Chapter 2 PDF

Chapter 2
Fundamentals of Iron and

Steelmaking
E.T. Turkdogan, Consultant: Pyrometallurgy & Thermochemistry
R.J. Fruehan, Professor, Carnegie Mellon University
There have been tremendous improvements in iron and steelmaking processes in the past twenty years.
Productivity and coke rates in the blast furnace and the ability to refine steel to demanding specifica-
tions have been improved significantly. Much of this improvement is based on the application of fun-
damental principles and thermodynamic and kinetic parameters which have been determined.
Whereas, many future improvements will be forthcoming in steelmaking equipment, process improve-
ments resulting from the application of fundamental principles and data will likewise continue.
In this chapter the basic principles of thermodynamics and kinetics are reviewed and the relevant
thermodynamic data and properties of gases, metals and slags relevant to iron and steelmaking are
presented. These principles and data are then applied to ironmaking, steelmaking and secondary
refining processes. These principles and data are also used in subsequent chapters in this volume.
In writing this chapter, an attempt has been made to limit the discussion to an average level suit-
able for the students of metallurgy pursuing graduate or post-graduate education as well as for
those with some scientific background engaged in the iron and steel industry. It is assumed that the
reader has some basic knowledge of chemistry, physics and mathematics, so that the chapter can
be devoted solely to the discussion of the chemistry of the processes.
2.1 Thermodynamics
2.1.1 Ideal Gas
A gas which obeys the simple gas laws is called an ideal gas satisfying the following relation:
PV=nRT (2.1.1)
where n is the number of mols and R the universal molar gas constant.
For one mol of an ideal gas at 273.16K and l atm pressure, the value of the molar gas constant is:
1 ´ 22.414
R= = 0.08205 l atm mol –1 K –1
273.16
For pressure in Pa (º Nm–2 º Jm–3) and volume in m3,
1.01325 ´ 105 ´ 22.414 ´ 10 ±3

R= = 8.314 J mol ±1 K ±1
273.16
Copyright © 1998, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved. 13
Steelmaking and Refining Volume
In a gas mixture containing n1, n2, n3... number of mols of gases occupying a volume V at a total
pressure P, the partial pressures of the constituent gaseous species are as given below.
n1
p1 = ´P (2.1.2)
n1 + n 2 + n 3 + K
P = p1 + p2 + p3 + ... (2.1.3)
The following equations are for a given mass of gas at constant pressure, volume and temperature:
V1 V2 V3
Constant pressure (isobaric) = = L (2.1.4)
T1 T2 T3
P1 P2 P3
Constant volume (isochoric) = = L (2.1.5)
T1 T2 T3
Constant temperature (isothermal) P1V1 = P2V2 = P3V3... (2.1.6)
Generally speaking, deviation from the ideal gas equation becomes noticeable with easily liquefi-
able gases and at low temperatures and high pressures. The behavior of gases becomes more ideal
with decreasing pressure and increasing temperature. The nonideality of gases, the extent of which
depends on the nature of the gas, temperature and pressure, is attributed to two major causes: (1)
van der Waals’ forces and (2) chemical interaction between the different species of gas molecules or
atoms.
2.1.2 Thermodynamic Laws

2.1.2.1 The First Law
The first law of thermodynamics is based on the concept of conservation of energy. When there is
interaction between systems, the gain of energy of one of the systems is equal to the loss of the
other system. For example, the quantity of heat required to decompose a compound into its ele-
ments is equal to the heat generated when that compound is formed from its elements.
2.1.2.1.1 Enthalpy (heat content) The internal energy of a system includes all forms of energy other
than the kinetic energy. Any exchange of energy between a system and its surroundings, resulting
from a change of state, is manifested as heat and work.
When a system expands against a constant external pressure P, resulting in an increase of volume
DV, the work done by the system is
w = PDV = P(VB – VA)
Since this work is done by the system against the surroundings, the system absorbs a quantity of
heat q and the energy E of the system increases in passing from state A to state B.
DE = EB – EA = q – PDV = q – P(VB – VA)
Upon re-arranging this equation, we have
(EB + PVB) – (EA + PVA) = q
The quantity E + PV is represented by a single symbol H, thus
DH = q = (EB + PVB) – (EA + PVA) (2.1.7)
The function H is known as enthalpy or heat content.
There are two fundamental thermochemical laws which express the first law specifically in terms of
enthalpy. The first principle derived by Lavoisier and Laplace (1780) states that “the quantity of heat
required to decompose a compound into its elements is equal to the heat evolved when that com-
pound is formed from its elements”; i.e. the heat of decomposition of a compound is numerically
14 Copyright © 1998, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
Fundamentals of Iron and Steelmaking
equal to its heat of formation, but of opposite sign. The second principle is that discovered by Hess
(1840); it states that “the heat of reaction depends only on the initial and final states, and not on the
intermediate states through which the system may pass.”
2.1.2.1.2 Heat Capacity The heat capacity of a substance is defined as the quantity of heat required to
raise the temperature by one degree. The heat capacity of 1 g of a substance is called the specific
heat. The heat capacity of 1 g-molecule (abbreviated as mol) is called the molar heat capacity.
The variation of energy, at constant volume, and of enthalpy, at constant pressure, with tempera-
ture gives the heat capacity of the system, thus
æ ¶E ö
CV = ç ÷ (2.1.8)
è ¶T ø V
æ ¶H ö
Cp = ç ÷ (2.1.9)
è ¶T ø p
For an ideal gas the difference between the molar heat capacities at constant pressure, Cp, and con-
stant volume, CV, is equal to the molar gas constant.
Cp – CV = R (2.1.10)
Because of experimental convenience, the heat capacity is determined under conditions of constant
pressure (usually atmospheric).
From the temperature dependence of heat capacity at constant pressure, the enthalpy change is
obtained by integrating equation 2.1.9.
T2
H oT2 - H oT1 ò
= C pdT
T1
(2.1.11)
Above 298 K, the temperature dependence of Cp is represented by:

Cp = a + bT – cT–2 (2.1.12)
T
ò (a + bT - cT ) dT
-2
DH = (2.1.13)
298
where the coefficients, a, b and c are derived from Cp calorimetric measurements at different tem-
peratures.
In recent compilations of thermochemical data, the H values are tabulated at 100 K intervals for
the convenience of users.
2.1.2.1.3 Standard State The enthalpy is an extensive property of the system, and only the change
in heat content with change of state can be measured. A standard reference state is chosen for
each element so that any change in the heat content of the element is referred to its standard
state, and this change is denoted by DH°.
The natural state of elements at 25°C and 1 atm pressure is by convention taken to be the reference
state. On this definition, the elements in their standard states have zero heat contents.
The heat of formation of a compound is the heat absorbed or evolved in the formation of 1 g-mol
of the compound from its constituent elements in their standard states, denoted by DH°298.
2.1.2.1.4 Enthalpy of Reaction The change of enthalpy accompanying a reaction is given by the dif-
ference between the enthalpies of the products and those of the reactants.
For an isobaric and isothermal reaction,

A+B=C+D (2.1.14)
the enthalpy change is given by:
DH = (DH°C + DH°D) – (DH°A + DH°B) (2.1.15)
By convention, H is positive (+) for endothermic reactions, i.e. heat absorption, and H is negative
(–) for exothermic reactions, i.e. heat evolution.
Temperature effect:
ò [SC p ( products) - SC p æè reactantsöø ] dT

T
DHoT = SDH298
o
( products) - SDH298
o
(reactants) + (2.1.16)
298
T
DH oT = DH o298 + ò (DC p ) dT (2.1.17)
298
The following are some examples of the special terms of the heat of reaction.
Enthalpy or heat of formation Fe + 1⁄2O2 ® FeO
Heat of combustion C + O2 ® CO2
Heat of decomposition 2CO ® C + CO2
Heat of calcination CaCO3 ® CaO + CO2
Heat of fusion (melting) Solid ® Liquid
Heat of sublimation Solid ® Vapor
Heat of vaporization Liquid ® Vapor
Heat of solution Si(l) ® [Si] (dissolved in Fe)
2.1.2.1.5 Adiabatic Reactions When a reaction occurs in a thermally insulated system, i.e. no heat
exchange between the system and its surroundings, the temperature of the system will change in
accordance with the heat of reaction.
As an example, let us consider the internal oxidation of unpassivated direct reduced iron (DRI) in
a stockpile, initially at 25°C. The enthalpy of reaction at 298K is
o
Fe + 1
2 O2 ® FeO, DH298 = -267 kJ mol-1 (2.1.18)
The heat balance calculation is made for 1000 kg Fe in the stockpile with 150 kg FeO formed in
oxidation. The heat absorbed by the stockpile is 150 ´ 103/72 ´ 267 kJ and the temperature rise is
calculated as follows:
Q = [nFe(Cp)Fe + nFeO (Cp)FeO](T - 298)
nFe = 17,905 g-mol for 1000 kg Fe
nFeO = 2087.7 g-mol for 150 kg FeO
Cp (Fe) = 0.042 kJ mol–1K–1
Cp (FeO) = 0.059 kJ mol–1K–1
\Q = 557,416 = (752 + 123) (T - 298)
With this adiabatic reaction, the stockpile temperature increases to T = 935K (662°C).
The moisture in the stockpile will react with iron and generate H2 which will ignite at the elevated
stockpile temperature. This has been known to happen when DRI briquettes were not adequately
passivated against oxidation.
2.1.2.2 The Second Law

The law of dissipation of energy states that all natural processes occurring without external inter-
ference are spontaneous (irreversible processes). For example, heat conduction from a hot to a cold
part of the system. The spontaneous processes cannot be reversed without some change in the sys-
tem brought about by external interference.
2.1.2.2.1 Entropy The degree of degradation of energy accompanying spontaneous, hence irreversible,
processes depends on the magnitude of heat generation at temperature T and temperatures between
which there is heat flow.
The quantity q/T is a measure of degree of irreversibility of the process, the higher the quantity
q/T, the greater the irreversibility of the process. The quantity q/T is called the increase in entropy.
In a complete cycle of all reversible processes the sum of the quantities Sq/T is zero.
The thermodynamic quantity, entropy S, is defined such that for any reversible process taking place
isothermally at constant pressure, the change in entropy is given by
dH Cp
dS = = dT = Cp d( ln T) (2.1.19)
T T
2.1.2.3 The Third Law

The heat theorem put forward by Nernst (1906) constitutes the third law of thermodynamics: ‘the
entropy of any homogeneous and ordered crystalline substance, which is in complete internal equi-
librium, is zero at the absolute zero temperature.’ Therefore, the integral of equation given above
has a finite value at temperature T as shown below.
T
ST = ò Cp d( ln T ) (2.1.20)
0
The entropy of reaction is

DS =SS(products) – SS(reactants) (2.1.21)
and the entropy of fusion at the melting point Tm is
DH m
DS m = (2.1.22)
Tm
2.1.2.4 Gibbs Free Energy

From a combined form of the first and second laws of thermodynamics, Gibbs derived the free
energy equation for a reversible process at constant pressure and temperature.
G = H – TS (2.1.23)
The Gibbs free energy is also known as the chemical potential.
When a system changes isobarically and isothermally from state A to state B, the change in the free
energy is
GB – GA = DG = DH – TDS (2.1.24)
During any process which proceeds spontaneously at constant pressure and temperature, the free
energy of the system decreases. That is, the reaction is thermodynamically possible when DG < 0.
However, the reaction may not proceed at a perceptible rate at lower tempertures, if the activation
energy required to overcome the resistance to reaction is too high. If DG > 0, the reaction will not
take place spontaneously.
As in the case of enthalpy, the free energy is a relative thermodynamic property with respect to the
standard state, denoted by DG°.
The variation of the standard free energy change with temperature is given by:
T T
DC p
DG oT = DH o298 + o
ò DC p dT - TDS 298 - T ò dT (2.1.25)
298 298 T
2.1.2.4.1 Generalization of Entropy Change of Reaction

1. When there is volume expansion accompanying a reaction, i.e. gas evolution, at con-
stant pressure and temperature the entropy change is positive, hence, DG decreases
with an increasing temperature.
C + CO2 = 2CO
(2.1.26)
DG° = 166,560 – 171.0TJ
2. When there is volume contraction, i.e. gas consumed in the reaction, at constant pres-
sure and temperature the entropy change is negative, hence DG increases with an
increasing temperature.
H2 + 1⁄2S2 = H2S
DG° = –91,600 + 50.6TJ (2.1.27)
3. When there is little or no volume change the entropy change is close to zero, hence
temperature has little effect on DG.
C + O2 = CO2
DG° = 395,300 – 0.5TJ (2.1.28)
2.1.2.4.2 Selected Free Energy Data For many reactions, the temperature dependence of DH° and DS°
are similar and tend to cancel each other, thus the nonlinearity of the variation of DG° with the tem-
perature is minimized. Using the average values of DH° and DS°, the free energy equation is sim-
plified to
DG° = DH° – DS°T (2.1.29)
The standard free energies of reactions encountered in ferrous metallurgical processes can be com-
puted using the free energy data listed in Table 2.1.
2.1.3 Thermodynamic Activity

The combined statement of the first and second laws for a system doing work only against pres-
sure gives the following thermodynamic relation.
dG = VdP – SdT (2.1.30)
At constant temperature G = VdP and for 1 mol of an ideal gas V = RT/P; with these substituted
in equation (2.1.30) we obtain
= RT d ( ln P )
dP (2.1.31)
dG = RT
P
Similarly, for a gas mixture
dGi = RT d(ln pi) (2.1.32)
where pi is the partial pressure of the ith species in the gas mixture, and Gi partial molar free energy.
In a homogeneous liquid or solid solution, the thermodynamic activity of the dissolved element is
defined by the ratio
æ ö
ai = ç vapor pressure of component (i) in solution÷ (2.1.33)
è vapor pressure of pure component øT
In terms of solute activity, the partial molar free energy equation is
dGi = RT d(ln ai) (2.1.34)
Integration at constant temperature gives the relative partial molar free energy in solution
Gi = RT ln ai (2.1.35)
In terms of the relative partial molar enthalpy and entropy of solution

Gi = Hi - Si T (2.1.36)
which gives
Hi S
ln ai = - i (2.1.37)
RT R
or
Hi Si
log ai = - (2.1.38)
2.303 RT 2.303 R
2.1.3.1 Solutions
A solution is a homogeneous gas, liquid or solid mixture, any portion of which has the same
state properties. The composition of gas solution is usually given in terms of partial pressures
of species in equilibrium with one another under given conditions. For liquid solutions, as liq-
uid metal and slag, the composition is given in terms of the molar concentrations of compo-
nents of the solution.
The atom or mol fraction of the component i in solution is given by the ratio
n
Ni = i
Sn
where ni is the number of g-atoms or mols of component i per unit mass of solution, and n the total
number of g-atoms or mols. Since the metal and slag compositions are reported in mass percent,
ni per 100g of the substance is given by the ratio
%i
ni =
Mi
where M i is the atomic or molecular mass of the component i.
Noting that the atomic mass of iron is 55.85g, the atom fraction of solute i in low alloy steels is
given by a simplified equation
%i
Ni = ´ 0.5585 (2.1.39)
Mi
In low alloy steelmaking, the composition of slag varies within a relatively narrow range, and the
total number of g-mol of oxides per 100g of slag is within the range Sn = 1.6 ± 0.1. With this sim-
plification, the mol fraction of the oxide in the slag is given by
%i
Ni = (2.1.40)
1.6 Mi
2.1.3.1.1 Ideal Solutions – Raoult’s Law: The solutions are said to be ideal, if the activity is equal to
the mol or atom fraction of the component i in solution,
ai = Ni (2.1.41)
A thermodynamic consequence of Raoult’s law is that the enthalpy of mixing for an ideal solution,
HM,id, is zero. Substituting a i = Ni and HM,id = 0 in the free energy equation gives for the entropy
of formation of an ideal solution.
SM,id = R(N1 ln N1 + N2 ln N2 + N3 ln N3 +...) (2.1.42)
Table 2.1 The Standard Free Energies of Formation of Selected Compounds from Com-
piled Thermochemical Data
Notations: < > solid, { } liquid, ( ) gas, d decomposition, m melting, b boiling.
DG° = DH° – DS°T

–DH° –DS° DG° Temp.Range
kJ mol-1 J mol-1K-1 ±kJ °C
<Al> = {Al} –10.8 11.5 0.2 660m
2{Al} + 3/2(O2) = <Al2O3> 1683.2 325.6 8 660–1700
{Al} + 1/2(N2) = <AlN> 328.3 115.5 4 660–1700
<C> + 2(H2) = (CH4) 91.0 110.7 2 25–2000
<C> + 1/2(O2) = (CO) 114.4 –85.8 2 25–2000
<C> + (O2) = (CO2) 395.3 –0.5 2 25–2000
<Ca> = {Ca} –8.5 7.7 0.5 842m
{Ca} = (Ca) 153.6 87.4 0.5 842–1500b
{Ca} + 1/2(O2) = <CaO> 900.3 275.1 6 842–1500b
{Ca} + 1/2(S2) = <CaS> 548.1 103.8 4 842–1500b
<CaO> + <Al2O3> = <CaAl2O4> 19.1 – 17.2 8 25–1605m
<CaO> + (CO2) = <CaCO3> 161.3 137.2 4 25–880d
2<CaO> + <SiO2> = <Ca2SiO4> 118.8 – 11.3 10 25–1700
<CaO> + <SiO2> = <CaSiO3> 92.5 2.5 12 25–1540m
<Cr> = {Cr} –16.9 7.9 – 1857m
2<Cr> + 3/2(O2) = <Cr2O3> 1110.3 247.3 2 900–1650
<Fe> = {Fe} –13.8 7.6 1 1537m
0.947<Fe> + 1/2(O2) = <Fe0.947O> 263.7 64.3 4 25–1371m
{Fe} + 1/2(O2) = {FeO} 225.5 41.3 4 1537–1700
3<Fe> + 2(O2) = <Fe3O4> 1102.2 307.4 4 25–1597m
2<Fe> + 3/2(O2) = <Fe2O3) 814.1 250.7 4 25–1500
<Fe> + 1/2(S2) = <FeS> 154.9 56.9 4 25–988m
{Fe} + 1/2(O2) + <Cr2O3>=<FeCr2O4> 330.5 80.3 2 1537–1700
2<FeO> + <SiO2> = <Fe2SiO4> 36.2 21.1 4 25–1220m
(H2) + 1/2(O2) = (H2O) 247.3 55.9 1 25–2000
(H2) + 1/2(S2) = (H2S) 91.6 50.6 1 25–2000
3/2(H ) + 1/2(N ) = (NH ) 53.7 32.8 0.5 25–2000
2 2 3
{K} = (K) –84.5 82.0 0.5 63–759b

(K) + <C> + 1/2(N2) = {KCN} 171.5 93.5 16 622–1132b
{KCN} = 1/2(KCN)2 109.2 76.7 4 622–1132b
<Mg> = {Mg} –9.0 9.7 0.5 649m
{Mg} = (Mg) 129.6 95.1 2 649–1090b
(Mg) + 1/2(O2) = <MgO> 759.4 202.6 10 1090–2000
(Mg) + 1/2(S2) = <MgS> 539.7 193.0 8 1090–2000
2<MgO> + <SiO2> = <Mg2SiO4> 67.2 4.3 8 25–1898m
<MgO> + <SiO2> = <MgSiO3> 41.1 6.1 8 25–1577m
<MgO> + (CO2) = MgCO3) 116.3 173.4 8 25–402d
<Mn> = {Mn} –14.6 9.6 1 1244m

<Mn> + 1/2(O2) = <MnO> 391.9 78.3 4 25–1244m
{Mn} + 1/2(O2) = <MnO> 406.5 87.9 4 1244–1700
{Mn} + 1/2(O2) = {MnO}** 352.2 61.5 4 1500–1700
** supercooled liquid below the melting point 1785°C
<Mn> + 1/2(S2) = <MnS> 277.9 64.0 4 25–1244m
{Mn} + 1/2(S2) = <MnS> 292.5 73.6 4 1244–1530m
{Mn} + 1/2(S2) = {MnS} 265.0 66.1 4 1530–1700
<MnO> + <SiO2) = <MnSiO3) 28.0 2.8 12 25–1291m
<Mo> = {Mo} –27.8 9.6 6 2620m
<Mo> + (O2) = <MoO2> 578.2 166.5 12 25–2000
<Mo> + 3/2(O2) = (MoO3) 359.8 59.4 20 25–2000
Table 2.1 (continued)

DG° = DH° – DS°T
–DH° –DS° DG° Temp.Range
kJ mol-1 J mol-1K-1 ±kJ °C
1/2(N ) + 3/2(H2) = (NH3) 53.7 116.5 0.5 25–2000
2
1/2(N ) + 1/2(O2) + (NO) –90.4 –12.7 0.5 25–2000
2
1/2(N ) + (O2) + (NO2) –32.3 63.3 1 25–2000
2
{Na} = (Na) –101.3 87.9 1 98–883b

(Na) + <C> + 1/2(N2) = {NaCN} 152.3 83.7 16 833–1530b
2(Na) + 1/2(O2) = {Na2O} 518.8 234.7 12 1132–1950d
<Nb> = {Nb} –26.9 9.8 – 2477m
2<Nb> + 1/2(N2) = <Nb2N> 251.0 83.3 16 25–2400m
<Nb> + 1/2(N2) = <NbN> 230.1 77.8 16 25–2050m
2<Nb> + 5/2(N2) = <Nb2O5> 1888.2 419.7 12 25–1512m
<Ni> = {Ni} –17.5 10.1 2 1453m
<Ni> + 1/2(O2) = <NiO> 235.6 86.1 2 25–1984m
<Ni> + 1/2(S2) = <NiS> 146.4 72.0 6 25–600
3<Ni> + (S2) = <Ni3S2> 331.5 163.2 8 25–790m
1/2(S ) + (O2) = (SO2) 361.7 72.7 0.5 25–1700

2
<Si> = {Si} –49.3 30.0 2 1412m

{Si} + 1/2(O2) = (SiO) 154.7 –52.5 12 1412–1700
<Si> + (O2) = <SiO2> 902.3 172.9 12 400–1412m
{Si} + (O2) = <SiO2> 952.5 202.8 12 1412–1723m
<Ti> = {Ti} –18.6 9.6 – 1660m

<Ti> + 1/2(N2) = <TiN> 336.3 93.3 6 25–1660m
<Ti> + (O2) = <TiO2> 941.0 177.6 2 25–1660m
<V> + {V} –22.8 10.4 – 1920m

<V> + 1/2(N2) = <VN> 214.6 82.4 16 25–2346d
2<V> + 3/2(O2) = <V2O3> 1202.9 237.5 8 25–2070m
{Zn} = (Zn) –118.1 100.2 1 420–907b

(Zn) + 1/2(O2) = <ZnO> 460.2 198.3 10 907–1700
{Zn} + 1/2(S2) = <ZnS> 277.8 107.9 10 420–907b
(Zn) + 1/2(S2) = (ZnS) –5.0 30.5 10 1182–1700
<Zr> = {Zr} –20.9 9.8 – 1850m

