Melting and Heat Transfer Behavior of Fluorine-Free Mold Fluxes for Casting Medium Carbon Steels

Transcription

1 , pp Melting and Heat Transfer Behavior of Fluorine-Free Mold Fluxes for Casting Medium Carbon Steels Lejun ZHOU, 1) Wanlin WANG, 1) * Juan WEI 3) and Kechao ZHOU 2) 1) School of Metallurgy and Environment, Central South University, No.932, South Lushan Road, Changsha , P. R. China. 2) State Key Laboratory of Power Metallurgy, Central South University, No.932, South Lushan Road, Changsha , P. R. China. 3) Department of Metallurgical and Materials Engineering, Colorado School of Mines, 1600 Illinois Street, Golden, USA. (Received on November 12, 2014; accepted on January 7, 2015) As fluxing agent, fluorine is important to the properties of mold flux; however, it tends to cause serious environment and health problems. In this paper, the melting and heat transfer behaviors have been studied by using Single Hot Thermocouple Technique (SHTT) and Infrared Emitter Technique (IET). The results show that the melting temperature range of F-free mold flux decreases with the addition of Na 2O/Li 2O and B 2O 3 contents; and the heat flux decreases with the increase of basicity and Na 2O/Li 2O, as well as the decrease of B 2O 3 contents. The analysis of EDS and XRD show that Ca 11Si 4B 2O 22 and Ca 14Mg 2(SiO 4) 8 are the two main precipitated crystalline phases in F-free mold fluxes, and the Ca 11Si 4B 2O 22 is common and stable crystalline phase in the designed F-free mold fluxes system that shows the potential to replace Ca 4Si 2O 7F 2 in the conventional F-containing mold fluxes. Besides, through the comprehensive comparison, Sample E (Basicity 1.15, Na 2O 7.92, Li 2O 1.97 and B 2O ) shows the closest performances with the benchmark conventional commercial mold flux and has the potential to be used for casting medium carbon steels. KEY WORDS: F-free mold fluxes; medium carbon steel; continuous casting; melting; heat transfer behavior. 1. Introduction Mold flux plays important roles in steel continuous casting process. The melting temperature range of mold flux directly determines the liquid thickness of mold flux on top surface of molten steel and the size/shape of rim in the vicinity of meniscus which will indirectly influence the infiltration and consumption of mold flux. 1,2) While, the heat transfer performance of mold flux will affect the horizontal heat transfer of mold greatly. Uneven heat transfer and rapid cooling of solidified shell can cause enormous thermal stress on the initial shell, especially for the casting of medium carbons steels, as they are crack-sensitive due to the δ to γ phase transformation that leads to a relatively larger volumetric shrinkage, which consequently causes cracks and other surface defects on the shell. 3,4) In conventional commercial mold flux, about 7 13 mass% fluorides are usually added to control the melting and heat transfer performance of mold flux. 5,6) Most of fluorides are acting as fluxing agent as they can provide the F to break down the polymerization of molten slag. 7) Meanwhile, fluorine has a big impact on heat transfer performance of mold flux due to the formation of crystalline phase of cuspidine (Ca 4Si 2O 7F 2). 8,9) Although fluorides are important to mold flux, it still causes some problems such as the volatilization of SiF 4, NaF, AlF 3 and HF etc. which will pollute the air and * Corresponding author: wanlin.wang@gmail.com DOI: cooling water, hurt health of operators, and also corrode equipment (waterborne fluoride HF) in plant. 