Influence of CaF₂ on the Viscosity and Structure of Manganese Ferroalloys Smelting Slags

Abstract

Addition of CaF₂ to the CaO-SiO₂-MnO (CaO/SiO₂ = 0.5) system, which corresponds qualitatively to a silicomanganese ferroalloy smelting slag, affected not only the critical (crystallization) temperature (T _CR) but also the viscosity at high temperatures, and its influence on slag properties was strongly dependent on the content of MnO in the slag. The viscosity of CaF₂-free 10 mass pct MnO slag was relatively high, i.e., about 10 dPa s at 1773 K (1500 °C), but decreased continuously upon addition of CaF₂ to the system. In contrast, the viscosity of the 40 pct MnO system was very low, i.e., 1 dPa s at 1773 K (1500 °C), and CaF₂ did not have a large effect. This indicates that Mn²⁺ is a strong network modifier in manganese ferroalloy smelting slags. Nevertheless, CaF₂ addition was very effective at decreasing the viscosity of low MnO slags at low temperatures. The activation energy for the viscous flow of silicate melts decreased linearly in response to CaF₂ addition, but this tendency was less pronounced in the more basic composition of the slag. The effect of CaF₂ on the viscosity and activation energy for viscous flow of melts was analyzed quantitatively using micro-Raman spectra of quenched glass samples and the silicate polymerization index, i.e., Q³/Q² ratio. The polymerization index decreased continuously with increasing CaF₂ content in less basic (10 pct MnO or C/S = 0.5) slags, whereas it was not affected by CaF₂ content in highly basic (40 pct MnO and C/S = 1.0) slags. Bulk thermophysical properties of the CaO-SiO₂-MnO-CaF₂ slags were quantitatively correlated with the structural information of the slags.

Introduction

Viscosity of silicate melts at high temperatures has been widely investigated because of the close relationship between the macroscopic thermophysical properties of slags and their microscopic molecular structure, which strongly affects reaction efficiency in process metallurgy.[1–26] Recently, we thoroughly evaluated the novel structure–viscosity relationship of the MnO-CaO-SiO₂ slag system using Raman spectroscopy due to the importance of this system in smelting and refining processes for the production of manganese ferroalloys.[21,22]

Enhancement of slag-metal reaction kinetics is desirable to increase the production rate and decrease the cost of smelting and refining processes for Mn ferroalloys. Fluorspar (CaF₂) addition may improve slag-metal reaction kinetics of MnO-containing slags at high temperatures by decreasing the viscosity of the slags, because CaF₂ is well known to decrease the viscosity of silicate melts.[3–12,15,19,20,24,25] The effect of CaF₂ on the viscosity of various silicate melts has been investigated experimentally by several researchers for CaO-SiO₂ (-MgO, -MnO)-CaF₂, [3–5,10–12,24,25] FeO(-CaO)-SiO₂-CaF₂,[6–8] CaO-SiO₂-Na₂O-Li₂O-CaF₂, [19] and CaO-Al₂O₃-SiO₂-MgO-CaF₂ systems.[20]

The above studies showed that viscosity itself and the activation energy of Newtonian flow of silicate melts generally decreased with increasing CaF₂ content due to depolymerization reactions of silicate networks, indicating that the addition of CaF₂ is more effective at higher silica concentrations. However, the effect of CaF₂ on the viscosity of the FeO-SiO₂ slags was not significant, in contrast to its effects on the CaO-SiO₂ slags.[6] In MO (M = Mg, Na₂)-containing slags, the addition of CaF₂ had a limited effect on viscosity within a specific range of CaF₂ concentration due to the contribution of basic oxides such as MgO and Na₂O in addition to CaO to silicate depolymerization.[10–12,19] Furthermore, for highly basic slags with a silica content lower than about 10 mass pct, CaF₂ suppressed the precipitation of solid phases at lower temperatures, whereas the effect of CaF₂ on viscosity itself was less significant when the slags were completely liquid at high temperatures.[20,26]

Little experimental work has been done to determine the effect of CaF₂ on the viscosity of MnO-containing silicate melts,[25] even though there is some experimental data regarding the viscosity of the CaO-MnO-SiO₂ ternary slags.[1,2,7] Therefore, in the current study, we measured the viscosity of the CaO-SiO₂-MnO-CaF₂ slags with the composition of CaO/SiO₂ = 0.5 (mass pct ratio) to clarify the effect of CaF₂ on the viscous flow of molten slags at high temperatures. In addition, the Raman spectra of quenched glass samples were quantitatively analyzed to investigate the structural role of CaF₂ in depolymerization of silicate networks with reference to our previous studies. [21–23,25–27]

