*3.5.1 Melts of Fe-C-O at EML*

*Magnetic Materials and Magnetic Levitation*

sion in the liquid slag phase.

temperature.

and experiment [7].

~5 g with an opening in which a slag powder weighing 0.35 g was laid; the hole was tightly closed with a thin foil lid of the same alloy. The initial sample was placed in an inductor for subsequent levitation, which, depending on the task, was carried out in vacuum, inert gases and carbon monoxide. After the generator was turned on, levitation and melting of the sample took place with heating to the experimental temperature, and the metal melt took an egg-shaped form, on the lower half of which slag was collected. At the beginning of the experiment, when the capsule was melted, the slag was in the form of a thin film enveloping the entire drop. The experimental time required to reach chemical equilibrium was determined by diffu-

Using levitation, the sulfur distribution between iron-carbon melts and lime-alumina slag was studied [7, 28, 32]: the dependence of the equilibrium sulfur distribution coefficient on the C content, gas phase composition and

Experimental results are shown in **Figure 14**. Comparison of the reduced coefficients of the equilibrium distribution of sulfur, obtained by numerical modeling using known thermodynamic data and the experimental equilibrium values of the reduced coefficient of the sulfur distribution, is shown in **Figure 15**. Since the known data for the equilibrium constant of the desulfurization reaction differ by one and a half orders of magnitude, this noticeably affects the reduced sulfur distribution coefficient. On the whole, it is necessary to recognize such a discrepancy as completely admissible and justified. As can be seen in **Figure 15**, in the logarithmic coordinates, the calculated and experimental curves slightly differ in slope, which is due to tolerances in the calculation

*Dependence of the experimental (2) and calculated (1) coefficients of the distribution of sulfur between liquid* 

*Experimental dependence of the reduced coefficient of distribution of 35S between Fe-C and oxide slag melts on* 

*the activity of C in iron. 1—PCo = 1 atm, 2000°C; 2—PHe = 1 atm, 1750°C; 3—PAr = 1 atm, 2000°C.*

*Fe-C and slag on the activity of C in iron at Pco = 1 atm and 2000°C.*

**182**

**Figure 15.**

**Figure 14.**

For comparing EML with crucible melting, the study was made of the influence of the refractory lining on the decarburization kinetics of Fe-C-O samples, which were carried out on melts in corundum crucibles, degassed in vacuum at 1700°С. The data obtained showed that the composition of the products of the decarburization reaction is close to equilibrium; however, the total volume of gases released from the metal depends on the melting method. In melts in a crucible, the gas evolution of CO and CO2 always exceeded the gas evolution during levitation of similar samples. For stable levitation of samples of this system, a setup was used with an inductor inside the chamber; the working pressure in which was 10<sup>−</sup><sup>7</sup> atm and a metal temperature of 2000 ± 30°C. The received dependence (see **Figure 16**, area 6) is qualitatively confirmed by the data obtained from the smelting of Fe-C-O alloys in an electron beam setup at 1550°C and a vacuum of 10<sup>−</sup><sup>7</sup> atm (see **Figure 16**, area 3). The calculated dependence of the O content on C for 2000°C confirms the disproportionate relationship between the deoxidation capacity of carbon and the partial pressure of CO in the gas phase. The observed decrease in the O concentration in the metal is due to a change in the deoxidizing ability of C due to a change in Pco. Apparently, the data obtained for melting in an inert atmosphere characterize the maximum possible increase in the deoxidizing ability of C at 2000°C.