<Zr> + 1/2(N2) = <ZnN> 363.6 92.0 16 25–1850m
<Zr> + (O2) = <ZrO2> 1092.0 183.7 16 25–1850m
<Zr> + (S2) = <ZrS2> 698.7 178.2 20 25–1550m
<ZrO2> + <SiO2> = <ZrSiO4> 26.8 12.6 20 25–1707m
* References to the compiled thermochemical data used in deriving DH° and DS° values are given in Ref. 27 cited in
Section 2.2.2.4
1.0
0.8
w
la
t's
ul
ao
0.6
Activity
R
aCu aFe
0.4
w
la
t's
0.2
ul
ao
R
aSi aFe
0
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
Silicon (atom fraction) Copper (atom fraction)
Fig. 2.1 Activities in liquid Fe–Si and Fe–Cu alloys at 1600°C showing strong negative and positive depar-
tures from Raoult’s law.
2.1.3.1.2 Nonideal Solutions Almost all metallic solutions and slags exhibit nonideal behavior.
Depending on the chemical nature of the elements constituting a solution, the activity vs com-
position relation deviates from Raoult’s law to varying degrees, as demonstrated in Fig. 2.1 for
liquid Fe–Si and Fe–Cu systems at 1600°C.
2.1.3.2 Activity Coefficient

The activity coefficient of solute i is defined by the ratio
ai
gi = (2.1.43)
Ni
1.0
If the activity is relative to the pure component i, it
follows from Raoult’s law that as Ni®1, g®1.
2.1.3.2.1 Henry’s Law for Dilute Solutions In infinitely 0.8

dilute solutions, the activity is proportional to the
concentration
0.6
ai = g°i Ni (2.1.44)
aC
The approach to Henry’s law at dilute solutions is
demonstrated in Fig. 2.2 for the activity of carbon 0.4
(relative to graphite) in austenite at 1000°C. The
austenite containing 1.65%C (NC = 0.072) is satu- Henry's law
rated with graphite at 1000°C for which the carbon 0.2
activity is one relative to graphite.
Since Henry’s Law is valid at infinite dilution only,
the ration gi/g °i is used as a measure of departure 0
0 0.02 0.04 0.06 0.08
from Henry’s Law for finite solute contents in NC
dilute solutions. For solute concentration in terms Fig. 2.2 Activity of carbon in austenite (relative
of mass percent, Henry’s activity coefficient is to graphite) at 1000°C, demonstrating deviation
defined by the ratio from Henry’s law.
gi
fi = (2.1.45)
g io
such that fi®1 when %i®0.
2.1.3.2.2 Interaction Coefficients Over several mass percentages of the solute, the composition
dependence of activity coefficient fi of solute i in binary systems is represented by the following
relation in terms of mass %i.
log fi = eii[%i] (2.1.46)
where e is called the solute interaction coefficient. For multi-component solutions, the following
summation is used
log fi = eii[%i] + S eij[%j] (2.1.47)
where eij is the effect of the alloying element j on the activity coefficient of solute i.
2.1.3.3 Conversion from One Standard State to Another

In steelmaking processes we are concerned with reactions involving dissolved elements at low con-
centrations in liquid steel. Therefore, it is convenient to express the solute activity relative to Henry’s
law. The free energy change accompanying the isothermal transfer of solute from state A to state B is
æa ö
DG = RT ln ç B ÷ (2.1.48)
è aA ø
Taking a pure component for state A, i.e. aA = 1 and Henry’s law for state B, i.e. a B = g°i Ni
DG = RT ln(g°i Ni) (2.1.49)
For one mass percent solute in iron,
0.5585
Ni » (2.1.50)
Mi
where Mi is the atomic mass (g) of the solute. Assuming that Henry’s law holds at 1 mass %, the
standard free energy of solution of pure component i in iron at 1 mass % is
æ 0.5585 o ö
DG s = RT ln ç gi ÷ (2.1.51)
è Mi ø
Since Henry’s law is valid at infinite dilution only, appropriate correction must be made for non-
ideal behavior when using the above equation.
2.1.4 Reaction Equilibrium Constant

Let us consider the following reaction equilibrium occurring at constant temperature and pressure.
mM + nN = uU + vV (2.1.52)
The state of equilibrium is defined by the following thermodynamic relation involving the activi-
ties of reactants M and N and the activities of the products U and V.
(aU ) (aV )
u v
K=
(a M ) (a N ) (2.1.53)
m n
where the equilibrium constant K can be derived from the standard free energy change accompa-
nying the reaction thus,
DG° = –RT ln K (2.1.54)
Since DG° is a function of temperature only, the equilibrium constant is also a function of temper-
ature only.
DG o
ln K = - (2.1.55)
RT
In terms of enthalpy and entropy changes
DH o DS o
ln K = - + (2.1.56)
RT R
In exothermic reactions DH is negative, therefore the equilibrium constant K decreases with an
increasing temperature. The temperature effect on K is the opposite in endothermic reactions.
For DG° in J, R = 8.314 J mol–1K–1 and substituting log for ln,
DG o (2.1.57)
log K = -
19.144 T
As an example, let us consider the following reaction equilibrium between liquid slag and low alloy
steel.
(MnO) + [C] = [Mn] + CO(g) (2.1.58)
where the parenthesis ( ) and [ ] represent respectively the oxides dissolved in slag and elements
dissolved in iron. From the free energy data in Table 2.1 and those in Table 2.8 in section 2.4 for
the dissolution of graphite and liquid manganese in liquid iron, the standard free energy change
accompanying reaction 2.1.58 is
DG° = 230,560 – 147.32TJ (2.1.59)
which gives for the equilibrium constant K
12 , 043 (2.1.60)
log K = - + 7.695
T
[% Mn] pCO
K=
[ ]
(aMnO ) %C (2.1.61)
where pCO is in atm, the oxide activity is with respect to pure molten MnO and at low concentra-
tions, the mass percentages of Mn and C in the steel are essentially equivalent to their activities,
i.e. fMn and fC are close to unity.
For a steelmaking slag of basicity 3.0 containing 5% MnO, aMnO = 0.088; at 1600°C the equilib-
rium constant is K = 18.42 for which equation 2.1.61 gives for the slag-metal equilibrium
[%Mn]/[%C] = 1.62 at 1 atm pressure of CO.
2.2 Rate Phenomena

There are many different facets of rate phenomena involving homogeneous or heterogeneous
chemical reactions, mass transfer via atomic or molecular diffusional processes, viscous flow, ther-
mal and electrical conduction and so on. The concepts of diffusion, mass transfer and chemical
kinetics presented briefly in this section of Chapter 2, are confined to cases which are relevant to
the study of the rates of reactions in the iron and steelmaking processes.
2.2.1 Diffusion
2.2.1.1 Fick’s Diffusion Laws
The first law is for the steady state diffusion.
The quantity of diffusing substance which passes per unit time through unit area of a plane per-
pendicular to the direction of diffusion, known as the flux J, is proportional to the concentration
gradient of the diffusing substance
dC
J = -D (2.2.1)
dx
The coefficient D is the diffusivity of the substance in the medium; C is the concentration of the
substance per unit volume and x the distance in the direction of diffusion.
The second law is for the nonsteady state diffusion.
The rate of accumulation of diffusing substance in a given volume element is the difference between
the inward and outward flux. In other words, what goes in and does not come out, stays there.
The rate of concentration change -dC/dt resulting from flux over a distance dx is
dC (2.2.2)
J=- dx
dt
Hence, the change in flux with distance is
dJ dC (2.2.3)
=-
dx dt
Invoking the first law and re-arranging gives the second law.
dJ d æ dC ö
= - çD ÷ (2.2.4)
dx dx è dx ø
The solution of this equation depends on the geometry and on the boundary conditions of the
medium in which the dissolved substance is diffusing.
2.2.1.2 Chemical and Self-Diffusivities

There are two kinds of diffusivities:
(i) The chemical diffusivity is the coefficient of diffusion defined by Fick’s law
Di = -Ji(¶C/¶x) for the single diffusing species i.
(ii) The self-diffusivity is the coefficient of diffusion measured in a homogeneous
medium with the radioactive or stable isotope (tracer) of the diffusing species. The
self-diffusion is a consequence of random movements of atoms or molecules.
For an ideal solution, or an infinitely dilute solution, we have the equality
Di = D*i (2.2.5)
For nonideal solutions in general (ionic or metallic), Onsager and Fuoss1 derived the following
relation for the chemical diffusivity corrected for departure from ideal behavior.
æ ¶ ln ai ö *æ ¶ ln g i ö
Di = Di* ç ÷ = Di ç 1 + ÷ (2.2.6)
è ¶ ln C i ø è ¶ ln C i ø
where a i is the activity and g i the activity coefficient of the single diffusing species i.
2.2.1.3 Types of Diffusional Processes
There are various types of diffusional processes, depending on the medium in which the diffusion occurs.
2.2.1.3.1 Diffusion in Molten Slags and Glasses Because of the ionic nature of the medium, the diffu-
sion in molten slags and glasses is by the ions. The ionic conduction is a purely random jump
process. For the limiting case of purely free and random movements of ions, the self-diffusivity of
the ionic species i is related to the electrical conductivity li by the Nernst-Einstein2 equation
RT
Di* = li (2.2.7)
F Zi2 C i
2
where Zi is the valence and Ci the concentration (mol cm-3) of the species i and F = 96,489 C mol–1
the Faraday constant.
Reference may be made to a previous publication by Turkdogan3 on the review of interrelations
between ionic diffusion and conduction in molten slags.
2.2.1.3.2 Diffusion in Porous Media The rate of a heterogeneous reaction between a gas and a porous
medium is much affected by the counter-diffusive flow of gaseous reactants and products through
the pores of the medium. There are two types of effective gas diffusivities, depending on the pore
structure of the medium:
(i) Molecular diffusivity in large pores and at ordinary to high pressures.
(ii) Knudsen diffusivity in small pores and at low pressures.
The pore diffusion is discussed in more detail in Section 2.3.
2.2.1.3.3 Eddy Diffusivity The mass transfer occurring within a turbulent fluid flow is
characterized by the diffusion coefficient called eddy diffusivity which is a function of the
time-averaged and fluctuating velocity components in turbulent flow. This subject is covered in
detail in a text book on fluid phenomena by Szekely 4.
2.2.2 Mass Transfer

In high temperature processes as in pyrometallurgy, the rates of interfacial chemical reactions
are in general much faster than the rates of transfer of the reactants and reaction products to and
from the reaction site. The formulations of rate equations for transport controlled reactions vary
considerably with the physical properties and type of fluid flow over the surface of the reacting
condensed phase. Then, there are formulations of rate equations for different regimes of gas
bubbles in liquid metal and slag. For a comprehensive discussion of the transport controlled
rate phenomena in metallurgical processes, reference may be made to textbooks by Szekely and
Themelis5 and by Geiger.6
There are numerous computer software packages, as for example PHOENICS, FLUENT, CFDS-
FLOWS3D, GENMIX, TEACH(2D, 3D), 2E/FIX, SOLA/VOF, METADEX(R) and FID, which are
used in the steel industry in the process and design engineering of plant facilities and in the con-
trol of steelmaking processes. References may also be made to the publications by Szekely et al.7-9
on the mathematical and physical modelling of metallurgical processes involving fluid flow, heat
transfer, mass transfer and gas-slag-metal reactions.
There are of course many other publications on this subject which cannot all be cited here as the
subject matter is well outside the scope of this book.
The rate phenomena discussed in this chapter are on chosen subjects which are pertinent to vari-
ous aspects of ferrous-pyrometallurgical processes, in addition to oxygen steelmaking, ladle refin-
ing and degassing.
2.2.2.1 Parabolic Rate of Oxidation of Iron

The scale forming in the oxidation of iron, or low alloy steels, consists of three layers of oxides:
wustite on the iron surface followed by magnetite then hematite on the outer surface of the scale.
From several experimental studies it was found that the average wustite:magnetite:hematite thick-
ness ratios were about 95:4:1. In iron oxides, the diffusivity of iron is greater than the oxygen
diffusivity. The rate of oxidation is controlled primarily by diffusion of iron through the wustite
layer from the iron-wustite interface to the wustite-gas or wustite-magnetite interface, then through
the thin layers of magnetite and hematite.
X2 = lt (2.2.8)
where l is the parabolic rate constant, cm2 (scale) s–1
If the measurement of the rate of oxidation is made by the thermogravimetric method, then the par-
abolic rate constant kp would be in units of (gO)2cm–4s–1. From the compositions and densitites of
the oxides, with the relative thickness ratios of wustite:magnetite:hematite = 95:4:1, the values of
kp and l are related as follows.
kp (gO)2 cm–4 s–1 = 1.877l, cm2 (scale) s–1 (2.2.9)
Many experimental studies have been made of the rate of oxidation of iron in air and oxygen at
temperatures 600 to 1300°C. The temperature dependence of the parabolic rate constant is shown
in Fig. 2.3; the references to previous studies denoted by different symbols are given in a paper by
Sheasby et al.10
From theoretical considerations, Wagner11 derived the following equation for the parabolic rate
constant in terms of the activity of oxygen and self-diffusivities of the mobile species in the scale.
For the case of wustite with D*Fe >> D*O Wagner’s equation is simplified to
aO¢¢
DFe d ( ln aO )
NO *
l=2ò (2.2.10)
N
aO¢ Fe
Temperature, °C
1300 1200 1100 1000 900 800 700 600
10–5
10–6
Kp, (g O)2 cm–4 s–1
10–7
10–8
10–9
10–10
6 8 10 12
104(K –1)
T
Fig. 2.3 Temperature dependence of the parabolic rate constant for oxidation of iron in air or oxygen with or
without H2O. From Ref. 10.
30
Himmel et al12
Hembree & Wagner13

Self diffusivity, D*Fe x 108 (cm2/sec.)
20
1100°C
equilibrium
with Fe
10
983°C
897°C equilibrium
with
Fe304
800°C
0 0.05 0.10 0.15

(1-NFe/NO)
Fig. 2.4 Self-diffusivity of iron as a function of (1 – NFe/NO) fraction of iron deficiency.
where aO¢ and a²O are oxygen activities at the iron-wustite and wustite-magnetite interfaces, respec-
tively. The diffusion of iron ions in the nonstoichiometric phase wustite FeXO, is via the cation
vacant sites in the wustite lattice. It is for this reason that the self-diffusivity of iron increases with
an increasing O/Fe atom ratio in wustite, i.e. with increasing iron deficiency, as shown in Fig. 2.4
The values of kp = 1.877l calculated from the measured tracer diffusivity of iron using Wagner’s
equation, are consistent with the average experimental values in Fig. 2.3 which is represented by
the following equation.
8868
Log k p = - + 0.9777 (2.2.11)
T
2.2.2.2 Oxidation of Carbon in CO2–CO Mixtures

Most forms of carbon are porous, therefore the rate of oxidation, i.e. gasification of carbon, is deci-
sively affected by the pore structure, internal pore surface area, effective gas diffusivity in the pores
and particle size.
In a critical review of the oxidation of carbon, Walker et al.14 gave a detailed and a comprehensive
account of (up to 1959) experimental and theoretical studies on this subject. In the late 1960s, Turk-
dogan and co-workers15–17 pursued further experimental work on the oxidation of several types of
carbon in CO2–CO mixtures at temperatures up to 1300°C and at pressures of 0.01 to 10 atm. A
few salient features of their findings are briefly given here.
Depending on the particle size, pore structure, temperature and gas pressure, there is either com-
plete internal burning or external burning of carbon. The rate equations are of the following forms
for these two limiting cases:
(i) For small porous carbon particles and at low temperatures and pressures, there is rapid
counter current pore diffusion of CO2 and CO, and the rate of oxidation is controlled by
the chemical kinetics on the pore walls of the carbon. For this limiting case of complete
internal burning, the rate equation is
ln(1 – F) = –fSCit (2.2.12)
where
F = the mass fraction of carbon oxidized,
f = specific isothermal rate constant of chemical reaction per unit area of the pore wall,
S = connected internal pore surface area per unit mass cm2mol–1,
Ci = molar concentration of CO2 per cm3 in the gas stream,
t = oxidation time
(ii) For large carbon particles and at high temperatures and pressures, the gas diffusion is
confined to the pore mouths on the surface of the carbon particle. For this limiting
case of external burning the rate equation is
( f r S De )
1
2
1 - (1 - F )
1
3
= Ci t (2.2.13)
r ro
where
De = effective CO2–CO pore diffusivity,
r = molar bulk density of the carbon,
ro = initial radius of carbon particle.
The effects of the particle size and gas pressure on the rate of oxidation of electrode graphite is
shown in Fig. 2.5
A mathematical analysis was made by Tien and Turkdogan18 to formulate the pore diffusion effect
on partial internal burning of relatively large
carbon particles. The experimentally determined 10–2
Experimental
initial rates of oxidation of coke and graphite
Rate, fractional mass loss per minute
Calculated
spheres (» 2 cm dia.) in 1:1 CO2:CO mixture at
1 atm pressure are compared in Fig. 2.6 with the
6 10–3
R, rate per unit mass of
carbon (min–1 x 103)
Atm. CO2
Coke
4
30
10–4
Graphite
2 5
0.96
0 10–5
0 1 2 3 900 1000 1100 1200
Particle diameter (mm) Temperature (°C)
Fig. 2.5 Effect of particle size on the rate of oxidation Fig. 2.6 Effect of temperature on rate of oxidation of
of electrode graphite in pure carbon dioxide at coke and graphite (1.9 and 2.2 cm dia. spheres,
1000°C. From Ref. 16. respectively) in 50:50 CO2:CO at 1 atm. From Ref. 18.
1400
III
1200 III
II
II
1000 I
I
Temperature (°C)
(a) (b)
800
1400
III
Pore III
1200 diffusion
II control II
Mixed
control
I
1000 Reaction I
control
(c)
800
10–2 10–1 100 101 10–2 10–1 100 101
Radius (cm)
Fig. 2.7 Critical radius and temperature for the limiting rate controlling processes at 1 atm pressure. (a) Coke,
100% CO2; (b) Coke, 1:1 CO:CO2; (c) Graphite, 100% CO2; (d) Graphite, 1:1 CO:CO2. From Ref. 18.
values calculated from the mathematical analysis, which is summarized in Fig. 2.7. In each dia-
gram, the lower curve is for 80% internal burning; therefore, in the region below this curve there
is almost complete pore diffusion. The upper curve is for 20% internal burning; therefore, in the
region above this curve, the pore diffusion control predominates.
Reference should also be made to a paper by Aderibigbe and Szekely19 for a more detailed math-
ematical formulation of the rate of pore-diffusion and reaction controlled gasification of metallur-
gical coke in CO2–CO mixtures.
2.2.2.3 Reduction of Iron Oxides

In view of its practical importance to the understanding and control of ironmaking processes, a
great deal of research has been done on the gaseous reduction of iron oxides and iron ores.
Because of the porous nature of iron oxides and the reduction products, the interpretation of the
reduction rate data is inherently complex.
The formation of product layers during the gaseous reduction of dense sintered hematite and magnetite
pellets or natural dense iron ore particles is a well-known phenomenon, as shown in Fig. 2.8.
In several studies made in the early 1960s20–22 it was found that the thickness of the reduced iron
layer, encasing the iron oxide core of the pellet, increased linearly with the reduction time. The mea-
sured rates were interpreted in terms of the rate-controlling chemical reaction at the iron wustite
interface; the diffusive fluxes of gases through the porous layers were assumed to be relatively fast.
On the other hand, Warner23 and Spitzer et al.24 have expressed the view that the rate of gaseous
reduction is much affected by the gaseous diffusional processes, e.g. the gas-film resistance at the
pellet surface and particularly the resistance to dif-

fusion in the porous product layers. The rate mea-
surements made by Turkdogan and Vinters25 on the
reduction of hematite pellets in H2–H2O and
CO–CO2 mixtures have clearly demonstrated that
the rate-controlling effects of gas diffusion into the
pores of the oxide granules or pellets and through
the porous iron layer dominate the reaction kinetics.
The sketch in Fig. 2.9 demonstrates three limiting
rate controlling processes as outlined below.
(a) With fine granules there is internal reduc-
tion producing rosettes or platelets of
Iron metallic iron within the oxide particle; the
rate in terms of mass fraction reduced, F, is
independent of particle size.
Wustite
(b) With large and dense oxide particles and
Magnetite at high temperatures, the gas diffusion is
slow and the reaction is confined essen-
Hematite tially to pore mouths on the outer surface
of the particle, bringing about the devel-
opment of a porous iron layer around the
Fig. 2.8 Formation of reaction product layers in pellet. Because of the layer formation, this
the gaseous reduction of hematite pellets. mode of reduction is often called topo-
chemical reduction. In the early stages of
reduction, the porous iron layer is sufficiently thin for rapid gas diffusion, therefore the
initial rate of reduction is controlled jointly by (i) gas diffusion into the pore mouths of
the oxide and (ii) reaction on the pore walls of the wustite.
dF f S De
µ (2.2.14)
dt r
(c) When the porous iron layer becomes sufficiently thick, the rate of reduction will be
controlled essentially by the counter current gas diffusion (H2–H2O and CO–CO2)
for which the limiting rate equation, for a given temperature is as follows.
é 2 ù De æ pi - ( pi ) eq ö
êë 3 - 2 F - 3 (1 - F ) 3 = Y=
úû ç
rr 2 çè RT ÷ø
÷ t +C (2.2.15)
a b c
Fig. 2.9 Schematic representation of (a) uniform internal reduction of wustite to iron, (b) limiting mixed con-
trol (partial internal reduction) and (c) diffusion in porous iron as the rate-controlling step.]
where pi is the H2 or CO partial pressure in the gas stream and (pi)eq that for the iron-
wustite equilibrium and C a constant (a negative number) that takes account of all
early time departures from the assumed boundary conditions for this limiting case.
(d) At high reduction temperatures, with large oxide pellets and low velocity gas flows,
the rate of reduction is controlled primarily by mass transfer in the gas-film layer at
the pellet surface. In this limiting case the rate is inversely proportional to the
square of the particle diameter.
The experimental data are given Fig. 2.10 showing the particle size effect on the initial rate of
reduction of hematite granules or pellets in hydrogen.
When the thickness of the reduced porous iron layer exceeds 1 mm, the subsequent rate of reduc-
tion is controlled primarily by gas diffusion through the porous iron layer. The reduction data plot-
ted in accord with equation 2.2.15 usually give elongated S-shaped curves as in Fig. 2.11 for 15
mm diameter spheroidal hematite ore reduced in H2 at 1 atm. From about 50% to 95% or 99%
reduction, data are well represented by straight lines. The effective H2–H2O diffusivities in the
pores of the iron layer are derived from the slopes of the lines; details are given in Section 2.3.3.
The effect of reaction temperature on the pore surface area of iron and wustite formed by the reduc-
tion of hematite is shown in Fig. 2.12. The higher the reduction temperature, the smaller the inter-
nal pore surface area, i.e. the coarser the pore structure. The iron oxides reduced at temperatures
below 800°C are known to be pyrophoric, which is a consequence of a fine pore structure with a
large pore surface area as noted from the data in Fig. 2.12.
10
1100°C
1000°C
900°C
1200°C
800°C
minute
1
700°C
per
loss per
oxygenloss minute
Fig. 2.10 Initial rate in hydrogen 600°C

at 0.96 atm. as a function of par-
oxygen
dF ,
dt O
ticle radius at indicated temper-

atures. From Ref. 25.
Fractional
fractional
0.1
Limiting
Slope –1
0.01
Radius
0.1 (cm) 1
Radius (cm)
1.0 100
1200°C
1100°C
0.8 98
1000°C 900°C 800°C 95