10,11) So, it is urgent to look for substitutes to replace the fluorides in mold flux. Many studies regarding fluorine free mold flux have been conducted, which can be divided into two categories: (1) Utilizing TiO 2 to substitute the fluorides, since the precipitated crystalline phase CaO SiO 2 TiO 2 was considered to be proper substitutes for cuspidine in commercial mold fluxes ) However, recently published paper from Q. Wang and K. C. Mills 16) pointed out that those fluorine free mold fluxes containing TiO 2 have the risk to increase the rate of breakout due to the possible formation of high melting point Ti(C, N) and CaTiO 4 that tends to deteriorate the heat transfer behavior in the mold. (2) Using oxides like Na 2O, K 2O, Li 2O, MgO, MnO, B 2O 3, BaO and rare earth elements oxides to modify the negative effects of the mold flux properties caused by the absence of fluorides; 17 22) however, there is no specific crystalline phase, which shows similar properties as cuspidine, has been reported. Recently, a potential crystalline phase with stoichiometry of Ca 11Si 4B 2O 22 was found in our previous studies during the investigations of the low fluorine mold fluxes for casting medium carbon steels, which shows similar crystallization behavior as cuspidine ) This study is a follow-up work to further design and develop F-free mold fluxes for casting medium carbon steels, and it was divided into two parts; the first part is about the study of melting and heat transfer behaviors of F-free mold fluxes; and the second part is about ISIJ

2 the viscosity and crystallization behavior of F-free mold fluxes. As the first part, the melting and heat transfer behaviors will be investigated in this article by using Single Hot Thermocouple Technique (SHTT) and Infrared Emitter Technique (IET), respectively. 2. Experiment 2.1. Preparation of Experimental Materials The benchmark Sample A used in this study is a commercial mold fluxes for casting medium carbon steel, and Sample B, C, D and E are designed F-free mold fluxes with varied Na 2O, B 2O 3, Li 2O contents and basicity based on the Sample A to modify the negative effects of the mold flux properties caused by the absence of fluorides. All those samples were synthesized by mixing reagent grade chemicals of CaO, SiO 2, Al 2O 3, MgO, Na 2O, Li 2O, CaF 2 and B 2O 3 through melting in an induction furnace at K (1 500 C) for 5 minutes to homogenize their chemical compositions, and then pouring onto a cool steel plate to quench to achieve a fully glassy phase. Meanwhile, a new cylindrical tube-like copper mold with the same diameter as the copper substrate was used to cast the mold flux before it solidified on the steel. After that, the cooled mold flux disks were polished by using the SiC sand papers with the grit size down to to control their surface roughness and thickness. The polished glassy samples were then placed on the top of the copper mold individually for heat transfer experiments through IET. Otherwise, a small amount of solidified glassy mold flux was crushed and ground into powders samples for melting tests through SHTT. The pre-melted slags were also analyzed by X-ray fluoroscopy (XRF, S4 Pioneer; Bruker AXS; GmbH Karlsruhe, Germany), and the results are shown in Table 1. It could be found that the evaporative loss of mold flux components is relatively small, which is consistent with our previous study. 24) 2.2. Melting Test The melting behaviors of F-free fluorine mold fluxes were observed and tested through Single Hot Thermocouple Technique (SHTT), which is schematically shown in Fig. 1, and the details about SHTT have been described by Prof. Kashiwaya. 27) During the melting behavior test, the mold flux sample was first mounted on a B type thermocouple. Then, the thermocouple together with the sample was heated with a rate of 15 K ( C)/s. The changing of mold flux sample would be observed and recorded through a CCD microscope connected to DVD recorder. Meanwhile, the corresponding temperature history was obtained by the temperature acquisition Table 1. The chemical compositions of pre-melted mold fluxes (in mass pct). Sample R CaO SiO 2 Al 2O 3 Na 2O B 2O 3 Li 2O F A B C D E Fig. 1. The schematic of SHTT experimental apparatus. 27) Fig. 2. Schematic illustration of the infrared emitter ISIJ 822

