Effect of Fluorine on Melt Structure for CaO–SiO₂–CaF₂ and CaO–Al₂O₃–CaF₂ by Molecular Dynamics Simulations

Abstract

The melt structure with fluorine addition has been investigated using molecular dynamics (MD) simulation for the CaO–SiO₂–CaF₂ and CaO–Al₂O₃–CaF₂ ternary slag system. By analyzing the coordination number of the network framework, the results indicate that the structure of Si–O tetrahedron is more stable than the structure of Al–O tetrahedron. F⁻ ions are dominantly coordinated with Ca²⁺, there is a dynamic equilibrium between Ca²⁺ and the coordination anions (O²⁻ and F⁻) in both systems, and the total coordination number (CN) is maintained between 6 and 7. The analysis of the distribution of oxygen types demonstrates the degree of polymerization (DOP) of network structure in CaO–SiO₂–CaF₂ system is lower than that in CaO–Al₂O₃–CaF₂ system. Although the addition of CaF₂ can lower the viscosity of both slag systems, the microscopic reasons for that are completely different: in CaO–SiO₂–CaF₂ system, CaF₂ actually acts as network diluent, and creates the space where particles can move longer. But in CaO–Al₂O₃–CaF₂ system, CaF₂ can depolymerize the network structure of the melt.

1. Introduction

In recent years, high-quality TRIP steel and RE steel have attracted extensive attention due to the good mechanical properties. However, the traditional CaO–SiO₂ based slag can react with these steels during the production process. As a result, the performance of the slag is extremely unstable, and even continuous casting of multiple furnaces cannot be achieved.^1,2) In order to solve the problem of slag-steel reaction, low-reactivity slag has become one of the research hotspots in the metallurgical industry. Low-reaction mould fluxes represented by CaO–Al₂O₃ based system has been widely developed and applied.^3,4) To satisfy the requirements of metallurgical production processes, other components shoud be added into the slag substrate to meet the comprehensive physical and chemical properties of the slag. As a traditional flux, CaF₂ is widely used to lower the melting point of slag and improve the fluidity of slag.

The influence mechanisms of CaF₂ on the melt structure of slags have been reported by many scholars using various spectroscopic methods. In CaO–SiO₂–CaF₂ system, Luth suggested through Raman spectroscopic study that when the CaO/SiO₂ ratio was fixed, the DOP did not change, while the DOP would increase with the increased CaF₂/CaO ratio.⁵⁾ Hayashi et al. found that fluorine was dominantly coordinated with calcium rather than silicon using X-ray photoelectron spectroscopy.⁶⁾ In contrast, Park et al. proposed that the DOP of the system had diverse trends with the ratio of CaF₂/CaO under different SiO₂ content, and provided evidence of Si–F complex with FT-IR spectrometry.⁷⁾ In (fluoro-) aluminate system, Park et al. studied the CaO–Al₂O₃–CaF₂ slag system using FT-IR spectra and found [AlO_nF_4−n]ⁱ⁻ complexes.⁸⁾ Gao et al. proposed that fluorine in the mold flux containing Al₂O₃ were classified into three distinct categories, according to function: Bridging F’s, Nonbridging F’s, and Free F’s.⁹⁾ Up to now, massive structural unit data in limited fluorine content range were obtained, while the effect of CaF₂ has not been considered in a wide range of components. Moreover, previous approaches have given contradictory results, and the mechanism of fluorine on the melt structure of slags remains unclear.

Another effective method for studying the melt structure is computer simulation method such as molecular dynamics, which has been applied to the field of metallurgical slag system to explain the microstructure changes that are not easily explained by experimental methods. However, few studies on MD simulations of F-containing slag systems were reported. Asada et al.¹⁰⁾ and Sasaki et al.¹¹⁾ discussed the changes in silicate melt structure at varied CaF₂/CaO ratio, similar conclusions with experiment⁵⁾ were drawn. Fan et al.¹²⁾ reported that network modifier cations were competing with Al ions for fluorine ions in CaO–Al₂O₃–SiO₂–CaF₂ system, more work is required to evaluate the change of DOP after the addition of CaF₂. Structure properties of CaO–Al₂O₃–CaF₂ slag by MD have not been well studied yet. Based on the extensive application of fluorine in metallurgical slag system, it is of great significance to systematically compare the mechanism of action with CaF₂ addition of CaO–SiO₂ based and CaO–Al₂O₃ based slag systems that have important applications in metallurgy.