3 - 2F - 3(1 - F)2/3
% reduction
0.6
90 Fig. 2.11 Diffusion plot of reduc-
tion data for 15 mm diameter
spheroidal hematite ore. From
0.4 Ref. 25.
75
0.2
50
25
0
0 10 20 30 40
Time (min)
100
10
S (m2g-1) (iron or wustite)
1.0
0.1
0.01
200 400 600 800 1000 1200
Reduction temperature (°C)
Fig. 2.12 Connected internal pore surface area of iron and wustite formed by reduction of hematite ores’ A
and B; (a) ore A reduced to iron in H2 (à); (b) ore B reduced to iron in H2(Ñ) and reduced to wustite in an
H2–H2O mixture (D); (c) ore B reduced to iron in a CO–CO2 mixture (❏) and reduced to wustite in another
CO–CO2 mixture (❍). From Ref. 25.
2.2.2.4 Mass Transfer with Gas Bubbles in Oxygen Steelmaking

In most practical applications of gas injection into liquids, we are concerned mainly with rates of
reactions in the swarm of bubbles. Mass-transfer phenomena in the swarm of bubbles have been
studied extensively with aqueous solutions; accumulated knowledge is well documented in a
review paper by Calderbank.26 Also, several studies have been made of the bubble size and veloc-
ity of rise of single bubbles in liquid metals. Szekely presented a comprehensive review of these
studies in his book on Fluid Flow Phenomena in Metals Processing.4
In his recent book Fundamentals of Steelmaking, Turkdogan27 made an assessment of the size
range of dispersed gas bubbles and the corresponding mass transfer in liquid steel for the condi-
tions prevailing in the BOF and Q-BOP steelmaking processes. In this assessment, the bubbles are
assumed to be generated from a multitude of points across the bottom of the vessel, as though the
entire vessel bottom were in fact one big porous plug.
2.2.2.4.1 Gas Holdup and Superficial Gas Velocity The volume fraction of gas holdup (e) is given by
the following ratio:
Vb V
e= = 1- s (2.2.16)
Vb + Vs HA
where
Vb = transitory bubble volume in the bath at any time,
Vs = steel volume in the vessel,
H = average bath depth,
A = average bath cross sectional area.
The velocity of bubble rise, Ub, in a swarm of bubbles is given by the ratio
Us V˙ (2.2.17)
Ub = =
e eA
where
.
V = gas flow rate,
Us = superficial gas velocity.
In the studies of Calderbank and co-workers26 with the air-water system, the gas was injected into
the column of water through a sieve plate on the bottom of the column. The fractional holdup of
gas was reported as shown in Fig. 2.13. The units in F factor are Us, m s–1 and rg (the gas density),
kg m–3.
The dot-dash curve is taken to represent the average values. It appears that at F ³ 1, the gas holdup
reaches an essentially constant value of e = 0.6.
In the estimation of fractional gas holdup in the steel bath, during decarburization or argon stir-
ring, the gas flow rates taken are for the melt temperature
. and average
. gas bubble pressure. The
values used are for 1600°C and 1.5 atm so that VT (m3s–1) = 4.57 ´ Vo(Nm3s–1); for these condi-
tions the gas densities are 0.27 kg m-3 for N2 and CO and 0.39 kg m-3 for Ar; an average value is
used, rg = 0.33 kg m–3.
The average bath cross sectional area A in BOF or Q-BOP vessel increases with an increasing
gas flow rate. Keeping in mind the inner dimensions
. of the BOF or Q-BOP vessel for 200-240
.
ton heats, the area values used are: A = 22 m2 at VT = 60 m3s–1 decreasing to A = 16 m2 at VT = 1
m3s–1. On this
. basis, the following equation is obtained for the superficial gas velocity Us as a
function of VT.
( )
˙T
V
Us m s-1 = (2.2.18)
14.55 + 0.089V˙T
The volume fraction gas holdup and bubble velocity of rise calculated using equation 2.2.18 and
Fig. 2.13 are plotted in Fig. 2.14 as a function of the gas flow rate. The gas holdup in the slag layer
is a more complex phenomenon, because of slag foaming and injection of large amounts of steel
droplets into the slag layer during decarburisation with oxygen blowing.
2.2.2.4.2 Estimation of Bubble Size for Uniformly Dispersed Bubbles in the Steel Bath Many estimates
were made in the past of gas bubble diameters in steelmaking processes, the estimates varying over
a wide range from 1 to 8 cm or more. Bubbles with diameters >1cm acquire a spherical cap shape
as they rise in liquids. The apparent diameter is that of a sphere which has a volume equivalent to
that of the spherical cap shaped bubble.
Calderbank and co-workers26 and Leibson et al.28 have found that in aqueous solutions the bub-
ble size becomes smaller with an increasing gas flow rate, ultimately reaching a minimum size of
about 0.45 cm diameter at large gas flow rates. Relatively large gas bubbles in motion are subject
to deformation and ultimately to fragmentation into smaller bubbles. The drag force exerted by
the liquid on a moving bubble induces rotational and probably a turbulent motion of the gas within
the bubble.
This motion creates a dynamic pressure on the bubble surface; when this force exceeds the surface
tension, bubble breakup occurs. Because of the large difference between the densitites of the gas
and liquid, the energy associated with the drag force is much greater than the kinetic energy of the
gas bubble. Therefore, the gas velocity in the bubble will be similar to the bubble velocity. On the
basis of this theoretical reasoning, Levich29 derived the following equation for the critical bubble
size as a function of bubble velocity.
1.0
Average
0.8
e, volume fraction gas holdup
0.6
0.4
0.2
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
F = ÖÏrw (ms–1) (kg m–3) 1/2
–1 –3 1/2
F=U s r (ms ) (kg m )
Fig. 2.13 Experimental data of Calderbank26 on gas holdup in sieve plate columns for the air-water system.
6 0.6
5 0.5
e
4 0.4
Ub (ms-1)
3 0.3
e
2 0.2 Ub
1 0.1
0 0
1 2 4 6 8 10 20 40 60 80 100
•
VT (m3 S–1)
Fig. 2.14 Gas holdup and bubble velocity as a function of gas flow rate. From Ref. 27.
1
æ 3 ö
3
2s
dc = ç 2÷
(2.2.19)
è C d r g rl ø U 2b
where s is the surface tension and Cd the drag coefficient, which is close to unity. As the bubble
size decreases with increasing bubble velocity, the gas circulation within the bubble diminishes,
thus becoming less effective in bubble fragmentation. Another view to be considered is that, as the
gas holdup increases with increasing gas flow rate, the liquid layer separating the bubbles becomes
thinner. Consequently, the dynamic gas pressure exerted on the bubble surface becomes nullified
by similar forces exerted in the neighboring bubbles, hence the cessation of bubble fragmentation
at high values of e.
The bubble fragmentation with increasing e, hence increasing Ub, is calculated for bubbles in water
with s = 0.072 Nm–1 and in liquid steel with (a) s = 1.7 Nm–1 in the presence of O < 4 ppm and 40
ppm S, and (b) s = 1.3 Nm-1 in the presence of 600 ppm O and 120 ppm S. The results are shown
in Fig. 2.15. As noted earlier, the limiting bubble size in air-water system is d = 0.45 cm which inter-
sects the fragmentation curve at e = 0.41, depicting the cessation of bubble fragmentation.
In an attempt to estimate the minimum bubble size in the swarm of bubbles in liquid steel, Turk-
dogan proposed the following hypothesis: ‘the surface energy of bubbles per unit mass of liquid
which the bubbles displace, are equal for all liquids.’ Thus, surface energy Eb of dispersed bubbles
of uniform size per unit mass of liquid which they displace is given by:
6e æ s ö -1
Eb = ç ÷ J kg liquid (2.2.20)
1- e è rd ø
4
Equation 2.2.19
Uniform limiting
bubble diameter
Equivalent bubble diameter (cm)
Liquid steel
b a
Water
0
0 0.2 0.4 0.6
e
Fig. 2.15 Calculated bubble diameter for bubble fragmentation mechanism equation 2.2.19, compared with
uniform limited diameter at high values for water and liquid steel: (a) for O < 4 ppm and 40 ppm S (b) for 600
ppm O and 120 ppm S. From Ref. 27.
This conjectured statement leads to the following similarity relation for non-foaming inviscid liquids.
s1 s2
= = L J kg -1 liquid (2.2.21)
r1 d 1 r2 d 2
With s1 = 0.072 Jm-2 and d1 = 0.45 cm for gas-water system and s2 = 1.3 Jm–2 and r2 = 6940 kg m–3,
we obtain d2 = 1.16 cm for the limiting bubble diameter in liquid steel containing » 600 ppm O and
120 ppm S. For the melt containing O < 4 ppm and 40 ppm S, d2 = 1.52 cm. These limiting values
intersect the fragmentation curves for liquid steel at about e = 0.40, similar to that for water.
A uniform constant bubble size and essentially constant e = 0.6 at high gas flow-rates, means a con-
stant number of bubbles per unit volume of the emulsion. Noting that for the close-packed (fcc)
arrangement of bubbles, the packing fraction (gas holdup) is e = 0.74, the thickness of the liquid film
at the nearest approach of the bubbles is derived from the following equality of the bubble number.
0.6 0.74
= (2.2.22)
( d + d 2)
3 3
d
Noting that for water the experimental data give d = 0.45 cm and that estimated for liquid steel
d = 1.16 cm, equation 2.2.22 gives d = 0.65 mm for bubbles in aqueous solutions and d = 1.68 for
bubbles in liquid steel at the limiting value of e = 0.6 for non-foaming inviscid liquids.
2.2.2.4.3 Rate Equation for Transport Controlled Gas Bubble Reactions in Liquid Steel For mass-transfer
controlled reactions of gas bubbles in liquid steel, the rate equation has the following form:
%X - %X e
ln = -Sok mt (2.2.23)
%Xo - %X e
where Xo and X e are the initial and gas-metal equilibrium concentrations of the diffusing reactant
in the melt, km the mass-transfer coefficient and So the bubble surface area which is in terms of e.
6e 2 3
So = , m m gas-melt
gas - melt emulsion
emulsion (2.2.24)
d
Bubble surface area with respect to unit mass of liquid is:
6e æ 1 ö
Sm = 2
ç ÷ , m /kg liquid (2.2.24a)
1- e è rd ø
From many studies of gas bubble reactions in nonfoaming aqueous solutions, the following for-
mulation have been derived for the liquid-phase mass-transfer coefficient for the regime of surface-
renewal at the gas-liquid interface.26
1
æ DU b ö 2
k m = 1.28 ç ÷ (2.2.25)
è d ø
1.0
where D is the diffusivity of the reactant in the
liquid-film boundary layer.
In the presence of surface active solutes such as
oxygen and sulphur in liquid steel, the bubble sur-
face will be less mobile hence the rate constant km
will be somewhat less than that given by equation
0.1 2.2.25 for the mobile surface, which is a neces-
sary condition for surface-renewal. On the other
Sokm (s-1)
hand, the surface active solutes decrease the bub-

ble diameter, hence increase the bubble surface
area for a given gas holdup. It appears that the
product Sokm may not be too sensitive to the pres-
0.01 ence of surface active solutes in the liquid.
The rate constant Sokm for the transport controlled
reaction is obtained from the combination of equa-
tions 2.2.17, 2.2.24 and 2.2.25 as given below for
an average solute diffusivity D = 5 x 10 –9 m2s–1.
5.43 ´ 10 -4 (e U s )
1
2
0.001
1
·
4 10 40 100 So k m = 3
(2.2.26)
VT (m3 s-1) at 1600°C and 1.5 atm. d 2
Fig. 2.16 Calculated rate constant for mass-trans- The rate constant thus calculated• is plotted in Fig.
fer controlled reaction of gas bubbles in 220 + 20 2.16 against the gas flow rate VT at 1600°C and
ton steel bath in the BOF or OBM (Q-BOP) vessel. for an average gas pressure of 1.5 atm in the steel
From Ref. 27. bath for 220 + 20 ton heats.
2.2.2.4.4 Maximum Rate of Degassing of Liquid Steel with Argon Purging If the rates of diffusional
processes and chemical reactions are sufficiently fast, the argon bubbles traversing the melt will be
saturated with N2 or H2 to the corresponding equilibrium values for the concentrations present in the
melt. It is for this limiting case that the following rate equation is applicable.
Denoting N and H by X, and using molar contents in the gas bubbles,
d ( ppm X ) 10 -6
nÇ X 2 = ´ 10 6 g-mol
g - molX2X 2 min ton
/min ton (2.2.27)
dt M
V˙
n˙ Ar = g-mol
g - molAr/min
Ar mintonton (2.2.28)
22.414 ´ 10 -3
•
where M is the molecular mass of X2 and V the argon blowing rate in Nm3 min–1t –1
. In the X2-sat-
urated argon bubbles for n× Ar >> n× X2
n˙ X 2 d ( ppm X ) 22.414 ´ 10 -3
pX 2 = P= P (2.2.29)
n˙ Ar dt MV˙
_
where P is the average bubble pressure in the melt.
Since the gas bubbles are in equilibrium with the melt
[ppm X ] d ( ppm X ) 22.414 ´ 10 –3

2
pX 2 = = P (2.2.30)
K2 dt MV˙
where K is the equilibrium constant. The integration of the above rate equation gives with Xo being
the initial concentration,
1 1 MV
- = t (2.2.31)
ppm X ppm X o 22.414 ´ 10 –3 K 2 P
For the average liquid steel temperature of 1650°C at turn down, the values of the equilibrium con-
stants are as given below.
For N 2 : K = 459 ü
ý with p X 2 in atm
For H 2 : K = 27.2þ
_
For an average gas bubble pressure of P = 1.5 atm in the melt, the following rate equations are
obtained.
1 1
- = 0.00395 V˙ t (2.2.32)
ppm N ppm N o
1 1
- = 0.0804 V˙ t (2.2.33)
ppm H ppm H o
As discussed later in detail, the maximum rates are not observed in steelmaking processes; the rates
are limited by mass transfer and or chemical kenetics. Nevertheless equations 2.2.32 and 2.2.33
indicate, even for the limiting case, large quantities of purging gas are required and the processes
are normally not feasible.
2.2.3 Chemical Kinetics

The theory of the absolute reaction rates is based on the concept of the formation of an activated
complex as an intermediate transition state, Activated complex
which has an infinitesimally short life time of
the order of 10–15 second. For an indepth study
of the theory of reaction kinetics, to which out-
standing contributions were made by Eyring DG°
and co-workers, reference may be made to the
Reactants
Energy
classical text books by Glasstone et al.30 and by

Hinshelwood.31
As illustrated in Fig. 2.17, there is a change in
energy profile accompanying the reaction that DG° Products
involves the formation and decomposition of an
activated complex. While the change in free
energy accompanying reaction is DG° < 0, the
Reaction path
activation energy DG* > 0.
In almost all pyrometallurgical processes, we Fig. 2.17 Change in energy profile during the for-
are concerned with heterogeneous reactions mation and decomposition of the activated complex
involving an interface between two reacting involved in the reaction.
phases, e.g. solid-gas, solid-liquid, liquid-gas and two immiscible liquids (slag-liquid metal). For the
case of a fast rate of transport of reactants and products to and from the reaction site, the rate is con-
trolled by a chemical reaction occurring in the adsorbed layer at the interface. The reaction between
adsorbed species L and M on the surface producing product Q occurs via the formation of an acti-
vated complex (LM)*.
L + M = (LM ) ® Q
ì stage I ü ì activated ü ì stage II ü (2.2.34)
í ý í ý í ý
îreactantsþ î complex þ îproducts þ
The theory of the absolute rates states that the activated complex is in equilibrium with the reac-
tants for which the equilibrium constant K* for constant temperature is
a*
K* = (2.2.35)
aL a M
where a’s are the thermodynamic activities.
Next to be considered is the specific rate of decomposition of the activated complex to the overall
reaction product Q, represented by
dn æ kT ö
= ç ÷ Go q* (2.2.36)
dt è h ø
where
dn/dt = the reaction rate, mol cm–2s–1,
k = the Boltzmann constant, 1.380 ´ 10–23 J K–1,
h = the Planck constant, 6.626 ´ 10–34 J s,
T = temperature, K,
G° = total number of adsorption sites on the surface, 1015 mole cm–2,
q* = fractional coverage by the activated complex.
For single site occupancy by the activated complex in the adsorbed layer, the activity of the com-
plex is represented by
q*
a * = j* (2.2.37)
1- q
where q is the total fractional occupancy of the sites by the adsorbed species and j* is the activity
coefficient of the complex in the chemisorbed layer. Combining equations 2.2.35 - 2.2.37 gives the
rate of forward reaction in terms of the absolute reaction rate theory.
dn æ kT ö æ K * ö
Rf = = ç ÷ Go
dt è h ø çè j * ÷ø
(1 - q) {aL aM} (2.2.38)
The thermodynamics of the chemisorbed layer at the interface, i.e. value of Go, K* and ,j are not
known, therefore the isothermal rate equation is given in a simplified general form thus
Rf = Ff (1 – q){a L a M} (2.2.39)
where Ff is the isothermal rate constant of the forward reaction.
As the reaction progresses, concentrations of the reactants L and M decrease while the concentra-
tion of the product Q increases. Because of these composition changes and the influence of the
reverse reaction Q®L + M, the rate decreases with an increasing reaction time. The rate of the
reverse reaction is represented by
Rr = – Fr (1 – q){a Q} (2.2.40)
where Fr is the rate constant of the reverse reaction. Therefore, the net overall rate of reaction is
dn
dt
= F f (1 - q){a L a M } - F r (1 - q) {a } Q
(2.2.41)
When the rates of forward and reverse reactions are the same, i.e. dn/dt = 0, the reaction is said to
be at an equilibrium state. It follows that the ratio of the rate constants Ff/ Fr is the equilibrium
constant of the reaction.
F f æ aQ ö
K= =ç ÷
F r è a L a M ø eq (2.2.42)
In terms of a single rate constant, the net reaction rate is formulated as

dn
dt
{
= Ff (1 - q ) a L a M - ( aL a M )eq } (2.2.43)
where (a La M)eq is the equilibrium value for the activity aQ in stage II at any given reaction time.
For a given surface coverage, the temperature dependence of the rate constant is represented by
é æ - DH * ö ù
F = êF o expç ÷ú (2.2.44)
êë è RT ø úû q
where Fo is a pre-exponential constant and

DH* is the apparent heat of activation for the
)
reaction at a given site fillage q.

Since the activation energy DG* is always
In rate constant (increasing
positive, the enthalpy of activation for the DH*

Slope = -
reaction is also positive. As shown in Fig. R
2.18 in the plot ln F vs 1/T the slope of the
line gives DH*. It should be noted that
since the thermodynamic quantities for the
activated complex (K*/j*) in the chemi-
sorbed layer are not known, the DH*
derived from the rate measurements over a
sufficiently large temperature range, is the
apparent heat of activation.
I
Another aspect of the kinetics of interfacial (increasing )
T
reactions is the rate of chemisorption or des-
orption at the reaction surface as given by Fig. 2.18 Effect of temperature on rate constant for
the following theoretical equation for an reaction involving an activated complex.
uncontaminated surface.
æ - DH * ö
F f = (2 p Mi R T )
- 12
exp ç ÷ (2.2.45)
è RT ø
where Mi is the molecular mass of the adsorbed species. This equation is transformed to the fol-
lowing form for the maximum rate of vaporization, i.e. free vaporization, from an uncontaminated
surface at low pressures
pi
Ratemax = (2.2.46)
2 p Mi R T
where pi is the vapor pressure in atm, for which the equation is reduced to the following form.
–1⁄2
Ratemax (g-mol cm–2 s–1) = 44.3 pi (MT) (2.2.47)
2.2.3.1 Examples of Experiments on Rates of Interfacial Reactions

2.2.3.1.1 Nitrogen Transfer Across Iron Surface The rate of reaction of gaseous nitrogen with liquid
and solid iron, in the presence of surface active alloying elements, has been studied by many inves-
tigators since the late 1950s. It was in the late 1960s32–34 that the rate controlling reaction mecha-
nism was resolved, which was further confirmed in the subsequent experimental work done by
Fruehan and Martonik35, Byrne and Belton36, and Glaws and Fruehan37.
In particular the latter investigators used an isotope exchange technique which measured the rate
of dissociation of the N2 molecule. They demonstrated the rate of dissociation was the same as the
rate of absorption of nitrogen, providing strong evidence that the absorption rate is controlled by
dissociation.
When the rate of reaction is not hindered by slow nitrogen transport to and from the gas-metal
interface, the rate of nitrogenation
N2 (g) = N2* ® 2[N] (2.2.48)
is controlled either by the rate of chemisorption or dissociation of N2 molecules on the metal sur-
face. The rate of reverse reaction, i.e denitrogenation, is of course a second order type with respect
to nitrogen dissolved in the metal. The equation below represents the rate of nitrogen transfer from
gas to liquid iron
d[%N ] 100 A
dt
=
rV {
F f (1 - q ) p N 2 - p N 2 ( ) eq } (2.2.49)
where r is the density of liquid iron, A the surface area of the melt on which the nitrogen stream
is impinging and V the volume of the melt. The rate constant Ff, in units of g N cm–2min–1atm–1N2,
is for the forward reaction 2.2.48. The equilibrium partial pressure (pN2)eq corresponding to the
nitrogen content of the melt at the reaction time t, is that given by the equilibrium constant K for
nitrogen solubility.
[% N]
2
(p )
N 2 eq =
K
(2.2.50)
With this substitution the isothermal rate equation is
d[%N ] 100 A ì [%N]2 ü