3 system. Combining the in-situ video and variation of temperature history, the melting behavior of mold fluxes could be well investigated Heat Transfer Experiment The heat transfer behaviors of F-free fluorine mold flux were investigated by using the infrared Emitter Technique (IET), which was schematically shown in Fig. 2. The details about the IET have been described elsewhere. 28,29) This experiment apparatus mainly included a power controller, an infrared radiant heater capable of emitting 2.0 MW/m 2 heat flux, a data-acquisition system, and a command-and-control unit. The copper mold was simulated by a one-end water cooled copper cylinder, which acted as the radiation target, and its schematic figure was shown in Fig. 3. As the radiation energy was applied to the top surface of the copper mold, which was covered with mold flux disk, the responding temperatures inside the mold could be measured by the subsurface thermocouples. The heating profile for IET tests was shown in Fig. 4, where the copper mold system was firstly heated up under the incident thermal energy increased linearly to 800 kw/m 2 ; and then it was maintained for 300 seconds. After that, the incident heat flux was increased to kw/m 2, which is in the magnitude of real caster, and maintained for 800 seconds. Figure 5(a) shows the subsurface response temperatures when the heating profile in Fig. 4 was applied to a bare copper mold system. The thermocouple placed at 2, 5, 10 and 18 mm below the irradiated surface (Fig. 3) were recorded as T 1, T 2, T 3 and T 4 respectively. The cooling water inlet and outlet temperature were recorded as T in and T out. As the thermal properties are functions of temperature and time, the 1- dimensional inverse heat conduction developed by Beck, 30) which is sufficient to use only one internal body temperature and one boundary condition to determine the unknown boundary condition and estimate the heat flux by utilizing measured transient interior temperature, 31) is adapted.... (1) = ( x k T i ) Ti ρcp x t Where x is the distance below the top surface, k is the thermal conductivity, T is the temperature, ρ is the mass density of copper, t is the time, and c p is the copper heat capacity. Then the calculated heat flux histories for the bare copper system was given in Fig. 5(b), and it could be observed that the measured heat flux increases linearly with the addition of the output power first, and then come into a steady state in a very short time Phase Analysis The structure (liquid, crystalline and glass) of the slag disks after the heat transfer experiment and the morphology of crystalline phases in the disks were further observed Fig. 3. Schematic figure of copper substrate used as the radiation target. Fig. 4. The heating profile for heat transfer test. Fig. 5. The responding temperatures (a) and heat flux (b) for bare copper system ISIJ

4 through Scanning Electron Microscope (Japanese Electronics Company JSM-6360LV) with an acceleration voltage of 20 kw and 50, 300, or 1 000, etc. times magnification. The crystalline phases developed during the test was analyzed by Energy Dispersive Spectrometer (America EDAX Corporation EDX-GENESIS 60S) and X-ray Diffractometer (Rigaku Corporation RIGAKU-TTR III) with Cu Kα ( nm). The XRD data were collected in a range of 2θ =10 80 with a step size of 10 /min. 3. Results and Discussions 3.1. Melting Behavior of F-free Mold Fluxes The details about melting process of mold flux through using SHTT have been described in previous paper. 