2.1.2 Deoxidation of Molten Steel2.2.2 Thermodynamic of DeoxidationThe dissolution of oxygen in molten steel may be represented by equation

where [O] denotes oxygen dissolved in the metal as atomic oxygen. For the above reaction,

where KO is equilibrium constant, for Reaction (1), pO2 denotes partial pressure of oxygen in the gas phase in atmosphere, and hO is the activity of dissolved oxygen in liquid steel with reference to the 1 wt.% standard state. KO is related to temperature as

Again,

hO = [ fO][WO]

where WO denotes the concentration of dissolved oxygen in weight percent, and fO is the activity coefcient of dissolved oxygen in steel in 1 wt.% standard state. In pure liquid iron,log fo = 0.17[WO]

The above relations would allow us to estimate WO in liquid iron at any value of pO with which the molten iron would be brought to equilibrium. This value of WO is nothing but solubility of [O] at that pO. However, oxygen tends to form stable oxides with iron. Therefore, molten iron becomes saturated with [O] when the oxide starts forming, i.e., when liquid iron and oxide are at equilibrium. This oxide, in its pure form, is denoted as FexO, where x is approximately 0.985 at 1600C. For the sake of simplicity we shall take x equal to 1 often and designate this compound as FeO.For the reaction FexO(1) = xFe(1) + [O]wt.%.

Where

Here, aFe = the activity of Fe in the metal phase in the Raoultian scale (approximately 1), and aFe O denotes the activity of FexO in oxide phase. If the FeO is not pure and is present in an oxide slag, then aFeO < 1, and h (i.e., solubility of [O] in equilibrium with the slag) would be less.

Example 1

Calculate the concentration of oxygen in molten iron at 1600C in equilibrium with (a) pure FexO, and (b) a liquid slag of FeO-SiO2 containing 40 mol.% SiO2.

Solution

or,logKFe = log fO + logWO logaFeO

= 0.17 + logWO logaFeO(8)

Again, at 1600C, from Eq. (5.6),

log KFe = 0.672(9)

(a) In pure FeO, aFeO = 1, and hence combining Eqs. (8) and (9) and solving,

WO = 0.233 wt.%(Ans.)

(WO at 1550C and 1650C are 0.185 and 0.29 wt.%, respectively)

(b) In liquid FeO-SiO2 slag at 1600C and at XSiO = 0.4 (XSiO denotes mole fraction of SiO2 in FeO-SiO2), aFeO = 0.43.

Solving Eqs. (E1.1) and (E1.2), we obtain:WO = 0.10 wt.%(Ans.)The traditional method of determination of oxygen in steel samples is chemical analysis by vacuum fusion or inert gas fusion apparatus. Here, a sample of solidied steel is taken in a graphite crucible and then heated to approximately 2000C under vacuum or under a highly puried inert atmosphere. The steel sample melts and the oxygen contained in it reacts very fast with the crucible and generates carbon monoxide. The quantity of CO is measured by a sensitive instrument such as an infrared analyzer, and from it the quantity of oxygen in the sample is estimated. This apparatus has been made quite accurate and reasonably fast.Analyses of alloying elements in steel are done very quickly and conveniently using an emission spectrometer. Commercial development of the instrument has recently been reported wherein the optical wavelength range has been extended to the ultraviolet region, enabling the determination of total oxygen as well. This would eliminate the need for separate sampling and analysis. However, the author is not aware of relative precision and reliability of these two techniques.In industrial melts, the bath not only contains dissolved oxygen but also oxide particles. During freezing, solidifying steel rejects most of its dissolved oxygen, which forms additional oxide particles, and these are also retained by the solid as inclusions. The above methods of determination give the total oxygen content, which is the sum of dissolved O and oxygen in inclusions. This hampered progress of our understanding about the behavior of oxygen in steel-making and deoxidation until the development of immersion oxygen sensors based on ZrO2 and doped with CaO or MgO during the decade of the 1960s. Thereafter, this has become quite a popular tool for the measurement of dissolved oxygen content in molten steel, both in the laboratory and in industry. Excellent reviews are available in the literature on the principles and details of such sensors.25For the sake of illustration, Figure 5.1 shows the sensor employed by Fruehan et al.3 schemat-ically. The ZrO2 (CaO) or ThO2 (Y2O2) disk served as the solid electrolyte, and at high temperature it is an ionic conductor with O2 as the only mobile ionic species. The Cr + Cr2O3 mixture is the

