Review on the High-Temperature Thermophysical Properties of Continuous Casting Mold Fluxes for Highly Alloyed Steels

Abstract

Several recently developed highly alloyed steel grades have shown unsurpassed performance in terms of physical, chemical, and electromagnetic properties. However, broader commercialization of these steels has been hampered by limitations in mold flux performance. Newly developed steels containing considerable amounts of dissolved Al, Mn, and Ti actively react with typical CaO-SiO₂-based mold fluxes, which severely changes the composition and subsequently the thermophysical properties of the mold flux that determine the external and internal quality of the as-cast steels. These dynamic changes result in nonuniform heat transfer, lubrication issues, surface defects, and caster breakouts. This work critically assesses the current status of the high-temperature thermophysical properties of CaO-SiO₂-based and CaO-Al₂O₃-based mold fluxes intended for use in casting highly alloyed steel grades. Thermophysical properties, including viscosity, crystallization, thermal conductivity, and heat flux, have been evaluated. The effect of compositional variables including CaO/SiO₂, CaO/Al₂O₃, and Al₂O₃/SiO₂ mass ratios and the additions of CaF₂, B₂O₃, Li₂O, K₂O, Na₂O, TiO₂, and BaO on these high-temperature thermophysical properties are discussed.

1. Introduction

With increased global urbanization, demand for renewable energy, lower energy dissipation during energy conversion, and fuel-efficient transportation, the steel industry has seen substantial growth for highly alloyed steels that can contain substantially higher Al, Mn, and Ti contents than typical carbon steels.^1,2,3,4,5,6) These new grades of steel seek to satisfy requirements for higher strength and formability with greater weight reductions, higher corrosion resistance at elevated temperatures, and higher relative magnetic permeability, resulting in lower hysteresis loss.^7,8,9) To be cost effective in producing these highly desired steels, continuous casting is necessary, and mold fluxes play an essential role in ensuring the external and internal product quality.^{10,11,12,13,14)} Optimal mold fluxes are designed to have specific physicochemical properties to enhance lubrication and provide uniform horizontal heat extraction of the superheat from the molten steel through the partially solidified shell in a water-cooled copper mold.^7,15) However, the aforementioned steel grades and the dissolved alloying elements can react with the typical CaO-SiO₂-based mold flux constituents due to the favorable thermodynamics of the redox reactions.^16,17,18,19)

Figure 1(a) summarizes the tensile strength and elongation characteristics for highly alloyed steels, including high-Al, high-Mn, high-Mn with high-Al, and high-Ti steels.²⁰⁾ High-Al transformation-induced plasticity (TRIP) steels with Al above 1 wt% have gained broad interest due to their combination of high strength-to-weight ratio and excellent ductility.²¹⁾ However, during continuous casting, the chemical reaction between SiO₂ in the mold flux with dissolved Al in the steel increases the Al₂O₃ to SiO₂ mass ratio (Al₂O₃/SiO₂), as shown in Fig. 1(b), which can cause casting issues and affect the quality of the semifinished products.^22,23)

Fig. 1.

(a) Tensile strength × elongation (MPa∙%) characteristics of highly alloyed steels and the compositional evolution in molten fluxes for (b) high-Al steels, (c) high-Mn steels, (d) high-Mn and high-Al steels, and (e) high-Ti steels. Adapted from published literature.^{18,19,20,22,23,25,26)} (Online version in color.)

Recently, commercialized high-Mn twinning-induced plasticity (TWIP) steels with Mn in excess of 10 wt% have been identified to have even greater strength and ductility than TRIP steels, thereby attracting the attention of the automotive, maritime, and offshore industries.²⁴⁾ As depicted in Fig. 1(c), the MnO content in the flux tended to increase with prolonged exposure to the steels and reached a steady state within a relatively short time.²⁵⁾ Further mechanical property enhancements, such as improving elongation uniformity and suppressing delayed fracture, can also be achieved with the addition of Al above 1 wt%.¹⁷⁾ Recent studies focusing on the reaction mechanism between the conventional CaO-SiO₂-based mold fluxes and molten steel with both high-Mn and high-Al have suggested that the Al content in the steel is the dominant factor determining the final composition and properties of the flux that results in excessive Al₂O₃ formation, as shown in Fig. 1(d).^18,26)

During casting of high-Ti ferritic stainless steels using existing calcium-silicate fluxes, increases in TiO₂ content can be observed in the mold flux. This phenomenon can be caused by the absorption of TiO₂ and TiN inclusions from the molten steel formed during secondary steelmaking and the oxidation of dissolved Ti in the steel by SiO₂ in the mold flux.²⁷⁾ Although the reaction of dissolved Ti with SiO₂ is not as pronounced as the reactions of Al and SiO₂, appreciable amounts of TiO₂ are observed, as depicted in Fig. 1(e).¹⁹⁾ This increase in the TiO₂ content also changes the thermophysical properties of the mold flux.^19,28)

Some basic theoretical slag-metal reactions already considered during the casting of these aforementioned highly alloyed steels using conventional CaO-SiO₂-based mold fluxes are given in Table 1. However, similar to the actual steelmaking slag reactions in refining, the reactions forming complex oxides have yet to be fully understood, which can fundamentally alter the thermodynamic driving forces that govern the kinetics of the reactions and the thermophysical properties of the mold flux.^1,27,29)

Table 1. Theoretical and actual slag/metal reactions during casting of highly alloyed steels.

Theoretical reactions	Actual reactions
(SiO2)+[Al]→(Al2O3)+[Si]	3(SiO2)+4[Al]+2(MgO) = 3[Si]+2MgAl2O4
(SiO2)+[Mn]→(MnO)+[Si]	2(SiO2)+2[Al]+[Mn] = 2[Si]+MnAl2O4
(SiO2)+[Ti]→(TiO2)+[Si]	3(SiO2)+4[Al]+(CaO) = [Si]+CaAl4O7
(MnO)+[Al]→(Al2O3)+[Mn]	(SiO2)+[Ti]+(CaO) = [Si]+CaTiO3
(TiO2)+[Al]→(Al2O3)+[Ti]	2(SiO2)+[Ti]+(CaO) = [Si]+CaSiTiO5
[Ti]+[N]→[TiN]	(TiO2)+[Al]+(CaO) = [Ti]+ Ca12Al14O33
[TiN]+(SiO2)→[Si]+(TiO2)+N2	[Ti]+[N]+C = (TiC)+N2
… …	… …

For high-Mn or high-Ti steels, the content of MnO or TiO₂ in the mold flux increases, and the corresponding SiO₂ content decreases. For high-Al steels, there is a rapid and significant amount of Al₂O₃ formation and subsequent SiO₂ reduction observed, which reach a steady-state within a relatively short residence time. These thermodynamically favorable reactions at steelmaking temperatures result in significant compositional changes in the mold flux that alter the intended thermophysical properties of the flux, which can affect the heat transfer and lubrication ability of the mold flux, subsequently decreasing the quality of the semifinished product. Thus, considerable efforts have been made to either inhibit or compensate for these reactions.

The chemical composition of the mold flux is constantly changing throughout the casting process, so the mold flux fed into the gap between the solidified shell and the mold has inconsistent properties, resulting in quality problems such as cracks or dents on the surface of the slab. Further deterioration conditions in the mold, including low mold flux consumption and heavily sintered layers, may also cause a breakout prevention alarm and the interruption of the continuous casting process.¹⁾ These are caused by the fact that the slag crystallization behavior and heat transfer uniformity cannot meet the current steel grade. Furthermore, regardless of the grade of steel, it is necessary to ensure good lubricity of the mold flux in the continuous casting process. Therefore, the present review is needed to summarize the properties of the modified mold flux to provide some reference and guidance for casting different highly alloyed steels.

This review attempts to highlight the results in the available literature on the thermophysical properties of CaO-SiO₂-based and CaO-Al₂O₃-based mold fluxes developed for highly alloyed steels. In particular, the viscosity, crystallization, thermal conductivity, and heat transfer behavior of the mold flux are assessed. By providing a critical evaluation of these properties for mold fluxes with different additives and processing conditions, technical guidance and possible research directions can be elucidated for the design and optimization of mold fluxes for these highly alloyed steel grades in the future.

2. Classification and Trends for Mold Flux Development

The typical chemical compositions of a multicomponent conventional mold flux for carbon steels are shown in Table 2.³⁰⁾ The traditional mold powders are based on the CaO–SiO₂ slag system, where a combination of fluidizers is added to optimize the thermophysical properties of the mold flux and ensure stable continuous casting. Among these fluidizing agents, fluoride, in the form of CaF₂, is commonly used as an additive because it can increase the initial crystallization temperature for nucleation while lowering the melting temperature and viscosity to improve the fluidity of the mold flux.^31,32) As the crystallization temperature is increased, the primary crystalline phase of cuspidine (Ca₄Si₂O₇F₂) is easily formed and uniformly dispersed, which can be used to effectively control the horizontal heat transfer between the molten steel and water-cooled copper mold.

Table 2. Compositions of traditional mold flux and some typical newly developed mold fluxes for some special alloy steels.