= F f (1 - q ) íp N 2 - ý
dt rV î K þ (2.2.51)
For constant N2 pressure and temperature, the integration of equation 2.2.51 gives for %N = 0 at
t = 0,
ln
[ ] =2p
K p N 2 + %N 100 A
F f (1 - q) t
- [% N]
N2 (2.2.52)
K pN2 rV
Byrne and Belton made an accurate determination of the rate constant Ff for reaction of N2 with
high purity iron and Fe-C alloys at 1550–1700°C, by measuring the rate of 15N®14N isotope
exchange that occurs on the iron surface, as represented by
-6340 ± ( 710)
log F f = + 1.85( ±0.38) (2.2.53)
T
where the rate constant Ff is in units of g N cm–2min–1atm–1. The apparent heat of activation =
121.4 kJ mol–1 is much lower than the value expected for the rate of dissociation of N2. As
pointed out by Byrne and Belton, the rate of chemisorption of N2 is presumably controlling the
reaction mechanism.
The surface active elements dissolved in iron, e.g. O, S, Se, Te, are known to lower the rate of
nitrogen transfer across the iron surface. On the basis of the experimental rate data with liquid
iron containing O and S, as given in various publications (Refs. 33–38) and the surface tension
data, the effects of O and S on the fraction of vacant sites, 1- q, in the chemisorbed layer may
be represented by
1
1- q =
æ %S ö
1 + 260ç %O + ÷ (2.2.54)
è 2 ø
which is a slightly simplified form of the equation that was derived by Byrne and Belton.
For the chemical reaction-controlled nitrogen removal from liquid iron (or steel) in reduced pres-
sures or in an inert gas stream with very low N2 pressure
2[N] ® N2 (g) (2.2.55)
the integrated form of the rate equation is
F r (1 - q) t
1 1 100 A
- = (2.2.56)
%N %No rV
where %No is the initial nitrogen content and Fr the rate constant = Ff /K.
The solubility of N2 in liquid iron or low alloy steel is given by
[% N ] = - 376 - 2.48
2
log K =
p ( atm )
(2.2.57)
N2 T
Combining this with equation 2.2.53 gives for the rate constant Fr in gN cm–2min–1%N–1
-5964 (2.2.58)
log F r = + 4.33
T
Now let us consider the chemical-reaction controlled rate of denitrogenation. It will be assumed
that the rate is controlled by only chemical kinetics. In actuality it is controlled by mass transfer
and chemical kinetics in series which is discussed later in detail. From equation 2.2.58 the rate for
1600°C Fr = 0.233 g N cm-2s–1%N–1. For liquid steel containing
. 600 ppm O and 120 ppm S, (1 –
q) = 0.055 from equation 2.2.54. For the gas flow rate VT = 34 m3s–1, e = 0.6 and the bubble sur-
face area is Sm = 1.11 cm2g-1 liquid steel. Inserting these numbers in equation 2.2.56 for the limit-
ing case of %Ne » 0, gives
1 1
- = 100 ´ 1.11 ´ 0.233 ´ 0.055t = 1.42 t (2.2.59)
%N %N o
In 60 seconds time of purging, 0.0030% No will be reduced only to 0.0024% N. In actual fact, the
argon bubbles traversing the melt will contain some N2; for example, if the nitrogen partial pres-
sure in gas bubbles were 0.001 atm N2, the equilibrium content in the melt would be 0.0015% Ne.
Therefore, with %Ne > 0, the nitrogen removal will be much less than that calculated above for the
chemical-reaction controlled rate for the hypothetical limiting case of %Ne » 0.
The foregoing predication from the rate equation is in complete accord with the practical experience that
even at high rates of argon purging of the melt, as in OBM(Q-BOP), there is no perceptible nitrogen
removal primarily because of the presence of surface active solutes, oxygen and sulfur, in the steel bath.
2.2.3.1.2 CO2 and H2O Reaction with Fe–C Melts Another example of chemical kinetics con-
trolling reaction rates is the decarburization of Fe–C melts.
C + CO2 = 2CO (2.2.60)
Sain and Belton39 and later Mannion and Fruehan40 showed that the rate was affected by surface
active elements and was consistent with CO2 dissociation on the surface controlling the rate.
Nagasaka and Fruehan41 measured the rate of decarburization by H2O.
H2O + C = CO + H2 (2.2.61)
They also found that the rate was controlled by chemical kinetics and most likely dissociation of H2O
on the surface. For both the CO2 and H2O reaction, the rate constant could be expressed by the rela-
tionship
kp
kC = + kr (2.2.62)
1 + KS [%S]
where k C, k p and k r are the rate constants, k p is the rate constant for pure iron and k r is the residual
rate at high sulfur contents respectively and K s is the absorption coefficient for sulfur on liquid iron.
The first term is based on the analysis by Belton42 and the residual rate which is relatively small is
discussed by Nagasaka et al in detail.41 The overall rate constants are given as a function of sulfur
content in Fig. 2.19.
100
k x 104 (mol/cm2 s atm)
H2O
10
Fig. 2.19 Rates of dissociation

of CO2 and H2O on iron as a
function of sulfur content. From CO2
Ref. 41.
1
Fe-Csat. Melts
T=1873K
0.1
0 0.005 0.01 0.015 0.02 0.025 0.03
[mass % S]
2.2.3.1.3 Chemical Kinetics of Carbon Oxidation in CO2–CO Mixtures Examples of the experimental data
on the initial rate of internal oxidation of granular (0.5 mm dia.) electrode graphite and metallur-
gical coke are given in Figs. 2.20 and 2.21, reproduced from a previous publication.16
In a subsequent publication, Turkdogan27 re-assessed the interpretation of the rate data on the
assumption that both CO and CO2 are chemisorbed on the pore walls of the carbon and the rate of
oxidation is due to the rate of dissociation of chemisorbed CO2. For the limiting case of complete
internal burning, the rate equation will be in the following form
F ¢ p CO 2
Rate = t (2.2.63)
1 + k 1 p CO + k 2 p CO 2
10–1

1200°C
10–2
1100°C
10–3
10–4 1000°C
10–5
0 20 40 60 80 100
CO2 (%)
Fig. 2.20 Effects of temperature and gas composition on the rate of oxidation of electrode graphite granules
(~0.5 mm dia.) in CO2–CO mixtures at 0.96 atm total pressure. From Ref. 16.
100
°C
1400
10–1 1300
1200
1100
10–2
1000
10–3
900
10–4
800
10–5
700
10–6
0 25 50 75 100
CO2 (%)
Fig. 2.21 Effect of temperature and gas composition on the rate of oxidation of granular metallurgical coke
(~0.5 mm dia.) in CO2–CO mixtures at 0.96 atm total pressure. From Ref. 16.
where F¢ is the rate constant (min–1atm–1CO2) for a given temperature and a particular type of car-
bon; the constants k1 and k2 are associated with the chemisorption of CO and CO2. It should be
noted that in compliance with the notation in equation 2.2.64 for the ideal monolayer, the constants
k1 and k2 are reciprocals of the respective activity coefficients of the absorbed species, i.e. k1 =
1/jCO and k2 = 1/jCO2; the fraction of vacant sites being (1 - q ) = (1 - qCO - qCO2).
The fraction of surface sites qi covered by the adsorbed species is related to the solute activity as
given in the following Langmuir adsorption isotherm.
qi
ai = ji (2.2.64)
1 - qi
From the cited experimental data for the oxidation of graphite and coke the following equations are
obtained for the temperature dependence of F¢, j CO = 1/k1 and jCO2 = 1/k2.
For electrode graphite:
(
log F ¢ min –1 atm –1 CO2 = - ) 16 ,540
T
+ 10.75 (2.2.65)
log j CO ( atm ) = -
8719 (2.2.66)
+ 4.84
T
log j CO 2 ( atm ) = -
590 (2.2.67)
- 0.072
T
For metallurgical coke:
(
log F ¢ min –1 atm –1 CO2 = - ) 16 ,540
T
+ 11.37 (2.2.68)
log j CO ( atm ) = -
2117
+ 0.27 (2.2.69)
T
log j CO 2 ( atm ) =
3840 (2.2.70)
- 3.45
T
It should be noted that these equations fit the measured initial rates of oxidation within a factor of
about 1.5 for temperatures 800 to 1200°C and for CO2 pressures 0.03 to 3 atm. For easy compari-
son of the reactivities of electrode graphite and metallurgical coke, the numerical values are given
below from the above equations for 900 and 1200°C.
Electrode graphite Metallurgical coke
Temperature °C F¢ jCO jCO2 F¢ jCO jCO2
————————— —————————————————— —————————————————
900 4.5 x 10 –4 0.0025 0.27 –3
1.9 x 10 0.029 0.67
1200 3.3 x 10–1 0.083 0.34 1.4 x 10° 0.068 0.14
After about 3 to 5 percent of initial oxidation, the pore surface area of the coke samples were found
to be four to five times greater than the graphite samples; this is consistent with the F¢ values for
coke being greater than for graphite by a similar factor. The extent of CO adsorption on the pore
walls of electrode graphite or coke is greater than the CO2 adsorption.
There are variations in the reported values of the apparent heat of activation for oxidation of the
electrode (or reactor grade) graphite in CO2. For example, Gulbransen et al.43 found DH* = 368 kJ
mol-1 while according to Blackwood’s work 44 DH* = 260 kJ mol–1; in the present case DH* = 317 kJ
mol-1. Reference should be made also to the papers of Ergun, 45 Hedden and Lowe,46 and by
Grabke47 for various other interpretations of the kinetics of oxidation of carbons.
Aderibigbe and Szekely48 also investigated the oxidation of metallurgical coke in CO2–CO mix-
tures at 850 to 1000°C. Their rate constants K1 (º F¢) are similar to those given by equation 2.2.68
their estimate of the apparent heat of activation being DH* = 249±47 kJ mol–1.
Recently Story and Fruehan49 examined the oxidation of carbons in CO–CO2 gas mixtures at high
temperatures (>1300°C) under conditions of high external mass transfer. The gas mixture was jet-
ted at high velocities, reducing the resistance to mass transfer at the surface. They were in limiting
case of external oxidation and determined the parameter (frS De) from the data, and then using
published values of S and computed values of De, computed the rate constant. The results are com-
pared to the rates measured at lower temperatures in Fig. 2.22.
Temperature (°C)
1545 1393 1265 1155 1060 977
10–6
10–7
Grabke
(extrapolated)
Rate (mol/cm2 s atm)
10–8
Fig. 2.22 Arrhenius plot
of graphite–CO2 data.
10–9
This work
CP graphite
10–10 UHP graphite
Turkdogan et.al.
Guibrannsen et. al. (extrapolated)
10–11
0.55 0.6 0.65 0.7 0.75 0.8
1/T x 103 (K–1)
2.2.4 Mixed Control

In many metallurgical operations the rate is not controlled by a single reaction step such as mass
transfer or chemical kinetics but rather two steps in series such as mass transfer followed by chem-
ical kinetics. By equating the fluxes for each of the reaction steps it is possible to obtain expres-
sions for the overall rate, or a series of expressions, which can be solved numerically.
2.2.4.1 Mixed Control for Gas Phase Mass Transfer and Chemical Kinetics
In general the temperature dependence of chemical kinetics is greater than for gas phase mass
transfer. Consequently at low temperatures chemical kinetics is predominantly controlling the rate
while at high temperatures mass transfer dominates. However for many reactions over a large range
of conditions including temperature, fluid flow conditions and the concentration of surface active
elements which effect the chemical rate, the rate is controlled by both processes in series.
For example, the rate of dissociation of CO2 and H2O on liquid iron is controlled by both gas phase
mass transfer and chemical kinetics in series over a large range of temperature and fluid flow con-
ditions. In this case the pressure of the reacting gas at the surface is not the bulk composition. The
flux of the reacting gas and the rate of the reaction are given by 2.2.71 and 2.2.72 respectively,
m
J = i piB - pis
RT
( ) (2.2.71)
R = k (pis – pei ) (2.2.72)
where
mi = the gas phase mass transfer coefficient for i,
p Bi = the pressure of i in the bulk phase,
p si = the pressure of i at the surface,
p ei = the equilibrium pressure of i,
k = the chemical rate constant.
Equations 2.2.71 and 2.2.72 assume that the chemical rate is first order and the reaction is equil-
molar in terms of the moles of reactant and product gases or the pressure of the reaction gas is
low. As it is not possible to accumulate component i at the surface, the processes are assumed to
be at steady state and the flux and chemical rate are equated. It is then possible to calculate the
pressure at the surface and the rate in terms of the bulk composition. Further simplification gives
R=
1
(
p B - pie
1 RT i
)
+ (2.2.73)
k mi
Equation 2.2.73 is analogous to an electrical current with two resistances in series where k and
mi/RT are analogous to conductivities. The first term in equation 2.2.73 is often referred to as the
observed or overall rate constant.
For example Nagasaka and Fruehan41 measured the rate of dissociation of H 2O on liquid Fe-C-S
alloys from 1673 to 1873°K for varying sulfur contents. At 1873K and 0.1% S the measured over-
all rate constant was 2.5´10–4 mole/cm2 s atm for conditions where the mass transfer coefficient
was 100 cm/s. For this case the chemical rate was computed to be 3.9´10-4 mole/cm2 s atm.
Similarly the nitrogen and carbon dioxide reactions can be controlled by gas phase mass transfer
and chemical kinetics and equations 2.2.71 – 2.2.73 can be applied.
2.2.4.2 Mixed Control for Chemical Kinetics and Liquid Phase Mass Transfer
For many conditions in steelmaking and refining the rate of pick up or removal of nitrogen from
liquid iron is controlled by chemical kinetics and liquid phase mass transfer in series. Fruehan,
Lally and Glaws50 developed a mixed control model and applied it to several metallurgical
processes. For nitrogen gas reacting with iron the flux of nitrogen for mass transfer and the chem-
ical reaction are given by 2.2.74 and 2.2.75.
JN =
mr
100
[
%N s - %N ] (2.2.74)
æ
[ ] ö÷
2
%N s
JN = k çpN 2 -
ç K ÷ (2.2.75)
è ø
where
JN = flux of nitrogen (g/cm2 s),
m = mass transfer coefficient for nitrogen (cm/s),
k = chemical rate constant (g/cm2 s atm),
%Ns = nitrogen content at the surface,
K = equilibrium constant for nitrogen reaction with liquid iron,
r = density of iron (g/cm3).
Since no nitrogen can be accumulated at the surface it is possible to equate the two flux equa-
tions, 2.2.74 and 2.2.75, and solve for the nitrogen content at the surface. The equation is a qua-
dratic and is solved numerically for small time increments for specific conditions.. The surface
nitrogen content is then used to compute the rate of nitrogen absorption or removal. This tech-
nique was applied to nitrogen absorption into liquid iron when bubbling nitrogen in the ladle
and typical results are shown in Fig. 2.23. The rate of nitrogen pick up for low sulfur steel is
considerably faster than for high sulfur steel, as sulfur decreases the chemical rate. Below
0.005% S the rate does not increase further as liquid phase mass transfer then dominates and it
is not affected by sulfur content.
0.010
0.009
0.008
0.005% S
Nitrogen (%)
0.007
0.006
0.03% S
0.005
0.004
0 4 8 12 16 20
Time (minutes)
Fig 2.23 Rate of nitrogen pick-up in 200 metric tons of steel containing 0.005 and 0.030% S, using 10 scfm
of N2. From Ref. 50.
2.3 Properties of Gases

Of importance to the iron and steelmaking and related pyrometallurgical processes, the selected
thermochemical and transport properties of gases are presented in this section.
2.3.1 Thermochemical Properties

2.3.1.1 Heat Content
The molar heat contents of simple selected gases of interest to pyrometallurgical processes are
plotted in Fig. 2.24 over the temperature range 300–1900K.
80
SO2
CO2
H2S
60
DHT - DH298 (kJ mol-1) H2O
O2
CO
N2
40
H2
s
g ase
o mic
nat
20 Mo
0
300 700 1100 1500 1900
Temperature (K)
Fig. 2.24 Heat contents of simple gases.
2.3.1.2 Equilibrium States in Gas Mixtures

When assessing the state of reactions involving gas mixtures, the partial pressures of the gaseous
species should be evaluated from the reaction equilibrium data for all the pertinent gas reactions. The
principle of the computation is demonstrated by the following example for the gas mixture
CO–CO2–H2O–CH4.
This example is in relation to reheating a reformed gas in the direct iron ore reduction processes.
If there is complete gas equilibrium at the reheating temperature and pressure, the activity of car-
bon in the equilibrated gas mixture must satisfy the following three basic reaction equilibria:
2CO = C + CO2 (2.3.1)
aC p CO 2
K1 = (2.3.2)
p 2CO
H 2 + CO = C + H 2O (2.3.3)
aC pH 2O
K2 = (2.3.4)
p H 2 p CO
CH 4 = C + 2 H 2 (2.3.5)
a 2
C pH 2
K3 = (2.3.6)
p CH 4
For the predominant gaseous species the following equalities are derived from the mass balance for
(SC), (SO) and (SH).
SC æ pCO + pCO + pCH4 ö æ pCO2 + pCO + pCH4 ö

= 2
÷ = ç 2p + p + p ÷
SO çè 2pCO 2
+ p CO + p H 2 øi
O è CO2 CO H 2O ø e
(2.3.7)
SC æ pCO2 + pCO + pCH4 ö æ pCO2 + pCO + pCH4 ö

=ç ÷ =ç ÷
SH è 2pH 2O + 2pH 2 + 4pCH4 ø i è 2pH 2O + 2pH 2 + 4pCH4 ø e
(2.3.8)
where i and e indicate partial pressures in the ingoing and equilibrated gas mixtures, respectively.
The equation below gives the total pressure P of the equilibrated gas mixture.
(pCO + pCO2 + pH2 + pH2O +pCH4) = P (2.3.9)
e
The partial pressures of gaseous species and the activity of carbon in the equilibrated gas mixture
are computed by simultaneous solution of equations 2.3.1 to 2.3.9 with known values of the equi-
librium constants K1, K2, and K3.
Examples are given in Fig. 2.25 showing the calculated carbon activities for 4 atm total pressure
and for an initial gas mixture containing 73% H2, 18% CO, 8% CO2, 1% CH4 to which 0.2% to
6.0% H2O has been added. The carbon deposition is imminent when its activity in the equilibrated
gas exceeds unity, with respect to graphite. With the addition of 0.2% H2O to this gas mixture, there
should be no carbon deposition in the equilibrated gas mixture at temperatures below 512°C and
above 720°C. In an earlier study of the reduction of iron oxides in a similar gas mixture, carbon
deposition was observed at all temperatures below 1000°C (see Ref. 51), indicating lack of com-
plete gas equilibrium even at 1000°C. These experimental findings suggest that carbon deposition
may occur during reheating of reformed natural gas in industrial operations, even when the calcu-
lated carbon activity is less than unity in the gas mixture under equilibrium conditions.
1.2
H2O (%)
0.2
1.1
2.0
Activity of carbon
3.0
1.0
4.0
6.0
0.9
0.8
400 500 600 700 800
Temperature (°C)
Fig. 2.25 Calculated activity of carbon for complete gas equilibrium at 4 atm and indicated temperatures for
reformed natural gas containing 73% H2, 18% CO, 8% CO2 and 1% CH4 to which an indicated amount of H2O
is added. From Ref. 52.
The activity of carbon cannot be calculated for a 100

nonequilibrated gas mixture. However, for the 50
purpose of identifying the reactions that may be
responsible for carbon deposition, we may cal- CO + H2 = C + H2O
20
culate carbon activities for individual reactions
on the assumption that there is no change in the 10
Activity of carbon
composition of the inlet gas upon heating. As is
seen from the example given in Fig. 2.26 the car- 5
bon activities for individual reactions differ con-
siderably from those calculated for complete gas 2CO = C + CO2
2
equilibrium. For complete gas equilibrium, the
carbon activity is below unity at temperatures 1
below 540 and above 700°C, while for the indi-
vidual reactions the carbon activity is below 0.5 For complete
gas equilibrium
unity only within the temperature range
880–1060°C. Therefore, in the nonequilibrated 0.2 CH4 = C + 2H2
gas, there can be no carbon deposition only
0.1
within the temperature range 880–1060°C. The 400 500 600 700 800 900 1000
temperature at which carbon deposition may Temperature (°C)
start depends on the relative rates of reactions
2.3.1 and 2.3.3: below 730°C if reaction 2.3.1 is Fig 2.26 Calculated activity of carbon for complete
fast, and below 880°C if reaction 2.3.3 is fast. gas equilibrium at 4 atm. compared with the activity
of carbon for individual reactions in a reformed inlet
On the basis of the foregoing reasoning, we may gas containing 2% H2O. From Ref. 52.
compute a diagram of % H2O in the inlet
reformed gas versus temperature, as in Fig. 2.27, to delineate the regions where (I) carbon will not
deposit, (II) carbon may deposit, and (III) carbon will deposit. With increasing pressure, regions II and
III move to higher temperatures and I to higher concentrations of H2O, and chances of carbon deposi-
tion become greater. From the equilibrium constant for reaction 2.3.3, we can also compute the criti-
cal pH2O/pCO ratio for possible carbon deposition at any temperature and partial pressure of H2 in the
inlet gas. The possibility of carbon deposition in region II depends much on the catalytic behavior of
the inner surface of the reheater tubes.
4
H2O (volume %)
III II I II
Fig. 2.27 Regions of Carbon Carbon Carbon Carbon

will may will not may
carbon deposition at deposit deposit deposit deposit
reheating temperatures
of reformed natural gas
at 1 atm containing 2
For reaction
indicated amounts of H2 + CO = C + H2O
H2O. From Ref. 52.
For reaction
For complete CH4 = C + 2H2
gas equilibrium
0
600 700 800 900 1000
Temperature (°C)
2.3.1.3 Catalytic Decomposition of Carbon Monoxide