26) The responding temperature history and the snapshot of initial/ complete melting process of the benchmark sample A and the other four F-free mold fluxes are shown in Fig. 6. It can be found from Fig. 6 that there were temperature deviations appearing in the curves during the mold fluxes melting due to the endothermic reaction of melting process. Before the initial melting of the powder mold fluxes, the temperature increased linearly with a heat rate of 15 K ( C)/s. Then, the temperature was gradually deviated from the preset heating rate line when the melting process initiated, and the melting of mold fluxes could be observed through the snapshots as shown in Fig. 6. The temperature would then increase linearly again when the melting process completed. Those temperature deviations would last for a period, which can be called as the melting temperature range. In order to compare the melting behavior of the benchmark sample A and the four designed F-free mold fluxes, the initial and complete melting temperatures of above five mold fluxes are shown in Fig. 7. It can be seen that the melting temperature range of benchmark sample A was K ( C), and it was K ( C), K ( C), K ( C) and K ( C) for F- free mold fluxes Sample B, C, D, E, respectively, which suggests that the combination of Na 2O, Li 2O, B 2O 3 and basicity could be used for replacement of fluoride, such that the melting temperature ranges of those four F-free mold fluxes are close to the benchmark sample A. Besides, comparing with Sample B, the initial melting temperature of Sample C decreased due to the slight increase of Na 2O/Li 2O contents in Sample C (from 10.92% to 11.89% for Na 2O and from 1.96% to 2.92% for Li 2O). However, once the basicity reduces to 1.15, and the Na 2O and B 2O 3 contents reduce to 9.91% and 7.95% respectively, the melting temperature range of Sample D became higher than that of Samples B and C. While comparing with Sample D, the melting temperature zone of sample E decreased, as the reduction of Na 2O and B 2O 3 in Sample E (from 9.91% to 7.92% for Na 2O and from 7.95% to 5.98% for B 2O 3). Above phenomena could be explained as the fact that CaO, Na 2O, Li 2O and B 2O 3 are very good fluxing agency for slag system, which can form low melting temperature substances with other components in the F-free mold fluxes. 32,33) Through careful comparison, it found that the melting behavior of Samples C and E are closer to the benchmark sample A Heat Transfer Behavior of F-free Mold Fluxes The prepared disks of benchmark Sample A and the four designed F-free mold fluxes were individually placed on the top of copper mold and subjected to the heating profile (Fig. 4) for the heat transfer tests. Here, the temperature history and snapshots of typical states of benchmark Sample A was chosen to describe the heat transfer test process, which was shown in Fig. 8. The whole heat transfer experiment could Fig. 7. Melting temperature ranges of benchmark sample A and the four designed F-free mold fluxes. Fig. 6. Temperature history and melting process of mold fluxes. Fig. 8. The measured heat flux and phase variation of Sample A at different stages ISIJ 824

5 be divided into six stages. Stage I and stage II were the preheating stages where the heat flux first increased linearly with the addition of thermal radiation in stage I, and then kept constant about 300 s at stage II. The disks were kept as glassy without phase transformation during the pre-heating stages. After that, the heat flux was increasing linearly during stage III, and the deviation of heat flux occurred due to the initiation of mold flux crystallization where the opaque crystals formed on the top of the disk when time went to stage IV as shown in Fig. 8. The heat flux then continuously increased and reached its peak value, where the top of disk begun to be melted (the shape of the disks gradually got rounded) in stage V due to the accumulation of heat. Finally, the heat flux would step into steady state when the evolution of flux film structure (liquid, crystalline and glass layers) became stable and the heat flux keeps relative constant, as shown in stage VI in Fig. 8. Figure 9 shows the measured heat fluxes and final crosssection views of benchmark sample A and the four designed F-free sample