FIGURE 5.1 Sketch of an oxygen senso

reference electrode. This assembly is immersed into liquid steel. Molten steel constitutes the other electrode. A molybdenum-Al2O3 cermet was dipped into it, and the electrical circuit was completed by platinum lead wires connected to the measuring circuit. These sensors can be used only once, i.e., they are a disposable type. Immersion time required is less than a minute. Efforts are going on to develop sensors that can be continuously immersed in liquid steel for a longer period. Laboratory successes have been reported.Such sensors behave as reversible galvanic cells. Since the solid electrolyte conducts oxygen ions only, the cell electromotive force (EMF) is related only to the difference of the chemical potentials of oxygen at the two electrodes.

O2(liquid steel) O2(reference) = ZFE

where O2 designates the chemical potential of oxygen, F is Faradays constant, Z is valence (4, here) and E is cell EMF.

The galvanic cell in Figure 5.1 may be represented as

Cr(s) + Cr2O3| ZrO2 + CaO| [O](in liquid steel)(reference)(solid electrolyte)With reference to Section 2.7,O2(reference) = RTlnpO2 (reference) = for formation of Cr2O3 per mole O2 = (Cr2O3)and

O2 (liq. steel) = RTlnpO2(in equlibrium with liq. steel) = 2RTln [

Combining the above equations,

= -4FE

Therefore, knowing (Cr2O3) and KO from the literature, the cell EMF allows us to calculate [hO]. With reference to Section 2.6.2, [fO] can be estimated from chemical analysis of steel. Therefore, the content of dissolved oxygen (i.e., WO) can be obtained from Eq. (5.4).Several designs of commercial oxygen sensors are now on the market. A popular one is CELOX, marketed by Electro-Nite n.v., Belgium. It has been jointly developed by CRM, Belgium, and Hoogovens Ijmuiden B.V., along with Electro-Nite. The cell is

Mo|Cr + Cr2O3||ZrO2(MgO)||liq.steel|Fe

The solid electrolyte is in the form of a tube with one end closed.All such sensors also contain immersion thermocouples as well so that the temperature of molten steel is also recorded simultaneously. At steelmaking temperatures, the solid electrolyte exhibits partial electronic conduction, especially at a low level of dissolved oxygen. The measured cell voltage of the cell of type illustrated by Expression (5.13) would also include thermo-EMF due to use of dissimilar leads, viz., Mo and Fe. The manufacturer provides correction terms for it.In pure liquid iron, the solubility of oxygen is governed by either Eq. (5.2) or (5.7). However, in molten steel, there are other more reactive alloying elements such as C, Si, and Mn. The oxygen solubility is governed by reaction with one or more of these elements. It has been well established that the carbon content of steel has a considerable inuence on bath oxygen content at the end of heat in steelmaking furnaces. The reaction is

[C] + [O] = CO(g) ;

The value of equilibrium constant (KCO) is given as.

Figure 5.2 shows the relationship between dissolved carbon and dissolved oxygen in a molten steel bath in a 100 kVA induction furnace. The equilibrium line corresponds to pCO = 1 atm at 1600C. Dissolved oxygen contents were measured by a solid electrolyte oxygen sensor with two types of reference electrodes.

a. Thermodynamics Of Simple DeoxidationDeoxidation of liquid steel is carried out mostly via ladle, tundish, and mold. Even in a furnace, deoxidizers are often added directly into the metal bath. In all these cases, the product of deoxi-dation, which is an oxide or a solution of more than one oxide, forms as precipitates.Deoxidation never occurs at a constant temperature. The temperature of molten steel keeps dropping from furnace to mold. The addition of a deoxidizer also causes some temperature change due to heat of reaction. However, we shall consider it as isothermal. This will not affect our