The increase in the deoxidizing ability of C dissolved in liquid iron, which occurs as a result of a decrease in Pco over the melt, is observed to a certain limit determined by the kinetics of the CO bubble growth. Since the deoxidizing ability of C is usually expressed by the product of the concentrations of C and O in the metal m = [% C] × [% O], the dependence of the deoxidizing ability on the partial pressure of CO can be expressed by a parabola. Under levitation conditions, when liquid metal does not come into contact with a refractory lining and there is no influence of the lining on the formation of CO bubbles, volume decarburization should prevail, not excluding the evaporation of CO molecules from the surface of a liquid droplet. It is likely that the contribution of evaporation to the removal of CO from a levitated drop of liquid metal should decrease as O adsorption in the surface layer decreases. In low-carbon iron, the formation of CO bubbles in a metal volume is facilitated by thermodynamic and kinetic factors. If the nucleus is comparable in size to large nonmetallic inclusions, the value 2*δ*/*r* is much lower than atmospheric pressure. In this regard, as the partial pressure of CO in the gas phase decreases, the

#### **Figure 16.**

*Solubility O in Fe-C-O melts. 1—Pco = 1 atm, 2000°C; 2—Pco = 1 atm, 2000°C; 3—Pco = 10<sup>−</sup><sup>7</sup> atm, 1550°C, EBM; 4—Pco = 10<sup>−</sup><sup>7</sup> atm, 1600°C (refractory crucible); 5—Pco = 1 atm, 1600°C (refractory crucible); 6— Pco = 10<sup>−</sup><sup>7</sup> atm, 2100°C.*

decarburization reaction intensifies until this pressure is equal to the partial pressure of CO in the gas nucleus and the deoxidizing ability of C becomes constant. The surface decarburization of carbon iron cannot be decisive due to the low diffusion rates of C and O to the metal-gas interface. When bubbles form in the metal volume, the pressure necessary to overcome the forces of surface tension should be two to three orders of magnitude higher than atmospheric. It follows that in iron-carbon melts, in particular at contents >1.5% C, with a decrease in the partial pressure of CO in the gas phase, the deoxidation ability of C in iron should not change. The description of EML experiments on Co and Ni is not given here because of limited space in the chapter.

### *3.5.2 Melts of Nb-C-O at EML*

The interaction of C and O should be taken into account when obtaining pure metals and studying their surface and bulk properties. However, the study of the behavior of these impurities in solid and liquid metals was often qualitative because of the complex nature of physical and chemical processes in the bulk and on the surface of the metal and because of significant experimental difficulties. A model was created within the framework of which a closed system of equations was obtained [34, 35]. The model allows describing the kinetics of the interaction of C and O during maintaining them in a liquid state in vacuum. It was found that if C and O impurities with initial concentrations of *N*0(0) and *N*c(0) are uniformly dissolved in the metal volume, then by thermal desorption of CO and MO molecules in vacuum (at high contents of oxygen, MO2), since the average О concentration *N*0(t) decreases infinitely, then, in accordance with the proposed model, the average С concentration *N*c(t) should reach a definite threshold level *N*c(∞). It is shown that the ratio between the average concentrations of O and C is uniquely fixed by two parameters, *N*\* and *S*. The first of them, *N*\* = *ω*/*G*, is determined by the ratio of the constant of the MO desorption rate *ω* to the effective constant of the CO desorption rate *G* and has a certain critical value, which sets the characteristic concentration scale. The effect of one impurity on another becomes high only when its concentration beats this critical value. The parameter *S* is a dimensionless indicator of the relative intensity of diffusion or surface processes.

The relative simplicity of the model makes it possible to experimentally check the behavior of the average concentrations of carbon *N*c(*t*) and oxygen *N*0(*t*) in time, as far as the created model [34, 35] corresponds to reality and determines the parameters *N* and *S*. So, if the model is correct, then in the coordinates ∂∆/∂ (In *N*c) − ∆, regardless of the ratio of the initial concentrations of *N*c(0) and *N*o(0), the experimental points should be on a common line with the fixed slope and ordinate.

The comparison of the results with the theoretical ones becomes much easier if the *S* values in the experiment are sufficiently large, which means that diffusion is more intensive than surface processes. In this case, the dependence In *N*c = *f* (∆) should be close to linear (**Figure 17**). Thus, the experimental points for different ratios of the initial concentrations O and C in the In *N*<sup>c</sup> − ∆ coordinates should fit on parallel straight lines with a slope 1/*N*.