disks. The measured heat flux curves of those five samples were overlapped with each other, which indicate that the general heat transfer behaviors of designed F- free mold fluxes are similar. Also, it can be observed from the final cross-section views of those mold fluxes disks that there were glass layer, crystalline layer and liquid layer from the bottom to top, except Sample C that only formed crystalline and liquid layer. In order to investigate the variation of heat flux and the three layer distribution in more details, the final steady state heat flux of mold fluxes in Fig. 9 was shown in Fig. 10, and the specific thickness of different layer of slag disks was measured and shown in Table 2. It can be seen from Fig. 10 that the steady state heat flux of benchmark sample A was 499 Kw/m 2, while the other four designed F-free mold fluxes were 511 KW/m 2 (Sample B), 486 Kw/m 2 (Sample C), 524 KW/m 2 (Sample D) and 505 KW/m 2 (Sample E), respectively. Besides, the specific thickness of crystalline layer of benchmark Samples A was 3.51 mm, and it was 2.57 mm, 3.71 mm, 2.09 mm and 3.43 mm respectively for Samples B, C, D and E. The structure of mold flux disks obtained here is similar to the mold flux in the gap of shell/ mold wall in the real continuous caster, as the glass mold flux usually forms near the cooled mold wall; the liquid mold flux layer exists near the shell due to the high temperature; and the crystalline layer develops between of them. Through comparing the general distribution of mold fluxes structure in Table 2, it could be found that Sample B showed relatively weak crystallization ability, while Sample C with basicity 1.25 and higher content of Na 2O and Li 2O showed strongest crystallization ability, such that its crystalline layer thickness would be thicker than others and no glass layer formed. When the basicity reduced to 1.15 and the Na 2O decreased to 9.91%, the crystalline layer thickness of Sample D attenuated to 2.09 mm. But, it increased to 3.43 mm again for Sample E as both B 2O 3 and Na 2O content reduced about 2% based on the Sample D. The reasons for above phenomena is because that CaO, Na 2O/Li 2O are network breakers while B 2O 3 is network former in the silicate structure system, and the CaO, Na 2O/Li 2O can provide O 2 to break down the Bridging Oxygen Si O Si into Non Bridging Oxygen Si O, then simplify the silicate structure and make the molten slag easer to get crystallized; while the B 2O 3 works in the opposite way, it will absorb the O 2 and make the silicate structure more complicated, then inhibited the precipitation of crystalline phases. 34,35) Meanwhile, the increase of crystalline layer thickness of Sample E compared with Sample D suggested that the B 2O 3 had more competitive influence on crystallization of F-free mold fluxes than Na 2O. Besides, it can also be found in Table 2 and Fig. 10 that the crystalline layer thicknesses of the mold flux disks was inverse proportion to the final steady state heat flux. The main reason for that is because the crystallization of mold flux can reduce the heat flux across the disk due to more incident radiation reflected and scattered from the crystals surface, grain boundary as well as defects, leading to less energy would be absorbed and conducted to the mold. 36,37) Through comprehensive comparison of the heat flux and the disk structure of those F-free mold fluxes, it can be Fig. 10. The final steady state heat flux of mold fluxes. Table 2. Thickness of different layers of slag disks. Fig. 9. The whole measured heat fluxes of all samples. Sample A B C D E Total/mm Liquid layer/mm Crystal layer/mm Glass layer/mm ISIJ