The catalytic decomposition of carbon monoxide has been the subject of much study, because of the
occurrence of carbon deposition in many processes, e.g., blast furnaces, the Stelling process, the Fis-
cher-Tropsch process, nuclear reactors, and so on. References to early work on this subject are listed
in an annotate bibliography by Donald53. The consensus is that iron, cobalt, and nickel are the most
effective catalysts for the decomposition of carbon monoxide. Kehrer and Leidheiser54 reported vir-
tually no catalytic effect of copper, silver, chromium, molybdenum, palladium, and rhenium.
In a comprehensive study of carbon deposition in CO–H2 mixtures at temperatures of 450–700°C,
Walker et al.55 observed that the C/H atom ratio in the deposit increased from 10 to 50 with increas-
ing reaction temperature and increasing CO/H2 ratio in the gas. They also found that the properties
of carbon deposit were affected by the amount accumulated on the catalyst, e.g., with increasing
thickness of the carbon layer, the crystallinity became poorer, the surface area increased, the elec-
trical conductivity decreased, and the C/H2 ratio increased. Another important finding was that car-
bon deposition ultimately ceased when most of the iron was converted to cementite. Upon
hydrogen treatment, cementite decomposed to iron and graphite, but the regenerated iron lost most
of its reactivity as a catalyst. Similar observations were made by Turkdogan and Vinters51 in sub-
sequent studies of the catalytic decomposition of carbon monoxide in the presence of porous iron.
It is generally agreed that iron, not cementite, catalyzes the decomposition of carbon monoxide. The
chemisorption of H2 and CO on the surface of iron is believed to approach equilibrium rapidly. The
carbon formed by reactions 2.3.1 and 2.3.3 migrates across the surface to a nucleating center where
cementite and free carbon are deposited. Based on x-ray diffraction and electron microscopic stud-
ies, Ruston et al.56 suggested that although the decomposition of carbon monoxide was catalyzed by
iron, an iron carbide Fe7C3 formed as an intermediate step in the decomposition of the activated
reaction product to graphite.
6
(a)
4 600°C
Rate x 103 (mole C min-1)
400°C
2
800°C
0
4
(b)
600°C
2
400°C
800°C
0
0 20 40 60 80 100
H2 (volume %)
Fig. 2.28 Effect of gas composition (H2–CO) on rate of carbon deposition on 660-mg porous iron granules at
1 atm(a) by reaction 2CO ® C + CO2 and (b) by reaction CO + H2 ® C + H2O. From Ref. 57.
As is seen from the experimental results of Olsson and Turkdogan 57 in Fig. 2.28, the rate of car-
bon deposition on an iron catalyst in CO–H 2 mixtures is a complex function of temperature and
gas composition. From the measured rate of carbon deposition and the measured CO 2/H2O in
the exhaust gas, the rates of reactions 2.3.1 and 2.3.3 were determined. These measured rates
are shown in Fig. 2.29 as a function of gas composition at 400, 600 and 800°C. The hydrogen
appears to have a dual role: at low concentrations the hydrogen catalyzes reaction 2.3.1, and at
high concentrations of hydrogen reaction 2.3.3 contributes directly to carbon deposition.
600°C
Rate x 103 (mole C min-1)
400°C
800°C
0
0 20 40 60 80 100
H2 (volume %)
Fig. 2.29 Effect of gas composition (H2–CO) on rate of carbon deposition on 660–mg porous iron granules
at atmospheric pressure and indicated temperatures. From Ref. 57.
Although H2O is expected to retard or inhibit carbon deposition by virtue of reverse reaction
2.3.4, it has been found 57 that an addition of 0.5%–2% H 2O to CO–CO2 mixtures enhances the
rate of carbon deposition at 400–600°C. A similar catalytic effect of H 2O has been observed for
carbon deposition from methane in the presence of nickel 58. We might mention in passing that
the rate of carbon formation by the disproportionation of carbon monoxide is much faster than
the pyrolysis of methane under similar conditions. 59 Ever since the first observations made by
Byrom60 and soon after by Carpenter and Smith 61, it has become generally recognized that sul-
fur in the gas retards or inhibits the catalytic action of iron in the decomposition of carbon
monoxide. This is due to the coating of iron surface with cementite, which does not decompose
readily in the presence of sulfur. In fact, in the Stelling process 62 the iron ore is reduced to
cementite in a CO–CO 2 atmosphere without the formation of free carbon, probably because of
the presence of sulfur in the system.
The nitrogen-bearing gaseous species such as NH3 and (CN)2 are also known to retard the decom-
position of carbon monoxide.62–63 Again, this may be attributed to extended metastability of cemen-
tite in the presence of nitrogen. The retarding effect of sulfur and nitrogen on carbon deposition is
similar to sluggish graphitization of steel (by decomposition of cementite) in the presence of sul-
fur and nitrogen in solution in steel.
The carburization of iron in gas mixtures containing H2 and CO following reactions similar to 2.3.1
and 2.3.3 occurs on iron surfaces. These reactions are relevant to the early stages of metal dusting
and were studied by Fruehan and Martonik.64 They also found that the rate of reaction 2.3.3 is
faster than 2.3.1 and contributes significantly to the carburization of iron64 as shown in Fig.2.30.
When CO is replaced by H2 the rate increases significantly. The H2–CO reaction rate constant is
approximately tenfold greater than that for CO for carburization.
2.3.2 Transport Properties

The molecular transfer of mass, momentum and energy are interrelated transport processes of dif-
fusion under a concentration gradient, viscous flow in a velocity gradient and heat conduction in a
thermal gradient.
The derivation of the transport properties from the rigorous kinetic theory of gases, described in
depth by Chapman and Cowling65 and Hirschfelder et al.66 is based on the evaluation of the inter-
molecular energy of attraction, the collision diameter and the collision integral involving the
dynamics of a molecular encounter, hence the intermolecular force law.
For a brief description of the theoretical equations for the calculation of the transport properties of
gases, reference may be made to a previous publication.52 For the present purpose, it is sufficient
to give selected numerical data on the transport properties of a few gases, relevant to ironmaking
and steelmaking processes.
1273 K
2 1188 K
x 104
cm2 min
gC
i
dC
dt
0
0 20 40 60 80 100
Vol % CO
Fig. 2.30 Initial rates of carburization or iron in H2–CO mixtures (pT = 0.965 atm). From Ref. 64.
Table 2.2 Diffusivity D at 1 atm Pressure: cm2s–1 º 10–4m2s–1
Temperature, K
Gas 300 1000 1500 2000
H2 1.46 10.82 21.07 33.90
Ar 0.18 1.46 2.88 4.56
N2 0.20 1.59 3.05 4.98
O2 0.22 1.67 3.27 5.30
CO 0.20 1.63 3.20 5.19
CO2 0.11 0.93 1.87 3.00
H2O 0.28 2.83 5.79 9.57
Table 2.3 Viscosity h (poise) º g cm–1 s–1 º 0.1N s m–2 º 0.1J m–3s
Temperature, K
Gas 300 1000 1500 2000
H2 0.89 1.95 2.53 3.06
Ar 2.50 5.34 6.98 8.27
N2 1.77 4.07 5.11 6.30
O2 2.07 4.82 6.20 7.64
CO 1.79 4.15 5.34 6.53
CO2 1.49 3.85 5.01 6.05
H2O 1.57 4.77 6.44 7.84
Table 2.4 Thermal Conductivity k 3 104, J cm–1s–1K–1 º 3 102, kg m s–3K–1 º 3 102,

Wm–1K–1
Temperature, K
Gas 300 1000 1500 2000
H2 17.45 39.96 53.87 67.64
Ar 1.97 4.18 5.44 6.44
N2 2.50 6.18 8.15 10.53
O2 2.57 6.68 9.02 11.62
CO 2.53 6.35 8.57 10.96
CO2 1.62 5.49 7.71 9.95
H2O 3.84 13.55 20.20 26.93
2.3.2.1 Interrelations Between Transport Properties

2.3.2.1.1 Viscosity/Thermal Conductivity
For monatomic gases,
15 R (2.3.10)
k= h
4M
where
R = 8.314 J mol–1K–1 (molar gas constant),
M = molecular mass, kg mol–1,
h = viscosity, Nsm–2 º Jm–3s.
For polyatomic gases,

15 R ì 4 C V 3 ü
k= hí + ý (2.3.11)
4 M î15 R 5 þ
where CV is the molar heat capacity at constant volume and
ì 4 C V 3ü
í + ý
î15 R 5 þ
is the Eucken67 correction factor for thermal conductivity of polyatomic gases.
For monatomic gases CV = 3R/2; with this substitution, equation 2.3.11 is reduced to equation
2.3.10. In terms of the molar heat capacity at constant pressure, CV = Cp - R, equation 2.3.11 is
transformed to the following form.
æ 5 Rö h
k = çC P + ÷ (2.3.12)
è 4 øM
2.3.2.1.2 Thermal Diffusivity/Thermal Conductivity The thermal diffusivity, DT, is analogous to mass
diffusivity and is given by the ratio
k
DT = (2.3.13)
rC P
where r is the molar density of the gas.
2.3.2.1.3 Temperature and Pressure Effects: The diffusivity, viscosity and thermal conductivity
increase with an increasing temperature, thus
3
⁄
DµT2 (2.3.14)
1
⁄2
hµT (2.3.15)
1
⁄2
kµT (2.3.16)
The viscosity and thermal conductivity are independent of pressure; however, the diffusivity is
inversely proportional to pressure.
2.3.2.1.4 Molecular Mass Effect
–1⁄
DµM 2 (2.3.17)
1
⁄
hµM2 (2.3.18)
k µ M–1 (2.3.19)
2.3.3 Pore Diffusion

When the pores are small enough such that the mean free path of the molecules is comparable to
the dimensions of the pore, diffusion occurs via the collision of molecules with the pore walls.
Knudsen68 showed that the flux involving collision of molecules with reflection from the surface
of the capillary wall is represented by the equation
2 dC (2.3.20)
J=- vr
3 dx
where –v is the mean thermal molecular velocity = (8kT/pm)1/2, r the radius of capillary tube and
dC/dx the concentration gradient along the capillary. For r in cm and the Knudsen diffusivity DK
in cm2s-1
1
æTö 2
DK = 9.7 ´ 10 r ç ÷
3
(2.3.21)
è Mø
The Knudsen diffusivity being independent of pressure but molecular diffusivity inversely propor-
tional to pressure, it follows that in any porous medium, the Knudsen diffusion predominates at low
pressures where the mean free path is larger than the dimensions of the pore.
For the mixed region of molecular and Knudsen diffusion in capillaries, Bosanquet69 derived the
following expression based on the random walk model in which the successive movements of mol-
ecules are terminated by collision with the capillary wall or with other molecules:
1 1 1
= - (2.3.22)
D DKi Dii
where Dii is the molecular self-diffusivity. For a porous medium of uniform pore structure with
pores of equal size, the Bosanquet interpretation formula gives for the effective diffusivity of com-
ponent i
e D12 DKi
Dei = (2.3.23)
t D12 + DKi
where e is the volume fraction of connected pores, t the tortuosity factor, D12 the molecular diffu-
sivity for a binary mixture 1–2 and Dki the Knudsen diffusivity of component i for a given uniform
pore radius r.
2.3.3.1 Examples of Experimental Data

The experimental data are shown in Fig. 2.31 for effective diffusivity He–CO2 at 20°C. As dis-
cussed in Section 2.2.2.3 the pore structure of iron becomes finer at a lower reduction temperature.
It is for this reason that the Knudsen diffusion effect at lower pressures becomes more pronounced
for porous iron reduced at 800°C as compared to that reduced at 1000°C. More detailed informa-
tion is given in the previous publications on the subject of gas diffusion in porous media.51, 70
1.0
Experimental
Calculated
Limiting slope -1
De (cm2/s)
0.1
800°C 1000°C
0.01
0.1 1.0 10 0.1 1.0 10
Pressure (atm.)
Fig. 2.31 Pressure dependence of effective diffusivity He–CO2 at 20°C in iron reduced from hematite ore in
hydrogen at indicated temperature. From Ref. 70.
It is seen from the data in Fig. 2.32 that the effective diffusivities for H2–H2O measured directly or
calculated from the pore structure considerations, agree well with those derived from the rate mea-
surements within the regime of pore-diffusion control, i.e. from the slopes of the lines in Fig. 2.11.
The temperature has a small effect on gas diffusivities. A marked effect of temperature on the effec-
tive diffusivity seen in Fig. 2.32 is due mainly to the coarseness of the pore structure of iron when
the oxide is reduced at higher temperatures.
Direct measurements
Calculated from pore structure
Derived from reduction data
3
De (cm2/s)
0
400 600 800 1000 1200
Reduction temperature (°C)

Fig. 2.32 H2–H2O effective diffusivity derived from data for rate of reduction of hematite to iron is compared
with that obtained by direct measurements and calculated from pore structure. From Ref. 70.
In the case of internal burning of carbon, the pore structure becomes coarser as the gasification pro-
gresses, resulting in an increase in the effective gas diffusivity. A few examples of the measured
CO–CO2 effective diffusivities for electrode graphite are given in Table 2.5, reproduced from Ref. 17.
Table 2.5 Measured CO–CO2 Effective Diffusivities for Electrode Graphite
Temperature % oxidized De(CO–CO2), cm2 s–1

500 0 0.006
500 3.8 0.016
500 7.8 0.020
500 14.6 0.045
700 0 0.009
700 3.8 0.023
700 7.8 0.029
700 14.6 0.067
800 0 0.012
900 0 0.014
Recently Story and Fruehan49 demonstrated the effect of Knudsen diffusion in the oxidation of car-
bon in CO2–Ar and CO2–He gas mixtures. Normally one would expect the rate to be significantly
faster in CO2–He and CO2–Ar due to their higher molecular diffusivities. In general the diffusiv-
ity is given by 2.3.22. Knudsen diffusion is important when the pore diameter is smaller than the
mean free path of the molecules, where the molecules strike the pore walls and diffusivity does not
depend on the molecule size. For ultra pure graphite the overall rate constant for external oxida-
tion, (SDe) as discussed in Section 2.2.2.2, was the same for CO2–He and CO2–Ar gas mixtures.
For this particular material the pore size was small and Knudsen diffusion predominates. On the
other hand for coke with large pores the rate is significantly faster with He as the dilution gas as
the pores are large and molecular diffusion dominates.
2.4 Properties of Molten Steel

2.4.1 Selected Thermodynamic Data
The thermodynamic properties of liquid iron alloys have been studied extensively, both experi-
mentally and theoretically. Most of these studies were made over a period from 1950 through 1970.
For the present purpose, only a selected set of activity data need to be considered, in relation to the
study and understanding of the chemistry of iron and steelmaking reactions.
2.4.1.1 Solute Activity Coefficients

For low solute contents, as in low-alloy steels, the activity is defined with respect to Henry’s law
and mass percent of the solute
a i = fix[%i] (2.4.1)
such that fi®1.0 as [%i]®0.
Up to several percent of the solute content, log f i increases or decreases linearly with an increas-
ing solute concentration.
log fi = ei[%i] (2.4.2)
The proportionality factor ei is known as the interaction coefficient.
The binary interaction coefficients eii for dilute solutions in liquid iron are listed in Table 2.6, taken
from the data compiled by Sigworth and Elliott.71
Table 2.6 Values of e ii for Dilute Solutions in Liquid Iron at 1600°C. From Ref. 71
Element i e ii Element i eii

Al 0.045 O –0.10
C 0.18 P 0.062
Cu 0.023 S –0.028
Cr 0.0 Si 0.11
Mn 0.0 Ti 0.013
Ni 0.0 V 0.015
In multicomponent melts, as in alloy steels, the activity coefficient of solute i is affected by the
alloying elements for which the formulation is
log fi = eii[%i] + S eij[%j] (2.4.3)
where eij is the interaction coefficient of i as affected by the alloying element j.
d log fi
eji =
[ ]
(2.4.4)
d %j
Selected interaction coefficients in dilute solutions of ternary iron base alloys for C, H, N, O and
S at 1600°C are given in Table 2.7.
Table 2.7 Selected Interaction Coefficients in Dilute Solutions of Ternary Iron Base
Alloys for C, H, N, O and S at 1600°C. From Ref. 71.
Element j eCj eHj eNj eOj eSj

__________________________________________________________________________________________
Al 0.043 0.013 –0.028 –3.9 0.035
C 0.14 0.06 0.13 –0.13 0.11
Cr –0.024 –0.002 –0.047 –0.04 –0.011
Mn –0.012 –0.001 –0.02 –0.021 –0.026
N 0.13 0 0 0.057 0.007
O –0.34 –0.19 0.05 –0.20 –0.27
P 0.051 0.011 0.045 0.07 0.029
S 0.046 0.008 0.007 –0.133 –0.028
Si 0.08 0.027 0.047 –0.131 0.063
For the mass concentrations of carbon and silicon above 1%, as in the blast furnace and foundry
irons, the following values of fsj should be used.
Mass % C or Si:
2.0 2.5 3.0 3.5 4.0 4.5 5.0
C
f S
: 1.79 2.14 2.53 3.05 3.74 4.56 5.75
f SiS : 1.37 1.50 1.64 1.78 1.95 2.10 2.32
2.4.1.2 Free Energies of Solution in Liquid Iron

For the solution of element Xi in liquid iron at mass % X i,
Xi (pure) = [X i] (1 mass %)
the free energy of solution is
æ 0.5585 o ö
DG S = RT ln ç gi ÷ (2.4.5)
è Mi ø
where Mi is the atomic mass (g-atom) and g ºi the activity coefficient (with respect to pure element)
at infinite dilution (%Xi® 0).
For DGS in J mol-1 and substituting log for ln,
æ 0.5585 o ö
DG S = 19.144 T log ç gi ÷ (2.4.6)
è Mi ø
The free energies of solution of various elements in liquid iron are listed in Table 2.8.
2.4.2 Solubility of Gases in Liquid Iron

Diatomic gases such as O2, S2, N2 and H2 dissolve in liquid and solid metals in the atomic form
/2 X2(g)
1
= [X] (2.4.7)
for which the isothermal equilibrium constant is
K=
[%X] (2.4.8)
( )
1
2
pX 2
For ideal solutions, the concentration of X is directly proportional to the square root of the equi-
librium gas partial pressure; this is known as the Sievert’s law.
Table 2.8 Free Energies of Solution in Liquid Iron for 1 mass %: (g) gas, (l) liquid,
(s) solid. From Ref. 71.
Element i g ºi DGs, Jmol-1

Al(l) 0.029 – 63,178 – 27.91T
C(gr) 0.57 22,594 – 42.26T
Co(l) 1.07 1,004 – 38.74T
Cr(s) 1.14 19,246 – 46.86T
Cu(l) 8.60 33,472 – 39.37T
1/2H (g) – 36,377 + 30.19T
2
Mg(g) – –78,690 + 70.80T
Mn(l) 1.30 4,084 – 38.16T
1/2N (g) – 3,599 + 23.74T
2
Ni(l) 0.66 – 20,920 – 31.05T
1/2O (g) – – 115,750 – 4.63T
2
1/2P (g) – – 122,173 – 19.25T
2
1/2S (g) – – 135,060 + 23.43T
2
Si(l) 0.0013 – 131,500 – 17.24T
Ti(s) 0.038 – 31,129 – 44.98T
V(s) 0.10 – 20,710 – 45.61T
W(s) 1.20 31,380 – 63.60T
Zr(s) 0.043 – 34,727 – 50.00T
2.4.2.1 Solubilities of H2, N2 and O2

For the solute content in ppm (by mass) and the gas pressure in atm., the temperature dependence
of the equilibrium constants for H2, N2 and O2 solubilities are given in Table 2.9 from the compiled
data cited in Ref. 27.
Table 2.9 Equilibrium Constants of Solubilities of H2, N2 and O2 in Liquid Iron.
log
[ppm H ] = - 1900 + 2.423 (2.4.9)
(p )
1
2 T
H2
log
[ppm N ] = - 188 + 2.760 (2.4.10)
(p )
1
2 T
N2
log
[ppm O] = 6046 + 4.242 (2.4.11)
( )
1
pO 2
2 T
2.4.2.2 Solubility of Gaseous Oxides

2.4.2.2.1 Solubility of CO Carbon monoxide dissolves in liquid iron (steel) by dissociating into
atomic carbon and oxygen
CO(g) = [C] + [O] (2.4.12)
for which the equilibrium constant (which is not too sensitive to temperature) is given by the fol-
lowing equation for low alloy steels containing less than 1% C.
K=
[%C ][ppm O] = 20 (2.4.13)
p ( atm )
CO
At higher carbon contents, a correction should be made to K regarding the activity coefficients
f OC (log f OC = - 0.13[%C]) and f CC(log f CC= 0.18[%C]).
For the reaction
CO2 (g) = CO(g) + [O] (2.4.14)
[ ]
p CO
K= ppm O = 1.1 ´ 10 4 at 1600o C (2.4.15)
p CO 2
For 800 ppm O in low carbon steel at tap, the equilibrium ratio pCO/pCO2 is 13.75. For this state of
equilibrium, the gas mixture contains 6.8% CO2 and 93.2% CO.
2.4.2.2.2 Solubility of H2O From the free energy of formation of water vapor and the solubilitites of
hydrogen and oxygen in liquid iron, the following equilibrium constant is obtained for the reaction
of water vapor with liquid iron for 1600°C.
H2O(g) = 2[H] + [O] (2.4.16)
[ppm H ] [ppm O] = 1.77 ´ 10

2
K= 6
at 1600o C
p (atm )
H 2O
(2.4.17)
The hydrogen and oxygen contents of low alloy liquid steel in equilibrium with H2–H2O mixtures
at 1 atm pressure and 1600°C are shown in Fig. 2.33.
100
80
60
40
H2 + H2O = 1 atm.
20
0.2
Hydrogen (ppm)
0.1
10 0.05
8
0.01
6
PH O atm. 0.003
2
4
1
0.001 0.004 0.01 0.04 0.1 0.2
Oxygen (wt. %)
Fig. 2.33 Concentrations of hydrogen and oxygen in liquid iron at 1600°C in equilibrium with indicated com-
positions of H2–H2O mixtures.
2.4.2.2.3 Solubility of SO2 From the free energy of formation of SO2 and the solubilities of gaseous
sulfur and oxygen in liquid iron, the following equilibrium constant is obtained for the reaction of
SO2 with liquid iron at 1600°C.
SO2(g) = [S] + 2[O] (2.4.18)
[% S][%O]
2
K= = 1558 at 1600o C (2.4.19)

p ( atm )
SO 2
For the concentrations of sulfur and oxygen present in liquid steel, it is seen that the correspond-
ing equilibrium pressure of SO2 is infinitesimally small. It is for this reason that no sulfur can be
oxidized to SO2 during steelmaking with oxygen blowing.
2.4.3 Iron-Carbon Alloys

2.4.3.1 Fe-C Phase Equilibrium Diagram
Since carbon is one of the most important ingredients of steel, the study of the iron-carbon phase
equilibrium diagram has received much attention during the past several decades. The phase dia-
gram is given in Fig. 2.34.
1600
1537°C d + liquid
0.52%
0.10%
1499°C
d Graphite
0.16% +
liquid
1400 Liquid
1392°C
g + liquid
1200
Temperature (°C)
2.01% 1152°C
g 2.04% 1146°C 4.27%
1000
g + graphite (stable phase)

910°C
g + cementite (metastable phase)
0.68%
800
a+g 738°C
0.020%
0.76% 727°C
a + graphite
0.021% a + cementite
600
0 1 2 3 4 5
Carbon (wt. %)
Fig. 2.34 Iron-carbon phase equilibrium diagram. Dashed lines represent phase boundary for metastable
equilibrium with cementite.
There are three invariants in this system; peritectic at 1499°C, eutectic at 1152°C and eutectoid at
738°C. The phase boundaries shown by broken lines are for the metastable equilibrium of cementite,
Fe3C, with austenite. During heat treatment of steel if sufficient time is allowed, iron-carbon alloys
containing austenite and cementite decompose to austentite and graphite. The solubilitites of graphite
and of cementite in a-iron below the eutectoid temperature are given by the following equations
log %C ( graphite) = -
5250 (2.4.20)
+ 3.53
T
log %C ( cementite) = -
3200 (2.4.21)
+ 1.50
T
The solubility of graphite in pure liquid iron is well established through many independent stud-
ies; the experimental data are summarized by the equation
[%C] = 1.30 + 2.57 x 10–3T(°C) (2.4.22)
If atom fraction NC is used, the same set of data can be represented by the following equation in
terms of log NC and the reciprocal of the absolute temperature.
560 (2.4.23)
log N C = - - 0.375
T
The effect of silicon, phosphorus, sulfur, manganese, cobalt and nickel on the solubility of
graphite in molten iron was determined by Turkdogan et al.72, 73 and graphite solubility in iron-
silicon and iron-manganese melts by Chipman et at. 74, 75 Similar measurements with iron-
chromium melts were made by Griffing and co-workers.76 The experimental data are given
graphically in Fig. 2.35 for 1500°C; the solubility at other temperatures can be estimated from
this plot by using the temperature coefficient given in equation 2.4.22 for binary iron-carbon
melts. In the iron-sulfur-carbon system there is a large miscibility gap. For example, at 1500°C
the melt separates into two liquids containing
phase (I) 1.8% S and 4.24% C and phase (II) 6
26.5% S and 0.90% C. Cr
Mn
2.4.3.2 Activity Coefficient of Carbon