Experimental

All slag samples were prepared using reagent grade chemicals such as MnO, SiO₂, CaF₂, and CaO calcined from CaCO₃ at 1273 K (1000 °C) for 10 hours, and pre-melted slags were used in experiments. Pre-melting was carried out using a Pt-10 mass pct Rh alloy crucible (diameter 40 mm, depth 60 mm) under a purified Ar atmosphere in a Super Kanthal electric furnace.

The rotating cylinder method was used in the current study and a schematic diagram of the experimental apparatus is shown in Figure 1. A rotating viscometer (Brookfield, model LV-DV II+ Pro) was set on the Super Kanthal electric furnace. The temperature was measured using an R-type (Pt/Pt-13 mass pct Rh) thermocouple. The experimental temperature range was about 1373 K to 1873 K (1100 °C to 1600 °C).

Fig. 1

Experimental apparatus used to measure slag viscosity

Viscosity was measured in the current study using a Pt-10 mass pct Rh alloy spindle, crucibles, and suspending wire, the dimensions of which are listed in Table I. The experiment was initiated by placing the crucible, containing pre-melted slag, inside the reaction chamber at 1873 K (1600 °C). Then, the spindle, rotating at a speed of 100 rpm, was lowered into the slag. The tip of the bob was placed about 5 mm above the base of the crucible and about 5 mm of the shaft was immersed in the melt. The equilibration time was approximately 20 minute at each temperature. Standard oil samples with a viscosity of 0.0486, 0.0952, 0.50, and 0.985 Pa s (0.486, 0.952, 5.0, and 9.85 poise) were used to calibrate the spindle.

Table I Dimensions of the Crucible, Spindle, and Suspending Wire Used to Measure Slag Viscosity (Material; All Pt-10 pct Rh Alloy)

In addition, the CaF₂ content of the slag samples was analyzed using X-ray fluorescence spectroscopy (Bruker, model S4 Explorer) after experiments, because evaporation of fluoride species from the melts was expected.[28] We confirmed that the weight loss of CaF₂ by fluoride emission during viscosity measurements was within about 5 pct point of the initial content, which is similar to what was reported in previous studies.[4,10–12,19,25,26] Hence, the composition in the present article refers to initial content.

The procedures used to prepare glass samples, and quantitative analysis of Raman spectra have been described in detail in previous articles.[21–23,25–27]

Results and Discussion

Effect of CaF₂ Addition on the Viscosity and Critical Temperature of MnO-CaO-SiO₂ Slags

The viscosities of the MnO-CaO-SiO₂ (-CaF₂) slags (CaO/SiO₂ = 0.5) are shown as a function of temperature in Figure 2 for the 10 mass pct MnO system and in Figure 3 for the 40 mass pct MnO system. The viscosity values of the CaF₂-free system at 1673 K, 1773 K, and 1823 K (1400 °C, 1500 °C, and 1550 °C) measured by other researchers are also shown in Figures 2 and 3 as well as are listed in Table II for comparison.[1,2,7] Previously reported values and our present results are in good agreement and within the experimental scatter.

Fig. 2

Viscosities of the CaO-SiO₂-CaF₂-10 pct MnO (C/S = 0.5) slags as a function of temperature at different CaF₂ contents

Fig. 3

Viscosities of the CaO-SiO₂-CaF₂-40 pct MnO (C/S = 0.5) slags as a function of temperature at different CaF₂ contents

Table II Viscosities of the CaO-SiO₂-MnO-CaF₂ [CaO/SiO₂ = 0.5(±0.03)] Slags at Different CaF₂ Contents

The viscosity of the 10 mass pct MnO system was significantly higher than that of the 40 mass pct MnO system at a given CaO/SiO₂ (=0.5) ratio and CaF₂ content. Viscosity gradually increased with decreasing temperature, and this tendency was emphasized as the CaF₂ content decreased when the MnO content was about 10 mass pct (Figure 2). In contrast, the viscosities were generally lower than 5 dPa.s as temperature decreased to about 1523 K (1250 °C) irrespective of the CaF₂ content when the MnO content was 40 mass pct. For the latter in Figure 3, the viscosity drastically increased at a specific temperature, which is referred to as the critical (or crystallization) temperature (T _CR).[10–12,18,19,25]

The viscosities of the 10 mass pct MnO system at various temperatures are plotted against the content of CaF₂ in Figure 4. The viscosity of slags decreased continuously as the content of CaF₂ increased at a given temperature, and the effect of CaF₂ addition was more pronounced at lower temperatures. The same results were obtained in the 40 mass pct MnO system (Figure 5), even though the viscosity value of the latter was about a tenth of the former at a given temperature.