Mold flux classification	CaO	SiO2	Al2O3	Na2O	K2O	MgO	MnO	Fe2O3	F	Li2O	B2O3	BaO	ZrO2	TiO2
Conventional mold flux	22–45	17–56	0–13	0–25	0–2	0–10	0–5	0–6	2–15	0–5	0–19	0–10	–	–
Mold flux for high–Al steel	19–46	9–42	1.5–46	9–13	–	0–2	–	–	0–15	0–7	0–17	0–10	0–5	–
Mold flux for high–Mn steel	17–42	5–54	0–49	0–14	–	0–15	3–14	0–2	0–20	0–1	0–25	–	–	–
Mold flux for high–Ti steel	27–43	28–43	2–7	4–23	–	0–4	0–3	–	0–17	0–2	2–6	–	–	0–30

On the other hand, excessive fluorine contents generate environmentally harmful HF, SiF₄ and NaF, increasing the corrosion of the continuous casting machine.^33,34,35) During common casting operations, up to 30 wt% of the fluoride in the mold flux has been estimated to dissolve with the secondary cooling water discharge, increasing water pollution and equipment corrosion.³⁶⁾ Considering the emergence of green metallurgical processing, there has been considerable drive to develop low-fluorine or fluorine-free mold fluxes while maintaining or improving the steel quality. Other oxides, such as Na₂O, K₂O, Li₂O, MgO, MnO, B₂O₃ and TiO₂, have also been added into multicomponent mold flux systems to compensate for the negative impact on the mold flux properties with lower fluoride contents, resulting in primary crystalline phases of calcium borosilicate (Ca₁₁Si₄B₂O₂₂) and perovskite (CaTiO₃) to substitute the typical cuspidine phase in CaO-SiO₂-based mold fluxes.^37,38,39,40)

As a result of the recently developed highly alloyed steels, greater customer expectations, and increased green processing needs, the requirements and compositions of mold fluxes have undergone significant changes. Aggressive reactions of the slag and the dissolved elements in the metal often occur for these special steels, resulting in significant compositional changes. In particular, mold flux composition changes from SiO₂-rich to Al₂O₃-rich, TiO₂-rich, and MnO-rich compositions have been observed. Several studies have attempted to develop nonreactive CaO-Al₂O₃-based or CaO-TiO₂-based mold fluxes, but problems with controlling the crystallization, heat transfer, and subsequent detrimental impact on the lubrication require significant amounts of additives, such as B₂O₃, Li₂O, Na₂O and BaO, to compensate for the adverse effects on the thermophysical properties.^{15,41,42,43,44)}

Within the mold fluxes, the constituents will dissociate into their respective anions and cations to intrinsically interact and form coordinated bonds, resulting in short-range ordered structural units. These constituents comprising the flux can either act as network formers (SiO₂, B₂O₃, TiO₂, Fe₂O₃ and others) or network modifiers (CaO, BaO, FeO, Na₂O, Li₂O and others).^11,45,46) Al₂O₃ is an amphoteric oxide, which can behave as a network former or modifier depending on the overall ratio between the existing network forming and modifying oxides present.^47,48) The structural units formed through the interaction between the network forming and modifying cations and anions can be correlated to the viscosity and crystallization behavior of the slags, which subsequently affects the lubrication and heat transfer behavior of the flux and the surface quality of the semifinished product. The typical effects on the viscosity, crystallization temperature, thermal conductivity and heat flux of some typical mold fluxes are summarized in Table 3.^{28,32,39,42,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69, 70,71,72,73,74,75,76,77,78,79,80,81,82,83,84)}

Table 3. Viscosity, crystallization temperature, thermal conductivity and heat flux of some typical CaO-SiO₂-based and CaO-Al₂O₃-based mold fluxes (“↓”: decrease; “↑”: increase; “↓↑”: initial decrease and then increase; and “↑↓”: initial increase and then decrease).

Thermophysical properties	Slag system	C/S	C/A	A/S	Al2O3	CaF2	Na2O	K2O	MgO	MnO	B2O3	Li2O	TiO2	BaO	Reference
Viscosity	CaO-SiO2-based mold flux	↓				↓	↓	↓	↓	↓	↓	↓	↑	↓	28,39,42, 50,51,52, 53,54,55, 56,57,58, 59)
	CaO-Al2O3-based mold flux		↓↑	↓↑		↓	↓	↓	↓	↓	↓	↓	↑
Crystallization temperature	CaO-SiO2-based mold flux	↑			↑	↑	↑	↓		↓	↓	↑	↓	↑	32,60,61, 62,63,64, 65,66,67, 68,69,70)
	CaO-Al2O3-based mold flux		↓↑	↑		↓	↓↑			↓	↓	↑	↓
Thermal conductivity	CaO-SiO2-based mold flux	↑		↓	↓	↓	↓						↑	↑	66,71,72, 73,74,75, 76,77,78)
	CaO-Al2O3-based mold flux		↑↓	↓↑		↑	↑			↑	↓		↑
Heat flux	CaO-SiO2-based mold flux	↓		↑		↓	↓			↑		↓	↑	↓	66,75,76, 77,78,79, 80,81,82, 83,84)
	CaO-Al2O3-based mold flux		↑↓	↓↑		↑	↑			↑	↓		↑

3. Mold Flux Behavior for Different Highly Alloyed Steels

3.1. Flux Developments for High-Al Steels

3.1.1. Viscous Behavior

There have been two main approaches in designing mold fluxes for high-Al steels. One approach was based on retaining the CaO-SiO₂-based slag system, where the necessary mold flux properties are ensured by adding different fluxing agents. By estimating the expected compositional changes at the thermodynamic equilibrium state for the slag-steel reaction during the casting of high-Al steel, the thermophysical properties required for acceptable lubrication and heat transfer can be met with the addition of appropriate fluidizers. The other approach was based on developing a CaO-Al₂O₃-based slag system, which is nonreactive with dissolved Al, by adding various fluidizers resulting in thermophysical properties comparable to the existing mold fluxes but with different primary crystalline phases beyond the cuspidine phase.⁸⁵⁾

Blazek et al.⁸⁶⁾ and Kim et al.¹⁸⁾ studied the utilization of CaO-SiO₂-based mold fluxes for casting high-Al steels, where the content of Al₂O₃ continued to increase in the liquid flux pool as long as a sufficient SiO₂ content above a critical amount was present in the slag. Others^22,23,41) have also observed this trend, and according to empirical evidence, the critical SiO₂ content seems to be approximately 10 wt% Considering the need to lower the melting point of the flux, SiO₂ additions up to 10 wt% may be possible. CaO-SiO₂-based mold flux having higher CaO to SiO₂ (CaO/SiO₂) mass ratios with lower viscosities and melting temperatures can suppress the steel-slag reaction and lower the thermodynamic driving force of the reactions.²³⁾ Due to the redox reactions between Al and SiO₂, several studies have focused on the influence of the Al₂O₃ to SiO₂ (Al₂O₃/SiO₂) mass ratios on the thermophysical properties of CaO-Al₂O₃-SiO₂-based slags. The viscosities of mold fluxes with different Al₂O₃/SiO₂ ratios in different slag systems are shown in Fig. 2(a). Liao et al.⁸⁷⁾ observed a slight decrease in the viscosity of the CaO–SiO₂–Al₂O₃–MgO slag system as the Al₂O₃/SiO₂ ratio increased within the range of 0.11 to 0.56. In contrast, Li et al.⁵³⁾ and Yu et al.⁵⁷⁾ identified multivariant behaviors of the viscosity in CaO–SiO₂–Al₂O₃–CaF₂ and CaO–SiO₂–Al₂O₃–CaF₂–MgO–Na₂O–MnO–Li₂O slag systems, where the viscosity initially decreased followed by an increase as the Al₂O₃/SiO₂ ratio increased. However, Zhang et al.⁸⁸⁾ observed an increase in the viscosity as the Al₂O₃/SiO₂ ratio increased within the range of 0.06 to 0.57 in the CaO–SiO₂-Al₂O₃-CaF₂-based mold flux system, which was comparable to the results reported by Zhang et al.⁸⁹⁾ for the CaO–SiO₂–Al₂O₃–CaF₂ slag system with an Al₂O₃/SiO₂ ratio ranging from 0.14 to 0.46. The discrepancies in the viscosity with different Al₂O₃/SiO₂ ratios can be ascribed to three competitive factors.^53,57) First, the amphoteric behavior of Al₂O₃ can result in the formation of either network forming [AlO₄]⁵⁻ tetrahedral structural units or network modifying [AlO₆]⁹⁻ octahedral structural units. Second, more Ca²⁺ ions are necessary to act as charge compensating cations with greater Al₂O₃ content, where newly formed [AlO₄]-tetrahedral units exist. This phenomenon decreases the amount of network modifying Ca²⁺ ions and accelerates the conversion of nonbridging oxygen (O^–) to bridging oxygen anions (O^o) for electroneutrality. Subsequently, greater O^o would enhance the polymerization of the network structure and promote higher viscosity. Third, the replacement of stronger Si–O bonds (bond lengths of 1.61×10⁻¹⁰ m) with weaker Al–O bonds (bond lengths of 1.75×10⁻¹⁰ m) can lead to a weaker network structure and lower the forces required for melt shearing.^55,90) Thus, an initial increase in the Al₂O₃ content tends to rely on the Al–O and Si–O bonding characteristics lowering the viscosity, and further Al₂O₃ additions tend to form complex structures with [AlO₄]-tetrahedral structures that would promote O^o formation and increase the viscosity. Considering these factors and the components comprising mold fluxes, the amphoteric Al₂O₃ can behave differently according to the slag system.

Fig. 2.

