Effect of Slag Compositions and Additive on Heat Transfer and Crystallization of Mold Fluxes for High-Al Non-magnetic Steel

Transcription

1 ISIJ International, Vol. 55 (2015), ISIJ International, No. 5 Vol. 55 (2015), No. 5, pp Effect of Slag Compositions and Additive on Heat Transfer and Crystallization of Mold Fluxes for High-Al Non-magnetic Steel Wei YAN, 1,2) * Weiqing CHEN, 1) Yindong YANG, 2) Carsten LIPPOLD 3) and Alex MCLEAN 2) 1) State Key Laboratory of Advanced Metallurgy, University of Science and Technology Beijing, 30 Xueyuan Road, Haidian District, Beijing, China. 2) Department of Materials Science and Engineering, University of Toronto,184 College Street, Toronto, Ontario M5S 3E4 Canada. 3) Qingdao Stollberg & Samil Co. Ltd., Qingdao, China. (Received on December 2, 2014; accepted on January 19, 2015) Intense reaction between silica in mold fluxes and aluminium in liquid steel during casting of high-al non-magnetic steel 20Mn23AlV ( Al in mass percent) would significantly alter both chemical compositions and properties of mold fluxes. This would subsequently lead to severe casting problems such as lots of slag rims, breakout and poor surface quality. Investigation carried out in this paper started with plant sampling, followed by a look at how the variation of Al 2 O 3 /SiO 2 ratio with reaction time can affect the casting process and product quality. Thus, this work focuses on the study of increasing Al 2 O 3 /SiO 2 and partial substitution of CaO with BaO in CaO SiO 2 system mold fluxes in terms of heat transfer and crystallization behavior. The techniques implemented are heat flux simulator and single hot thermocouple technique (SHTT). The results showed that an increase in Al 2 O 3 /SiO 2 inhibits heat transfer, increases crystallization temperature and critical cooling rate while shortens incubation time, additionally, accelerates precipitation of phase with high melting temperature. However, greater substitution of CaO with BaO accelerates heat transfer, reduces crystallization temperature and critical cooling rate at the cost of longer incubation time even at elevated Al 2 O 3 /SiO 2. Eventually, partial substitution of CaO with BaO, to some extent, counteracts the effect of increasing Al 2 O 3 /SiO 2 on heat transfer and crystallization properties of mold fluxes for casting of high-al steels. KEY WORDS: high-al steel; heat transfer; crystallization; SHTT; CCT; TTT. 1. Introduction Development of continuous casting technology had transformed the production model of high-al alloy steels such as non-magnetic steel 20Mn23AlV ( Al in mass percent) used for structural parts of transformer and transformation induced plasticity (TRIP) steel (0.5 3 Al in mass percent) for production of lighter and stiffer car bodies from ingot casting to continuous casting. Mold fluxes play key role during continuous casting by providing lubrication and controlling heat transfer which are mainly affected by varying chemical compositions. Undesirably, dissolved Al in molten steel are prone to react with SiO 2 in mold fluxes seriously, 1,2) which leads to considerable changes in chemical compositions SiO 2 and Al 2 O 3 and thus causes change in thermo-physical properties of mold fluxes. Furthermore, a series of problems such as formation of a large number of slag rims, poor lubrication, uneven heat transfer, and surface defects in the form of longitudinal and transverse depressions and cracks, even breakouts are found. 3,4) These are indeed the issues that need to be overcome by clear and essential understanding the effect mechanism on slag properties of slag compositions changes due to slag/metal * Corresponding author: yanweimetal@gmail.com DOI: reaction. Considerable efforts and optimization 5 10) have been carried out to study and suppress these changes in compositions and properties of mold fluxes for high-al steel based on conventional CaO SiO 2 system mold fluxes. However, no effective solutions to these problems are proposed based on the conventional CaO SiO 2 system. Ba and Ca belong to the same group elements whose oxides have similar properties, but BaO has lower melting temperature than CaO. Substituting CaO with some BaO in the CaO SiO 2 system mold fluxes may suppress the formation of high-melting temperature phase gehlenite caused by strong reaction between dissolved Al and SiO 2. Meanwhile, appropriate BaO in mold fluxes is beneficial for strengthening the capacity of absorbing inclusions like Al 2 O 3 and improving lubrication between strand shell and mould. 