefficient n in Eq. (3) as a function of the capillary number Ca, which is a dimensionless number that represents the ratio of the viscous force to the surface tension acting on the boundary between different fluids. The capillary number is given by
| (4) |
where ηL, U, σL,
| (5) |
We attempted to obtain the coefficient n in the Einstein–Roscoe equation by substituting the Ca number and the gas phase fraction calculated from the experimental conditions of the present study (i.e., liquid phase viscosity, shear rate, bubble diameter, and surface tension) into Eq. (5) and reproducing the relative viscosity of the CaO–SiO2–Fe2O3 foaming slag. Figure 10 shows an example of the results, where the relative viscosity was calculated for a rotational speed of 20 rpm. The results showed that the calculated values shown by the solid line, dotted line, and dash-dot line were significantly lower than the measured values shown in the plot; this tendency was similar for the other rotational speeds.
Coefficient n in the Einstein–Roscoe equation was calculated from the relative viscosity and gas phase fraction of the CaO–SiO2–Fe2O3 foaming slag measured in the present study, and we investigated its relationship with the Ca number calculated from the other experimental conditions. The results are presented Fig. 11, which indicate that there was a large amount of variation and coefficient n in the Einstein–Roscoe equation tended to decrease with a decreasing Ca number. In other words, this implies that the relative viscosity increased. The relationship between the coefficient n shown by the dotted line in Fig. 11 and the Ca number is given as
| (6) |
This regression equation was used to reproduce the relative viscosity of the CaO–SiO2–Fe2O3 foaming slag measured in the present study. The results of calculating the relative viscosity of the 30CaO-30SiO2-40Fe2O3 (mass%) slag are shown in Fig. 12. The results show that the calculated values indicated by the solid, dotted, dashed, and double-dot-dash lines reproduce the general tendency of the relative viscosity to increase with increasing gas phase fraction and decreasing rotational speed; however, the measured values shown in the plot cannot be reproduced. Although this is also the case for other slag compositions, it is attributed to the low coefficient of determination (R2 = 0.237) between coefficient n and the Ca number in Eq. (6). We presumed that, among the factors necessary for calculating the Ca number from the experimental conditions (liquid phase viscosity, surface tension, shear rate, and bubble diameter), the liquid phase viscosity and shear rate remain constant in the experimental system of the present study, whereas the surface tension and bubble diameter can vary. Thus, in the present study, the interfacial reaction between the foaming agent Fe-0.53mass%C powder and molten CaO–SiO2–Fe2O3 slag generated either CO or CO2 gas, and the foaming slag suitable for the viscosity measurement was prepared so that the surface tension of the generated CO bubble surface could decrease because of the reaction. Tanaka et al. investigated the change in interfacial tension over time for various combinations of molten slag and liquid Fe alloy at 1823 K to clarify the mechanism of the decrease in interfacial tension between the liquid Fe alloy and molten slag caused by the reduction and oxidation reactions. They concluded that the behavior of the change in interfacial tension over time could be explained by the adsorption of oxygen at the interface and the diffusion of oxygen from the interface to the liquid Fe alloy and molten slag.47) In this study, if the reaction CO+(O)=CO2 occurs between the slag containing FexO and the surface of the generated CO bubbles, the surface tension of the CO bubbles dispersed in the molten slag phase are assumed to have decreased by a similar mechanism. Meanwhile, the bubble diameter of the foaming slag shown in Fig. 7 is measured using the bubbles that appeared on the slag surface; when considering that fine bubbles that were generated at the bottom of the crucible coalesced as they rose,48) the possibility that the viscosity of foaming slag in which smaller bubbles were dispersed was measured also needs to be considered.
Thus, the surface tension and bubble diameter that reproduce the relative viscosity of CaO–SiO2–Fe2O3 foaming slag were estimated using the surface tension and bubble diameter in Eq. (4) as fitting parameters. Figures 13 and 14 show the estimated surface tension and bubble diameter, respectively. Figure 13 shows that the surface tension reported in the literature increased with increasing iron oxide content because of a decrease in the polymerization of silicate, which increased the unsaturated bonds on the surface while decreasing the surface tension obtained by the fitting. This was attributed to the reaction on the CO bubble surface becoming more active after considering that the activity of (O) increases with increasing FexO content in the slag.34,35) In addition, as shown in Fig. 14, the bubble diameter obtained by fitting ranged from 0.01 to 0.3 mm, which was smaller than the average bubble diameter of 0.4–0.6 mm obtained by the surface observation shown in Fig. 7. Here, Tanaka et al. reported that the contact angle greatly decreased with decreasing interfacial tension because of the interfacial reaction.47) Based on a comparison of the results of physical models as well as cold and hot experiments, Ogawa et al. reported that a smaller contact angle resulted in a finer bubble size of the bubbles generated at the interface when they detached.49) Even in this study, if a reaction occurs at the interface between the slag containing FexO and the added Fe–C alloy powder, finer bubbles could be dispersed in the molten slag compared to the bubble diameter observed because of a similar mechanism. Further, the results indicate that the estimated bubble diameter increases with an increasing gas phase fraction and rotational speed. This can be attributed to the increased frequency of bubble collisions and coalescence in the slag. The abovementioned estimation results are based on the assumption that the viscosity of the foaming slag estimated using the Einstein–Roscoe equation and Ca number is correct, and therefore, future research is required to develop technology that enables visualization and sensing of such reactions at high-temperature interfaces with higher accuracy.
In this study, we expanded the basic knowledge for understanding the flow behavior of multiphase melts in various steel refining processes involving foaming slags. We prepared CaO–SiO2–FexO foaming slags at 1673 K and systematically measured their viscosities by the rotational method. The findings of the study are listed below:
(1) Adding Fe–C alloy powder to CaO–SiO2–FexO slags continuously generated bubbles in the molten slag, thereby enabling the preparation of foaming slags suitable for viscosity measurements.
(2) A higher additive content of Fe–C alloy powder generated a greater number of generated bubbles, resulting in a larger gas phase fraction of the foaming slag. Further, we found that a lower Fe2O3 content resulted in a higher gas phase fraction because of the higher viscosity of the slag liquid phase, as predicted by the foaming index.
(3) A greater gas phase fraction of the foaming slag generated shear rate and stress in the direction opposite to the main flow direction near the dispersed bubbles, resulting in a greater relative viscosity.
(4) Shear-thinning properties were observed wherein a greater rotational speed resulted in a smaller relative viscosity of the foaming slag; i.e., we observed the promotion of the rearrangement of dispersed bubbles and propagation of flow with lower stress.
(5) We did not observe general non-Newtonian fluid behavior, e.g., that of Bingham fluids, in the relationship between the shear rate and shear stress calculated from the viscosity of the foaming slag and experimental conditions.
(6) We could not reproduce the viscosity of the foaming slag at 1673 K using either the Einstein–Roscoe equation or the empirical equation determined through the regression of the viscosity values of a simulated foaming slag at room temperature.