(a) Viscosity of CaO-SiO₂-based mold fluxes with different Al₂O₃/SiO₂ ratios and (b) viscosity of CaO-Al₂O₃-based mold fluxes with different CaO/Al₂O₃ ratios. Adapted from published literature.^{53,54,57,66,87,88,89,93)}

Kim and Sohn⁵²⁾ studied the effect of CaF₂ on the viscosity of CaO-SiO₂-based mold flux containing Na₂O for application to TRIP steels. CaF₂ additions of up to 8 wt% decreased the viscosity but had no further effects beyond 8 wt% CaF₂ and NaF have similar effects on depolymerizing the slag network, which lowers the viscosity and improves the diffusion kinetics to accelerate the formation of the primary cuspidine phase during cooling. However, the reactions of CaF₂ and NaF with SiO₂ and the volatilization of the fluorides at high temperatures may result in other operational problems at the casters that would need to be considered.³²⁾

To circumvent some of the concerns of fluorides, some studies have focused on developing low-fluorine or fluorine-free mold fluxes with the addition of B₂O₃ and Li₂O in CaO-SiO₂-based mold fluxes for high-Al TRIP steels.^28,52,57) Both the network forming B₂O₃ and the network modifying Li₂O can lower the viscosity, but with different roles. Although [BO₄]-tetrahedral structural units are initially formed, B–O has lower bond energy than Si–O, which can result in easy dissociation of the B–O bonds at high temperatures, promoting the transformation of the unstable [BO₄]-tetrahedral structural units to the looser [BO₃]-triangular structural units, simplifying the slag structure and decreasing the viscosity. The highly basic oxide of Li₂O can provide free oxygen ions (O²⁻) to depolymerize the network structure and lower the viscosity, but has a limited effect when Li₂O content is above 2 wt% and sufficient fluidizers of Na₂O and CaF₂ already exist in the flux system. Na₂O additions in mold fluxes typically lower the viscosity, and the effect of these additions is more pronounced when sufficient complex structural units are present in the melt.⁹¹⁾ However, CaO-SiO₂-based flux systems containing both B₂O₃ and Na₂O have observed considerable volatilization as NaBO₂ gases, which would mandate a thick powder layer to minimize volatilization.⁹²⁾ From the aforementioned works, developments in the CaO-SiO₂-based fluxes concentrate on compensating for the compositional changes by adding various fluidizers and focus on their effects on the thermophysical properties. Within these works, the Al and SiO₂ reactions are assumed to be unavoidable, and dynamic compositional changes are considered.

A different approach is the development of nonreactive CaO-Al₂O₃-based mold fluxes.⁴²⁾ Kim and Sohn⁵⁴⁾ investigated the viscosity of the CaO–Al₂O₃–Na₂O–B₂O₃ system, where the viscosity decreased as the CaO/Al₂O₃ mass ratios (CaO/Al₂O₃) increased from 0.8 to 1.1 at B₂O₃ contents from 8 to 18 wt%, as shown in Fig. 2(b). However, as the CaO/Al₂O₃ ratio continued to increase above 2, the viscosity began to increase due to unmelted CaO.⁹³⁾ A similar “V” shaped trend of the viscosities with increasing CaO/Al₂O₃ ratio was also obtained by Yan et al.⁶⁶⁾ in the CaO–Al₂O₃–SiO₂–Na₂O–B₂O₃–CaF₂ slag system at lower temperatures. These changes in viscosity are closely related to the network structure of the melt at higher temperatures, where the degree of polymerization (DOP) decreased for slags with higher CaO/Al₂O₃ ratios. Since the crystallization temperature will increase with increasing CaO/Al₂O₃ ratios, the crystals precipitate more readily at higher CaO/Al₂O₃ ratios, and greater undercooling will enhance crystal growth, thereby increasing the viscosity.

With increasing B₂O₃ content, the viscosity decreased in the CaO-Al₂O₃-based mold flux, which is similar to the behavior exhibited by CaO-SiO₂-based mold fluxes.^54,56) The structural analysis suggested that increases in the B₂O₃ content accelerates the construction of B–O bonds within [BO₃]-triangular structural units at the expense of unstable [BO₄]-tetrahedral structural units, resulting in the dominant role of [BO₃]-triangular structural units in the slag structure. Therefore, the accelerated linkage between [BO₃]-triangular structural units and [AlO₄]-tetrahedral structural units causes the network structure to change from a 3-D structure to a partial 2-D structure, thereby causing looseness of the structure and a decrease in viscosity. With regard to Li₂O, the viscosity of the CaO-Al₂O₃-based mold flux tended to decrease with increasing substitutions of CaO with Li₂O.⁹⁴⁾ As Li₂O dissociates into its constituent cations and anions in the ionic melts, O²⁻ can break the Al–O covalent bonds to lower the viscosity.⁹⁵⁾ However, the viscosity seems to slightly increase for Li₂O contents in excess of 7 wt% due to the formation of the lithium aluminate (LiAlO₂) crystal phase, which has a high melting point.⁹⁴⁾

According to Wang and Sohn,⁵⁵⁾ the substitution of CaO with equal amounts of BaO can increase the viscosity of CaO–BaO–SiO₂–Al₂O₃–MgO slags, which is consistent with the trends observed by others in the CaO–SiO₂–Al₂O₃–BaO slag system⁹⁵⁾ and CaO–SiO₂–Al₂O₃–MgO–Na₂O–Li₂O–CaF₂–BaO slag system.⁵⁸⁾ The increasing viscosity trend, especially at higher BaO contents, can be attributed to the significantly different molecular mass between BaO (153.33 g/mol) and CaO (56.08 g/mol), which results in a dramatic decrease in the number of network modifiers, thereby increasing the DOP and the viscosity.

3.1.2. Crystallization Behavior

The most important parameter for the crystallization behavior is the crystallization temperature, which may depend on the experimental methods employed. For the confocal laser scanning microscopy (CLSM) and single hot thermocouple technology (SHTT), where the crystallization behavior can be observed, the crystallization temperature corresponds to the temperature at which the initial crystal nucleation can be observed. For the differential scanning calorimetry (DSC) and differential thermal analysis (DTA), the crystallization temperature corresponds to the temperature taken from the intersection of the tangent of the heat flow curve before the exothermic peak and the tangent of the steep exothermic peak curve. The volumetric heat analyzed from the DSC and DTA require significant crystal nucleation and growth to occur before appreciable changes can be detected. Higher crystallization temperatures were observed in the CaO–Al₂O₃–SiO₂–MgO slag system with higher Al₂O₃/SiO₂,⁶⁴⁾ as shown in Fig. 3(a). The primary phase changed from akermanite (Ca₂MgSi₂O₇) to merwinite (Ca₃MgSi₂O₈) and then to gehlenite (Ca₂Al₂SiO₇) as the Al₂O₃/SiO₂ ratio increased. A similar increase in the crystallization temperature was also observed in the CaO-SiO₂-Al₂O₃-CaF₂-based mold flux system.⁶⁸⁾ However, Jung et al.¹³⁾ observed the opposite trend in the initial crystallization temperature as the Al₂O₃/SiO₂ ratio increased in the CaO–Al₂O₃–SiO₂–Na₂O–Li₂O slag system with Na₂O and Li₂O contents of 17 wt% This phenomenon can be attributed to the change in the primary crystalline phase of sodium calcium silicate (Na₂Ca₃Si₂O₈) to caswellite (Ca₃Al₂Si₃O₁₂), which has a lower crystallization temperature. In the CaO–SiO₂–Al₂O₃–B₂O₃–Na₂O flux system, a combination of wollastonite (CaSiO₃) and combeite (Na₂Ca₂Si₃O₉) crystallizes at a low CaO/SiO₂ ratio of 0.8, which transitions to Ca₃Si₂O₇ and Ca₁₁Si₄B₂O₂₂ phases above a CaO/SiO₂ ratio of 1.5.⁹¹⁾ Greater B₂O₃ additions in CaO-SiO₂-based flux systems inhibit crystallization by lowering the crystallization temperature during continuous cooling and extend the incubation time for nucleation during isothermal cooling.⁹⁶⁾ As expected, depending on the components comprising the mold flux system, varying crystallization behaviors can occur.

Fig. 3.

(a) Crystallization temperature of mold fluxes with different Al₂O₃/SiO₂ ratios and (b) incubation time of mold fluxes with different BaO contents. Adapted from published literature.^{13,64,68,69,74)}

Seo et al.⁶³⁾ performed DSC tests and found a lower crystallization temperature with a 1.8 wt% Li₂O addition in CaO-SiO₂-based slags, similar to the findings of Yang et al.⁹⁷⁾ However, greater additions of Li₂O may promote the formation of LiAlO₂,⁹⁸⁾ which increases the driving force of the initial nucleation in liquid slags. Due to the lower viscosity with Li₂O additions, diffusion of ionic species is promoted, which can increase the growth rate of the primary crystals.⁹⁹⁾

Jiang et al.¹⁰⁰⁾ used SHTT to construct the time-temperature-transformation (TTT) and continuous-cooling-transformation (CCT) curves for CaO-Al₂O₃-based mold fluxes, and they found that Al₂O₃ additions seemed to enhance crystallization with CaO/Al₂O₃ ratios between 0.8 and 1.2. In addition, the critical cooling rate and crystallization temperature increased with shorter incubation times with lower CaO/Al₂O₃ ratios. Although the increased viscosity observed at higher Al₂O₃ contents can still meet the fluidity requirements for mold fluxes during continuous casting,⁹⁴⁾ the melting temperature and initial crystallization temperature for nucleation in CaO-Al₂O₃-based mold fluxes dramatically increase with higher Al₂O₃ contents, which typically exceed the necessary conditions for optimal casting of high-Al steels.^93,94)

To resolve the problems of strong crystallization and subsequent insufficient lubrication for CaO-Al₂O₃-based mold fluxes during the casting of high-Al steels, additives, such as B₂O₃, Li₂O, Na₂O and BaO, have been used both independently and in combination.