11) Some investigations 12 16) have been carried out to study the effects of BaO on melting behavior, viscosity, and crystallization of fluoride-free mold fluxes. However, these studies are very limited in clear presenting the effects of BaO additions on heat transfer and crystallization properties of BaO-containing mold fluxes, especially for high-al steels. Some studies on heat transfer of mold fluxes or steelmaking slags have been conducted using infrared emitter technique, 17,18) hot wire method 19) or heat transfer simulator employed in our previous, Yamauchi s, Hao s and Wen s studies, 2,20 23) and crystallization by 2015 ISIJ 1000

2 employing single hot thermocouple technique (SHTT) or double hot thermocouple technique (DHTT) ) As a part of comparative study on CaO SiO 2 and CaO Al 2 O 3 based mold fluxes for high-al steels, in this paper, plant sampling and sample analysis, heat transfer simulator and SHTT were employed to carry out comparative study on heat transfer and crystallization characteristics with varying Al 2 O 3 /SiO 2 of conventional and industrial CaO SiO 2 based mold fluxes, as well as experimental CaO BaO SiO 2 based mold fluxes prepared by partial substitution of CaO with BaO. 2. Problem Found during the Casting of High-Al Steel 2.1. Plant Sampling High-Al non-magnetic steel 20Mn23AlV studied in the current work was cast by a vertical continuous caster. A large number of slag rims along the meniscus were formed during its casting. Molten mold fluxes and slag rims were taken from the mould during continuous casting of 20Mn23AlV to examine their changes in compositions and properties. The molten mold flux samples were collected using a small wooden scoop after pushing away the powder layer from the same position, which is about midway between the submerged entry nozzle and the narrow side of the mold. In addition, the slag rims were taken using a wooden clamp. Sampling was carried out every ca. 10 min since the beginning of casting and all samples were air cooled to room temperature. The original mold fluxes and the spent mold flux samples were subjected to chemical analysis Molten Fluxes and Slag Rims Analysis Chemistry changes of mold fluxes for high Al non-magnetic steel 20Mn23AlV taken during continuous casting are presented in Fig. 1(a). It can be seen that Al 2 O 3 /SiO 2 mass ratio of molten fluxes has demonstrated a great increase from 0.2 to ca. 1.3 at approximately 15 min into the casting of 20Mn23AlV due to violent reaction between Al and SiO 2. Greater increase of Al 2 O 3 /SiO 2 mass ratio from 0.2 to ca. 1.7 was found in slag rims. Therefore, chemistry change of conventional CaO SiO 2 based mold fluxes can be represented by the Al 2 O 3 /SiO 2 mass ratio. Significant changes in chemical compositions of mold fluxes during continuous casting of high-al steels would introduce a series of severe consequences such as rapid increase in melting temperature, as shown in Fig. 1(b), and dramatic change in heat transfer and crystallization properties, which would in turn deteriorate the casting process and surface quality of cast slabs. Hot-stage-microscope method obtained elsewhere 27,28) was used to measure melting temperature. Mold flux columns were prepared with size Φ 3 mm 3 mm and then placed onto the alumina sheet to heat up. Change in dimensions of mold fluxes were recorded. The hemispherical temperature of the slag column represents the melting temperature of mold fluxes. Previous study of this author also revealed the effect of the slag/metal reaction on crystalline phases. 29) These effect and mechanism would be studied and discussed in detail in the following sections Fig. 1. (a) Changes of Al 2O 3/SiO 2 and (b) its effect on melting temperature of mold fluxes. 3. Experiments 3.1. Experimental Mold Fluxes Preparation Experimental mold fluxes were designed according to the analysis results of molten mold fluxes and slag rim samples taken from the caster. On this basis, CaO was substituted with BaO partly to evaluate the effect of BaO/CaO on heat transfer and crystallization behavior of spent CaO SiO 2 based mold fluxes with high Al 2 O 3 contents. The chemical compositions of the designed mold fluxes are listed in Table 1. All experimental mold fluxes were synthesized using reagent grade chemicals CaO, SiO 2, Al 2 O 3, CaF 2, where Na 2 O and BaO were added in the form of carbonates: Na 2 CO 3 and BaCO 3. The synthetic mold fluxes were pre-melted in graphite crucible at C for 1 h with protecting gas Ar and then quenched into water to form bulk glass fluxes. The water-quenched mold fluxes were dried in drying oven at 120 C for 4 h and then crushed into powder with size less than 75 μm for following experiments. The chemical compositions of selected pre-melted mold fluxes determined by XRF are listed in Table 2. It can be seen that the chemical changes of each composition are within 2 wt-%. The changes are relatively small compared with the designed compositions. Thus, it is reasonable to assume that the chemical changes will introduce negligible effect on experimental results ISIJ