6 known that the Sample E shows more similar heat transfer and crystallization behaviors with the benchmark sample A Phase Analysis of the F-free Mold Fluxes The structure (liquid, crystalline and glass) of the slag disks after the heat transfer experiments and the morphologies of crystalline phases was further observed through SEM, and the phase compositions of crystalline phases were determined by EDS and XRD, which are shown in Figs Among them, Fig. 11 is the SEM and EDS of benchmark Sample A, and Fig. 11(a) shows the cross-section views of Sample A, which is constituted of glass layer on the top (originated from quenching of liquid layer), crystalline layer in the middle, and glass layer on the bottom. Figures 11(b) and 11(c) are 50 times magnification of boundaries of liquid/crystalline layer and crystalline/glass layer in Fig. 11(a). Spherical holes with different size were observed in the Fig. 11(b), which might originate from bubbles of volatiles or shrinkage cavity of liquid mold flux after solidification. In order to further observe and indentify the morphologies and composition of crystalline phases, the crystals close to top liquid layer, middle crystalline part, and bottom glass layer were amplified and shown in Figs. 11(d), 11(e), and 11(f), respectively. There were two types of crystals in Fig. 11(d), one was lighter particles with composition close to cuspidine (Ca 4Si 2O 7F 2), and the other was darker particles with a composition close to melilite (Ca 4Al 2MgSi 3O 14); while in Figs. 11(e), and 11(f), there only cuspidine crystals occurred. The details about compositions of those crystalline phases obtained by EDS are listed in the table of Fig. 11. Figure 12 shows the SEM and EDS of F-free mold flux Fig. 11. SEM and EDS of Sample A. (a) Cross-section views, (b) SEM of boundary of liquid/crystalline layer, (c) SEM of positions of crystalline/glass layer, (d) Crystalline phases close to liquid layer, (e) Crystalline phases in the middle of crystalline layer, (f) Crystalline phases close to glass layer. Fig. 12. SEM and EDS of Sample B. (a) Crystalline phases close to liquid layer, (b) Crystalline phases in the middle of crystalline layer, (c) Crystalline phases close to glass layer ISIJ 826

7 Fig. 13. SEM and EDS of sample C. (a) Crystalline phases close to liquid layer, (b) Crystalline phases in the middle of crystalline layer. Fig. 14. SEM and EDS of sample D. (a) Crystalline phases close to liquid layer, (b) Crystalline phases in the middle of crystalline layer, (c) Crystalline phases close to glass layer. Sample B, where Fig. 12(a) is close to liquid layer, Fig. 12(b) is in the middle of crystalline layer, and the Fig. 12(c) is close to glass layer on the bottom of the disk. There were also two typical crystals occur in the disk, one was the very fine dendritic crystals, and the other was larger blocky crystals. The results of EDS analysis suggest that the composition of the fine dendritic crystals was similar to the calcium borosilicate (Ca 11Si 4B 2O 22); while the blocky crystals was similar to the Ca 14Mg 2(SiO 4) 8. Besides, through comparing the Figs. 12(a), 12(b) and 12(c), it can be found that the amount of Ca 11Si 4B 2O 22 crystals increased while Ca 14Mg 2(SiO 4) 8 decreased with the decrease of temperature, as the Fig. 12(a) is close to the higher temperature liquid layer on the top of the disk, while the Fig. 12(c) is close to lower temperature glass layer on the bottom of the disk. The right side of Fig. 12(c) is part of glass layer where crystals rarely to be observed. The SEM and EDS of F-free mold flux Sample C is shown in Fig. 13. Also, Fig. 13(a) shows crystals close to the top liquid layer of the disk, while the Fig. 13(b) shows the crystals close to the bottom of the disk. The morphologies of crystalline in Fig. 13(a) were small particles with the composition similar to Ca 11Si 4B 2O 22, and lager blocky crystals with the composition similar to Ca 14Mg 2(SiO 4) 8. The EDS analysis shows the long-bar crystals in Fig. 13(b) were also Ca 11Si 4B 2O 22. It seems that the morphologies of Ca 11Si 4B 2O 22 in Figs. 13(a) and 13(b) were different at first glance, but if observing the crystals in Fig. 13(b) carefully, it will be found that long-bar crystals were constituted by small particles just like those Ca 11Si 4B 2O 22 in Fig. 13(a), which also suggests that the variation of morphology of certain crystal may occur under different thermal conditions. Figures 14 and 15 show the SEM and EDS of F-free mold flux Sample D and Sample E, respectively. The structure of ISIJ