Fig. 4

Effect of CaF₂ addition on the viscosities of the CaO-SiO₂-10 pct MnO (C/S = 0.5) slags at temperatures from 1623 K to 1873 K (1350 °C to 1600 °C)

Fig. 5

Effect of CaF₂ addition on the viscosities of the CaO-SiO₂-40 pct MnO (C/S = 0.5) slags at temperatures from 1573 K to 1873 K (1300 °C to 1600 °C)

The results shown in Figures 4 and 5 confirmed that CaF₂ improves the fluidity of the MnO-CaO-SiO₂ slags with CaO/SiO₂ = 0.5. Furthermore, the effect of CaF₂ on the fluidity of slags was highly significant in the high SiO₂ and low MnO systems at low temperatures. The viscous flow behavior of MnO-based slags shown in Figures 4 and 5 is important for several operating stages during Mn ferroalloys production. For example, interfacial reaction kinetics between the FeMn slag and SiMn melt can be improved by addition of small amounts of CaF₂, resulting in the production of a high purity FeMn alloy by desiliconization.[29]

Activation Energy for the Viscous Flow of MnO-CaO-SiO₂-CaF₂ Slags

The viscosity of the silicate melts is strongly dependent on the degree of polymerization, which is a function of temperature and composition. The relationship between them has generally been discussed by taking the activation energy for viscous flow into account. The activation energy for the viscous flow of the silicate melts can be calculated using the following Arrhenius equation:

$$ \eta = \eta_{\text{o}} { \exp }\left( {\frac{{E_{\eta } }}{RT}} \right), $$

(1)

where η, η _o, E _η , R, and T are viscosity, a pre-exponent constant, the activation energy, a gas constant, and an absolute temperature, respectively. Hence, it is possible to calculate E _η using a plot of ln η vs 1/T. The results are listed in Table III.

Table III Activation Energy for Viscous Flow of CaO-SiO₂-MnO-CaF₂ [CaO/SiO₂ = 0.5(±0.03)] Slags

The Arrhenius plot for the MnO-CaO-SiO₂ (-CaF₂) slags (C/S = 0.5) is shown in Figure 6 as a function of 1/T and the contents of MnO and CaF₂, respectively. Here, the viscosities of the partially crystallized melts in non-Newtonian regions were excluded; that is, only the viscosity of melts that exhibited Newtonian flow was taken into account. The viscosity, ln η, of the melts increased linearly with increasing 1/T and the viscosity of the 40 mass pct MnO system was unambiguously lower than that of the 10 mass pct MnO system at a given temperature and CaF₂ content.

Fig. 6

Arrhenius plot of the CaO-SiO₂-MnO-CaF₂ (C/S = 0.5) slags in the Newtonian flow region

The slope of the line generally decreased with increasing CaF₂ content in both systems, indicating that F⁻ ions contributed to effective depolymerization of the silicate networks. From the phase diagram of the MnO-CaO-SiO₂ ternary slag, the 10 mass pct MnO system with C/S = 0.5 is on the tridymite-type silica primary area, which has a highly polymerized three-dimensional tectosilicate framework structure. This fully polymerized network can be further depolymerized into more simple [Si₂O₄F]-sheets and/or [SiO₂F]-chain structures by addition of CaF₂, for example, as follows:[10–12,15,25]

$$ 2\left[ {{\text{SiO}}_{ 2} } \right]_{(3{\text{D-network}})}+ {\text{F}}^{-} = \left[ {{\text{Si}}_{ 2} {\text{O}}_{ 4} {\text{F}}} \right]_{{({\text{sheet}})}}^{-} $$