3 Table 1. Chemical compositions of the designed mold fluxes (in mass percent). No. Al 2O 3/ SiO 2 BaO/CaO CaO SiO 2 Al 2O 3 Na 2O CaF 2 BaO R R R R R R R Table 2. Chemical compositions of the selected pre-melted mold fluxes (in mass percent). No. Al 2O 3/ SiO 2 BaO/CaO CaO SiO 2 Al 2O 3 Na 2O CaF 2 BaO R R R R Fig. 2. Schematic diagram of experimental apparatus Heat Transfer Experiments Vertical tube furnace was employed to carry out heat transfer experiments. The schematic diagram of apparatus was shown in Fig. 2. The apparatus was mainly made up of heating unit, pure iron crucible, water-cooling copper column, thermocouples and gasket. Casting slab and watercooling copper wall found in the mold were replicated using pure iron crucible and water-cooling copper column respectively. Gasket was employed to adjust distance between pure iron crucible and water-cooling copper column. The outer diameter of pure iron crucible was 70 mm with inner diameter of 60 mm, overall height of 25 mm and bottom thickness of 5 mm. The thickness of flux film was kept at a constant of 2 mm. Ni Cr/Ni Si thermocouples (type K) A, B and WRe 5/26 thermocouple (type C) C were employed to measure the temperature of different positions within the experimental apparatus during experiments. Among them, the distance from thermocouple A and B to the base of copper column were 10 mm and 2 mm respectively, and thermocouple C was in physical contact with the base of pure iron crucible. The vertical tube furnace adopts carbon tube and argon gas (99.99% pure) as heating unit and protecting gas respectively. To eliminate the effect of radiation heat transfer as much as possible, pure iron crucible was placed in the upper region of constant temperature zone, and cooling water tube for the copper column were wrapped by heat insulating material to achieve one-dimensional steady-state heat transfer across flux film to the base of water-cooling copper column ISIJ 1002

4 At the beginning of this experiment, 30 g pre-melted mold fluxes were fed into pure iron crucible when the temperature of thermocouple C was heated up and kept at K (1 350 C). After about 30 min, once the mold fluxes have become molten and the temperature of position C has stabilized at K (1 350 C) again, water-cooling copper column with flow rate 120 L/h was inserted into the crucible. Meanwhile, the temperature of A, B and C were recorded. In the case of the aforementioned, heat transfer across flux film could be viewed as one-dimensional steadystate heat transfer, the temperature distributions of points A, B and C were shown in Fig. 3. Heat transfer parameters can be calculated by the following equations according to temperature of A, B, C and the given parameters. Total heat flux q T (Wm 2 ) across flux film can be calculated by Eq. (1). qt = kcu ( TB TA) / dab... (1) The base temperature T Cu-base (K) of water-cooling copper column can be calculated by Eq. (2). TCu base = TA + qda / kcu... (2) Interfacial temperature T Fe-S (K) between flux film and pure iron crucible can be calculated by Eq. (3). TFe S = TC qdc / kfe... (3) The overall thermal resistance R T (m 2 KW 1 ) can be calculated by Eq. (4). RT = ( TFe S TCu base )/ q = ( TC TA) dab / ( TB TA) kcu... (4) d / k + d / k ( C Fe A Cu ) Where k Cu, k Fe are thermal conductivity of copper column and pure iron crucible respectively. According to the measured results, temperature of water-cooling copper column was lower than 373 K (100 C) and temperature of pure iron crucible was about K (1 200 C). Values of k Cu at 373 K (100 C) and k Fe at K (1 200 C) were found to be 393 Wm 1 K 1 and 31.6 Wm 1 K 1, 30) respectively. It is therefore assumed k Cu and k Fe are constants in the range of experiment temperatures CCT and TTT Tests SHTT was employed to construct the CCT and TTT diagrams to investigate