The activity of carbon in liquid iron was mea- 5
Co
sured in many independent studies. The data
Ni
compiled and re-assessed by Elliott et al.77 are
given in Figs. 2.36 and 2.37 as activity coeffi-
Carbon (wt.%)
S
cients gC and f C for two different standard
states. 4
Miscibility
2.4.3.3 Peritectic Reaction gap Al
P
The peritectic reaction occurring in the early
stages of solidification of low carbon steels is Si
3
of particular importance in the continuous
casting of steel. The peritectic region of the
Fe–C system is shown on a larger scale in Fig.
2.38. As the temperature decreases within the
two phase region, d+liquid, the carbon con-
2
tents of d-iron and residual liquid iron 0 5 10 15
increase. At the peritectic temperature 1499°C, Alloying element (wt.%)
d-iron containing 0.10% C reacts with liquid
iron containing 0.52% C to form g-iron with Fig. 2.35 Solubility of graphite in alloyed iron melts
0.16% C. at 1500°C.
C(%)
0 1 2 3 4 5 6
0.8
1260°C
Graphite saturation
1360°C
0.6
Austenite saturation 1460°C
1560°C
0.4
1660°C
1760°C
log gC
0.2
0 Standard state:
ac = 1 for pure graphite
–0.1
–0.2
–0.3
0 0.04 0.08 0.12 0.16 0.20 0.24
NC
Fig. 2.36 Activity coefficient (gC) of carbon in liquid iron.
NC
0 0.04 0.08 0.12 0.16 0.20
1.0
Reference state:
fC = 1 when %C®0 1460°C 1560°C
Graphite
0.8
Eutectic
1360°C 1660°C
1260°C 1760°C
0.6
log fC
Austenite saturation
0.4
0.2
0
0 1 2 3 4 5 6
C (mass%)
Fig. 2.37 Activity coefficient (fC) of carbon in liquid iron.
1550
Liquid
d + liquid
Temperature (°C)
0.16 0.52
1500 Fig. 2.38 Peritectic region of
0.10
binary iron-carbon system.
d +g
d
g + liquid
1450
0 0.2 0.4 0.6
Mass (%C)
The X-ray diffraction data give for the densities of iron-carbon alloys: 7.89 g cm-3 for d–Fe with
0.10% C and 8.26 g cm-3 for g–Fe with 0.16% C. Hence, the d to g phase transformation is accom-
panied by 4.7% volume shrinkage. Because of this shrinkage, the thin solidified shell in the mould
of the caster will contract, producing a gap between the shell surface and mould wall. This situation
leads to an uneven surface which is in partial contact with the mould wall, hence resulting in reduced
heat flux at contracted areas. A reduced solidification growth rate and a nonuniform shell with thin
spots, lowers the resistance of the steel to cracking which my cause a breakout in the mould. This phe-
nomenon was well demonstrated experimentally by Singh and Blazek78 using a bench scale caster;
this subject is discussed further in a
paper by Wolf and Kurz.79
In low alloy steels containing 0.10 to
0.16% C, the solid/liquid ratio at the
peritectic invariant is higher than for Liquid
steels containing more than 0.16% C.
Therefore, due to the peritectic reac-
tion, low alloy steels with 0.10 to 0.16% d + liquid
C are more susceptible to the develop-
Temperature
ment of surface cracks in continuous d

casting than steels with higher carbon
contents. d + g + liquid
2.4.3.3.1 Effect of Alloying Elements on
d+g
Peritectic Invariant As shown schemat-
ically in Fig. 2.39 for the ternary
Fe–C–X, or multicomponent alloy steels, g + liquid
the peritectic reaction occurs over a
g
temperature and composition range.
From the experimental data on the liq-
uidus and solidus temperatures of alloy
steels, the carbon and temperature 0 %C
equivalents of alloying elements have
been evaluated in three independent Fig. 2.39 Schematic representation of phase boundaries in
investigations.80–82 low alloy steels as pseudo binary Fe–C system.
The carbon equivalents of the alloying elements for the peritectic reaction
C(d – Fe) + C (l – Fe) = C (g–Fe) (2.4.24)
are usually formulated as follows
Liquid phase: D %C = 0.52 + SDC XPl [ %X ] (2.4.25)
Delta phase: D %C = 0.10 + SDC XPd [ %X ] (2.4.26)
The changes in peritectic temperatures are formulated as

Liquid phase: DT = 1499o C + SDTPXl [ %X ] (2.4.27)
SDTPXd [ %X ]
(2.4.28)
Delta phase: DT = 1499o C +
The coefficients DCXP and DTXP can be positive or negative, depending on the alloying element X. The
peritectic temperature and carbon coefficients determined by Yamada et al.81 are listed in Table 2.10
with minor numerical adjustments to some of the parameters in accord with the empirical correlations
given in Fig. 2.40, reproduced from Ref. 27.
10
Ni
0 Cr Mn
DTPl (°C/%X)
-10 V
Mo
-20 Si
S P
-30
-60 -50 -40 -30 -20 -10 0 10
DTPd (°C/%X)
0.1
Si Mo
V
0
Cr
Mn
-0.1 Ni
DCPl (%C/%X)
-0.2
P
-0.3
-0.4
S
-0.5
-0.10 -0.08 -0.06 -0.04 -0.02 0 0.02 0.04
DCPd (%C/%X)
Fig. 2.40 Interrelations between peritectic temperature and composition changes. From Ref. 27.
Table 2.10 Peritectic Temperature and Composition Parameters in Accord with the
Curves in Fig. 2.40.
Alloying Composition D°C/%X

______________________ D%C/%X
______________________
Element range, wt.% DTPl DTPd DCPl DCPd
Cr <1.5 –3 –5 +0.022 +0.006
Mn <1.5 +3 +7 –0.085 –0.029
Mo <1.5 –10 –18 +0.055 +0.012
Ni <3.5 +4 +9 –0.082 –0.027
P <0.05 –24 –56 –0.250 –0.080
S <0.03 –28 –64 –0.500 –0.105
Si <0.6 –16 –32 –0.035 –0.010
V <1.0 –6 –11 +0.045 +0.010
The phase boundaries in the peritectic region of the Fe–C–X alloys, projected on the Fe–C–X com-
position diagram, are shown schematically in Fig. 2.41 for alloys where X is: (a) ferrite stabiliser
and (b) austenite stabiliser. In the Fe–C–Mn (or Ni) system, the peritectic reaction occurs at all com-
positons between the peritectic regions of the binaries Fe–C and Fe–Mn (Ni) systems. In the Fe–C
alloys with ferrite stabilisers, the peritectic reaction will not occur beyond a certain concentration of
the alloying element X as shown in Fig. 2.41. This limiting case applies only to high alloy steels.
2.4.4 Liquidus Temperatures of Low Alloy Steels

The liquidus temperatures of low alloy steels are derived from the binary Fe–X systems on the
assumption that the coefficients a = DT/%X are additive in their effects on the melting point of iron.
Below 0.5% C, where the solidification begins with the formation of delta (d) iron, the following
equation would apply
Liquidus T (°C) = 1537 – 73.1[%C] + Sa[%X] (2.4.29)
For the carbon contents within the range 0.5 to 1.0% C, where the solidification begins with the
formation of gamma (g) iron, the following equation is recommended.
Liquidus T (°C) = 1531 – 61.5[%C] + Sa[%X] (2.4.30)
The same coefficients a are used in both equations.
Alloying element X
_________________ Coefficient a, °C/%X
___________________
Al –2.5
Cr –1.5
Mn –4.0
Mo –5.0
Ni –3.5
P –30.0
Si –14.0
S –45.0
V –4.0
2.4.5 Solubility of Iron Oxide in Liquid Iron

Subsequent to earlier studies by various investigators, Taylor and Chipman83 made the most reli-
able measurement of the oxygen solubility in liquid iron in equilibrium with essentially pure liq-
uid iron oxide at temperatures of 1530–1700°C. In a later study, Distin et al.84 extended the
solubility measurements up to 1960°C. The solubility data are represented by the equation
[ ]
6380
log %O sat . = - + 2.765 (2.4.31)
T
The oxygen content of liquid iron oxide in equilibrium with liquid iron decreases with an increas-
ing temperature as given below and reaches the stoichiometric composition (22.27%) at about
2000°C.
Temperature °C
___________________ %O in iron
__________ %O in liquid iron oxide
____________________
1527 (eutectic) 0.16 22.60
1785 0.46 22.40
1880 0.63 22.37
1960 0.81 22.32
2.4.6 Elements of Low Solubility in Liquid Iron

A few elements of low solubility in liquid iron play some role in the steelmaking technology; a
brief comment on the chemistry of such elements in liquid iron is considered desirable.
2.4.6.1 Lead
The break out of the furnace lining is often blamed
on the presence of lead in the melt. Small amounts
of metal trapped in the crevices of the lining are (a) X: Ferrite stabilizer
likely to get oxidized subsequent to tapping. The
lead oxide together with iron oxide will readily
flux the furnace lining, hence widening the cracks
which ultimately leads to failure of the lining. The
solubility of lead in liquid iron is sufficiently high
Liquid
that in normal steelmaking practice there should
be no accumulation of lead at the bottom of the
%X
melt, except perhaps in the early stages of the d + liquid

melting of lead-containing scrap.
At steelmaking temperatures the vapor pressure d+g
d
of lead is about 0.5 atm and the solubility85 in + liquid
liquid iron is about 0.24% Pb at 1500°C increas-
ing to about 0.4% Pb at 1700°C. The free- d+g g + liquid
machining leaded steels contain 0.15 to 0.35% 0 %C
Pb. Evidently lead added to such steels is in solu-
tion in the metal prior to casting and precipitates (b) X: Austenite stabilizer
as small lead spheroids during the early stages of
freezing.
2.4.6.2 Calcium
The boiling point of calcium is 1500°C and its
solubility in liquid iron is very low. Sponseller Liquid
d+
and Flinn86 measured the solubility of calcium in

g+
%X
liqu
iron at 1607°C for which the calcium vapor pres-

id
sure is 1.69 atm. Under these conditions at

1607°C, 0.032% Ca is in solution. That is, the sol-
d+
ubility at 1 atm pressure is 189 ppm Ca. They also g + liquid

g
investigated the effect of some alloying elements d

on the solubility; as seen from the data in Fig.
2.42, C, Si, Ni and Al increase markedly the cal-
cium solubility in liquid iron. In melts saturated 0 %C
with CaC2 the solubility of calcium of course Fig. 2.41 Phase boundaries in the peritectic region of
decreases with an increasing carbon content. the Fe-C-X alloys projected on the composition diagram.
2.4.6.3 Magnesium
The solubility of magnesium in iron-carbon alloys determined by Trojan and Flinn87, is shown is
Fig. 2.43 as functions of temperature and carbon content. Subsequently, Guichelaar et al.88, made
similar measurements with liquid Fe-Si-Mg alloys. Their data have been used in numerous studies
to derive the equilibrium relations for the solubility of magnesium in liquid iron. However, there
are some variations in the interpretation of the above mentioned experimental data. A reassessment
of these experimental data is considered desirable.
The equilibrium relation for the solubility of Mg (in units of mass % atm-1) is represented by
[ ]
Mg ( g ) = Mg
K Mg =
[% Mg ] f Mg
(2.4.32)
P Mg
where fMg is the activity coefficient affected by the alloying elements. In the experiments with the
Fe–C–Mg melts coexistent with liquid Mg, the latter contained less than 2 percent Fe, therefore,
the Mg vapor pressure prevailing in the reactor would be essentially the same as that for pure Mg
for which the following is obtained from the data of Guichelaar, et al.
6730
o
log PMg (atm) = - + 4.94 (2.4.33)
T
0.3
0.2 Si
Ca (wt.%)
Al
0.1
Ni
C
CaC2 - saturated
0
0 2.5 5.0 7.5 10.0
Alloying elements (wt.%)
Fig. 2.42 Effect of alloying elements on the solubility of liquid calcium in liquid iron at 1607°C, correspond-
ing to 1.69 atm pressure of calcium vapor. From Ref. 87.
3
Magnesium (wt. %)
2
°C
26
14
°C
43 Carbon saturation
13
°C
60
12
8.7 5.5 3.2 atm. Mg vapor
0
2 3 4 5
Carbon (wt. %)
Fig. 2.43 Solubility of magnesium in liquid iron–carbon alloys at indicated temperatures and pressures of
magnesium vapor. From Ref. 88.
2.4.7 Surface Tension

The experimental data for the surface tension of liquid iron and its binary alloys have been com-
piled recently by Keene.89 For purified liquid iron, the average value of the surface tension at tem-
perature T(°C) is represented by
sFe = (2367 ± 500) – 0.34(°C), mNm–1 (2.4.34)
Keene derived the following weighted average limiting values of (mNm-1) for dilute solutions of X
in Fe-X binary alloys.
Fe–C: Virtually no effect of C on sFe
Fe–Ce: s = sFe - 700[%Ce]
Fe–Mn: s = sFe - 51[%Mn]; ¶s/¶T = - 0.22
Fe–N: s = sFe - 5585[%N]
Fe–P: s = sFe - 25[%P]
Fe–S: (Discussed later)
Fe–Si: s = sFe - 30[%Si]; ¶s/¶T = - 0.25
Gibbs’ exact treatment of surface thermodynamics gives, for fixed unit surface area and constant
temperature and pressure,
k
ds = - RT S Gi d ( ln ai ) (2.4.35)
i
where Gi is the surface excess concentration of the ith component and ai its activity. For a ternary
system, equation 2.4.35 is reduced to
ds = –RT(G2d(ln a2) + G3d(ln a3)) (2.4.36)
Since carbon dissolved in iron has virtually no effect on the surface tension of liquid iron, for the
ternary system Fe–C–S, equation 2.4.36 is simplified to
ds = –RTGsd(ln as) (2.4.37)
As is seen from the compiled data in Fig. 2.44, the experimental results of various investigators for
the Fe–S melts are in close agreement. References to these data are given in an earlier publication.90
2000
1800
1600
Surface tension (mN m–1)
1400
1200
1000
800
600
0 0.2 0.4 0.6 0.8 1.0
S (mass %)
Fig. 2.44 Surface tension of Fe–S alloys at 1550-1600°C. From Ref. 90.
The classical work of Kozakevitch91 on the surface tension of Fe–C–S melts at 1450°C are repro-
duced in Fig. 2.45
The points read off from the curves in Figs. 2.44 and 2.45 are plotted in Fig. 2.46 as s versus log
aS. It is seen that for aS (%S x fS) > 0.01, s is a linear function of log aS, a limiting case for almost
complete surface coverage with chemisorbed S. The shaded area represents the data of Selcuk and
Kirkwood92 for 1200°C.
1300
1100
Surface tension (mNm–1)
900
700
500
0 0.4 0.8 1.2 1.6 2.0 2.4
Sulfur activity
Fig. 2.45 Surface tension of liquid Fe–C–S alloys at 1450°C. From Ref. 91.
2000
1800
1600
Surfact tension (mNm–1)
(Fe–S) 1575°C
1400
Fig. 2.46 Surface tension of Fe–S,
Fe–C–S and Fe–C–Si–S alloys related to
sulfur activity; aS®%S when %C and 1200
%Si®0. From Ref. 90.
(Fe–C–Si–S) 1200°C
1000
(Fe–C–S) 1450°C
800
600
–3 –2 –1 0
Log aS
Numerous studies have been made of the surface tension of the Fe–O melts. The more recent exper-
imental data93,94 are reproduced in Fig. 2.47. At oxygen activities aO (%O x fO) > 0.1, there is essen-
tially complete surface coverage with the chemisorbed oxygen.
2200 Jimbo and Cramb93
Kasama et al94
2000
Surface tension (mNm –1)
1800
1600
1400
1200
° » 1.1 x 1019 mol m–2

GO
1000
800
–9 –8 –7 –6 –5 –4 –3 –2 –1
In a O
Fig. 2.47 Effect of oxygen on surface tension of liquid iron at 1550°C.
2.4.8 Density
The temperature dependence of the densities of liquid iron, nickel, cobalt, copper, chromium, man-
ganese, vanadium and titanium are given by the following equations as a linear function of tem-
perature in °C in g cm–3:
Iron: 8.30 – 8.36 ´ 10–4T
Nickel: 9.60 – 12.00 ´ 10–4T
Cobalt: 9.57 – 10.17 ´ 10–4T
Copper: 9.11 – 9.44 ´ 10–4T
Chromium: 7.83 – 7.23 ´ 10–4T
Manganese: 7.17 – 9.30 ´ 10–4T
Vanadium: 6.06 – 3.20 ´ 10–4T
Titanium: 4.58 – 2.26 ´ 10–4T
The specific volume and density of liquid iron-carbon alloys are given in Fig. 2.48 for various tem-
peratures. It should be noted that the density of the liquid in equilibrium with austenite does not
change much over the entire liquidus range.
2.4.9 Viscosity
The viscosity is a measure of resistance of the fluid to flow when subjected to an external force. As
conceived by Newton, the shear stress e, i.e. force per unit area, causing a relative motion of two adja-
cent layers in a fluid is proportional to the velocity gradient du/dz, normal to the direction of the
applied force
du
e=h (2.4.38)
dz
where the proportionality factor h is viscosity of the fluid (liquid or gas).
0.155
6.5
Specific volume (cm3/g)

6.6
Density (g/cm3)
0.150
6.7
1600°C
1500°C 6.8
0.145 1400°C 6.9

Fig. 2.48 Density of iron-carbon 1300°C
alloys. From Ref. 95. 7.0
7.1
0.140 Austenite liquidus
7.2
0 1 2 3 4 5
wt. (%)
0 2 4 6 8 10 12 14 16 18 20
Carbon concentrations / atom %
Viscosity unit: poise (g cm-1s–1) º 0.1N s m–2.

The viscosities of Fe–C alloys determined by Barfield and Kitchener96 are given in Fig. 2.49. In the
iron-carbon melts the coefficient of viscosity is essentially independent of composition within the
range 0.8 to 2.5% C; above 2.5% C, the viscosity decreases continously with an increasing carbon
content.
2.4.10 Diffusivity, Electrical and Thermal Conductivity, and Thermal Diffusivity

The diffusivity is an exponential function of temperature,
æ E ö
D = Do exp ç - ÷ (2.4.39)
è RT ø
where Do is a constant for a given solute and E the activation energy (enthalpy) for the diffusion
process.
-1.0
2.53
0.08
2.10
1400°C
-1.1 3.43
Viscosity 0.1Ns m-2
0.07
log h (0.1Ns m-2)
1500°C 1450°C
X
1550°C -1.2
X
4.40
1600°C 0.75
0.06
X
-1.3
0.05
-1.4
0 1.0 2.0 3.0 4.0 5.0 5.5 6.0 6.5
Carbon wt % 1/T x 104 (K-1)
Fig. 2.49 Composition and temperature dependence of the viscosity coefficient of iron-carbon melts. From
Ref. 96.
The general trend in the temperature dependence of solute diffusivity in solid iron is shown in Fig.
2.50. The interstitial elements, e.g. O, N, C, B, H have diffusivities much greater than the substitu-
tional elements. Because of larger interatomic spacing, the diffusivities of interstitials in bcc-iron
are greater and the heats of activations (~ 85 kJ) are smaller than those in the fcc-iron with E ~ 170
kJ. With the substitutional elements, the heat of activation is within 210 to 250 kJ for bcc-iron and
within 250 to 290 kJ for fcc-iron.
In liquid iron alloys, diffusivities of elements are within 10–5 to 10–4 cm2s-1 with E within 15 to 50 kJ.
Some solute diffusivity data for liquid iron and iron-carbon alloys are given in Table 2.11.
Table 2.11 Selected Solute Diffusivities in Liquid Fe-C Alloys

Diffusing Concentration Temp. D Do E
element mass % Medium range °C cm2sec–1 cm2sec–1 kJ
C 0.03 Fe 1550 7.9x10–5 – –
C 2.1 Fe 1550 7.8x10–5 – –
C 3.5 Fe 1550 6.7x10–5 – –
Co Dilute sol. Fe 1568 4.7x10–5 – –
Co Dilute sol. Fe 1638 5.3x10–5 – –
Fe – Fe–4.6%C 1240–1360 – 4.3x10–3 51
Fe Fe–2.5%C 1340–1400 – 1.0x10–2 66
H Dilute sol. Fe 1565–1679 – 3.2x10–3 14
Mn 2.5 gr.satu.Fe 1300–1600 – 1.93x10–4 24
N Dilute sol. Fe 1600 1.1x10–4 – –
N Dilute sol. Fe–0.15%C 1600 5.6x10–5 – –
O Dilute sol. Fe 1600 5.0x10–5 – –
P Dilute sol. Fe 1550 4.7x10–5 – –
S <0.64 gr.satu.Fe 1390–1560 – 2.8x10–4 31
S ~1 Fe 1560–1670 – 4.9x10–4 36
Si <2.5 Fe 1480 2.4x10–5 – –
Si <1.3 Fe 1540 3.8x10–5 – –
Si 1.5 gr.satu.Fe 1400–1600 – 2.4x10–4 34
The electrical conductivity l in the units of W–1cm–1 is the reciprocal of the electrical resistivity. The
electrical conductivity of liquid low alloy steel is about l = 7140 W–1cm–1 at steelmaking temperatures.
From Fick’s law, the thermal conductivity is defined by
Q
k=- (2.4.40)
¶T ¶x
where Q is heat flux, energy per unit per unit time, ¶T/¶x the temperature gradient normal to the
direction of heat flow and k the thermal conductivity.
Units of k: J cm–1s–1K–1 º 10–2 kg m s–3K–1 º 10–2 Wm–1K–1
The effect of temperature on the thermal conductivities of iron, carbon steels and high alloy steels
are shown in Fig. 2.51.
Analogous to mass diffusivity, the thermal diffusivity is defined as
k
a= cm 2 s -1 (2.4.41)
rC p
where r is density and Cp molar heat capacity. In metals, the electrons migrate at much faster rates
than the atoms; therefore the thermal diffusivity is much greater than the mass diffusivity.
For low–alloy steels at 1000°C:
a » 0.04 cm2s–1
D » 3 ´ 10–12 cm2s–1 (substitutional)
D » 3 ´ 10–7 cm2s–1 (interstitial)
1600 1400 1200 1000 800 600°C

10–4
Int
10–5 ers
titi
als
in
Int
bc
c-F
ers
e
titia
ls i
n
10–6
fcc
-Fe
Diffusivity (cm /sec.)
Fig. 2.50 Range of diffusivities of
2
interstitial and substitutional ele- 10–7

ments in bcc and fcc iron.
10–10
Su
Su
bs
bst
tit
itut
ut
io
ion
na
l in
al i
10–11
bc
nf
c-
cc-
Fe
Fe
10–12
5 6 7 8 9 10 11
104/T (K–1)
80
Purified iron
Thermal conductivity (Wm–1K–1)
64
Fig. 2.51 Temperature

dependence of thermal 48
Carbon steels
conductivity of purified
iron and iron alloys.
32
High alloy steels
16
0 200 400 600 800 1000
Temperature (°C)
2.5 Properties of Molten Slags