The kinetics of the interaction of C and O was investigated in a wide range of ratios of their initial concentrations during levitation of Nb and Mo in vacuum. The temporal relationship between the average concentrations of C and O in a wide range ∆ follows a simple law. The kinetics of conducting of the average concentrations of C and O was calculated numerically, which showed a good alignment between the digital simulation and the experiment for all studied series. The effective constant of the CO desorption rate and the high-temperature sticking

**185**

**4. Conclusions**

**Figure 17.**

*Electromagnetic Levitation of Metal Melts DOI: http://dx.doi.org/10.5772/intechopen.92230*

coefficient of CO to Nb and Mo were received. The description of EML experiments

The main advantage that attracted the attention of researchers was the lack of contact of liquid metal with refractory lining, which ensured the elimination of one of the main sources of metal contamination by such a harmful impurity, such as oxygen. This is especially important for melting refractory highly reactive metals and semiconductors. Compared to other melting methods, which also ensured the absence of contact of liquid metal with the crucible, EML of liquid metals has a number of significant advantages: adjustable residence time of a drop of metal in a liquid state; controlled gas atmosphere and slag phase; controlled metal temperature (from melting temperatures to boiling); ability to use an additional heat source (electron beam, laser beam or plasma); vigorous stirring of metal by electromagnetic field; possibility of introducing alloying additives into a liquid drop; a favorable ratio between the surface of the droplet and its volume for the passage of heterogeneous

> –106 °C/s.

*(∞), plotted in coordinates In* N*<sup>c</sup> − ∆ (lines A* 

on Mo is not given here because of limited space in the chapter.

*and D) and In* N*<sup>c</sup> − In(1 + ∆/*N*′) (lines C and B) on the basis experimental data.*

*Mean C concentrations as a function of the value ∆ =* N*<sup>o</sup> −* N*c +* N*<sup>c</sup>*

reactions; and achieving extremely high crystallization rates up to 105

The noncontact of a liquid sample is the essence of EML, combined with an ultra-clean environment, which is an excellent instrument for researches. In addition, ELM is one of the oldest noncontact methods of levitation used in materials science experiments for decades. EML is the most mature of all noncontact melting methods and has been used for decades in ground-based experiments, as well as in microgravity experiments with a wide range of alloys. EML in gravitational conditions has some problems associated with gravitational forces, so if levitation is performed under microgravity conditions, then only small levitation forces are

*Electromagnetic Levitation of Metal Melts DOI: http://dx.doi.org/10.5772/intechopen.92230*

**Figure 17.**

*Magnetic Materials and Magnetic Levitation*

space in the chapter.

*3.5.2 Melts of Nb-C-O at EML*

relative intensity of diffusion or surface processes.

on parallel straight lines with a slope 1/*N*.

decarburization reaction intensifies until this pressure is equal to the partial pressure of CO in the gas nucleus and the deoxidizing ability of C becomes constant. The surface decarburization of carbon iron cannot be decisive due to the low diffusion rates of C and O to the metal-gas interface. When bubbles form in the metal volume, the pressure necessary to overcome the forces of surface tension should be two to three orders of magnitude higher than atmospheric. It follows that in iron-carbon melts, in particular at contents >1.5% C, with a decrease in the partial pressure of CO in the gas phase, the deoxidation ability of C in iron should not change. The description of EML experiments on Co and Ni is not given here because of limited