(2a)

$$ \left[ {{\text{Si}}_{ 2} {\text{O}}_{ 4} {\text{F}}} \right]_{{({\text{sheet}})}}^{-} + {\text{F}}^{-} = {2}[{{\text{SiO}}_{ 2} {\text{F}}}]_{{({\text{chain}})}}^{-} $$

(2b)

However, the 40 mass pct MnO system with C/S = 0.5 is on the wollastonite (CaSiO₃) primary area in the CaO-SiO₂-MnO phase diagram. The main structural unit of wollastonite (inosilicate or metasilicate subclass) melt is a [SiO₃]-chain structure,[21,30–35] which can be further depolymerized into more simple [Si₂O₆F]-dimers and/or [SiO₃F]-monomers by addition of CaF₂, for example, as follows:[10–12,15,25]

$$ 2\left[ {{\text{SiO}}_{ 3} } \right]_{{({\text{chain}})}}^{{ 2{-}}} + {\text{ F}}^{-} = \, \left[ {{\text{Si}}_{ 2} {\text{O}}_{ 6} {\text{F}}} \right]_{{({\text{dimer}})}}^{{ 5{-}}} $$

(3a)

$$ \left[ {{\text{Si}}_{ 2} {\text{O}}_{ 6} {\text{F}}} \right]_{{({\text{dimer}})}}^{{ 5{-}}} + {\text{ F}}^{-} = { 2}\left[ {{\text{SiO}}_{ 3} {\text{F}}} \right]_{{({\text{monomer}})}}^{{ 3{-}}} $$

(3b)

The depolymerization reactions mentioned above for 10 pct and 40 pct MnO systems are illustrated schematically in Figures 7 and 8, respectively, from the perspective of structural similarity among silicate minerals and melts concluded by Mysen et al.[21,35–39] There were far fewer bond-breaking events that occurred due to supply of F⁻ ions when the MnO content was relatively high in the system, which consisted predominantly of two-dimensional chain structures. This provides us with very important insights into the structural role of F⁻ in silicate melts. Some studies have reported that F⁻ ions are very effective at depolymerizing silicate networks, while other studies have reported that F⁻ ions behave only as a diluent. This inconsistency in the reported role of F⁻ ions is due to the distribution of silicate structural units as discussed above.

Fig. 7

Schematic illustration of the depolymerization reaction of the 10 pct MnO slag system by addition of CaF₂

Fig. 8

Schematic illustration of the depolymerization reaction of the 40 pct MnO slag system by addition of CaF₂

The activation energy for the viscous flow of the CaO-SiO₂-MnO-CaF₂ slags is plotted against the content of CaF₂ at different C/S ratios and MnO contents in Figure 9. Activation energy decreased linearly with increasing content of CaF₂, irrespective of C/S ratio and MnO content. Furthermore, the effect of CaF₂ on activation energy became less significant as the C/S ratio and MnO content increased. This is not only because the dominant fractions of silicates were depolymerized into simpler forms due to the network-modifying roles of Ca²⁺ and Mn²⁺ ions before F⁻ ions were introduced, but also because the absolute amounts of silicates broken into flow decreased as the basicity increased. [21]

Fig. 9

Activation energy for viscous flow of the CaO-SiO₂-MnO-CaF₂ slags as a function of CaF₂ content

Structural Changes in Silicate Melts Due to CaF₂ Addition: Micro-Raman Spectra Analysis

Raman spectroscopy has been used to probe the local anionic structure of silicates.[21,35,40] The bands assigned to antisymmetric stretching of Si-O⁻ (non-bridging oxygen; NBO) and Si-O⁰ (bridging oxygen; BO) bonds occur in the 850 to 1200 cm⁻¹ region, whereas Si-O-Si bending modes are found between about 500 and 700 cm⁻¹.[21,35,41] The frequencies of the stretching modes decrease with decreasing degree of polymerization, viz. increasing NBO/Si. There are several types of units; SiO₂ (fully polymerized), Si₂O₅ (sheet), SiO₃ (chain), Si₂O₇ (dimer), and SiO₄ (monomer) groups.[35–45] In a nuclear magnetic resonance (NMR) study of silicates,[46,47] the stoichiometric notations for each unit were replaced by the so-called Qⁿ concept, wherein the groups described above correspond to Q⁴ (NBO/Si = 0), Q³ (NBO/Si = 1), Q² (NBO/Si = 2), Q¹ (NBO/Si = 3), and Q⁰ (NBO/Si = 4), respectively. We use the Qⁿ concept in the present paper.