FIGURE 5.2 Dissolved oxygen content of liquid iron as a function of bath carbon at 1873 K in a 100 kVA induction furnace considerations of deoxidation equilibria, since only the nal temperature at which the equilibrium is supposed to be attained is of importance. Thermodynamically, it would not make any difference if the process were presumed to take place at that temperature.Deoxidation may be carried out by addition of one deoxidizer only. This is known as simple deoxidation. In contrast, we may use more than one deoxidizer simultaneously and, in that case, it will be termed a complex deoxidation. In this section, we will discuss simple deoxidation. A deoxidation reaction may be represented asx[M] + y[O] = (MxOy)where M denotes the deoxidizer, and MxOy is the deoxidation product. The equilibrium constant (KM ) for reaction (5.16) is given as

Again, on the basis of Eq. (2.46), hM = fM WM and hO = fO WO.If the deoxidation product is pure, then = 1. Also, in very dilute solutions, fM and fO may be taken as 1. Hence, Eq. (5.17) may be rewritten as

where A and B are constants. Equation (5.19) shows that as T increases, log KM and hence KM also increase. In other words, the solubility of MxOy in molten steel increases with temperature. Since, in deoxidation, we are interested in lowering the concentration of oxygen with the addition of as little deoxidizer as possible, an increase in temperature would adversely affect the thermodynamics of the process.Experimental determination as well as thermodynamic estimation of KM for various deoxidizers have been going on for the last four or ve decades. With advancements in science and technology, more accurate values are being found with the passage of time. This has led to a number of compilations, some old and some new, where efforts have been made to record the most acceptable values. The exercise is still going on, and discrepancies still exist, especially with more reactive elements such as Al, Zr, Ce, Ca, etc. presents such a compilation taken from that of the 19th Steelmaking Committee of the Japan Society for Promotion of Science, as well as from other sources.It may be noted that all oxide products are denite compounds except for deoxidation by manganese, where the product is either a solid or a liquid solution of FeO-MnO of variable composition. The underlying reason for this behavior is the fact that manganese is a weak deoxidizer, since the stability of MnO, although greater than that of FeO, is not drastically different from that of the latter (Figure 2.1).For deoxidation by Mn, it is in a way more appropriate to consider the reaction

(MnO) + [Fe] = [Mn] + (FeO)Fe and Mn form an ideal solution (i.e., one that obeys Raoults law). The same is true of the MnO-FeO slag. Therefore, aMnO = XMnO, aFeO = XFeO, and hMn = WMn. Noting that aFe = 1, the equilibrium constant for Reaction (5.20) is

where X denotes mole fraction. Equation (5.21) shows that XMnO /XFeO in the deoxidation product would be proportional to WMn at constant temperature. Figure 5.3 shows the relationship. The oxide product is liquid at low and solid at high values.

Example 2

Consider deoxidation of molten steel by aluminum at 1600C. The bath contains 1% Mn and 0.1% C. The nal oxygen content is to be brought down to 0.001 wt.%. Calculate the residual aluminum content of molten steel assuming that [Al] [O] Al2O3 equilibrium is attained. Also take into account all interaction coefcients.

Solution

=

=

Noting that WMn = 1, WC = 0.1, and WO = 0.001, and substituting the values in Eq. (E4.1) and taking KAl value from Appendix 5.1 and values of e from Appendix 2.3, we obtain

log [2.5 x 1014] = 2 log WAl + 3 log 0.001 3.42 WAl 0.06 (E4.2)

Taking a rst guess as WAl = 0.01, a trial-and-error solution yields

WAl = 5.36 x 10-3 wt.% as the residual aluminum in the bath (Ans.)

b. Thermodynamics of Compleks Deoxidation

As already stated, if more than one deoxidizer is added to the molten steel simultaneously, it is known as complex deoxidation. Some important complex deoxidizers are Si-Mn, Ca-Si, Ca-Si-Al, etc. Complex deoxidation offers the following advantages and is being employed increasingly for a better quality product.

1. The dissolved oxygen content is lower in complex deoxidation as compared to simple deoxidation from equilibrium considerations. Consider deoxidation by silicon.