Using SHTT, the incubation time of the primary phase in CaO-Al₂O₃-based fluxes containing 9 wt% CaF₂ initially increased as the Li₂O content increased from 1 to 4.5 wt% and subsequently decreased as the Li₂O content increased up to 6 wt% due to the precipitation of LiAlO₂.¹⁰¹⁾ Due to the lower crystallization temperature of fluorite (CaF₂) at higher Li₂O contents, a larger undercooling was required for the primary phase,⁹⁸⁾ and the incubation time was extended. Lu et al.⁹⁸⁾ found that Li₂O and Na₂O in CaO-Al₂O₃-based mold flux played similar roles in affecting the crystallization behavior, where the additions of these fluidizers would tend to inhibit crystallization by lowering the crystallization temperature and increasing the incubation time. However, when the Li₂O and Na₂O contents exceeded 7 and 13 wt%, respectively, the crystallization was accelerated by the formation of the LiAlO₂ and carnegieite (Na_xAl_ySi_zO₄) crystal phases.

Other works have identified BaO to potentially control the crystallization rate of fluxes^{69,74,102,103)} because the melting point of BaO (2196 K) is much lower than that of CaO (2846 K). Lu and Wang⁶⁹⁾ and Yan et al.⁷⁴⁾ observed that the partial substitution of CaO with BaO can inhibit crystallization in CaO-Al₂O₃-based mold fluxes by increasing the incubation time, as shown in Fig. 3(b). This approach can increase the potential vitrification fraction of the mold flux, which is more conducive to lubrication in continuous casting.^94,104) Although the substitution of CaO with BaO can inhibit crystallization, the barium calcium aluminate (BaCa₂Al₈O₁₅) primary phase has been known to precipitate during continuous cooling in CaO-Al₂O₃-based fluxes.¹⁰⁵⁾ However, according to Wang and Sohn,^55,90) although the substitution of CaO with equimolar amounts of BaO can reduce the liquidus temperature of CaO-Al₂O₃-based mold flux to inhibit crystallization, the substitution of CaO with BaO cannot occur without restrictions. As BaO gradually replaces CaO, there is also a significant increase in the viscosity of the mold flux, which should also be considered, and the optimal amount of BaO used to inhibit crystallization must be balanced with the combined effect of higher flux viscosities.^55,58,90)

In addition, Fig. 3(b) also shows that lower CaF₂ contents can promote crystallization by decreasing the incubation time and accelerating the formation of gehlenite with a higher nucleation temperature.^65,69) Furthermore, the incubation time sharply decreased with higher Al₂O₃/SiO₂ ratios in the fluorine-containing mold flux, as shown in Fig. 3(b);⁷⁴⁾ this phenomenon can be ascribed to the acceleration of the precipitated primary phases, which changed from cuspidine to gehlenite, as shown in Fig. 4.^63,78,106)

Fig. 4.

(a) Morphology of cuspidine (Ca₄Si₂O₇F₂) obtained by CLSM, (b) morphology of cuspidine obtained by SEM, (c) morphology of cuspidine obtained by SHTT, and (d) morphologies of cuspidine and gehlenite (Ca₂Al₂SiO₇) obtained by SEM. Adapted from published literature.^63,78,106) (Online version in color.)

3.1.3. Heat Transfer Behavior

The overall heat transfer through the copper mold is the sum of the conductive and radiative heat terms, which is affected by the characteristics of the mold flux film.¹⁰⁷⁾ This film is comprised of the liquid, glassy, and crystalline phase situated between the water-cooled copper mold and the partially solidified steel shell. The partition between these phases depend on the flux properties including the crystallization behavior.¹⁰⁷⁾ The total heat transfer (q_Tot.) can be expressed by the following equation,^10,107)   

$$q_{Tot.}=q_{Rad.}+q_{Cond.}=q_{Tot.}^{Glass}+q_{Tot.}^{Cryst.}+q_{Tot.}^{Liq.}$$(1)

where q_Rad. and q_Cond. correspond to the heat transfer though slag film via the radiation and conduction mechanism, respectively.

Due to the crystallization of the mold flux, the uneven air gaps can be formed at the interface between the copper mold and the mold flux to increase the thermal resistance, thereby reducing the heat transfer. In addition, the scattering caused by the crystallites present in the slag film can also inhibit the radiation.^10,107)

The overall resistance to heat transfer (R_Tot.) between the shell and mold can be regarded as a series of resistances as shown in the following equation,^15,107)   

$$\begin{array}{l}{R}_{Tot.}^{}=R^{Int.}+R^{Liq.}+R^{Glass.}+R^{Cryst.}=\\ R^{Int.}+{(d/q)}^{Liq.}+{(d/q)}^{Glass.}+{(d/q)}^{Cryst.}\end{array}$$(2)

where R^Int. is the interfacial resistance, R^Liq., R^Glass. and R^Cryst. are the resistance in the liquid, glassy and crystalline phases, and d is the thickness of different phases. Since R^Int. was also found to increase with increasing solid slag thickness and increasing crystallinity, the key factor affecting heat transfer is the thickness of the solid slag film and the fraction of the crystalline.¹⁵⁾ Thus, understanding the crystallization behavior of mold fluxes is essential in controlling the overall heat transfer in the mold.

For CaO-SiO₂-based mold fluxes, the precipitated cuspidine primary phase is generally used to effectively control the heat transfer between the molten steel and the mold.¹⁰⁸⁾ However, with changes in the Al₂O₃/SiO₂ and CaO/SiO₂ ratios, the formation and growth kinetics, morphology, and distribution of cuspidine can be greatly affected, resulting in unoptimized and uneven heat transfer in the mold. Therefore, several works have been carried out to optimize the heat transfer performance of mold flux for casting high-Al steels.

As the Al₂O₃ content increases from 7 to 30 wt% in the CaO-Al₂O₃-based flux containing 6 wt% CaF₂, the primary phase changes from cuspidine to carnegieite and then to fluorite, where the crystallization temperature decreases. As the Al₂O₃ content increases further to 40 wt%, the primary phase becomes gehlenite with a higher crystallization temperature. With the transformation of the primary phase and the decrease in crystallization temperature, the heat flux gradually decreases as the Al₂O₃ content increases from 7 to 30 wt% and then increases as the Al₂O₃ content increases to 40 wt%, which is inversely proportional to the trends exhibited by the crystalline layer thickness and the crystalline fraction with increasing Al₂O₃ content.⁸³⁾

For Li₂O additions ranging from 2 to 5 wt% in CaO-Al₂O₃-based mold fluxes, the shortened incubation time and increased crystallization temperature resulted in lower heat flux densities.⁷⁶⁾ For the same slag system, Wang et al.⁷⁷⁾ observed that the incubation time of nonreactive mold flux increased with increasing Na₂O content. The restrained crystallization behavior leads to inhibited growth of slag rims and increases the heat flux during continuous casting of high-Al steels.

3.2. Flux Developments for High-Mn Steels

3.2.1. Viscous Behavior

For high-Mn steels with low Al content, the interfacial reaction between the slag and metal is predominantly controlled by the dissolved Mn in the molten steel. Yang and Zhu²⁵⁾ studied the reaction between high-Mn steel (Mn=5 to 18 wt%; Al=0.004 to 0.01 wt%) and the CaO-SiO₂-based flux system, where MnO in the mold flux reached 12 wt% after 30 min; moreover, they found that higher CaO/SiO₂ ratios in the mold flux helped to reduce the accumulation of MnO. The slag viscosity was found to decrease continuously with increasing MnO content and decreasing SiO₂ content in the melt. This behavior is due to an increase in O²⁻ and decrease in complex oligomer structures formed from [SiO₄]-tetrahedral units in the melt, which decreases the DOP of the slag and lowers the shear stress required for viscous flow. A decreasing break temperature was also observed with increasing MnO content and decreasing SiO₂ content, which has been known to be beneficial in reducing slag rims at the meniscus, resulting in improved lubrication and decreased thermal resistance.

Due to the excellent mechanical properties of high-Mn steels with Al additions, numerous studies have been carried out on the thermophysical properties of mold flux systems for high-Mn and high-Al steels. Kim et al.^18,26) studied the reaction between high-Mn and high-Al steel (Mn=13 to 20.8 wt%; Al=0.4 to 5.2 wt%) with CaO-SiO₂-based mold fluxes, where Al₂O₃ increased and MnO sharply increased at the beginning of the reaction followed by a rapid decrease. Mn was initially oxidized by SiO₂, which was then reduced by the high Al content in the molten steel. In addition, lower CaO/SiO₂ ratios in the mold flux, lower Al content in the steel, and higher temperatures could enhance the MnO content in the molten flux with a MnO content of approximately 7 wt% at a CaO/SiO₂ ratio of 0.3.¹⁸⁾ Although both Al₂O₃ and MnO accumulate in the flux, the flux reaction with Al has a greater thermodynamic favorability than the reaction with Mn, which results in a higher Al₂O₃ content balanced with MnO at steady state.^{17,18,109,110)}

To optimize the viscosity of mold fluxes for casting high-Mn and high-Al steels, several reports have observed the effect of CaO/SiO₂ ratio, Al₂O₃/SiO₂ ratio or the additions of fluidizers.^{14,42,52,53,57,111)} Since many of the optimized mold flux viscosity studies in the aforementioned section were based on high-Al TRIP steels and high-Mn and high-Al TWIP steels, most of the viscosity behavior of mold fluxes for high-Mn and high-Al steels can be referred to as the viscosity behavior for high-Al steels. In addition, He et al.¹¹²⁾ clarified that the multicomponent and proportional compositions of each component can be used to lower the content of CaO and SiO₂ and increase other fluxing agents to ensure the stability of the mold flux performance. It was found that adding MnO can accelerate the reaction between Al and MnO and partially restrain the reaction between Al and SiO₂. Since MnO has less influence on the slag performance than SiO₂, using MnO may reduce the demerits caused by large changes in SiO₂ content on the thermophysical properties of the mold flux.