The interaction of C and O should be taken into account when obtaining pure metals and studying their surface and bulk properties. However, the study of the behavior of these impurities in solid and liquid metals was often qualitative because of the complex nature of physical and chemical processes in the bulk and on the surface of the metal and because of significant experimental difficulties. A model was created within the framework of which a closed system of equations was obtained [34, 35]. The model allows describing the kinetics of the interaction of C and O during maintaining them in a liquid state in vacuum. It was found that if C and O impurities with initial concentrations of *N*0(0) and *N*c(0) are uniformly dissolved in the metal volume, then by thermal desorption of CO and MO molecules in vacuum (at high contents of oxygen, MO2), since the average О concentration *N*0(t) decreases infinitely, then, in accordance with the proposed model, the average С concentration *N*c(t) should reach a definite threshold level *N*c(∞). It is shown that the ratio between the average concentrations of O and C is uniquely fixed by two parameters, *N*\* and *S*. The first of them, *N*\* = *ω*/*G*, is determined by the ratio of the constant of the MO desorption rate *ω* to the effective constant of the CO desorption rate *G* and has a certain critical value, which sets the characteristic concentration scale. The effect of one impurity on another becomes high only when its concentration beats this critical value. The parameter *S* is a dimensionless indicator of the

The relative simplicity of the model makes it possible to experimentally check the behavior of the average concentrations of carbon *N*c(*t*) and oxygen *N*0(*t*) in time, as far as the created model [34, 35] corresponds to reality and determines the parameters *N* and *S*. So, if the model is correct, then in the coordinates ∂∆/∂ (In *N*c) − ∆, regardless of the ratio of the initial concentrations of *N*c(0) and *N*o(0), the experimental points should be on a common line with the fixed slope and ordinate.

The comparison of the results with the theoretical ones becomes much easier if the *S* values in the experiment are sufficiently large, which means that diffusion is more intensive than surface processes. In this case, the dependence In *N*c = *f* (∆) should be close to linear (**Figure 17**). Thus, the experimental points for different ratios of the initial concentrations O and C in the In *N*<sup>c</sup> − ∆ coordinates should fit

The kinetics of the interaction of C and O was investigated in a wide range of ratios of their initial concentrations during levitation of Nb and Mo in vacuum. The temporal relationship between the average concentrations of C and O in a wide range ∆ follows a simple law. The kinetics of conducting of the average concentrations of C and O was calculated numerically, which showed a good alignment between the digital simulation and the experiment for all studied series. The effective constant of the CO desorption rate and the high-temperature sticking

**184**

*Mean C concentrations as a function of the value ∆ =* N*<sup>o</sup> −* N*c +* N*<sup>c</sup> (∞), plotted in coordinates In* N*<sup>c</sup> − ∆ (lines A and D) and In* N*<sup>c</sup> − In(1 + ∆/*N*′) (lines C and B) on the basis experimental data.*

coefficient of CO to Nb and Mo were received. The description of EML experiments on Mo is not given here because of limited space in the chapter.

### **4. Conclusions**

The main advantage that attracted the attention of researchers was the lack of contact of liquid metal with refractory lining, which ensured the elimination of one of the main sources of metal contamination by such a harmful impurity, such as oxygen. This is especially important for melting refractory highly reactive metals and semiconductors. Compared to other melting methods, which also ensured the absence of contact of liquid metal with the crucible, EML of liquid metals has a number of significant advantages: adjustable residence time of a drop of metal in a liquid state; controlled gas atmosphere and slag phase; controlled metal temperature (from melting temperatures to boiling); ability to use an additional heat source (electron beam, laser beam or plasma); vigorous stirring of metal by electromagnetic field; possibility of introducing alloying additives into a liquid drop; a favorable ratio between the surface of the droplet and its volume for the passage of heterogeneous reactions; and achieving extremely high crystallization rates up to 105 –106 °C/s.

The noncontact of a liquid sample is the essence of EML, combined with an ultra-clean environment, which is an excellent instrument for researches. In addition, ELM is one of the oldest noncontact methods of levitation used in materials science experiments for decades. EML is the most mature of all noncontact melting methods and has been used for decades in ground-based experiments, as well as in microgravity experiments with a wide range of alloys. EML in gravitational conditions has some problems associated with gravitational forces, so if levitation is performed under microgravity conditions, then only small levitation forces are

required to compensate for residual accelerations. With an inductor optimized for microgravity, heating and positioning of the samples can be carried out almost independently.