8 Fig. 15. SEM and EDS of sample E. (a) Crystalline phases close to liquid layer, (b) Crystalline phases in the middle of crystalline layer, (c) Crystalline phases close to glass layer. Fig. 16. XRD of (a) benchmark sample A and (b) the designed F-free mold fluxes. slag Sample D and Sample E disks was similar to the Sample B disk which could be divided into liquid, crystalline and glass layers. In Fig. 14, there were two main crystal phases which are lager blocky with the composition similar to Ca 14Mg 2(SiO 4) 8, and needle crystals with composition similar to Ca 11Si 4B 2O 22. The needle crystals in Fig. 14(c) grew toward the glass layer inverse of the heat transfer direction. In Fig. 15, there are also two main crystal phases, the composition of the lager blocky is close to Ca 14Mg 2(SiO 4) 8 and the composition of the elongated crystals is close to Ca 11Si 4B 2O 22. From both Figs. 14 and 15, it also can be found that the amount of Ca 14Mg 2(SiO 4) 8 reduced while the amount of Ca 11Si 4B 2O 22 increased with the decrease of temperature, which indicates that the crystallization behavior of Ca 11Si 4B 2O 22 could precipitate among the disk especially in the low crystallization temperature zone that is very similar to the cuspidine (Ca 4Si 2O 7F 2) through comparing with the previous study. 27) In order to further indentify the crystalline phases in the benchmark sample A and the designed F-free mold fluxes, the mold flux disks were ground into powders and analyzed by using XRD. The X-ray diffraction patterns of those mold fluxes are shown in Fig. 16. It can be seen from Fig. 16(a) that the major characteristic peaks of Sample A are cuspidine (Ca 4Si 2O 7F 2) and melilite (Ca 4Al 2MgSi 3O 14). The main characteristic peaks of the other four F-free mold fluxes are shown in Fig. 16(b), and it could be observed that Ca 11Si 4B 2O 22 and Ca 14Mg 2(SiO 4) 8 are the two major phases crystallized in the F-free samples, which is consistent with the SEM and EDS results above. Therefore, it could be concluded that the calcium borosilicate (Ca 11Si 4B 2O 22) is common and stable crystalline phase in the designed F-free mold fluxes system just like the cuspidine (Ca 4Si 2O 7F 2) in the conventional F-containing mold fluxes. 4. Conclusions The melting and heat transfer behaviors of F-free mold fluxes have been studied by using the SHTT and IET, and the structure of the slag disks and morphologies of crystalline phases have been observed through SEM. Meanwhile, the phase compositions of crystalline phases were deter ISIJ 828

9 mined by EDS and XRD. The main conclusions were summarized as follows: (1) The melting temperature ranges of those four designed F-free mold fluxes are close to the benchmark conventional F-containing mold fluxes, and the melting temperature range decreases with the increase of CaO, Na 2O/Li 2O and B 2O 3. (2) The heat transfer behaviors of designed F-free mold fluxes flux are similar to the benchmark Sample A. The final steady state heat flux of Sample A is 499 KW/m 2 ; while it is 511 KW/m 2, 486 Kw/m 2, 524 KW/m 2 and 505 KW/m 2 for the F-free mold fluxes Sample B, C, D and E, respectively. (3) The crystalline layer thicknesses of mold flux disks is inversely proportional to the final steady state heat flux due to the fact that the crystallization of mold flux can reduce the heat flux across the disk through reflecting and scattering incident radiation from the crystals surface, grain boundary as well as defects. (4) The results of EDS and XRD show that