the crystallization behavior of mold fluxes. Minor slag samples about 10 mg was mixed with alcohol, and was subsequently mounted on the tip of a B-type thermocouple and was heated directly. Simultaneously, the temperature was recorded according to the thermal cycle, and the crystallization process was recorded by a CCD camera to collect images every one second. Figure 4 shows the thermal cycles for the construction of continuous cooling transformation (CCT) curves and time temperature transformation (TTT) curves. First, the thermocouple was heated at a rate of 30 C/s to C, and then the mold fluxes sample were held at C for 1 min to eliminate bubbles and to allow homogenization of composition and temperature. Finally, the sample was cooled continuously to 900 C at different cooling rates to construct CCT diagram by recording the start time and temperature of crystallization at different cooling rates. To construct the TTT diagram, the sample was rapidly cooled at a cooling rate about 80 C/s to various isothermal temperatures while recording the start time and the end time of crystallization. During CCT and TTT tests, 5% and 95% volume of crystallization were defined as start and end of crystallization, respectively. The volume ratio of crystallization were analyzed by an image analysis software XRD Analysis The water-quenched mold fluxes were treated in a muffle Fig. 3. Distributions of temperature during heat transfer. Fig. 4. Thermal cycles of (a) CCT and (b) TTT diagrams ISIJ

5 furnace for 2 h to achieve crystallization at their nose temperature determined by SHTT to investigate the crystal phases. In order to obtain the crystallization products as far as possible at the nose temperature, the mold fluxes were quenched into water immediately when they were taken out from the muffle furnace. Then, the mold fluxes were dried, crushed and ground to be subjected to XRD (Ultima IV, Rigaku Corporation, Japan) analysis. The XRD patterns were collected by Cu Kα radiation at a rate of 20 degree/min. 4. Results and Discussions 4.1. Heat Transfer Heat Flux and Heat Resistance Change Curve The measured heat flux density and heat resistance as a function of experiment time are shown in Fig. 5. It could be seen that from Fig. 5 the change of heat flux density with time could be divided into three stages which are the ascent stage (I) of heat flux density from 0 to ca. 50 s, the descent stage (II) from 50 to ca. 300 s, and the stable stage (III) are achieved after ca. 300 s. The ascent stage reflects the heat flux density change between copper column and molten fluxes, therefore the heat flux density reaches the maximum. The descent stage reflects the formation process of solid slag film therefore the heat flux density decreases with the formation and crystallization of solid slag film. Eventually, the solid slag film has formed completely and the heat flux density tends to be stable after ca. 300 s. The heat resistance change, however, presents an opposite tendency. Figure 6 gives the fitting curves of heat flux density and heat resistance using the following fitting Eq.(5) derived from a graphing and data analysis software. The high adj. R-square values are indicative of the excellent fit. p f () t = A2 + ( A1 A2) /( 1+ ( t / t0) )... (5) Where t is the immersion time of copper column, parameters A 1, A 2, t 0 and p can be determined by fitting the data. According to the experiment practice, the heat flux/thermal resistance tends to be in stable state after 300 s, so the fitting heat flux/thermal resistance f(t) in stable state can be derived as A 2 when the time t tends to be infinite. The mean deviation of mold fluxes R1 R4 from average heat flux in stable state (after 300 s) is 2.8%, which suggests the fitting result is fairly reliable Effect of Al 2 O 3 /SiO 2 on Heat Transfer of Mold Fluxes Figure 7 presents the fitting results and R-Square of heat flux density and thermal resistance from experimental values. The R-Square values indicate the reliability of fitting results. It could be seen that heat flux decreases and thermal resistance increases remarkably with increasing Al 2 O 3 /SiO 2 of mold fluxes for high-al steels. In other words, reaction between SiO 2 and Al inhibits heat transfer from the strand to Fig. 5. (a) Heat flux and (b) heat resistance change of sample R1 with time. Fig. 6. Fitting curve of (a) heat flux and (b) thermal resistance of sample R ISIJ 1004