In the present study, the slag systems of (1−x/2)CaO−(1−x/2)SiO₂−xCaF₂ and (1−x/2)CaO−(1−x/2)Al₂O₃−xCaF₂ were studied by MD. The different effects of fluorine on melt structure of the two base slag systems were compared and analyzed in a wide range of CaF₂ content. Furthermore, since the viscosity is very sensitive to the change of melt structure, the different mechanisms to lower viscosity in two different slag systems were discussed. The results would provide valuable guidance for the design of metallurgical slag systems.

2. Calculation

In the MD simulations, the approximation of the Buckingham form was used as the pair potential   

$$U_{ij}(r)=\frac{q_i{q}_j}{r_{ij}}+A_{ij}exp\left(-\frac{r_{ij}}{{\rho }_{ij}}\right)-\frac{C_{ij}}{r_{ij}^6}$$(1)

Where U_ij(r) is the interatomic pair potential, q_i and q_j are the selected charges, r_ij is the distance between particles i and j, A_ij and ρ_ij are repulsive potential parameters, C_ij is van der Waals force parameter. The parameters of the interaction potentials used in this study are shown in Table 1.

Table 1. Buckingham potential parameters of particle pairs in this study.

Atom 1	Atom 2	Aij/(eV)	ρij/(Å)	Cij/(eV·Å6)
Si	Si	4142.15	0.16	0
Si	Ca	26674.68	0.16	0
Si	O	62794.37	0.165	0
Si	F	43406.00	0.165	0
Ca	Ca	329051.6	0.16	4.355
Ca	O	717827.0	0.165	8.67
Ca	F	496191.5	0.165	8.67
O	O	1497049.0	0.17	17.34
O	F	1046135.4	0.17	17.34
F	F	730722.8	0.17	17.34
Al	Al	4142.149	0.16	0.0
Al	Ca	36918.57	0.16	0.0
Al	O	86057.58	0.165	0.0
Al	F	59481.584	0.165	0.0

Based on the ternary phase diagram of the research system,^13,14) a wide range of the components in the molten state at 1873 K was determined. Set the total number of particles in the MD simulations primitive cell to about 2000. The density of each sample was obtained from the measured data in the reference book.¹⁵⁾ The composition, atomic number, density, and box length of the research system are shown in Table 2.

Table 2. Composition, atomic number, density and box length of different samples at 1873 K.

Group	Mass fraction/(%)					Atomic number					Density/(g/cm3)	Box length/(Å)
	CaO	SiO2	Al2O3	CaF2	Ca	Si	Al	O	F	Total
CSF1	45	45	–	10	439	354	–	1087	120	2000	2.69	30.75
CSF2	40	40	–	20	462	318	–	976	244	2000	2.68	30.9
CSF3	35	35	–	30	486	281	–	863	370	2000	2.65	31.14
CSF4	30	30	–	40	510	243	–	746	500	1999	2.64	31.3
CSF5	25	25	–	50	535	205	–	630	630	2000	2.63	31.44
CSF6	20	20	–	60	560	166	–	510	764	2000	2.60	31.7
CAF1	45	–	45	10	444	–	420	1013	122	1999	2.81	30.42
CAF2	40	–	40	20	466	–	378	910	246	2000	2.76	30.7
CAF3	35	–	35	30	490	–	334	804	374	2002	2.71	31.0
CAF4	30	–	30	40	514	–	288	695	502	1999	2.65	31.3
CAF5	25	–	25	50	538	–	242	584	634	1998	2.60	31.6
CAF6	20	–	20	60	562	–	196	472	768	1998	2.56	31.9

The periodic boundary conditions were employed for the basic cells in NVT ensemble. The integration of the equation of motion was solved with a time step of 0.1 fs by the leap-frog algorithm. The long-range Coulomb forces were evaluated using the Ewald sum method, and the cut-off radius of the short-range repulsive force was set to 10 Å. For temperature and pressure control, the Nose-Hoover method¹⁶⁾ and Parrinello-Rahman method¹⁷⁾ were used.