If only ferrosilicon is added, then the product is pure SiO2, i.e., aSiO = 1. On the other hand, simultaneous addition of ferrosilicon and ferromanganese in a suitable ratio leads to the formation of liquid MnO-SiO2. Consequently, aSiO is less than 1, and hence [WSi] [WO]2 is less than that obtained by simple ferrosilicon addition. At a xed value of WSi, therefore, WO, would be less in complex deoxidation.2. The deoxidation product, if liquid, agglomerates easily into larger sizes and consequently oats up faster, making the steel cleaner. This is what happens in many complex deox-idation such as in the example presented above.3. Properties of inclusions remaining in solidied steel can be made better by complex deoxidation, thus yielding a steel of superior quality. This will be discussed again later, in an appropriate place.

Equilibrium calculations involving complex deoxidation require data on activity vs. composition relationships in the binary or ternary oxide systems of interest, besides values of KM and ei . These are available for many systems. Figure 5.5 presents the activity-composition data for a MnO-SiO2 system. The activities are in Raoultian scale, whereas the composition has been expressed in terms of weight percent of SiO2. Figure 2.3 has presented isoactivity lines for silica in the ternary CaO-SiO2-Al2O3 system at 1550C. Figure 5.6 shows the same for CaO and Al2O3. The activities were determined in the liquid slag region only.j

For activity in oxide (i.e., slag) systems, the general discussions in Chapter 2 may be consulted. For complex deoxidation, the desired product should be within this liquid eld. Thermodynamic calculations involving complex deoxidation should aim at the following:

Estimation of weight percentages of deoxidizing elements and oxygen remaining in molten steel when equilibrium is attained Estimation of the composition of the deoxidation product in equilibrium with the above

Rigorous calculation pose difficulties for two reasons. First of all, the activity vs. Composition data in oxide systems are not available in the form of equation. Secondly, interaction of more than

FIGURE 5.5 Activity vs. composition relationship in MnO-SiO2 melts; standard state are pure solid MnO and pure -crystobalite.FIGURE 5.6 Activities of CaO and Al2O1.5 in CaO-Al2O3-SiO2 system at 1823 K.

one deoxidizer calls for an iterative procedure for the solution of Eqs. (5.22) and (5.23). Therefore, it is necessary to use a computer-oriented method. A major challenge is the minimization of calculation errors. Turkdogan has carried out thermodynamic analysis for complex deoxidation by Si-Mn. Bagaria, Deo, and Ghosh have carried out thermodynamic analysis of simultaneous deoxidation by Mn-Si-Al. Ghosh and Naik have done the same for deoxidation systems: Ca-Si-Al and Mg-Si-Al. Readers may refer to those works for details. Some salient ndings by Ghosh and Naik are presented below.Calculations were performed in the range where the deoxidation product is liquid CaO-SiO2-Al2O3 slag in the ternary diagram (Figure 2.3) at two temperatures. Figure 5.7 presents some results of calculations for a Ca-Si-Al system as log WO vs. log WM (M = Si or Al) curves for three compositions of liquid deoxidation products. The dotted curves are based on rigorous calculations, taking into consideration all interaction coefcients. For the solid curves, h values were taken to be the same as weight percent, i.e., the interaction coefcients were ignored. The two curves differ by about 20%. Thermodynamically, the complex deoxidizer was found to be, at most, an order of magnitude more powerful than simple deoxidation by Al or Si.

The above exercise is important from the point of view of industrial application. Ignoring of interactions, i.e., taking hi = W, simplies the calculation procedure in a signicant way. The above analysis shows that the kind of error one may encounter is tolerable for many applications. It is also possible to predict thermodynamically the sequence of precipitation of deoxidation product, provided the process is treated as reversible. This issue is pertinent for deoxidation, where the product composition varies with time. An example of this approach is the work by Wilson et al. on a Fe-O-S-Ca system. Another is the analysis of a Fe-O-Ca-Al system by Faulring et al. Here, the hCa/hAl ratio in liquid iron determined the nature of the deoxidation product.