6 Fig. 7. Effect of Al 2O 3/SiO 2 on heat flux and thermal resistance of mold fluxes. flux density or decrease of total thermal resistance with increased substitution of CaO with BaO could help compensate for the decrease in heat flux density or increase in total thermal resistance due to the increase of Al 2 O 3 /SiO 2 during casting of high Al steels. Accordingly, to some extent, CaO BaO SiO 2 based mold fluxes have the potential to replace CaO SiO 2 based mold fluxes for casting of high-al steels from the perspective of balancing heat transfer. Contribution of BaO may be related to its suppression on precipitation of high-melting temperature phase such as gehlenite, which will be discussed below. It is well known, according to previous studies, 31 33) that crystallization of mold fluxes is the principal factor in controlling heat transfer during continuous casting. It could be concluded that heat transfer inhibition from increasing Al 2 O 3 /SiO 2, and heat transfer promotion from partial substituting CaO with BaO would suggest increasing Al 2 O 3 /SiO 2 and BaO/CaO may in fact enhance and suppress crystallization of mold fluxes for high-al steels, respectively. During casting of non-magnetic 20Mn23AlV, mold fluxes with a weak crystallization ability would promote heat transfer to achieve thick shell and high strength to avoid breakout. Therefore, substituting CaO with BaO in mold fluxes is expected to improve heat transfer by changing crystallization behavior, which would be discussed further in the following section. Fig. 8. Effect of BaO/CaO on heat flux and thermal resistance of mold fluxes. the mold. The heat transfer performance of mold fluxes must be coherent with the solidification characteristic of casting slab. To take high-al non-magnetic steel 20Mn23AlV for example, its solidification requires fast cooling to form thick initial shell and fine grain structure to increase strength of the shell to avoid breakout. Therefore, decrease in heat transfer will introduce unexpected effect on slab quality and casting stability during continuous casting of high-al nonmagnetic steel 20Mn23AlV. In silicate melts with less Al 2 O 3 content, SiO 2 plays role as networker former by forming [SiO 4 ], increase of Al 2 O 3 and decrease of SiO 2 due to reaction between Al and SiO 2 not only increases basicity (CaO/ SiO 2 ) but also forms [AlO 6 ] as network modifier and phase with high melting temperature, which all strengthen crystallization ability of mold fluxes to suppress heat transfer Effect of Partial Substitution of CaO with BaO on Heat Transfer of Mold Fluxes The heat flux and thermal resistance were measured after substituting CaO with some BaO in mold fluxes. Fitting results from heat flux density and thermal resistance give the more visual change as a function of BaO/CaO and are shown in Fig. 8. It could be seen that heat flux density increases with the increasing BaO/CaO mass ratio. On the contrary, the total thermal resistance shows a decreasing trend with increasing BaO/CaO mass ratio. Increase of heat 4.2. Crystallization Behavior CCT Diagrams SHTT was employed to study and compare crystallization characteristics of CaO SiO 2 based and CaO BaO SiO 2 based mold fluxes for high-al steels at various cooling rate by constructing CCT diagrams. Figures 9(a) through 9(d) show the CCT diagrams of CaO SiO 2 based mold fluxes R2 and R4, and CaO BaO SiO 2 based mold fluxes R41 and R43. It could be seen that mold fluxes R2 exhibits no crystallization even though at low cooling rate from C/s. Mold fluxes R4, R41 and R43 show various crystallization at different cooling rate. The crystallization temperature decreases with the increasing cooling rate which may be due to the fact the crystallization is dependent on the crystals nucleation and growth rate which is the function of thermodynamics and kinetics. Some researchers 17,26,34) considered that increase in viscosity increases diffusion resistance of ions, and hence restrains nucleation and crystal growth. But in the current study, although increasing basicity strengthens crystallization tendency of mold fluxes by depolymerizing the network structure, increase in Al 2 O 3 /SiO 2 increases viscosity of mold fluxes according to authors measurements, that is to say viscosity is not the dominated factor in controlling crystallization in this CaO SiO 2 system with high Al 2 O 3 /SiO 2. Increasing Al 2 O 3 /SiO 2 causes increase in liquidus temperature and correspondingly undercooling. It can be concluded high undercooling may promote crystallization. But for mold fluxes with fixed and high Al 2 O 3 / SiO 2, increasing BaO/CaO causes increase in viscosity and decrease in liquidus temperature, thus viscosity may play dominated role in controlling crystallization of flux system with increasing substitution of CaO with BaO. More time are also required for nucleation and crystal growth at increasing cooling rate ISIJ