In the simulation process, an appropriate number of atoms of each type were placed in the primary MD cell with a random initial state. At the beginning of the simulation, the initial temperature was performed at 4273 K for 20000 steps to mix the system completely and eliminate the effect of the initial distribution. Then, the temperature was decreased to 1873 K through 95000 steps. After equilibrium calculation, the systems relaxed for another 60000 steps. Finally, structure information of melts could be calculated and analyzed. All the calculations were carried out using LAMMPS program.

3. Results and Discussion

Figure 1 presented the radial distribution functions (RDFs) of all particle pairs in the CaO–SiO₂–CaF₂ and CaO–Al₂O₃–CaF₂ system at 1873 K while the content of CaF₂ equal to 40 mass% as an example. The abscissa corresponding to the first peak values of each curve represents the average bond length of the particle pair. The average bond lengths of Si–O, Al–O, Ca–O, Ca–F, and Al–F are around 1.615 Å, 1.745 Å, 2.325 Å, 2.345 Å, 1.895 Å, respectively. It gives a good agreement between the calculated results with the existing values.^11,18,19)

Fig. 1.

The RDFs curves of the research system at [CaF₂] = 40%. (Online version in color.)

3.1. Effect of CaF₂ on the Structure Around the Network Formers

Figure 2 shows the effect of CaF₂ on the bond length and coordination number around Si⁴⁺ which is regarded as the unique network former particle in CaO–SiO₂–CaF₂ system. The bond length of Si–O has not any obvious change with the CaF₂ addition in the system. The 4-coordinate structure of Si–O is very stable which is hardly affected by the content of CaF₂. The Si–F coordination structure in the system can be almost ignored. With the increased CaF₂ content, the distance between any two Si⁴⁺ which act as network former particles becomes longer, and the coordination number between oxygen particles which are the only ones that can coordinate with network former particles is decreasing. It shows that addition CaF₂ can relax the network structure. The calculated results with MD agree with the experimental study of Luth.⁵⁾

Fig. 2.

Effect of CaF₂ on the structure of Si⁴⁺ in CaO–SiO₂–CaF₂ system. (Online version in color.)

As shown in Fig. 3, there is almost no effect on the bond lengths of Al–O and Al–F with increasing content of CaF₂ in CaO–Al₂O₃–CaF₂ system. But the coordination number of Al–O gradually decreases and the average coordination number of Al–F is changed from 0.04 to 0.24. These findings indicate that F⁻ can replace O²⁻ to form a [AlO₃F]⁴⁻ structure in the local area which is consistent with the experimental results in the slag system through spectral by Gao et al.⁹⁾

Fig. 3.

Effect of CaF₂ on the structure of Al³⁺ in CaO–Al₂O₃–CaF₂ system. (Online version in color.)

3.2. Effect of CaF₂ on the Structure Around Network Modifiers

Ca²⁺ ion in slag system is well known for its role in modifying the network structure. The bond lengths of Ca–O and Ca–F do not change with the increased content of CaF₂ in the two research systems as shown in Fig. 4, but the average coordination number of Ca–O and Ca–F changes significantly following a same change rule. There is a dynamic equilibrium phenomenon between Ca²⁺ and the coordinated anions (O²⁻ and F⁻). When the coordination number of Ca–O decreases, the coordination number of Ca–F increases simultaneously, and the total coordination number between Ca²⁺ and coordinated anions always stays within 6–7. This regular change illustrates that Ca²⁺ ions originally exist in the network to modify the network and they are gradually surrounded by F⁻, distributed as Ca–F clusters in network structure after the CaF₂ content increases.

Fig. 4.

Effect of CaF₂ on the structure of Ca²⁺ in the research system. (Online version in color.)

3.3. Effect of CaF₂ on the Distribution of Oxygen Types

It can be seen from Fig. 5, in CaO–SiO₂–CaF₂ system, oxygen mainly exists in types of free oxygen (O_f), non-bridge oxygen (O_nb) and bridge oxygen (O_b). The proportion of various oxygen types vary little with the content of CaF₂, which is consistent with the XPS experimental data by Hayashi.⁶⁾ In CaO–Al₂O₃–CaF₂ system, a small amount of oxygen triclusters exists to balance the charge. The proportion of various oxygen types change obviously with the different content of CaF₂. The increased percentage of free oxygen and non-bridge oxygen indicates that CaF₂ can make complex network structures to be simple.

Fig. 5.