The Raman spectra of the CaO-SiO₂-CaF₂-10 pct MnO system (C/S = 0.5) as a function of wavenumbers (Raman shifts) ranged from 700 to 1300 cm⁻¹ are shown in Figure 10. The intensity of the high frequency Q³ (1050 to 1080 cm⁻¹) band was relatively strong in the CaF₂-free high silica system and decreased continuously as the content of CaF₂ increased. However, the relative intensity of the Q² (960 to 970 cm⁻¹) band, which was initially a weak shoulder in the CaF₂-free high silica system, increased significantly with increasing CaF₂ content. A similar tendency was observed for the Q⁰ (860 to 870 cm⁻¹) band, even though it was much less dominant compared with Q² band. A more quantitative analysis of structural changes induced by CaF₂ addition is provided below.

Fig. 10

Raman spectra of quenched CaO-SiO₂-CaF₂-10 pct MnO (C/S = 0.5) slags with CaF₂ content ranging from 0 pct to 15 pct

Relative fractions of silicate anionic units obtained from Gaussian deconvolution of Raman bands shown in Figure 10 are listed in Table IV and plotted against CaF₂ content in Figure 11. Details of the Gaussian deconvolution process are reported in our previous publications.[21–23,25–27] The fraction of the Q³ (Si₂O₅-sheet) unit decreased continuously from about 55 pct to 35 pct as the CaF₂ content increased, whereas the fraction of Q² (SiO₃-chain) and Q⁰ (SiO₄-monomer) units increased from about 40  pct to 52 pct and from about 2 to 12 pct, respectively, as the CaF₂ content increased. This indicated that the large Q³ unit was depolymerized into smaller Q² and Q⁰ units by addition of CaF₂. The relative fraction of the Q¹ (Si₂O₇-dimer) unit was not affected by CaF₂ content.

Table IV Peak Area Fraction Obtained from Gaussian Deconvolution of Raman Spectra

Fig. 11

Relative fractions of Qⁿ (n = 0, 1, 2, 3) structural units for the CaO-SiO₂-CaF₂-10 pct MnO (C/S = 0.5) system as a function of CaF₂ content

Raman spectra of the CaO-SiO₂-CaF₂-40 mass pct MnO system as a function of wavenumbers (Raman shifts) ranged from 700 to 1300 cm⁻¹ are shown in Figure 12. The intensity of the high frequency Q³ (1050 to 1100 cm^-1) band was negligible through the composition range investigated. The relative intensity of the Q⁰ [850(±5) cm⁻¹] band increased slightly, whereas that of the Q² [970(±5) cm⁻¹] band decreased as the CaF₂ content increased, but not significantly. There was no remarkable change in the relative intensity of Q¹ [910(±5) cm⁻¹] band at CaF₂ contents ranging from 0 to 15 mass pct. However, a more quantitative analysis of the structural changes induced by CaF₂ addition is provided below.

Fig. 12

Raman spectra of quenched CaO-SiO₂-CaF₂-40 pct MnO (C/S = 0.5) slags with CaF₂ content ranging from 0 pct to 15 pct

The relative fractions of silicate anionic units obtained from Gaussian deconvolution of Raman bands shown in Figure 12 are listed in Table IV and plotted against CaF₂ content in Figure 13. The fraction of the Q³ (Si₂O₅-sheet) unit was very small for the entire composition range investigated, i.e., C/S = 0.5 and MnO = 40 mass pct, and decreased slightly from about 5 to 1 pct as the CaF₂ content increased. Interestingly, the relative fraction of the Q² (SiO₃-chain) unit decreased slightly from about 75 to 60 pct, while the fractions of Q¹ (Si₂O₇-dimer) and Q⁰ (SiO₄-monomer) units increased from about 5 pct to 10 pct and 15 pct to 30 pct, respectively, as the content of CaF₂ increased. In Figures 11 and 13, the fractions of each Qⁿ unit for the CaF₂-free ternary system exhibited good agreement with the previous measurements,[21] indicating that the Raman analysis performed in this study was highly consistent.