7 Mold flux R2 does not express crystallization behavior at the cooling rate range C/s, while R4 with high Al 2 O 3 /SiO 2 shows the highest crystallization temperature C at the cooling rate 0.5 C/s. Mold flux R41 with BaO/CaO=0.33 shows a slightly lower crystallization temperature at C, obtained using a cooling rate of 0.5 C/s. With the increase of BaO/CaO, mold flux R43 with BaO/CaO=3 shows a larger decrease of crystallization temperature to C, at the same cooling rate 0.5 C/s. Critical cooling rate is a crucial parameter which reflects crystallization ability of mold fluxes. Larger critical cooling rate indicates strong crystallization ability. Mold fluxes could not crystallize when the cooling rate is larger than the critical cooling rate, which is 10 C/s for mold fluxes R4. The critical cooling rate of R41 and R43 are 4 C/s and 0.5 C/s respectively, which are far less than that of R4. Figure 10 gives the comparison of critical cooling rate of CaO SiO 2 based mold fluxes and CaO BaO SiO 2 based mold fluxes. Obviously, the critical cooling rate increases with increased Al 2 O 3 /SiO 2, but the critical cooling rate decreases with the increased BaO/CaO. It could be concluded that reaction between Al and SiO 2 during casting of high-al steel increases the critical cooling rate of mold fluxes, that is to say, the reaction strengthens the crystallization tendency of traditional CaO SiO 2 mold fluxes. This is consistent with the results of Zhang 6) who examined the precipitation of post-experimental slag samples with varying Al 2 O 3 /SiO 2 from viscosity measurement after furnace cooling. However, increase in BaO/CaO would decrease the critical cooling rate of CaO SiO 2 mold fluxes at high Al 2 O 3 / SiO 2. In other words, BaO/CaO weakens the crystallization ability of CaO SiO 2 mold fluxes with strong crystallization property at high Al 2 O 3 /SiO 2. Both high critical cooling rate and crystallization temperature decreases the tendency to form glassy slag layer that help promote heat transfer. Therefore, mold fluxes R4 with high Al 2 O 3 /SiO 2 tend to form thick crystallization layer to inhibit heat transfer during continuous casting, while mold fluxes with high BaO/CaO tend to form thick glassy layer to accelerate heat transfer, which are in good agreement with the aforementioned heat transfer results found in substituting CaO with BaO in CaO SiO 2 based mold fluxes with high Al 2 O 3 /SiO 2. Fig. 9. CCT diagrams of mold fluxes (a) R2, (b) R4, (c) R41, (d) R43. Fig. 10. Critical cooling rate and crystallization temperature (Cry. T) at cooling rate 0.5 C/s ISIJ 1006

8 TTT Diagrams TTT diagrams shown in Fig. 11 summarize the crystallization time during isothermal tests. The label 5% represents the 5% volume fraction of crystallization and is viewed as the start of crystallization. Similarly, 95% represents the end of crystallization. It could be seen that all TTT diagrams other than R43 appear to be typical C-shaped curve. Two noses shown in TTT diagram of R43 indicates it may have two crystallization processes. The TTT diagrams also suggest the following:1) mold flux R4 crystallizes at the highest start crystallization temperature C, 2) mold flux R41 and R43 have lower start crystallization temperatures at C and C, respectively, 3) mold flux R2 has the lowest start crystallization temperature at C. Similarly, mold fluxes R4, R41, R43 and R2 express gradually decreasing nose temperature, which are about C, C, C, and C, respectively. In addition, Figs. 11(a) through 11(d) indicate mold fluxes with highest crystallization temperature or nose temperature also have the shortest full-crystallization time (from initial crystallization to completion) at nose temperature, which further indicates the strongest crystallization ability of mold flux R4. Figure 12 presents the comparison of incubation time (the necessary time to initiate crystallization). It could be seen that mold flux R4 crystallizes almost immediately (mostly, less than 10 s) in the isothermal test range C. Mold flux R4 also displays the shortest incubation time of almost 0 s at its nose