Effect of CaF₂ on the distribution of oxygen types in the research system. (Online version in color.)

In this study, the ratio of non-bridge oxygen and bridge oxygen is approximately 3:1 in CaO–SiO₂–CaF₂ system, while the value over a wide range of CaF₂ content is approximately 1:1 in CaO–Al₂O₃–CaF₂ system. These results can be explained by assuming that the DOP in CaO–SiO₂–CaF₂ system is lower than that in CaO–Al₂O₃–CaF₂ system. If the network structure units are regarded as unique structure which is used to indicate the complexity of the system, the Si–O tetrahedrons mainly exist as a dimer while the Al–O tetrahedrons link with chain, as shown in Fig. 6. This is just a simple illustration, and the actual situation is much more complicated.

Fig. 6.

Main types of Si–O tetrahedron and Al–O tetrahedron in the research system. (Online version in color.)

3.4. Effect of CaF₂ on Distribution of Structural Units of Qⁿ

Figure 7 exhibits the effect of CaF₂ on distribution of structure units of Qⁿ in both systems. In CaO–SiO₂–CaF₂ system, the sum of the structural units Q⁰, Q¹ and Q² with a low DOP remains stable, while the sum of the structural units Q³ and Q⁴ with a high DOP also hardly changes with the CaF₂ content. It may be that the increased content of CaF₂ does not depolymerize the network structure of the CaO–SiO₂–CaF₂ system.

Fig. 7.

Effect of CaF₂ on distribution of structural units of Qⁿ in the research system. (Online version in color.)

In CaO–Al₂O₃–CaF₂ system, with the increased content of CaF₂, the structural units Q³ and Q⁴ with a high DOP will shift to the low DOP units such as Q⁰, Q¹, and Q². This finding shows that it is a significant depolymerization effect on the network structure of CaO–Al₂O₃–CaF₂ system with CaF₂ addition.

3.5. Mechanism of Lower Viscosity of Slags by CaF₂ Addition

It is well known that addition of CaF₂ could lower the viscosity of the slag system. Through our study, the mechanism of CaF₂ lowers viscosity in CaO–SiO₂–CaF₂ is different from that in CaO–Al₂O₃–CaF₂ slag systems. In CaO–SiO₂–CaF₂ system, Bills²⁰⁾ and Hayashi⁶⁾ had developed a structural model as follows: in pure silicate glass, divalent calcium ions bind the silicate anions together by electrostatic force, which plays a role in modifying the network structure. When CaF₂ is added, the added F⁻ will preferentially coordinate with Ca²⁺, and the Ca–F ion pair is added to the anion in Si–O tetrahedron, which means the original electrostatic binding is destroyed. As the silicate anions are reduced in electrostatic bonding through divalent calcium ions, the flow resistance is reduced, thus the viscosity is lowered, as shown in Fig. 8(a). Our results indicate that the fluorine in the slag composition is dominantly coordinated with calcium. Although the type of network structural unit has not changed, the relative proportion of network structural unit will decrease with the increased content of CaF₂. CaF₂ actually acts as network diluent, and creates the space where particles can move longer, which supports the hypothetical effect of CaF₂ proposed by Bills²⁰⁾ and Hayashi.⁶⁾ In CaO–Al₂O₃–CaF₂ system, with the increased content of CaF₂, the network structure will change to a simplification besides dilution network structure, at the same time, the transition from [AlO₄]⁵⁻ to [AlO₃F]⁴⁻ structure will occur. The synthetical effect of the two transformations can depolymerize the melt network structure, as shown in Fig. 8(b), which provides a reasonable micro-explanation for CaF₂ to improve the fluidity of CaO–Al₂O₃–CaF₂ slag.

Fig. 8.

Hypothetical effect of CaF₂ addition on network structure in (a) CaO–SiO₂ based and (b) CaO–Al₂O₃ based systems.

4. Conclusions

Molecular dynamic simulation was carried out in CaO–SiO₂–CaF₂ and CaO–Al₂O₃–CaF₂ systems to study the structure properties and mechanism of lower viscosity in these systems. The following conclusions could be obtained:

(1) The coordination number analysis of the network formers indicates that the structure of Si–O tetrahedron is more stable than that of Al–O tetrahedron.