the Ca 11Si 4B 2O 22 and Ca 14Mg 2(SiO 4) 8 are the two major crystalline phases in F-free mold fluxes, and the Ca 11Si 4B 2O 22 tend to precipitate in the low temperature zone that is similar to the cuspidine (Ca 4Si 2O 7F 2) in the conventional F-containing mold fluxes. (5) Although the melting temperature range of Sample C is closer to Sample A than that of Sample E, the heat flux of Sample C deceased dramatically and there was no glass layer formed in the final disk after the heat transfer tests due to its strong crystallization ability. Therefore, Sample E has the potential to be used for casting medium carbon steels. Acknowledgements The authors wish to thank the China Postdoctoral Science Foundation (2014M550423), and The Hunan Provincial Science and Technology Program (2014TT2030) and NSFC ( ) for the support of this research. REFERENCES 1) R. M. Mcdavid and B. G. Thomas: Metall. Mater. Trans. B, 27B (1996), ) K. C. Mills, A. B. Fox, Z. Li and R. P. Thackray: Ironmaking Steelmakking, 32 (2005), 26. 3) C. Li and B. G. Thomas: Metall. Mater. Trans. B, 35B (2004), ) S. Sridhar, K. C. Mills and S. T. Mallaband: Ironmaking Steelmaking, 29 (2002), ) K. C. Mills, A. B. Fox, R. P. Thackray and Z. Li: Proc. VII Int. Conf. on Molten Slags, Fluxes and Salts, South African Inst. Min. Metall., Johannesburg, (2004), ) J. H. Park, D. J. Min and H. S. Song: Metall. Mater. Trans. B, 33B (2002), ) S. Sridhar: JOM, 54 (2002), 46. 8) T. Watanabe, H. Fukuyama and K. Nagata: ISIJ Int., 42 (2002), ) T. Watanabe, H. Fukuyama, K. Nagata and M. Susa: Metall. Mater. Trans. B, 31B (2000), ) M. Persson, S. Seetharaman and S. Seetharaman: ISIJ Int., 47 (2007), ) S. Y. Choi, D. H. Lee, D. W. Shin, S. Y. Choi, J. W. Cho and J. M. Park: J. Non-Cryst. Solids, (2004), ) Z. Wang, Q. F. Shu and K. Chou: Can. Metall. Q., 52 (2013), ) Z. Zhang, G. Wen and Y. Zhang: Int. J. Miner. Metall. Mater., 18 (2011), ) H. Nakada and K. Nagata: ISIJ Int., 46 (2006), ) J. L. Klug, R. Hagenmann, N. C. Heck, A. C. F. Cilela and P. R. Scheller: Technol. Metal. Mater. Miner., 9 (2012), ) Q. Wang, Y. J. Lu, S. P. He, K. C. Mills and Z. S. Li: Ironmaking Steelmaking, 38 (2011), ) Y. Lu, G. Zhang, M. Jiang, H. Liu and T. Li: J. Environ. Sci., 23 (Supplement) (2011), ) E. Benavidez, L. Santini, M. Valentini and E. Brandaleze: 11th Int. Cong. Metallurgy & Materials, SAM/CONAMET, Curran Associates, Inc., Rosario, (2011), ) A. B. Fox, K. C. Mills, D. Lever, C. Bezerra, C. Valadares, I. Unamuno, J. J. Laraudogoitia and J. Gisby: ISIJ Int., 45 (2005), ) N. Takahira, M. Hanao and Y. Tsukaguchi: ISIJ Int., 53 (2013), ) G. Li, H. Wang, Q. Dai, Y. Zhao and J. Li: J. Iron Steel Res., Int., 14 (2007), ) A. B. Fox, M. E Valdez, J. Gisby, R. C. Atwood, P. D. Lee and S. Sridhar: ISIJ Int., 44 (2004), ) B. Lu, W. Wang, J. Li, H. Zhao and D. Huang: Metall. Mater. Trans. B, 44B (2013), ) L. Zhou, W. Wang, J. Wei and B. Lu: ISIJ Int., 53 (2013), ) J. Wei, W. Wang, L. Zhou, D. Huang, H. Zhao and F. Ma: Metall. Mater. Trans. B, 45B (2014), ) L. Zhou, W. Wang, B. Lu and G. Wen: Met. Mater. Int., (2014), accepted. 27) Y. Kashiwaya, C. E. Cicutti, A. W. Cramb and H. Ishii: ISIJ Int., 38 (1998), ) W. Wang, K. Blazek and A. Cramb: Metall. Mater. Trans. B, 39B (2008), ) W. Wang: PhD Thesis, Carnegie Mellon University, Pittsburgh, PA, (2007). 30) J. V. Beck: 1HCP1D: Version 5.31, Beck Engineering Consultants Company, Houston, TX, (1997). 31) H. Zhang, W. Wang, D. Zhou, F. Ma, B. Lu and L. Zhou: Metall. Mater. Trans. B, 45B (2014), ) V. D. Eisenhüttenleute: Slag Atlas, 2nd Edition, Verlag Stahleisen GmbH, (1995), ) H. Wang, T. Zhang, H. Zhu, G. Li, Y. Yan and J. Wang: ISIJ Int., 51 (2011), ) K. Mills: S. Afr. Pyrometallurgy, 111 (2011), 1. 35) W. Shi and P. F. James: J. Mater. Sci., 28 (1993), ) S. Ozawa, M. Susa, T. Goto, R. Endo and K. C. Mills: ISIJ Int., 46 (2006), ) J. Cho, H. Shibata, T. Emi and M. Suzuki: ISIJ Int., 38 (1998), ISIJ