temperature of C. Mold flux R41 shows a short incubation time of 17 s at its nose temperature of C, but it is still longer than that of R4. The incubation time was found with mold flux R43, in the range of s, and its incubation time at nose temperature C and C are 146 s and 102 s, respectively, far longer than the shortest incubation time of R4 and R41 at their nose temperatures. Mold flux R2 presents the longest incubation time of 353 s at its nose temperature of C. Incubation time is also an important parameter to evaluate the crystallization property of mold flux. A short incubation time indicates a strong crystallization tendency. Therefore, mold flux R4 has the strongest crystallization tendency while the mold flux R2 has the weakest crystallization tendency. In another word, increase in Al 2 O 3 /SiO 2 would result in improved crystallization tendency, while substitution of Fig. 11. TTT diagrams of mold fluxes (a) R2, (b) R4, (c) R41 and (d) R43. Fig. 12. The crystallization incubation time at various isothermal temperature ISIJ

9 CaO with BaO (BaO/CaO) decreases crystallization tendency, even though the mold fluxes may have high Al 2 O 3 / SiO 2. The TTT test results are in good agreement with that of CCT tests and heat transfer of mold fluxes XRD Analysis Figures 13(a) through 13(d) present the XRD patterns of mold fluxes R2, R4, R41 and R43 after heat treatment for 2 h at their nose temperature determined by the TTT tests. It could be seen clearly that CaF 2 phase exists in all mold flux samples. Additionally, for mold flux R2 with low Al 2 O 3 /SiO 2 of 0.43, cuspidine Ca 4 Si 2 O 7 F 2 and silicon-rich nepheline Na 6.8 Al 6.3 Si 9.7 O 32 were found as shown in Fig. 13(a). However, for mold flux sample R4, with Al 2 O 3 /SiO 2 increasing to 2.33, Ca 4 Si 2 O 7 F 2 and Na 6.8 Al 6.3 Si 9.7 O 32 disappeared but gehlenite Ca 2 Al 2 SiO 7 and wollastonite CaSiO 3 were found in Fig. 13(b). This suggests that increasing Al 2 O 3 /SiO 2 suppresses the precipitation of cuspidine Ca 4 Si 2 O 7 F 2, which is always present in common mold fluxes, 35) while accelerates the precipitation of gehlenite Ca 2 Al 2 SiO 7. Precipitation of gehlenite with high melting temperature of C 36) would deteriorate lubrication and form mass slag rims that block infiltration of liquid mold fluxes into the gap between mold and strand during casting of high-al steels according to our previous onsite study on industrial mold fluxes. 29) It could be seen from Fig. 13(c) that a few of substitution of CaO with BaO seems to cause no effect on crystallization phase species, but it decreases the peak intensity of mold flux R41. Further increase in substitution of CaO with BaO, R43 has a similar nose temperature to that of R41, but the peak intensity is lowered further, as shown in Fig. 13(d). The peak intensity is related to crystallinity to some extent. While at low nose temperature C, it is noticeable that some new crystallization phases Na 2 Ca 3 Si 6 O 16 have precipitated, whose melting temperatures are relatively lower than C according to phase diagrams. 36) Formation of crystalline phases with low-melting temperature is beneficial for lubrication function of mold fluxes. Thus, the following could be concluded: 1) increase in Al 2 O 3 /SiO 2 accelerates precipitation of gehlenite, which strengthens the crystallization ability, 2) increasing substitution of CaO with BaO suppresses precipitation of gehlenite, at the same time, accelerates precipitation of Na 2 Ca 3 Si 6 O 16 with low melting temperature, which consequently, weakens crystallization ability of mold fluxes for high-al steels. Additionally, it is interesting to note that no Ba-bearing crystal phase was observed. Similarly, Ba-bearing crystal phase was also not found in the study of Dong et al. 14) on BaO-bearing mold fluxes, the reason for this phenomenon needs to be further studied. 5. Conclusions Comparative study on effect of Al 2 O 3 /SiO 2 and substitution of CaO with BaO on heat transfer and crystallization of mold fluxes for high-al steels were conducted by plant sampling, heat flux simulator, SHTT and XRD techniques. Main conclusions could be drawn as follows: (1) Plant sampling shows reaction between mold fluxes and liquid steel during continuous casting of high-al steels have resulted in significant increases of Al 2 O 3 /SiO 2 ratios Fig. 13. XRD patterns of mold fluxes (a) R2, (b) R4, (c) R41 and (d) R43 at their nose temperature ISIJ 1008