(2) With the addition of CaF₂, there is a dynamic equilibrium between Ca²⁺ and the coordination anions (O²⁻ and F⁻) in both systems, and the total coordination number is maintained between 6 and 7. Ca²⁺ ions that originally existed to modify the network are gradually surrounded by F⁻, distributed as Ca–F clusters in network structure.

(3) The DOP of network structure in CaO–SiO₂–CaF₂ system is lower than that in CaO–Al₂O₃–CaF₂ system through the analysis of the distribution of oxygen types. If it could be simply regard as unique structural unit, the Si–O tetrahedrons mainly exists as a dimer while the Al–O tetrahedrons link with chain.

(4) In CaO–SiO₂–CaF₂ system, the structural units of the Si–O network will not change with the addition of CaF₂, but the Ca–F clusters can break the electrostatic bond of Ca–O connected to the network. CaF₂ actually acts as network diluent, and creates the space where particles can move longer, which the slag viscosity is lowered on a macro scale.

(5) In CaO–Al₂O₃–CaF₂ system, the Al–O tetrahedral structure in the system can not only be transformed from the complex (Q⁴ and Q³) to simple (Q² and Q¹) structure and also from the Al–O tetrahedral [AlO₄]⁵⁻ to [AlO₃F]⁴⁻ structure, resulted to depolymerize the network structure of the melt. These observations from atomic scale well explain that the addition of CaF₂ can improve the fluidity of CaO–Al₂O₃–CaF₂ slag.

Acknowledgment

The authors would like to deeply acknowledge the financial support by the National Natural Science Foundation of China (No. 51874082 and 51774087).

References

• 1)   K. H.  Moon,  M. S.  Park,  S.  Yoo,  J. K.  Park,  J. W.  Cho and  G.  Shin: Proc. 8th Pacific Rim Int. Cong. on Advanced Materials and Processing, ed. by F. Marquis, Springer, Cham, (2013), 735.
• 2)   S. C.  Zhou: Metallurgist, 57 (2013), 510.
• 3)   B.  Lu,  K.  Chen,  W.  Wang and  B.  Jiang: Metall. Mater. Trans. B, 45 (2014), 1496.
• 4)   J.  Qi,  C.  Liu and  M.  Jiang: Can. Metall. Q., 56 (2017), 212.
• 5)   R. W.  Luth: Am. Mineral., 73 (1988), 297.
• 6)   M.  Hayashi,  N.  Nabeshima,  H.  Fukuyama and  K.  Nagata: ISIJ Int., 42 (2002), 352.
• 7)   J. H.  Park,  D. J.  Min and  H. S.  Song: ISIJ Int., 42 (2002), 344.
• 8)   J. H.  Park,  D. J.  Min and  H. S.  Song: ISIJ Int., 42 (2002), 38.
• 9)   J.  Gao,  G.  Wen,  T.  Huang,  B.  Bai,  P.  Tang and  Q.  Liu: J. Am. Ceram. Soc., 99 (2016), 3941.
• 10)   T.  Asada,  Y.  Yamada and  K.  Ito: ISIJ Int., 48 (2008), 120.
• 11)   Y.  Sasaki,  M.  Iguchi and  M.  Hino: ISIJ Int., 47 (2007), 643.
• 12)   G.  Fan,  S.  He,  T.  Wu and  Q.  Wang: Metall. Mater. Trans. B, 46 (2015), 2005.
• 13)   A. K.  Chatterjee and  G. I.  Zhmoidin: J. Mater. Sci., 7 (1972), 93.
• 14)   I. D.  Sommerville and  D. A. R.  Kay: Metall. Trans., 2 (1971), 1727.
• 15)  VDEh, ed.: Slag Atlas, Verlag Stahleisen, Düsseldorf, (1995), 326.
• 16)   S.  Nosé: Mol. Phys., 52 (1984), 255.
• 17)   M.  Parrinello and  A.  Rahman: J. Appl. Phys., 52 (1981), 7182.
• 18)   K.  de Boer,  A. P. J.  Jansen and  R. A.  van Santen: Phys. Rev. B, 52 (1995), 12579.
• 19)   K.  Zheng,  Z.  Zhang,  F.  Yang and  S.  Sridhar: ISIJ Int., 52 (2012), 342.
• 20)   P. M.  Bills: J. Iron Steel Inst., 201 (1963), 133.