Some computational models have also been developed to estimate the changes in the composition of molten slag during continuous casting of high-Mn and high-Al steels.^110,113) Wang et al.¹¹⁰⁾ studied the reaction of CaO-SiO₂-based mold fluxes with a 20Mn23AlV (Mn=21−25wt%) steel and developed a model equation to predict the Al₂O₃ accumulation in the mold flux during casting. Kim and Kang¹¹³⁾ developed a multicomponent reaction model to describe the complex reactions between high-Mn and high-Al steels with CaO-SiO₂-based fluxes. Based on the impact factors of the model, the compositional changes in the steel and the flux during continuous casting can be simulated, which may be useful for predicting the viscosity and crystallization behavior of the flux and potentially identifying the components necessary for compensating for the potential deterioration in the flux performance.

3.2.2. Crystallization Behavior

For high-Mn steels with low Al content, the crystallization temperature initially increase and then decrease as the CaO/SiO₂ ratio and MnO content increase.²⁵⁾ As the MnO content increases, tephroite (Mn₂SiO₄) crystals are formed by consuming [SiO₄]-tetrahedral structural units within cuspidine, resulting in a reduced fraction of cuspidine crystals during continuous cooling.

As described previously for casting high-Mn and high-Al steel, the dominant component changes in the mold flux is the accumulation of Al₂O₃ and the consumption of SiO₂, where a sharp rise in the Al₂O₃/SiO₂ ratio can be detected, which affects the crystallization behavior of the mold flux similar to casting high-Al steels. Thus, the effect of the Al₂O₃/SiO₂ ratio and the addition of other fluxing agents to optimize the crystallization behavior can be found in the previous discussions on high-Al steels.

The crystallization behavior of the mold flux during casting of some specific high-Mn and high-Al steels is discussed below. Using the traditional CaO-SiO₂-based mold flux with low melting points and a low CaO/SiO₂ ratio for casting 20Mn23AlV (Mn=21−25wt%) nonmagnetic steels, Yu et al.¹¹⁴⁾ observed that an increase in the Al₂O₃/SiO₂ ratio promotes the nucleation of fluorite crystals in the slag melt. Considerable efforts have been made to optimize nonreactive CaO-Al₂O₃-based mold fluxes, which could moderate the metal/slag interaction and control the heat flux during casting of high-Mn and high-Al steels. For nonmagnetic 20Mn23AlV (Mn=21−25wt%) steels, the reaction between Al and SiO₂ results in higher melting temperatures and higher viscosities and provides a greater propensity for the nucleation and gradual growth of the gehlenite crystals in the flux.¹¹⁵⁾ Higher Al₂O₃/SiO₂ ratios also increase the crystallization temperature and shorten the incubation time, reducing the overall heat transfer.⁷⁴⁾ It can be inferred from the phase diagram that higher Al₂O₃ contents and increased CaO/SiO₂ ratios can promote the precipitated phase to shift from wollastonite with a low crystallization temperature to gehlenite with comparatively higher crystallization temperatures. This inference seems to coincide with the findings reported by Hiromoto et al.,¹¹⁶⁾ where an Al₂O₃ content greater than 20 wt% promoted the formation of gehlenite and inhibited the formation of cuspidine. The excess F⁻ that is not consumed by the formation of cuspidine will combine with Ca²⁺ to precipitate fluorite.

Yan et al.⁶⁶⁾ studied the effect of a wide range of CaO/Al₂O₃ ratios, from 0.6 to 3.2, on the crystallization behavior of mold fluxes for casting 20Mn23AlV (Mn=21−25wt%) steel and found that increasing the CaO/Al₂O₃ ratio initially inhibited and then enhanced crystallization by initially decreasing and then increasing the crystallization temperature. Moreover, they found that increasing the CaO/Al₂O₃ ratio initially increased and then decreased the incubation time. In addition, they found that a lower viscosity with a higher CaO/Al₂O₃ ratio could reduce the mass transfer resistance and accelerate the crystal growth rate.

The effect of substituting CaF₂ with B₂O₃ on the crystallization behavior of nonreactive CaO-Al₂O₃-based mold fluxes showed that CaF₂ inhibited the crystallization of CaO-Al₂O₃-based mold fluxes by lowering the crystallization temperature and prolonging the incubation time for crystallization,⁶⁵⁾ which was contrary to the behavior exhibited by CaO-SiO₂-based mold fluxes.³²⁾ The results also confirmed that with CaF₂ contents up to 20 wt%, cuspidine was not formed in the CaO-Al₂O₃-based mold flux containing 5 wt% SiO₂. As the CaF₂ content decreases, the primary crystal phase changes from calcium aluminum oxyfluoride (Ca₂Al₃O₆F) to gehlenite. The addition of B₂O₃ from 10 to 20 wt% in fluoride-free CaO-Al₂O₃-based mold fluxes inhibits the crystallization of gehlenite and promotes the formation of takedaite (Ca₃B₂O₆) at lower temperatures.⁶⁵⁾

3.2.3. Heat Transfer Behavior

For high-Mn steels with low Al content, the thermal conductivity of CaO-SiO₂-based mold flux decreased with increasing MnO contents.¹¹⁷⁾ With MnO additions, O²⁻ depolymerized the complex silicate networks, resulting in the disaggregation of the cuspidine and the generation of the tephroite (Mn₂SiO₄) phase, which has poor conductivity and a lower melting point, thereby leading to lower conductive heat transfer.¹¹⁸⁾

For high-Mn and high-Al steels, an increase in Al₂O₃/SiO₂ promotes the primary phase to change from cuspidine to gehlenite with a higher crystallization temperature, thereby inhibiting heat transfer by decreasing the thermal conductivity of the slag film, as shown in Fig. 5(a).^74,75) Therefore, the hindered heat transfer with a higher Al₂O₃/SiO₂ ratio will adversely affect the quality of the slab and the stability of the casting process.⁷⁴⁾

Fig. 5.

(a) Thermal conductivity of mold fluxes with different chemical composition ratios (R₁: Al₂O₃/SiO₂; R₂: BaO/CaO; and R₃: CaO/Al₂O₃) and (b) heat flux of mold fluxes with different TiO₂ contents. Adapted from published literature.^{66,74,75,79,129)}

Some studies have indicated that the substitution of CaO with BaO can address the concerns of the crystallization behavior in CaO-Al₂O₃-based fluxes by reducing the crystallization temperature and extending the incubation times at elevated Al₂O₃/SiO₂ ratios.^69,74) Thus, partial substitution of CaO with BaO could compensate for the reduced thermal conductivity caused by the increase in the Al₂O₃/SiO₂ ratio during casting of high-Mn and high-Al steels, as shown in Fig. 5(a).

Further investigations on the effect of a wide range of CaO/Al₂O₃ ratios, from 0.6 to 3.2, on crystallization behavior were conducted by Yan et al.,^66,75) where increases in the CaO/Al₂O₃ ratio initially increased and then decreased the thermal conductivity, as shown in Fig. 5(a). This finding correlated well with the initially inhibited and then enhanced crystallization behavior, as demonstrated by the changes in initial crystallization temperatures and incubation times.

Although B₂O₃ showed similar effects to CaF₂ on heat flux in CaO-Al₂O₃-based mold fluxes, the ability of B₂O₃ to enhance thermal conductivity was stronger than that of CaF₂, which can be attributed to the formation of Ca₃B₂O₆ instead of Ca₂Al₃O₆F in fluorine-containing CaO-Al₂O₃-based mold fluxes.^32,65)

Beyond compositional changes made to the fluxes, Cho et al.¹¹⁹⁾ observed significant improvements in the castability of high-Mn and high-Al (Fe-18Mn-0.6C-1.5Al) steel using liquid flux feeding. By ensuring a thick liquid pool, reactions between the flux and dissolved elements would be diluted and a relatively constant CaO/SiO₂ ratio of 0.6 with an Al₂O₃ pickup less than 10 wt% was maintained to ensure stability in the thermophysical properties of the mold flux. Some limitations of this liquid feeding approach would be the economics of melting slag, the additional nonreactive auxiliary equipment necessary, the greater volatilization of gaseous species, and the mold cover requirements to ensure vertical radiative thermal insulation.

3.3. Flux Developments for High-Ti Steels

3.3.1. Viscous Behavior

According to Mills,¹¹⁾ the problems encountered in the continuous casting process of high-Ti stainless steels are related to the increase in the TiO₂ content from the slag/metal reaction and the liquidus temperature associated with austenitic stainless steels, which is lower than that of typical carbon steels. Mukongo et al.¹⁹⁾ sampled the mold flux during casting of a 6 wt% Ti-stabilized austenitic stainless steel, where a TiO₂ content of 3–4 wt% was observed. This amount of dissolved Ti is uncommon, but several steel producers have been producing Ti-containing steels above 1 wt%, which should also result in significant reactions that form TiO₂.