10 in mold fluxes from 0.25 to ca. 1.5, which consequently increases the melting temperature of mold fluxes from C to more than C. (2) Heat transfer experiments indicated heat flux increases first and then decreases, before stabilizing after about 300 s, which signifies state change of mold flux film from liquid state to layered state. Increase in Al 2 O 3 /SiO 2 decreases the overall heat flux density but increases the overall thermal resistance. In contrast, increasing substitution of CaO with BaO increases the overall heat flux density. (3) CCT tests suggest increase in Al 2 O 3 /SiO 2 accelerates crystallization of mold fluxes by increasing both crystallization temperature and critical cooling rate. However, increased substitution of CaO with BaO at high Al 2 O 3 /SiO 2 can suppress crystallization of mold fluxes by decreasing the crystallization temperature and critical cooling rate. (4) TTT tests suggest increase in Al 2 O 3 /SiO 2 also accelerates crystallization of mold fluxes by increasing the initial crystallization temperature and nose temperature, and shortening the incubation time. In contrast, increased substitution of CaO with BaO can suppress crystallization by decreasing the initial crystallization temperature and nose temperature, and prolonging the incubation time. (5) XRD analysis indicates increases in Al 2 O 3 /SiO 2 can transform main crystal phases from cuspidine Ca 4 Si 2 O 7 F 2 to gehlenite Ca 2 Al 2 SiO 7, characterized by melting temperature of up to C. However, increased substitution of CaO with BaO not only decreases crystallization peak intensity and suppresses precipitation of gehlenite Ca 2 Al 2 SiO 7 but also accelerates precipitation of low-melting temperature phases Na 2 Ca 3 Si 6 O 16, which could ultimately retard the crystallization of mold fluxes. REFERENCES 1) M. S. Kim, S. W. Lee and J. W. Cho: Metall. Mater. Trans. B, 44B (2013), ) H. Wang, P. Tang and G. H. Wen: Ironmaking Steelmaking, 38 (2011), ) J. J. Becker, M. A. Madden and T. T. Natarajan: Proc. Iron & Steel Technology Conf., AIST, Warrendale, PA, (2005), 99. 4) S. Street, K. James and N. Minor: Iron Steel Tech., 7 (2008), 38. 5) G. A. Bezuidenhout and P. C. Pistorius: Ironmaking Steelmaking, 27 (2000), ) Z. T. Zhang, G. H. Wen and P. Tang: ISIJ Int., 48 (2008), ) W. Wang, K. Blazek and A. Cramb: Metall. Mater. Trans. B, 39 (2008), 66. 8) H. S. Park, H. Kim and I. Sohn: Metall. Mater. Trans. B, 42 (2011), ) T. Omoto, T. Suzuki and H. Ogata: Shinagawa Tech. Rep., 50 (2007), ) H. Todoroki, T. Ishii and K. Mizuno: Mater. Sci. Eng. A, (2005), ) J. H. Chi, B. Xie and Y. N. Gan: J. Chongqing Univ., 13 (1990), ) E. Brandaleze, A. Martin and J. Madias: Proc. Iron & Steel Technology Conf., AIST, Warrendale, PA, (2007). 13) Q. Wei, Y. Q. Ding and L. Y. Dong: J. Chin. Rare Earth Soc., 18 (2000), ) F. Dong, Y. C. Wang and B. F. Wang: Spec. Steel, 27 (2006), 7. 15) Y. Q. Lu, X. Fang and G. D. Zhang: Adv. Mater. Res., (2011), ) Q. Y. Huang, L. F. Zhang and X. Lv: Proc. Materials Science and Technology (MS&T), Pittsburgh, PA, (2008), ) B. X. Lu, W. L. Wan and J. Li: Metall. Mater. Trans. B, 44B (2013), ) W. L. Wang, K. Z. Gu and L. J. Zhou: ISIJ Int., 51 (2011), ) K. Nagata, M. Susa and K. S. Goto: Tetsu-to-Hagané, 69 (1983), ) W. Yan, W. Q. Chen and C. Lippold: Metall. Mater. Trans. B, 44B (2013), ) A. Yamauchi, K. Sorimachi and T. Sakuraya: ISIJ Int., 33 (1993), ) Z. Q. Hao, W. Q. Chen and C. Lippold: Metall. Mater. Trans. B, 41B (2010), ) P. Tang, C. S. Xu and G. H. Wen: J Iron Steel Res Int.,15 (2008), 7. 24) Y. Kashiwaya, C. E. Cicutti and A. W. Cramb: ISIJ Int., 38 (1998), ) H. Liu, G. H. Wen and P. Tang: ISIJ Int., 49 (2009), ) J. Wei, W. L. Wang and L. J. Zhou: Metall. Mater. Trans. B, 45B (2014), ) D. Singh, P. Bhardwaj and Y. D. Yang: Steel Res. Int., 81 (2010), ) S. K. Sahoo, J. N. Tiwari and U. K. Mohanty: Metall. Mater. Trans. B, 44B (2013), No.6, ) W. Yan, W. Chen and C. Lippold: Specl Steel, 34 (2013), ) Y. S. Shen, B. W. Li and M. L. Wu: Basic Principle of Metallurgical Transfer, Metallurgical Industry Press, Beijing, (2006). 31) H. Nakada, M. Suza and Y. Seko: ISIJ Int., 48 (2008), ) M. Susa, R. Endo and Y. Kobayashi: ISIJ Int., 49 (2009), ) M. Kawamoto, T. Tsukaguchi and N. Nishida: ISIJ Int., 37 (1997), ) C. B. Shi, M. D. Seo and J. W. Cho: Metall. Mater. Trans. B, 45B (2014), ) P. Grieveson, S. Bagha and N. Machingawuta: Ironmaking Steelmaking, 15 (1988), ) Verlag Stahleisen GmbH: Slag Atlas, 2nd Ed., Verlag Stahleisen GmbH, Düsseldorf, (1995) ISIJ