Fig. 13

Relative fractions of Qⁿ (n = 0, 1, 2, 3) structural units for the CaO-SiO₂-CaF₂-40 pct MnO (C/S = 0.5) system as a function of CaF₂ content

The Q³/Q² ratio, which was proposed as a quantitative polymerization index for silicates,[21–23,25,27] was plotted against CaF₂ content (Figure 14). The Q³/Q² ratio on a logarithmic scale decreased continuously with increasing CaF₂ content in less basic (C/S = 0.5 or 10 pct MnO) systems, whereas it did not change as CaF₂ content increased in the highly basic (C/S = 1.0 and 40 pct MnO) systems.[25] This means that CaF₂ contributed effectively to depolymerization of silicate networks in the former system, while it did not in the latter system, as discussed in the previous section.

Fig. 14

Effect of CaF₂ on the polymerization index, viz. Q³/Q² ratio, in the CaO-SiO₂-MnO-CaF₂ slags

In our earlier study,[21,25] we were able to predict the activation energy of the CaO-SiO₂-MnO (-CaF₂) slags as a linear function of the polymerization index, ln (Q³/Q²). The activation energy of the viscous flow of silicate melts in a Newtonian flow region is shown in Figure 15 as a function of ln (Q³/Q²). The activation energy required for the CaO-SiO₂-MnO-CaF₂ slags can be expressed as a linear function of ln (Q³/Q²) within some experimental scatter by linear regression analysis as follows:

Fig. 15

Activation energy for viscous flow of the CaO-SiO₂-MnO-CaF₂ slags as a function of Q³/Q² ratio

$$ E_{\eta } ({\text{kJ}}) = 22.0( \pm 2.1) \cdot \ln \frac{{{\text{Q}}^{ 3} }}{{{\text{Q}}^{ 2} }} + 172.3( \pm 5.1) \left( {r^{ 2} = \, 0. 8 6} \right) $$

(4)

Our Raman spectroscopy results for viscosity measurements and structural analysis of the CaO-SiO₂-MnO-CaF₂ (C/S = 0.5) slags, which correspond qualitatively to SiMn smelting slags, are consistent with those we reported previously for the CaF₂-free ternary system as well as for the CaO-SiO₂-MnO-CaF₂ (C/S = 1.0) slags, which correspond to FeMn smelting slags.[21,25]

Conclusions

The addition of CaF₂ to the CaO-SiO₂-MnO (CaO/SiO₂ = 0.5) system, which corresponds qualitatively to SiMn smelting slag, affected not only the critical (crystallization) temperature (T _CR) but also viscosity at high temperatures, and the influence of CaF₂ on slag properties was strongly dependent on the content of MnO in the slag. The viscosity of the CaF₂-free slag was relatively high, i.e., about 10 dPa s at 1773 K (1500 °C) and 30 dPa s at 1673 K (1400 °C), and decreased continuously by CaF₂ addition in the 10 mass pct MnO system, whereas the viscosity of the 40 pct MnO system was very low, viz. approximately 1/10 of the viscosity of the 10 pct MnO system at a given CaF₂ content. This indicates that Mn²⁺, like Ca²⁺, is a strong network modifier in manganese ferroalloys smelting slags. Nonetheless, CaF₂ addition was very effective at decreasing the viscosity of low MnO slags at low temperatures. For example, the viscosity of 10 pct MnO slag decreased from about 95 dPa.s to 7 dPa s at 1623 K (1350 °C) in response to addition of 15 pct CaF₂.

The activation energy for the viscous flow of silicate melts decreased linearly in response to CaF₂ addition, though less significantly the more basic composition of the slag, i.e., in the 40 pct MnO and C/S =1.0 system, which is consistent with the experimental results reported for various fluorosilicate melts by several research groups. This tendency can be understood using a structural model for the depolymerization process, e.g., “three-dimensional network → sheet → chain → dimer → monomer” in response to CaF₂ addition.

We quantitatively analyzed the effect of CaF₂ addition on the viscosity and activation energy for viscous flow of melts using micro-Raman spectra of quenched glass samples and the concept of a silicate polymerization index, the Q³/Q² ratio. The polymerization index decreased continuously with increasing content of CaF₂ in less basic (10 pct MnO or C/S = 0.5) slags, whereas it was not affected by CaF₂ content in highly basic (40 pct MnO and C/S = 1.0) slags. Consequently, the bulk thermophysical properties of the CaO-SiO₂-MnO-CaF₂ slags are quantitatively correlated to the microscopic molecular (or ionic) structure of the slags.