Substitution of SiO₂ with TiO₂ up to 10 wt% in mold fluxes for casting high-Ti-stabilized stainless steels at temperatures between 1563 and 1623 K showed the viscosity to decrease with increasing TiO₂ content.¹²⁰⁾ At lower temperatures, the viscosity of mold fluxes with 10 wt% TiO₂ was greater than that of TiO₂-free slag, which was attributed to the precipitation of solid phases. Wang et al.²⁸⁾ and Sun et al.¹²¹⁾ studied the viscosity of CaO-SiO₂-based mold fluxes and found lower viscosities with higher TiO₂ contents. These findings are comparable to the results of Zheng et al.¹²²⁾ in the CaO–SiO₂–TiO₂ slag system and those of Park et al.¹²³⁾ and Feng et al.¹²⁴⁾ in the CaO–SiO₂–Al₂O₃–MgO–TiO₂ slag system. Two factors seem to be prevalent regarding the addition of TiO₂, which are the decrease in the main constituents of SiO₂, where [SiO₄]-tetrahedral units are depolymerized, and the bond energy of Ti–O, which is weaker than that of Si–O.^{28,120,122,123)}

With regard to the existing literature on nonreactive CaO-Al₂O₃-based mold fluxes for casting high-Ti steels, there seems to be limited research on this particular subject and may require more in-depth attention in the near future.

3.3.2. Crystallization Behavior

For casting high-Ti steels with CaO-SiO₂-based mold fluxes, F has been replaced with TiO₂ to generate the perovskite primary crystal phase, thereby substituting the cuspidine phase to control the heat transfer.^27,125,126) It has been speculated that TiO₂ reduces the infrared transmittance of the slag film and inhibits radiation heat transfer. In addition, Nakada and Nagata¹²⁷⁾ also observed that the incubation time for titanite (CaSiTiO₅) nucleation in CaO–SiO₂–TiO₂ slags was comparable to that of the cuspidine phase in commercial mold fluxes, which suggests that the CaSiTiO₅ could also crystallize rapidly in slag films similar to cuspidine in commercial mold fluxes.

By comparing the crystallization temperature and evaluating the effective activation energy for crystal growth, the substitution of SiO₂ with TiO₂ can restrain the nucleation of the dominant cuspidine and favor the nucleation of perovskite in the CaO-SiO₂-CaF₂-based mold flux.^120,128) While cuspidine is the dominant crystal phase for TiO₂-free slags, a combination of the perovskite and cuspidine phase is detected with 5 and 10 wt% TiO₂, where a greater fraction of perovskite is observed. Sun et al.¹²¹⁾ also observed that TiO₂ additions can restrain the nucleation of cuspidine and wollastonite and promote the nucleation of perovskite and melilite. For mold fluxes with different CaO/SiO₂ ratios of 0.85 and 1.0, the predominant primary crystal phase transformed from cuspidine to perovskite as the TiO₂ content increased, and the nucleation amount and growth rate of cuspidine were restrained for slags with lower CaO/SiO₂ ratios.⁷⁹⁾ With fluorine-free CaO-SiO₂-based mold fluxes, the crystallization was promoted with shorter incubation times as the TiO₂ content increased from 5 to 10 wt%, which was ascribed to the lower slag viscosity and greater mass transport accelerating crystallization.²⁸⁾

3.3.3. Heat Transfer Behavior

For the CaO-SiO₂-based mold flux of some special high-Ti steels, such as 409 ferritic stainless steels (0Cr11Ti, Ti=6×Cwt%~0.75wt%) and 321 austenitic stainless steels (1Cr18Ni9Ti, Ti=5(Cwt%−0.02wt%)~0.80wt%), an increase in the heat flux and heat transfer coefficient of the slag film was observed with increasing TiO₂ content, and this effect was more pronounced for mold fluxes with lower CaO/SiO₂ ratios,⁷⁹⁾ as shown in Fig. 5(b). This phenomenon was speculated to be a suppression of the growth rate and lower crystallization temperature of cuspidine with higher TiO₂ contents and lower CaO/SiO₂ ratios. However, due to the enhanced crystallization of perovskite in fluorine-free TiO₂-bearing mold fluxes, a lower heat flux can be observed with a higher TiO₂ content,¹²⁹⁾ as shown in Fig. 5(b), which also suggests that perovskite is a potential replacement for cuspidine. The morphology of perovskite is also shown in Fig. 6,¹³⁰⁾ which can be compared with the morphology of cuspidine shown in Fig. 4.

Fig. 6.

(a) Morphology of perovskite obtained by SHTT and (b) morphology of perovskite obtained by SEM. Adapted from published literature.¹³⁰⁾ (Online version in color.)

During actual continuous casting of Incoloy-800 steels containing Al and Ti, surface cracking problems of slabs were found to be highly correlated to the heat transfer behavior of the mold flux and the solidification characteristics of Incoloy-800. Based on the compositions of spent mold fluxes, higher CaO/SiO₂ ratios, MnO contents and CaF₂ contents in mold fluxes could inhibit heat transfer, while higher BaO and Al₂O₃ contents tended to increase the heat transfer.⁸²⁾ Yan et al.⁸²⁾ also considered reducing the CaO/SiO₂ ratio to promote heat transfer and suppress the precipitation of gehlenite and perovskite with a higher nucleation temperature. By adjusting the corresponding compositions, such as a lower CaO/SiO₂ ratio, higher CaF₂, BaO and Al₂O₃ contents, and lower MnO content, the newly designed mold flux reduced the melting temperature and provided good lubrication properties with lower crystalline fractions, thereby promoting heat transfer to thicken and strengthen the partially solidified shell.

4. Outlook

The thermophysical properties of CaO-SiO₂-based and CaO-Al₂O₃-based mold fluxes for casting highly alloyed steels were critically assessed. By changing the composition of the mold flux in anticipation of the fundamental thermodynamic reactions and compensating the multicomponent flux system with various fluidizers, the necessary thermophysical properties of the mold flux can be met for continuous casting. For the reactive CaO-SiO₂-based mold flux, it is necessary to stabilize the thermophysical properties by adding different kinds and contents of fluxing agents to either inhibit or compensate for these slag/steel reactions, so that even if the slag/steel reaction increases the Al₂O₃/SiO₂ or TiO₂/SiO₂ ratios, the thermophysical properties can still satisfy the continuous casting requirements of highly alloyed steels. For nonreactive CaO-Al₂O₃-based mold fluxes, it is especially crucial to avoid excessively high crystallization temperatures and crystallization morphologies with high aspect ratios, which result in nonuniform heat transfer through the additions of varying fluxing agents.

While the current design philosophy has considered the lubrication ability and the crystallization kinetics of fluxes, control of the crystallization morphology to achieve uniform heat transfer has yet to be fully realized. Much of the current knowledge has been limited to the initial crystallization and the typical uniform crystallization observed for the primary cuspidine crystal phase during casting, whereas the behavior of the CaO–Al₂O₃ primary crystalline phases is yet to be fully understood. The perovskite phase, which does seem to nucleate rapidly, has limited growth, and excessive TiO₂ additions tend to increase the melting temperature of the flux, which is detrimental to the lubrication properties during casting. Thus, future attempts to accelerate uniform growth of the perovskite phase are critical to its eventual application.

In addition, the role of thermal conductivity for fluxes has yet to be clearly identified. According to recent works, flux crystallization typically inhibits radiative heat transfer but significantly increases conductive heat transfer. Thus, the total heat flux in terms of the radiative and conductive heat transfer should be considered when designing mold fluxes. Furthermore, as crystallization is realized, the viscosity of the fluxes tends to increase significantly, and lubrication may be lost. Thus, optimal mold flux development for highly alloyed steels still requires a balance between the heat transfer mechanisms and lubrication.

In terms of the fluid thermophysical properties that have been widely examined, the continuous casting mold fluxes exist not in a single liquid state but typically exist within the solid-liquid coexisting regions across the casting length of water-cooled copper molds. In particular, although the initial solidification within the caster can assume the flux to be mostly liquid, the total length of the mold is much longer than the meniscus. Liquid flux films extending beyond 300 mm from the meniscus have been identified.¹³¹⁾ Thus, while the break temperature of the mold fluxes is rightly examined and provided by most mold flux vendors, the next generation of nonreactive mold fluxes must be developed considering the solid-liquid non-Newtonian behavior of the flux melt, which should be explored in more detail. Furthermore, partially glassy fluxes, when exposed to high temperatures along the length of the mold, can crystallize and increase the fraction of crystalline phases.¹³²⁾ This phenomenon inherently affects the overall heat transfer in the mold, which should also be a topic of discussion in future works.

Acknowledgments

This work was supported by the third stage of the Brain Korea 21 Plus Project of the Division of Creative Materials in 2018 and was supported by the Korea Institute for Advancement of Technology grant, funded by the Korea Government (MOTIE) (P0002019), as part of the Competency Development Program for Industry Specialists.