2.5.1 Structural Aspects
Molten slags are ionic in nature consisting of positively charged ions known as cations, and nega-
tively charged complex silicate, aluminate and phosphate ions known as anions.
The fundamental building unit in solid silica and molten silicates is the silicate tetrahedron SiO44–.
Each silicon atom is tetrahedrally surrounded by four oxygen atoms and each oxygen atom is
bonded to two silicon atoms. The valency of silicon is +4 and that of oxygen is –2, therefore the
silicate tetrahedron has 4 negative charges.
The addition of metal oxides such as FeO, CaO, MgO, . . . to molten silica brings about a break-
down of the silicate network, represented in a general form by the reaction
-
æ | | ö æ | ö 2+
ç Si| O Si| ÷ + MO ® 2 ç Si| O÷ + M (2.5.1)
è ø è ø
The cations are dispersed within the broken silicate network. In MO–SiO2 melts the atom ratio
O/Si > 2, therefore part of the oxygen atoms are bonded between two silicon atoms and part to only
one silicon atom. Partial depolymerization of the silicate network with the addition of a metal oxide
MO is illustrated in Fig. 2.52.
Fig. 2.52 Schematic representation

of depolymerization of the silicate
network with the dissolution of metal
oxides in silicate melts.
In highly basic slags with molar ratio MO/SiO2 > 2, the silicate network completely breaks down
to individual SiO4 tetrahedra intermixed with cations M2+ and some oxygen ions O2–.
At low concentrations, Al2O3 behaves like a network-modifying oxide and forms aluminum cations
Al3+. At high concentrations, the aluminum enters the tetrahedral structure isomorphous with silicon.
This process may be schematically represented by the reaction
-
æ ö
æ Si O Si ö + MAlO ® ç Si| O Al
| | |
O
|
Si ÷ + M+ (2.5.2)
è | | ø 2
ç | | | ÷
è O ø
The cation M+ is located in the vicinity of Al–O bonding to preserve the local charge balance. At low
concentrations of phosphorus in steelmaking slags, the phosphate ions PO43– are incorporated in the
silicate network. In steelmaking slags, the sulfur exists as a sulfide ion S2–. The sulfate ions SO42– exist
in slags only under highly oxidizing conditions and in the absence of iron or any other oxidizable metal.
Although molten slags are ionized, the slag composition can be represented in terms of the con-
stituent oxides, e.g. CaO, FeO, SiO2, P2O5. The thermodynamic activity of an ion in the slag can-
not be determined. However, the activity of an oxide dissolved in molten slag, forming M2+ and
O2– ions, can be determined experimentally and the following equality can be written
MO ® M2+ + O2– (2.5.3)
a MO
=
(a M 2+
aO 2- )
(2.5.4)
(a MO ) (a )
o o
M 2+
aO 2-
where the superscript ° refers to the standard state which is usually pure solid or liquid oxide.
2.5.2 Slag Basicity

For steelmaking slags of low phosphorus content, the slag basicity has traditionally been repre-
sented by the mass concentration ratio
%CaO
V= (2.5.5)
% SiO2
For slags containing high concentrations of MgO and P2O5, as in some laboratory experiments, the
basicity may be defined by the following mass concentration ratio, with the assumption that on a
molar basis the concentrations of CaO and MgO are equivalent. Similarly, on a molar basis 1/2
P2O5, i.e. PO2.5, is equivalent to SiO2.
%CaO + 1.4 ´ % MgO
B= (2.5.6)
% SiO2 + 0.84 ´ % P2O5
For slags containing MgO < 8% and P2O5 < 5%, the basicity B is essentially directly proportional
to V.
B = 1.17V (2.5.7)
Another measure of slag basicity is the difference between the sum of the concentrations of basic
oxides and acidic oxides.
(%CaO + %MgO + %MnO) – (%SiO2 + P2O5 + TiO2) (2.5.8)
This formulation of slag basicity is not used very often.
For the calcium aluminate type of slags used in steel refining in the ladle furnace, the slag basic-
ity used in some German and Japanese publications, is defined by the ratio
%CaO
(2.5.9)
%SiO 2 ´ %Al 2 O3
However, such a ratio becomes meaningless at low concentrations of either SiO2 or Al2O3. For the
ladle furnace slag the basicity may be defined by the following mass concentration ratio, on the
assumption that on a molar basis Al2O3 is equivalent to SiO2.
%CaO + 1.4 ´ % MgO
B LF = (2.5.10)
% SiO2 + 0.6 ´ % Al 2O3
The slag basicities as defined above are for the compositions of molten slags. In practice, the steel-
making slags often contain undissolved CaO and MgO. The chemical analyses of such slag sam-
ples without correction for undissolved CaO and MgO, will give unrealistic basicities which are
much higher than those in the molten part of the slag.
There has been a trend in recent years to relate some physiochemical properties of slags, such as
sulfide capacity, phosphate capacity, carbide capacity, etc., to optical slag basicity. The optical
basicity is discussed briefly in Section 2.5.12.
2.5.3 Iron Oxide in Slags

Iron oxide dissolves in slags in two valency states: divalent iron cations Fe2+ and trivalent iron
cations Fe3+. The ratio Fe3+/ Fe2+ depends on temperature, oxygen potential and slag composition;
this is discussed later in this section. In the formulation of the equilibrium constants of slag-metal
reactions and the thermodynamic activities of oxides in slags, the total iron dissolved in the slag as
oxides is usually converted to the stoichiometric formula FeO and denoted by FetO, thus
%FetO = %FeO (analyzed) + 0.9 x %Fe2O3 (analyzed) (2.5.11)
or
%FetO = 1.286 x %Fe (total as oxides) (2.5.12)
For the ease of understanding, the subscript t will be omitted in all the subsequent equations and
diagrams.
2.5.4 Selected Ternary and Quaternary Oxide Systems

Most steelmaking slags consist primarily of CaO, MgO, SiO2 and FeO. In low-phosphorus steel-
making practices, the total concentration of these oxides in liquid slags is in the range 88 to 92%.
Therefore, the simplest type of steelmaking slag to be considered is the quaternary system
CaO–MgO–SiO2–FeO.
First let us consider the ternary system CaO·SiO2–FeO; the liquidus isotherms of this system is
shown in Fig. 2.53. The isothermal section of the composition diagram in Fig. 2.54 shows the phase
equilibria at 1600°C.
SiO2
10 90
20 80
170
0
30 70
170
0
Cristobalite
40 60
160
0
Troymite 1500
CaO.SiO2 Pseudowollastonite 140
0
130 50
0 0
150 120
0 FeO.SiO2
0
3CaO.2SiO2 1 4 0 Wollastonite 1150
0
Ran
kinit 130 250 0
40
e 1 120 1150 1100
50
2CaO.SiO2
11
00 CaO.FeO.SiO2 Olivine
21
1200
70 00 O 2FeO.SiO2
3CaO.SiO2 20
1200
00 1250
19
60 00 20
Ca 18 Ca2SiO4
22 3 Si
23 00 2 O Wustite
00 1 00 5 1700 130
20 0
90 24 00 1 1600 10
00 Lime 900 18
25 0 0 1 1500
00 70 1 1400
0 600 1
500 II-Ca2SiO4 1300
CaO 10 20 30 40 50 60 70 80 90 FeO
Weight %
Fig. 2.53 Liquidus isotherms of CaO–SiO2–FeO system.
SiO2
20 80
40 60
SiO
)
t%
Fig. 2.54 Phase equilibrium in
(w
2
(w
O
the system CaO–SiO2–FeO in
Ca
t%
equilibrium with liquid iron at
)
1600°C. 60 40
liquid
2CaO-SiO2
3CaO-SiO2
80 20
CaO 20 40 60 80 FeO
FeO (wt%)
There are four two-phase regions where, as depicted by dotted lines, the melt is saturated with
SiO2, 2CaO·SiO2, 3CaO·SiO2 or CaO; two three-phase regions (2CaO·SiO2 + 3CaO·SiO2 + liquid)
and (3CaO·SiO2 + CaO + liquid); and one liquid phase region.
Magnesia is another important ingredient of steelmaking slags, which are invariably saturated with
MgO to minimize slag attack on the magnesia refractory lining of the furnace.
The effect of MgO on the solubility of calcium silicates and calcium oxide is shown in
Fig 2.55 for the system (CaO + MgO)–SiO2–FeO, in equilibrium with liquid iron at 1600°C.
SiO2
20 80
SiO2 + liquid
)
wt%
O(
40 60
Mg
SiO
Fig. 2.55 Effect of MgO (wt.%) on

O+
the solubility isotherms at 1600°C

(w
in the system (CaO + MgO)–

Ca
t%
Liquid
)
SiO2–FeO in equilibrium with liq-

uid iron. 60 40
Silicate + liquid
10
80 Silicate + Ca(Mg)O + 8 20
6
Mg(Fe)O® liquid 4
2
0% MgO
Ca(Mg)O + Mg(Fe)O ® liquid
Cao + MgO 20 40 60 80 FeO

FeO (wt%)
The broken-line curve delineates the region of saturation of molten slag with solid calcium (magne-
sium) silicates and solid magnesio-wustite (MgO–FeO solid solution). Effects of the concentrations
of MgO and FeO on the solubility of CaO in 2CaO·SiO2-saturated slags are shown in Fig. 2.56.
Tromel et al.97 have made a detailed study of the solubility of MgO in iron–calcium silicate melts
in equilibrium with liquid iron at 1600°C. Below the dotted curve BACD in Fig. 2.57 for double
saturations, the curves for 10 to 60% FeO are the MgO solubilities in the slag.
60
55
MgO (%)
CaO (mass %)
2
Fig. 2.56 Solubility of CaO in
50 (CaO–MgO–SiO2–FeO) slags
saturated with 2CaO·SiO2 at
8 1600°C.
45
40
0 10 20 30
FeO (mass %)
32
olivine + 2MgO SiO2
olivine A
+ 10% FeO
Mg(Fe)O 20
30
24
Mg(Fe)O + Mg(Ca)O
40
MgO (wt %)
B C
16
60
Mg(Fe)O + 2Ca(Mg)O SiO2
8
Mg(Fe)O + Ca(Mg)O
D
0 1 2 3 4 5
%CaO
%SiO2
Fig. 2.57 Solubility of MgO, as magnesio–wustite, in the system CaO–MgO–SiO2–FeO at 1600°C as a func-
tion of slag basicity and FeO concentration. From Ref. 97.
SiO2
2CaO·SiO2
30
70
3CaO·SiO2 3Al2O3·2SiO2
80 20
liquid
90 10
CaO 10 20 30 40 50 60 70 80 90 Al2O3
Mass (%) CaO·2Al2O3 CaO·6Al2O3
Fig. 2.58 Phase equilibria in the system CaO–Al2O3–SiO2 at 1600°C.
Pertinent to the compositions of neutral ladle slags, the phase equilibria in part of the system
CaO–Al2O3–SiO2 at 1600°C is shown in Fig. 2.58.
2.5.5 Oxide Activities in Slags

In this section the activity data are given for a few ternary and multicomponent systems which are
closely related to the iron and steelmaking slags.
2.5.5.1 Activities in CaO – FeO–SiO2 System

The oxide activities in the CaO–FeO–SiO2 melts relevant to oxygen steelmaking in equilibrium
with liquid iron at about 1550°C are given in Fig. 2.59. The iso-activity curves in the left dia-
gram represent the experimentally determined activities of iron oxide with respect to liquid
FeO.98–99 The activities of CaO and SiO2 with respect to solid oxides were calculated from the
FeO activities by Gibbs-Duhem integration.99
0.2 0.8
SiO2¨ liq. aFeO(1)
0.4
0.1
0.2 0.6
a SiO2(s)
0.3
0.4 0.6 a CaO(s)
N Si
O
Ca
N
O2
0.4 (0.9)
0.00 0.00 (0.7)
0.6 0.5 0.4 6 2
O
0.6 0.01 0.4

Ca
0.6 (0.6)
N
N Si
0.7 (0.4)
O2
C2S ¨ liq. 0.1 (0.2)

0.8
0.8 0.2 0.8 0.2 0.2
0.9 (0.1)
0.4
(0.05)
(0.001)
(0.0005)
CaO 0.2 0.4 N 0.6 0.8 FeO CaO 0.2 0.4 0.6 0.8 FeO
FeO NFeO
Fig. 2.59 Activities of FeO(l) (experimental), CaO(s) and SiO2(s) (by Gibbs-Duhem integration) in
CaO–FeO–SiO2 melts in equilibrium with liquid iron at 1550°C. From Refs. 98, 99.
Recently new iron smelting processes have been developed and are briefly discussed here. The
slags are CaO–SiO2–FeO–A12O3–MgO (saturated) with low FeO contents (<5%). Knowledge
of the activity of FeO in these slags is critical and has been recently measured by Liu et al.100
The activity coefficient of FeO in these slags is given in Fig. 2.60 and is approximately 3.5. It
increases slightly with basicity as shown in Fig. 2.61 but is nearly constant with a value of 3.5.
10.00
· Liu et al
Temperature: 1550°C
8.00
Basicity (wt% CaO/ wt%SiO2): 0.79 – 1.23
Al2O3 Content (wt%): 6.31 – 8.04
6.00
FeO activity coefficient
4.00
2.00
0.00
0.00 0.01 0.02 0.03 0.04 0.05
FeO mole fraction
Fig. 2.60 The activity of FeO in CaO–SiO2–Al2O3–FeO–MgO (saturated) slags relevant to iron smelting. From
Ref. 100.
10.00
Liu et al
8.00 FeO content: 0.7 – 4.7
Al2O3 content: 6.3 – 8.0
FeO activity coefficient
6.00
4.00
2.00
0.00
1.50 2.00 2.50 3.00
Slag basicity, (CaO+MgO)
SiO2
Fig. 2.61 The effect of slag basicity on the activity coefficient of FeO in CaO–SiO2–Al2O3–FeO–MgO
(saturated) slags relevant to iron smelting. From Ref. 100.
2.5.5.2 Activities in CaO–Al2O3–SiO2 System

Rein and Chipman101 measured the activity of silica in the CaO–Al2O3–SiO2 system; from these
experimental data they calculated the activities of CaO and Al2O3 by Gibbs-Duhem integration.
The salient features of these oxide activities at 1600°C, with respect to solid oxides, are shown in
Fig. 2.62 for the mass ratios of CaO/Al2O3 = 2/3 and 3/2; the compositions of ladle slags are well
within the range given in Fig. 2.62.
Rein and Chipman also measured the activity of silica in the quaternary melts CaO–MgO–Al2O3–SiO2.
The activity coefficients of SiO2 derived from these data are given in Fig. 2.63 for melts contain-
ing 10% MgO and 0, 10 or 20% Al2O3.
10–1 0.5
CaO
CaO Al2O3
Al2O3
0.4
2/3
Fig. 2.62 Silica and alu- 10–2
SiO2 activity
Al2O3 activity
0.3
mina activities, with re- 2/3
spect to solid oxides, in
CaO–Al2O3–SiO2 melts
at 1600°C, derived from 0.2
experimental data of 10–3 3/2
Rein and Chipman.101 3/2
0.1
10–4 0
0 10 20 30 0 10 20 30
Mass % SiO2 Mass % SiO2
1.6
0
1.2 10
20
Activity coefficient of SiO2
Fig. 2.63 Activity coefficient of SiO2 with

respect to pure solid oxide, in CaO– 0.8
MgO–Al2O3–SiO2 melts at 1600°C, con-
taining 10% MgO and 0, 10 or 20%
Al2O3, derived from experimental data of
Rein and Chipman101.
0.4
0
0.4 0.8 1.2 1.6
% CaO + % MgO
% SiO2
The ratio of the activities (aAl2O3)1/3/aCaO within the entire liquid composition range up to 30% SiO2
is shown in Fig. 2.64 reproduced from Ref. 27.
2.5.5.3 Activities in MnO–Al2O3–SiO2 System

In the deoxidation of steel with the ladle addition of silicomanganese and aluminum together, the
deoxidation product is molten manganese aluminosilicate with the mass ratio MnO/SiO2 at about
1:1 and containing 10 to 45% Al2O3. The activities of oxides in the MnO–Al2O3–SiO2 melts at
1550 and 1650°C were computed by Fujisawa and Sakao102 from the available thermochemical data
on the system. They also determined experimentally the activities of MnO and SiO2 by the selected
slag-metal equilibrium measurements and found a close agreement with the computed data.
The activities of Al2O3 and SiO2, with respect to solid oxides, are plotted in Fig. 2.65 for melts with
mass ratio of MnO/SiO2 = 1. For melts containing up to 30% Al2O3 the activity of MnO remains
essentially unchanged at about 0.1 then decreases to about 0.05 at 40% Al2O3.
2.5.5.4 Activity Coefficient of FeO in Slags

With hypothetical pure liquid FeO as the standard state, the activity of iron oxide is derived from
the concentration of dissolved oxygen in liquid iron that is in equilibrium with the slag. For the
reaction equilibrium
FeO(l) = Fe + [O] (2.5.13)
KO =
[a ]
O
(2.5.14)
a FeO
where aO = [%O]fO, using log fO = – 0.1´[%O]. The temperature dependence of the equilibrium
constant KO is given below.
5370
log K O = - + 2.397 (2.5.15)
T
100
40 CaAl12O19
Saturated with
10 CaAl4O7
Molten aluminosilicate
(a Al2O3)1/3
4 Fig. 2.64 Effect of slag composition

on the activity ratio (aAl2O3)1/3/aCaO
(aCaO)
CaO/Al2O3 = 1.0 for the system CaO–Al2O3–SiO2 at

1600°C. From Ref. 27.
1.0
Saturated with
Ca2SiO4
0.4 Ca3SiO5
CaO
0.1
0 5 10 15 20 25 30
Silica (mass%)
1.0
a Al2O3
1550°C
0.8 1650°C a SiO2
Al2O3 and SiO2 activities
0.6
0.4
1650°C
0.2 1550°C
0
0 5 10 15 20 25 30 35 40 45
Mass % Al2O3
Fig. 2.65 Al2O3 and SiO2 activities in MnO–Al2O3–SiO2 system for mass ratio MnO/SiO2 = 1, derived from
data compiled by Fujisawa & Sakao102.
In the past a wide variety of formulations were generated to represent the composition dependence
of the iron oxide activity or activity coefficient in complex slag. On a recent reassessment of this
property of the slag, Turkdogan103 came to the conclusion that, within the limits of uncertainty of
the experimental data on slag-metal reaction equi-
librium, there is a decisive correlation between the 4
activity coefficient of FeO and the slag basicity as
shown in Fig. 2.66. The gFeO reaches a peak at a
basicity of about B = 1.8. It should be pointed out
once again that the concentration of iron oxide is
for total iron as oxides in the slag represented by 3
the stoichiometric formula FeO.
g FeO and g MnO
As discussed in Ref. 103, the experimental data

used in deriving the relation in Fig. 2.66 are for g FeO g MnO
simple and complex slags which differ consider- 2
ably in their compositions thus: in mass percent,
CaO 0–60, MgO 0–20, SiO2 0–35, P2O5 0–20,
FeO 5–40, MnO 0–15.
1
2.5.5.5 Activity of MnO in Slags
The activity coefficient ratio g FeO/g Mno, derived
from the slag-metal equilibrium data cited in Ref.
103, varies with slag basicity as shown in Fig.
2.67. There is a sharp decrease in the ratio of the 0
0 2 4 6
activity coefficients as the basicity B increases B
from 1.5 to 2.0. At basicities above 2.5, the ratio
gFeO/g MnO is essentially constant at about 0.63. The Fig. 2.66 Effect of slag basicity B on the activity
coefficients of FeO and MnO, with respect to pure
curve for g MnO in Fig. 2.66 is that derived from the liquid oxides, in simple and complex slags at tem-
combination of the curve for g FeO with the ratio in peratures of 1550 to 1700°C. From compiled data
Fig. 2.67. in Ref. 103.
5
2.5.6 Gas Solubility in Slags
2.5.6.1 Solubility of H2O
4 In acidic melts, H2O vapor reacts with double
bonded oxygen and depolymerizes the melt, thus
æ | | ö æ | ö
ç |
Si O Si ÷ + H 2O = 2 ç Si OH÷
3 è | ø è | ø
gMnO
gFeO
In basic melts, H2O reacts with free oxygen ions