References

• 1)   S.  He,  Z.  Li,  Z.  Chen,  T.  Wu and  Q.  Wang: Steel Res. Int., 90 (2019), 1800424.
• 2)   S.  Martin,  S.  Richter,  S.  Decker,  U.  Martin,  L.  Krüger and  D.  Rafaja: Steel Res. Int., 82 (2011), 1133.
• 3)   A.  Hamada and  L.  Karjalainen: Mater. Sci. Eng. A, 528 (2011), 1819.
• 4)   E.  Zinngrebe,  C.  Van Hoek,  H.  Visser,  A.  Westendorp and  I.  Jung: ISIJ Int., 52 (2012), 52.
• 5)   G.  Xu,  X.  Gan,  G.  Ma,  F.  Luo and  H.  Zou: Mater. Des., 31 (2010), 2891.
• 6)   M.  Garibaldi,  I.  Ashcroft,  M.  Simonelli and  R.  Hague: Acta Mater., 110 (2016), 207.
• 7)   A.  Fox,  K.  Mills,  D.  Lever,  C.  Bezerra,  C.  Valadares,  I.  Unamuno,  J.  Laraudogoitia and  J.  Gisby: ISIJ Int., 45 (2005), 1051.
• 8)   W.  Wang,  B.  Lu and  D.  Xiao: Metall. Mater. Trans. B, 47 (2016), 384.
• 9)   H.  Luo and  H.  Dong: Mater. Sci. Eng. A, 626 (2015), 207.
• 10)   K. C.  Mills and  A. B.  Fox: ISIJ Int., 43 (2003), 1479.
• 11)   K. C.  Mills: ISIJ Int., 56 (2016), 14.
• 12)   G.  Kim,  C.  Kim and  I.  Sohn: ISIJ Int., 53 (2013), 170.
• 13)   S. S.  Jung,  G. H.  Kim and  I.  Sohn: Trans. Indian Inst. Met., 66 (2013), 577.
• 14)   H. S.  Park,  H.  Kim and  I.  Sohn: Metall. Mater. Trans. B, 42 (2011), 324.
• 15)   K.  Mills,  A.  Fox,  Z.  Li and  R.  Thackray: Ironmaking Steelmaking, 32 (2005), 26.
• 16)   L.  Zhang and  W.  Wang: Steelmaking, 33 (2017), No. 3, 20 (in Chinese).
• 17)   Y.  Kang,  M.  Kim,  S.  Lee,  J.  Cho,  M.  Park and  H.  Lee: Metall. Mater. Trans. B, 44 (2013), 309.
• 18)   M.  Kim,  S.  Lee,  J.  Cho,  M.  Park,  H.  Lee and  Y.  Kang: Metall. Mater. Trans. B, 44 (2013), 299.
• 19)   T.  Mukongo,  P.  Pistorius and  A.  Garbers-Craig: Ironmaking Steelmaking, 31 (2004), 135.
• 20)   R.  Elliott,  K.  Coley,  S.  Mostaghel and  M.  Barati: JOM, 70 (2018), 680.
• 21)   E. M.  Bellhouse,  A.  Mertens and  J. R.  McDermid: Mater. Sci. Eng. A, 463 (2007), 147.
• 22)   X.  Fu,  G.  Wen,  Q.  Liu,  P.  Tang,  J.  Li and  W.  Li: Steel Res. Int., 86 (2015), 110.
• 23)   C.  Ji,  Y.  Cui,  Z.  Zeng,  Z.  Tian,  C.  Zhao and  G.  Zhu: J. Iron Steel Res. Int., 22 (2015), 53.
• 24)   S.  Lee,  Y.  Estrin and  B. C.  De Cooman: Metall. Mater. Trans. A, 45 (2014), 717.
• 25)   J.  Yang and  M.  Zhu: ISIJ Int., 56 (2016), 2191.
• 26)   M.  Kim,  M.  Park,  S.  Kang,  J.  Park and  Y.  Kang: ISIJ Int., 58 (2018), 686.
• 27)   Q.  Wang,  Y. J.  Lu,  S. P.  He,  K. C.  Mills and  Z. S.  Li: Ironmaking Steelmaking, 38 (2011), 297.
• 28)   Z.  Wang,  Q.  Shu and  K.  Chou: Steel Res. Int., 84 (2013), 766.
• 29)   J.  Li,  Q.  Shu,  X.  Hou and  K.  Chou: ISIJ Int., 55 (2015), 830.
• 30)   E.  Brandaleze,  G.  Di Gresia,  L.  Santini,  A.  Martín and  E.  Benavidez: Science and Technology of Casting Processes, Intech, Rijeka, (2012), 205.
• 31)   K.  Nagata and  H.  Fukuyama: Steel Res. Int., 74 (2003), 31.
• 32)   J. Y.  Park,  J. W.  Ryu and  I.  Sohn: Metall. Mater. Trans. B, 45 (2014), 1186.
• 33)   A. I.  Zaitsev,  A. V.  Leites,  A. D.  Lrtvina and  B. M.  Mogutnov: Steel Res., 65 (1994), 368.
• 34)   M.  Persson,  S.  Seetharaman and  S.  Seetharaman: ISIJ Int., 47 (2007), 1711.
• 35)   L.  Zhou,  W.  Wang,  B.  Lu,  G.  Wen and  J.  Yang: Met. Mater. Int., 21 (2015), 126.
• 36)   J.  Gao,  G.  Wen,  Q.  Liu,  W.  Tan and  P.  Tang: J. Non-Cryst. Solids, 409 (2015), 8.
• 37)   L.  Zhou,  W.  Wang and  K.  Zhou: ISIJ Int., 55 (2015), 1916.
• 38)   W.  Wang,  D.  Cai and  L.  Zhang: ISIJ Int., 58 (2018), 1957.
• 39)   J.  Li,  Q.  Shu and  K.  Chou: Can. Metall. Q., 54 (2015), 85.
• 40)   J.  Yang,  J.  Zhang,  Y.  Sasaki,  O.  Ostrovski,  C.  Zhang,  D.  Cai and  Y.  Kashiwaya: ISIJ Int., 56 (2016), 574.
• 41)   J.  Cho,  K.  Blazek,  M.  Frazee,  H.  Yin,  J. H.  Park and  S.  Moon: ISIJ Int., 53 (2013), 62.
• 42)   G. H.  Kim and  I.  Sohn: ISIJ Int., 52 (2012), 68.
• 43)   Q.  Shu,  Z.  Wang,  J. L.  Klug,  K.  Chou and  P. R.  Scheller: Steel Res. Int., 84 (2013), 1138.
• 44)   C.  Shi,  M.  Seo,  H.  Wang,  J.  Cho and  S.  Kim: Metall. Mater. Trans. B, 46 (2015), 345.
• 45)   I.  Sohn and  D. J.  Min: Steel Res. Int., 83 (2012), 611.
• 46)   L.  Zhang,  W.  Wang,  B.  Zhai and  I.  Sohn: J. Am. Ceram. Soc., 103 (2019), 112.
• 47)   Y.  Kang and  K.  Morita: ISIJ Int., 46 (2006), 420.
• 48)   J. H.  Park,  D. J.  Min and  H. S.  Song: Metall. Mater. Trans. B, 35 (2004), 269.
• 49)   H.  Kim,  W. H.  Kim,  I.  Sohn and  D. J.  Min: Steel Res. Int., 81 (2010), 261.
• 50)   G.  Kim and  I.  Sohn: J. Am. Ceram. Soc., 102 (2019), 6575.
• 51)   L.  Wang,  Y.  Cui,  J.  Yang,  C.  Zhang,  D.  Cai,  J.  Zhang,  Y.  Sasaki and  O.  Ostrovski: Steel Res. Int., 86 (2015), 670.
• 52)   H.  Kim and  I.  Sohn: ISIJ Int., 51 (2011), 1.
• 53)   J.  Li,  Q.  Shu and  K.  Chou: Ironmaking Steelmaking, 42 (2015), 154.
• 54)   G. H.  Kim and  I.  Sohn: Metall. Mater. Trans. B, 45 (2014), 86.
• 55)   Z.  Wang and  I.  Sohn: J. Am. Ceram. Soc., 101 (2018), 4285.
• 56)   X.  Huang,  J.  Liao,  K.  Zheng,  H.  Hu,  F.  Wang and  Z.  Zhang: Ironmaking Steelmaking, 41 (2014), 67.
• 57)   X.  Yu,  G.  Wen,  P.  Tang and  H.  Wang: Ironmaking Steelmaking, 36 (2009), 623.
• 58)   E.  Gao,  W.  Wang and  L.  Zhang: J. Non-Cryst. Solids, 473 (2017), 79.
• 59)   T.  Watanabe,  H.  Hashimoto,  M.  Hayashi and  K.  Nagata: ISIJ Int., 48 (2008), 925.
• 60)   J.  Wei,  W.  Wang,  L.  Zhou,  D.  Huang,  H.  Zhao and  F.  Ma: Metall. Mater. Trans. B, 45 (2014), 643.
• 61)   J.  Li,  W.  Wang,  J.  Wei,  D.  Huang and  H.  Matsuura: ISIJ Int., 52 (2012), 2220.
• 62)   L.  Zhou,  W.  Wang and  K.  Zhou: Metall. Mater. Trans. E, 2 (2015), 99.
• 63)   M. D.  Seo,  C. B.  Shi,  J. W.  Cho and  S. H.  Kim: Metall. Mater. Trans. B, 45 (2014), 1874.
• 64)   Z.  Wang and  I.  Sohn: Ceram. Int., 44 (2018), 19268.
• 65)   W.  Yan,  W.  Chen,  Y.  Yang,  C.  Lippold and  A.  McLean: Ironmaking Steelmaking, 43 (2016), 316.
• 66)   W.  Yan,  W.  Chen,  Y.  Yang,  C.  Lippold and  A.  McLean: Ironmaking Steelmaking, 42 (2015), 698.
• 67)   J.  Gao,  G.  Wen,  Q.  Sun,  P.  Tang and  Q.  Liu: Metall. Mater. Trans. B, 46 (2015), 1850.
• 68)   Z.  Zhang,  G.  Wen,  J.  Liao and  S.  Sridhar: Steel Res. Int., 81 (2010), 516.
• 69)   B.  Lu and  W.  Wang: Metall. Mater. Trans. B, 46 (2015), 852.
• 70)   S.  Ozawa and  M.  Susa: Ironmaking Steelmaking, 32 (2005), 487.
• 71)   S.  Choi,  D.  Lee,  D.  Shin,  S.  Choi,  J.  Cho and  J.  Park: J. Non-Cryst. Solids, 345–346 (2004), 157.
• 72)   M.  Hayashi,  R. A.  Abas and  S.  Seetharaman: ISIJ Int., 44 (2004), 691.
• 73)   K.  Nagata,  M.  Susa and  K. S.  Goto: Tetsu-to-Hagané, 69 (1983), 1417 (in Japanese).
• 74)   W.  Yan,  W.  Chen,  Y.  Yang,  C.  Lippold and  A.  Mclean: ISIJ Int., 55 (2015), 1000.
• 75)   W.  Yan: Ph.D. thesis, University of Science and Technology Beijing, (2016), http://cdmd.cnki.com.cn/Article/CDMD-10008-1016044564.htm, (accessed 2015-12-21) (in Chinese).
• 76)   H.  Wang,  P.  Tang and  G. H.  Wen: J. Univ. Sci. Technol. Beijing, 33 (2011), 544.
• 77)   H.  Wang,  P.  Tang,  G.  Wen and  X.  Yu: Ironmaking Steelmaking, 38 (2011), 369.
• 78)   W.  Wang,  X.  Yan,  L.  Zhou,  S.  Xie and  D.  Huang: Metall. Mater. Trans. B, 47 (2016), 963.
• 79)   Z.  Hao,  W.  Chen and  C.  Lippold: Metall. Mater. Trans. B, 41 (2010), 805.
• 80)   L.  Zhou,  W.  Wang,  J.  Wei and  B.  Lu: ISIJ Int., 53 (2013), 665.
• 81)   B.  Lu,  W.  Wang,  J.  Li,  H.  Zhao and  D.  Huang: Metall. Mater. Trans. B, 44 (2013), 365.
• 82)   W.  Yan,  W.  Chen,  Y.  Yang,  C.  Lippold and  A.  McLean: Steel Res. Int., 87 (2016), 173.
• 83)   L.  Zhou,  W.  Wang and  K.  Zhou: Metals, 6 (2016), 139.
• 84)   H.  Zhao,  W.  Wang,  L.  Zhou,  B.  Lu and  Y.  Kang: Metall. Mater. Trans. B, 45 (2014), 1510.
• 85)   L.  Zhang,  W.  Wang and  H.  Shao: J. Iron Steel Res. Int., 26 (2019), 336.
• 86)   K.  Blazek,  H.  Yin,  G.  Skoczylas,  M.  McClymonds and  M.  Frazee: AIST Trans., 8 (2011), 232.
• 87)   J.  Liao,  Y.  Zhang,  S.  Sridhar,  X.  Wang and  Z.  Zhang: ISIJ Int., 52 (2012), 753.
• 88)   Z.  Zhang,  G.  Wen,  P.  Tang and  S.  Sridhar: ISIJ Int., 48 (2008), 739.
• 89)   G.  Zhang,  Y.  Zhen and  K.  Chou: J. Iron Steel Res. Int., 23 (2016), 633.
• 90)   Z.  Wang and  I.  Sohn: J. Am. Ceram. Soc., 102 (2019), 5632.
• 91)   L.  Wang,  C.  Zhang,  D.  Cai,  J.  Zhang,  Y.  Sasaki and  O.  Ostrovski: Metall. Mater. Trans. B, 48 (2017), 516.
• 92)   L.  Wang,  J.  Zhang,  Y.  Sasaki,  O.  Ostrovski,  C.  Zhang and  D.  Cai: Metall. Mater. Trans. B, 48 (2017), 1055.
• 93)   Q.  Wang,  J.  Yang,  C.  Zhang,  D.  Cai,  J.  Zhang and  O.  Ostrovski: J. Iron Steel Res. Int., 26 (2019), 374.
• 94)   T.  Wu,  Q.  Wang,  S.  He,  J.  Xu,  X.  Long and  Y.  Lu: Steel Res. Int., 83 (2012), 1194.
• 95)   S.  Sukenaga,  N.  Saito,  K.  Kawakami and  K.  Nakashima: ISIJ Int., 46 (2006), 352.
• 96)   J.  Yang,  J.  Zhang,  Y.  Sasaki,  O.  Ostrovski,  C.  Zhang,  D.  Cai and  Y.  Kashiwaya: Metall. Mater. Trans. B, 48 (2017), 2077.
• 97)   C.  Yang,  G.  Wen and  P.  Tang: Steel Res. Int., 87 (2016), 880.
• 98)   B.  Lu,  K.  Chen,  W.  Wang and  B.  Jiang: Metall. Mater. Trans. B, 45 (2014), 1496.
• 99)   H.  Liu,  G.  Wen and  P.  Tang: ISIJ Int., 49 (2009), 843.
• 100)   B.  Jiang,  W.  Wang,  I.  Sohn,  J.  Wei,  L.  Zhou and  B.  Lu: Metall. Mater. Trans. B, 45 (2014), 1057.
• 101)   L.  Zhou,  H.  Li,  W.  Wang,  D.  Xiao,  L.  Zhang and  J.  Yu: Metall. Mater. Trans. B, 49 (2018), 2232.
• 102)   D.  Xiao,  W.  Wang and  B.  Lu: Metall. Mater. Trans. B, 46 (2015), 873.
• 103)   F.  Dong,  Y.  Wang and  B.  Wang: Special Steel, 27 (2006), No. 6, 7 (in Chinese).
• 104)   Z.  Li,  X.  You,  M.  Li,  Q.  Wang,  S.  He and  Q.  Wang: Metals, 9 (2019), 142.
• 105)   L.  Zhou,  H.  Li,  W.  Wang,  Z.  Wu,  J.  Yu and  S.  Xie: Metall. Mater. Trans. B, 48 (2017), 2949.
• 106)   J.  Park,  G. H.  Kim,  J. B.  Kim,  S.  Park and  I.  Sohn: Metall. Mater. Trans. B, 47 (2016), 2582.
• 107)   J.  Park and  I.  Sohn: Int. J. Heat Mass Transf., 109 (2017), 1014.
• 108)   T.  Watanabe,  H.  Fukuyama and  K.  Nagata: ISIJ Int., 42 (2002), 489.
• 109)   D. J.  Kim and  J. H.  Park: Metall. Mater. Trans. B, 43 (2012), 875.
• 110)   Q.  Wang,  S.  Qiu and  P.  Zhao: Metall. Mater. Trans. B, 43 (2012), 424.
• 111)   L.  Zhang,  W.  Wang,  S.  Xie,  K.  Zhang and  I.  Sohn: J. Non-Cryst. Solids, 460 (2017), 113.
• 112)   S.  He,  Q.  Wang,  J.  Zeng,  M.  Zhang and  B.  Xie: J. Iron Steel Res., 12 (2009), 59.
• 113)   M.  Kim and  Y.  Kang: Calphad, 61 (2018), 105.
• 114)   X.  Yu,  G.  Wen,  P.  Tang,  F.  Ma and  H.  Wang: J. Iron Steel Res. Int., 18 (2011), 20.
• 115)   W.  Yan,  W.  Chen,  L.  Carsten and  H.  Zheng: Special Steel, 34 (2013), No. 1, 45 (in Chinese).
• 116)   T.  Hiromoto,  T.  Shima and  R.  Sato: 62nd Steelmaking Conf. Proc., Iron and Steel Society, Warrendale, PA, (1979), 40.
• 117)   X.  Qiu,  B.  Xie,  X.  Qing,  J.  Diao,  Q.  Huang and  S.  Wang: J. Iron Steel Res. Int., 20 (2013), 27.
• 118)   J.  Diao,  B.  Xie,  N.  Wang,  S.  He,  Y.  Li and  F.  Qi: ISIJ Int., 47 (2007), 1294.
• 119)   J.  Cho,  S.  Yoo,  M.  Park,  J.  Park and  K.  Moon: Metall. Mater. Trans. B, 48 (2017), 187.
• 120)   Z.  Wang,  Q.  Shu,  X.  Hou and  K.  Chou: Ironmaking Steelmaking, 39 (2012), 210.
• 121)   L.  Sun,  H.  Wang,  M.  Jiang,  Q.  Lin,  C.  Liu and  Y.  Zou: Adv. Mater. Res., 189–193 (2011), 107.
• 122)   K.  Zheng,  Z. T.  Zhang,  L. L.  Liu and  X. D.  Wang: Metall. Mater. Trans. B, 45 (2014), 1389.
• 123)   H.  Park,  J. Y.  Park,  G. H.  Kim and  I.  Sohn: Steel Res. Int., 83 (2012), 150.
• 124)   C.  Feng,  M.  Chu,  J.  Tang,  J.  Qin,  F.  Li and  Z.  Liu: Int. J. Miner. Metall. Mater., 23 (2016), 868.
• 125)   G.  Wen,  S.  Sridhar,  P.  Tang,  X.  Qi and  Y.  Liu: ISIJ Int., 47 (2007), 1117.
• 126)   S.  He,  X.  Long,  J.  Xu,  T.  Wu and  Q.  Wang: Ironmaking Steelmaking, 39 (2012), 593.
• 127)   H.  Nakada and  K.  Nagata: ISIJ Int., 46 (2006), 441.
• 128)   Z.  Wang,  Q.  Shu and  K.  Chou: Metall. Mater. Trans. B, 44 (2013), 606.
• 129)   X.  Qi,  G.  Wen and  P.  Tang: J. Non-Cryst. Solids, 354 (2008), 5444.
• 130)   D.  Zheng,  J.  Li,  C.  Shi and  J.  Ju: Ironmaking Steelmaking, 45 (2018), 135.
• 131)   P.  Hu,  X.  Wang,  J.  Wei,  M.  Yao and  Q.  Guo: ISIJ Int., 58 (2018), 892.
• 132)   C.  Yang,  G.  Wen,  Q.  Sun and  P.  Tang: Metall. Mater. Trans. B, 48 (2017), 1292.