(O2–) + H2O = 2(OH)– (2.5.16)
2
Both for acidic and basic melts the overall reaction is
represented by
(O*) + H2O = 2(OH*) (2.5.17)
1
where O* represents double or single bonded oxygen,
or O2–, and OH* is single bonded to silicon or as a free
ion. The equilibrium constant for a given melt compo-
0 sition is
0 2 4 6
B
C OH =
(ppm H 2O) (2.5.18)
(p )
1/ 2
Fig. 2.67 Variation of activity coefficient ratio H 2O
with basicity. From Ref. 103.
where PH is the vapor partial pressure.
2O
The solubilities of H2O (in units of mass ppm H2O) at 1 atm pressure of H2O vapor in CaO–FeO–SiO2
at 1550°C, measured by Iguchi et al.104 are given in Fig. 2.68.
SiO2
SiO2
30 30
650
10 90
te
ali
40 40
600
ob
ist
650 20 80
Cr
50 50
60 60 30 70
1300 1050
500
70 70 40 60
10 20 30 40 50 60 Mullite
CaO FeO (wt%) FeO 600
50 50
Al2O3
60 650
Fig. 2.68 Hydroxyl capacity, COH of
CaO–FeO–SiO2 and CaO–Al2O3–SiO2
melts at 1500°C. From Ref.104. 70
1000
800 750 Corondum
80 Silicate liquid
90 2000 1400
2500 Al2O3
CaO 10 20 30 40 50 60 70
Al2O3 (wt%)
2.5.6.2 Solubility of N2
Nitrogen dissolves in molten slags as a nitride ion N3– only under reducing conditions
/2N2(g) + 3/2(O2–) = (N3–) + 3/4O2(g)
1
(2.5.19)
for which the equilibrium constant (known as nitride capacity) is
3
pO 2 4
C N = (% N) 1
(2.5.20)
pN2 2
Many studies have been made of the solubility of nitrogen in CaO–Al2O3 and CaO–Al2O3–SiO2 melts
in the 1970s. These were reviewed in a previous publication3. Reference should be made also to a sub-
sequent work done by Ito and Fruehan105 on the nitrogen solubility in the CaO– Al2O3–SiO2 melts. They
showed that the nitride capacity of the aluminosilicate melts increases with a decreasing activity of CaO.
2.5.6.2.1 Nitrogen Dissolution into Slags The nitride capacity in certain slag systems increases and in
others decreases with basicity. According to equation 2.5.19 one would expect it to increase with
increasing slag basicity, i.e. increasing concentration of free oxygen ion O2–. The reason for this
apparent contradiction is that nitrogen can enter the slag by replacing single bonded oxygen ions
in the SiO2 network. This phenomena is explained in detail by Ito and Fruehan.105
Briefly, spectroscopic research indicates that the nitrogen is combined with the network former in
silicate melts which can be represented by
(2O–) + 1/2N2 =(N–) + 1/2 (O2–) + 3/4 O2 (2.5.21)
where O– is a nonbridging system in the network. If SiO2 is the network former equation 2.5.21
corresponds to
æ - | -ö æ | -ö
ç O Si| O ÷ +
è ø
1
2 N 2 = ç Si| = N ÷ +
è ø
1
2 (O ) +
2- 3
4 O2 (2.5.22)
In this case the nitride capacity is given by
(% N )(p )
3
3–
aO
4
O2 -
C N 3- = = K (2.5.23)
( ) (aO )
1 1
2 2
pN2 fN – 2-
where K is an equilibrium constant for reaction 2.5.21, fi is an activity coefficient for species i. In this
case N3– does not mean a free
nitride ion but the nitrogen analyzed
–12
as a nitride which is associated with
the network former. Similar reac-
1823 K
tions can be written for bridging
oxygen, but then reactions are less
likely. In either case the nitride
log CN3–
capacity decreases with oxygen ion

activity, i.e. increasing slag basicity. –13
CaO – SiO – Al O 2 2 3
For the CaO–SiO2, CaO–Al2O3, CaO – SiO2
and CaO–SiO2–Al2O3 systems CaO 1
CaO – Al2O3
is the basic component. Therefore 2
the equations predict that a plot of
the logarithm of nitride capacity –14
versus the logarithm of the activity –3 –2 –1 0
of CaO should yield a straight line log a CaO
with a slope of –1⁄2 as shown in Fig. Fig. 2.69 The relation between the nitride capacity and the activ-
2.69. ity of lime for CaO–SiO2–Al2O3 slag at 1823 K. Ref. 105.
For highly basic oxide systems the nitrogen enters as a free nitride ion and the simple relationship
given by 2.5.19 is valid. In this case the free nitride concentration and the nitride capacity is pro-
portional to the oxygen ion activity to the 3/2 power (aO2–)3/2. Min and Fruehan106 have shown that
the nitride capacity does increase with basicity for highly basic slags, particularly when there is no
strong network former such as SiO2 present. In theory the nitride capacity should decrease with
basicity when the nitrogen is primarily in the network and increase when it is a free nitride. This
was observed for the BaO–B2O3 and CaO–B2O3 system as shown in Fig. 2.70.
: BaO–B2O3, 1698 K 2
–12 : CaO–B2O3, 1773 K 3

log CN3–
Fig. 2.70 Nitride capacity of the slags

–13 as a function of the activity of CaO or
BaO. From Ref. 106.
–14
–4 –3 –2 –1 0
log a MO
2.5.6.2.2 Nitrogen Removal by Slags For Al-killed steels the corresponding equilibrium partial pres-
sure of oxygen is
1.22 ´ 10-16
p O (atm) = at 1600o C
[%Al]
2
4
3 (2.5.24)
From the solubility data for nitrogen we have
( )
1
[
(atm) 2 = 21.9 % N at 1600o C ] (2.5.25)
2 1
pN
2
Substituting these in the equation for CN gives
CN =
(% N ) ´ 5.3 ´ 10-14
[% N ][% Al]
In lime-rich aluminate ladle slags of low SiO2 content, the lime activity is aCaO = 0.5 for which
CN = 10–13. These give the following equilibrium relation
(% N ) = 1.9 ´ % Al
[% N] [ ]
Dissolved [%Al]
______________ (%N) / [%N]
___________
0.005 0.0095
0.02 0.038
0.06 0.114
0.10 0.190
The equilibrium nitrogen distribution ratio between slag and steel is very low even at high alu-
minum contents, well above the practical range. The slag/steel mass ratio in the ladle is also low,
about 1/100. For these reasons, liquid steel in the ladle cannot be de-nitrogenized by the slag that
is usable in industry in steel refining.
As noted from the recent experimental studies of Fruehan and co-workers,106–108 that even with fluxes
of high nitride capacities (as with alkaline earth borates) it is not practically possible to remove nitro-
gen from the aluminum-killed steel in the ladle. The highest nitride capacity was for a slag containing
about 22% CaO – 27% BaO – 48% Al2O3 – 3% TiO2 and was about 5 ´ 10–12. For this case for a steel
containing 0.06% Al the nitrogen distribution ratio is about 5.5 which is still not high enough for effec-
tive nitrogen removal.
2.5.6.3 Solubility of S2
Sulfur-bearing gases dissolve in molten slags as sulfide ions (S2–) under reducing conditions, and
as sulfate ions (SO42–) under highly oxidizing conditions. In steelmaking the oxygen potential is
not high enough for the solution of sulfur as sulfate ions, therefore we need to consider only the
sulfide reaction.
Whether the sulfur-bearing species is primarily H2S or SO2, there is a corresponding equilibrium
value of pS2 depending on the temperature and gas composition. It is convenient to consider the
reaction in a general form as
1
/2S2(g) + (O2–) = (S2–) + 1/2O2(g) (2.5.26)
For a given slag composition the equilibrium relation is represented by
1
æ pO ö 2
C S = (% S) ç 2 ÷ (2.5.27)
è pS 2 ø
where p’s are equilibrium gas partial pressures. The equilibrium constant CS is known as the sul-
fide capacity of the slag. The value of CS depends on slag composition and temperature.
Experimentally determined sulfide capacities of binary oxide melts are shown in Fig. 2.71. Refer-
ences to experimental data are given in Ref. 3.
For the slag-metal system, the sulfur reaction is formulated in terms of the activities
(@ concentrations) of sulfur and oxygen dissolved in the steel.
[S] + (O2–) = (S2–) + [O] (2.5.28)
For low-alloy steels
(% S) %O
[% S] [ ]
kS = (2.5.29)
where the equilibrium constant kS depends on slag composition and temperature.

2.5.6.3.1 Conversion of pO2 / pS2 to [%O]/[%S] The free energies of solution of O2 and S2 in liquid low
alloy steel are given below; see Table 2.1 in section 2.1.2.4.2.
1
/2S2 = [S]; DGs = –135,060 + 23.43TJ (2.5.30)
1
/2O2 = [O]; DGo = –115,750 – 4.63TJ (2.5.31)
For the reaction equilibrium
/2S2 = [O] = 1/2O2 + [S]
1
the standard free energy change is

DG° = DGs – DGo = –19,310 + 28.06TJ (2.5.32)
[% S] = 1009 - 1.466
1
æ pO ö 2
log ç 2 ÷
è pS 2 ø [%O] T (2.5.33)
For steelmaking temperatures an average value of K is 0.133; with this conversion factor the fol-
lowing is obtained.
[%O]
1
æ pO 2 ö 2
çp ÷ = 0.133
è S2 ø [% S] (2.5.34)
With this substitution the values of CS are converted to kS.

(% S) %O
[% S] [ ]
7.5 ´ C S = k S = (2.5.35)
1622°C
10–1 'FeO'
CaO–FeOx
1350°C MnO–Al2O3
1500°C
1650°C
10–2 MnO–SiO2
CaO–Al2O3
CS (wt %)
Fig. 2.71 Sulfide capacities of

10–3 binary oxide melts. From Ref. 3.
Na2O–SiO2
CaO–P2O5
CaO
(cryst.)
10–4 1650°C
MgO–SiO2
CaO–SiO2
Al2O3–SiO2
10–5
0 20 40 60 80 100
Basic oxide (mol. %)
A slag of high kS value and steel deoxidation to low levels of [O] are necessary conditions for steel
desulfurization. The subject is discussed in more detail later.
2.5.6.4 Solubility of O2
Oxygen dissolves in molten slags by oxidizing the divalent iron ions to the trivalent state.
Gas–slag reaction: /2O2(g) + 2(Fe2+) = 2(Fe3+) + (O2–)
1
(2.5.36)
Slag–metal reaction: Fe° + 4(Fe3+) + (O2–) = 5(Fe2+) + [O] (2.5.37)
These reactions provide the mechanism for oxygen transfer from gas to metal through the over-
laying slag layer. Partly for this reason, the steel reoxidation will be minimized by maintaining a
low concentration of iron oxide in the ladle slag, tundish and mold fluxes.
Examples are given in Fig. 2.72 of variations of the ratio Fe3+/Fe2+ with slag composition of melts
co-existing with liquid iron, reproduced from a paper by Ban-ya and Shim109.
1600°C 0.12
0.20 40%
50% 0.10
Fe3+/Fe2+
20% liq. Fe-saturated

0.15 10% 0.08
60% 1600°C
0.06
0.10 70%
0.04
0.05 d-Fe-saturated
0.02
(a) 1400°C (b)
0 0
1 2 3 4 5 6 100 80 60 40 20
Molar ratio (CaO + MgO)/SiO2 FeO (mol. %)
0.3
1600°C
0.2
Fe3+/Fe2+
CaO
Al2O3
0.1
TiO2
P2O5 SiO2
(c)
00 10 20 30 40
MxOy (mol. %)
Fig. 2.72 Effect of slag composition on the ratio Fe3+/Fe2+ in melts saturated with metallic iron at 1600°C: (a)
CaO–MgO–SiO2–FeO melts at indicated molar concentrations of total iron oxide FeO; (b) MgO–SiO2–FeO
melts; (c) pseudobinary FeO–MxOy melts. From Ref. 109.
2.5.7 Surface Tension

Surface tensions measured by Kozakevitch110 are given in Fig. 2.73 for binary melts with iron
oxide, and in Fig. 2.74 for FeO–MnO–SiO2 and FeO–CaO–SiO2 melts at 1400°C. For additional
data on surface tension of a wide variety of slags and mold fluxes, reference may be made to a
review paper by Mills and Keene111.
600
Al2O3
CaO
Surface tension (mNm–1)
550 MnO
TiO2 Fig. 2.73 Surface tensions of

500 binary iron oxide melts at 1400°C.
From Ref. 110.
SiO2
450
P2O5
(1480°C) Na2O
400
0 4 8 12 16 20 24 28 32
Additions to FeO at 1400°C (mol.%)
SiO2
(a) 0.6
390
410 0.4
1
430
x SiO
m–
450
mN
470
490 0.2
510
530
550
FeO MnO
0.2 0.4 0.6 0.8
NMnO
SiO2
Fig. 2.74 Isosurface tension
curves in (a) FeO-MnO-SiO2 and
(b) 0.6 (b) FeO-CaO-SiO2 melts saturated
with iron at 1400°C. From Ref. 110.
1
m–
410 0.4
mN
430
x SiO
450
470
2
490 0.2
510
530
550
570
FeO 0.2 0.4 0.6 0.8 CaO
NCaO
2.5.7.1 Interfacial Tension

The slag-metal interfacial tensions have values between those for the gas-slag and gas-
metal surface tensions. Consequently, the addition of surface active elements to liquid iron
lowers the slag-metal interfacial tension.
Effects of sulfur and oxygen dissolved in iron on the interfacial tension between liquid iron and
CaO–Al2O3–SiO2 melts at 1600°C, determined by Gaye et al.112, are shown in Figs. 2.75 and 2.76;
compared to the surface tensions of Fe–S and Fe–O melts. The effect of oxygen on the interfacial ten-
sion is greater than sulfur, e.g. si = 600 mN m–1 with aO = 0.05 (0.05%) while at aS = 0.05 (0.05%),
si = 1000 mN m–1.
1300
Surface tension Fe–O
Interfacial tension (mNm–1)

1100
Fig. 2.75 Effect of oxygen in iron on the interfacial

tension between liquid iron and CaO–Al2O3–SiO2 900
melts, determined by Gaye et al.112 is compared to
the surface tension of Fe–O melt at 1600°C.
700
Interfacial tension
500
0 0.02 0.04 0.06 0.08
Oxygen activity in iron
1200
Surface tension Fe–S
1100
Interfacial tension (mNm–1)
1000
Fig. 2.76 Effect of sulfur in iron on the interfacial ten-
sion between liquid iron and CaO–Al2O3–SiO2 melts,
determined by Gaye et al.112 is compared to the sur-
face tension of Fe–S melts at 1600°C. 900
800
Interfacial tension
700
0.1 0.3 0.5 0.7
Sulfur activity in iron
As shown by Ogino et al.113 Fig. 2.77, a single curve describes adequately the effect of oxygen in
iron on the interfacial tension between liquid iron and a wide variety of simple and complex slags,
including those containing Na2O and CaF2. In the case of slags containing iron oxide, a decrease
in interfacial tension with an increasing iron oxide content is due entirely to the corresponding
increase in the oxygen content of the iron.
Recently Jimbo and Cramb114 accurately measured the interfacial tension between liquid iron and
CaO–Al2O2–SiO2 slags. As shown in Fig. 2.78 their results are slightly higher than those of Gaye et al.
Mass % oxygen in iron

0.0010 0.0032 0.0100 0.0316 0.1000 0.2512
1800
1400
Interfacial tension (mNm –1)
Fig. 2.77 General relation between the

oxygen content of iron and the interfacial
1000 tension between the metal and various
slag systems at 1580°C. From Ref. 113.
600
200
-3.0 -2.5 -2.0 -1.5 -1.0 -0.6
Log [%O]
2200
CaO/Al203 (1:1)
Interfacial tension (mNm –1)
CaO/Al203 (3:5)
1800
1400 Fig. 2.78 Interfacial tension between CaO–Al 2O3–

SiO2 and Fe at 1600°C. From Ref. 114.
1000
Data from Gaye et al.
600
0 20 40 60 80
SiO2 content [wt %]
2.5.8 Density
Many repetitive measurements of slag densities have been made. Only selected references are
given on the density data cited here for steelmaking type of slags. Since the density of silica
(2.15 g cm–3 at 1700°C) is much lower than the densities of other metal oxide components of
slags, densities of slags will decrease with an increasing silica content.
The density data for binary silicates are given in Fig. 2.79. The data in Fig. 2.80 are for
CaO–MgO–Al2O3–SiO2 melts, relevant to neutral slags for steel refining in the ladle.
FeO–SiO2 1400°C, (115, 116)
4
Density (g cm-3)
Fig. 2.79 Densities of MnO–SiO2 1475°C, (117, 118)

binary silicate melt. From
Refs. 115–119.
3
MgO–SiO2
1700°C, (119)
CaO–SiO2
1700°C, (119)
2
0 20 40 60 80
Mass % SiO2
2.8
Fig. 2.80 Densities of

Density (g cm–3)
CaO–MgO–Al2O3–SiO2
melt at 1550°C, using
2.6
data in Refs; 120–122.
a b c
(a) 0% MgO, CaO/Al2O3
= 1; (b) 0% MgO, 5%
Al2O3; (c) 5% MgO,
5%Al2O3.
2.4
0 10 20 30 40 50 60
Mass % SiO2
The density data in Fig. 2.81 compiled by Mills and Keene111 are for simple and complex slags
containing FeO, CaO, MgO, SiO2 and P2O5. Since the densities of FeO–SiO2 and MnO–SiO2 are
essentially the same, the average of the data in Fig. 2.81 is represented by the following equation
in terms of (%FeO + %MnO).
r,gcm–3 = 2.46 + 0.018 x (%FeO + %MnO) (2.5.38)
4.5
4.4
4.3
4.2
4.1
4.0
+5%
3.9
-5%
3.8
3.7
Density, r (g cm-3)
3.6
Fig. 2.81 Densities of
3.5 simple and complex slags
containing iron oxide at
about 1400°C. From Ref.
3.4 111.
3.3
3.2
3.1
3.0
2.9
2.8
2.7
2.6
2.5
0 10 20 30 40 50 60 70 80 90 100
Total mass % iron oxides
2.5.9 Viscosity
The size of the silicate and aluminosilicate network in molten slags becomes larger with increas-
ing SiO2 and Al2O3 contents, hence their mobility decreases resulting in a higher viscosity. The
addition of metal oxides or an increase in temperature leads to the breakdown of the Si(Al)O4 net-
work, resulting in lower melt viscosity.
Machin and Yee123 made an extensive study of the viscosity of CaO–MgO–Al2O3–SiO2 melts at
temperatures of 1350 to 1500°C. The data in Fig. 2.82 are for the ternary system at 1500°C. The
isokoms are approximately parallel to the binary side Al2O3–SiO2, indicating that Al2O3–SiO2 are
isomorphous in their effect on the slag viscosity. The isokoms in Fig. 2.83 are for the quaternary
system with 35% and 50% SiO2. In this case, the isokoms are approximately parallel to the binary
side CaO–MgO, indicating that Ca2+ and Mg2+ cations have similar effects on the breakdown of the
aluminosilicate network.
SiO2
0 90
10 80
Fig. 2.82 Isokoms (0.l N s m–2)

for CaO–Al2O3–SiO2 system at
1500°C. From Ref. 123.
SiO
20 70
)
t%
2
(w
(w
2000
O
t%
Ca
)
1000
30 60
600
400
200
100
40 50
80
60
2.5 3.0
50 3.5 40 40
4
30
18 20
5 6 7 8 9 10 16
60 12 14 30
CaO Al2O3
10 20 30 40 50 60 70
Al2O3 (wt%)
8
9
20 10 30 50% SiO2
12 7 5 4
14 6 3
3.5
3
2O
35% SiO2 16
25 15 5
Al
20
3 poise 2.5 2 30 18 20
Mg
4 26 24
Mg
28
O
3
30 22 20
2O
9 8 7 5
O
12 6
Al
35
35 10 10 40 40
10
14
16 45
1500°C
18 1500°C
45 0 0
20 30 40 50 60 10 20 30 40 50
Mass % CaO Mass % CaO
Fig. 2.83 Isokoms (0.l N s m–2) for CaO–MgO–Al2O3–SiO2 system at 1500°C for melts containing 35% and
50% SiO2. From Ref. 123.
90 10
80 20
70 30
60 40
Fe
)
ss%
O
Fig. 2.84 Viscosity (N s m–2) of
(m
a
50 50
(m
CaO–FeO–SiO2 melts in 1400°C.
a
ss%
0.4
2
From Ref. 110.

S iO
0.2
)
40 0.15 60
0.1
30 70
0.05
20 80
10 90
90 80 70 60 50 40 30 20 10
CaO (mass%)
Viscosities of steelmaking slags are well represented by the experimental data of Kozakevitch110
given in Fig. 2.84
In the study of viscosities of mold fluxes for continuous casting, the experimental results have been
represented as a function of temperature using the relation.
h = AT exp (B/T) (2.5.39)
where A and B are functions of slag composition. For the composition range (wt %) 33–56% SiO2,
12–45% CaO, 0–11% Al2O3, 0–20% Na2O and 0–20% CaF2, an interpolation formula has been derived
for the parameters A and B as a function of the mole fractions of the constituents as given below.
lnA = –17.51 – 35.76(Al2O3) + 1.73(CaO) + 5.82(CaF2) + 7.02(Na2O) (2.5.40)
B = 31,1140 – 68,833(Al2O3) – 23,896(CaO) – 46,351(CaF2) – 39,519(Na2O) (2.5.41)
where A is in units of 0.1 N s m–2 K–1 ( poise/deg.) and B in degrees Kelvin.
2.5.10 Mass Diffusivity, Electrical Conductivity and Thermal Conductivity

2.5.10.1 Mass Diffusivity
Because of the ionic nature of molten slags, the diffusive mass transfer is by ions. The ionic diffu-
sivities are measured using the radioactive tracer elements dissolved in an oxidized form in the
melt. Typical examples of ionic diffusivities in slags at 1600°C are given below.
Ion
_____________ D*i , cm2/s
_______________
Si , O2–
4+ 4 x 10–7 – 1 ´ 10–6
Al3+ »1 ´ 10–6
Ca , Mg2+, Fe2+
2+ 6 x 10–6 – 1 ´ 10–5
S2– » 4 ´ 10–6
Since the electroneutrality has to be maintained, diffusion of a cation is accompanied by diffusion
of the oxygen ion. The diffusion that occurs in the dissolution of a solid oxide in the slag is con-
trolled by the mobility of the O2– ion which is smaller than the divalent cations.
2.5.10.2 Electrical Conductivity

The electrical current in molten slags is carried by the cations. However, in slags containing high con-
centrations of FeO or MnO (> 70%) the electronic conduction becomes the dominant mechanism.
The ionic conductivity li is theoretically related to the self diffusivity of the ionic species i by the
Nernst-Einstein equation
RT
Di* = li (2.5.42)
F Zi2 C i
2
where
F = Faraday constant; 96,489 C mol–1,
Zi = valency of ion i,
Ci = concentration of ion i, mol cm–3,
li = specific conductivity, W–1cm–1.
For steelmaking slags: l = 0.5–1.5 W–1cm–1.
For ladle slags: l = 0.4–0.7 W–1cm–1.
The electrical conductivity increases with an increasing slag basicity and increasing temperature.
2.5.10.3 Thermal Conductivity

Because of the presence of iron oxide, the metallurgical slags are opaque to infrared radiation,
therefore the heat conduction is primarily thermal.
Thermal conductivity of slags and mold fluxes are in the range 0.5 to 1.2 Wm–1K–1. From experi-
mental data the following approximate empirical relation has been found
k(Wm–1 K–1) = 1.8 ´ 10–5 V–1 (2.5.43)
3
where V is the molar volume = M/r, m mol .–1
2.5.11 Slag Foaming

Slag foaming plays an important role in many steelmaking processes. In oxygen steelmaking
excessive foaming can lead to slopping. On the other hand controlled foaming in the electric arc
furnace is desirable to protect the refractories from the electrical arc radiation. In the direct iron-
making processes such as DIOS and AISI Direct Steelmaking, excessive foaming can cause oper-
ational problems. However some foam is desirable to help capture the energy from post
combustion. Slag foaming should not be confused with simple gas bubbling or hold up. Whenever
gas passes through a liquid, the liquid expands due to the presence of the gas. However in some liq-
uids a stable foam develops which consists of foam bubbles on the top of an unfoamed liquid.
Early work on foaming was limited to a qualitative understanding of this complex phenomenon.
Due to the added importance of foaming in the EAF and iron bath systems Ito and Fruehan124,125
and other researchers developed a quantitative measure of foaming and measured the foaming
characteristics of many important slag systems. The foam index (S) was defined by
Dh
S= (2.5.44)
Vgs
Q
Vgs = (2.5.45)
A
where
D = the increase in the height of the slag,
Vsg = the superficial gas velocity,
Q = the gas flow rate,
A = the area of the vessel.
It follows that the volume of foam is approximately given by
Vf = QS (2.5.46)
The foam index has the units of time and represents the average traveling time of the gas through
the foam.
The foam can also be characterized by the time for the foam to decay. The average foam life (t) is
defined by the first order decay
h t
ln =- (2.5.47)
ho t
where h and ho are the foam height and the foam height when the gas flow stops, respectively. It
can be shown that for an ideal foam, in which the gas fraction of the foam is constant, that the foam
index (S) and foam life (t) are equal.
The foam index for CaO–SiO2–FeO–Al2O3 system is shown in Fig. 2.85. The foam index decreases
initially with increasing basicity because the viscosity is decreasing. However when the solubility
limit is reached the foam index increases with basicity because the second phase particles increase
the bulk viscosity of the slag.
Zhang and Fruehan126 later demonstrated that the foam index on stability increased with decreas-
ing bubble size and developed a general correlation demonstrated in Fig. 2.86 from which the foam
index for many complex systems would be desired.
h1.2
S= (2.5.48)
s 0.2 r DB0.9
where
h = slag viscosity,
s = surface tension,
r = slag density,
DB = foam bubble diameter.
Takumetsu et al127 showed that in bath smelting processes char could reduce foaming significantly.
Zhang and Fruehan128 examined the fundamentals of slag control by carbonaceous materials. Due
to surface tension phenomenon the slag foam bubbles collapse and form longer bubbles resulting in
a less stable foam. Carbonaceous materials will cause the foam to collapse while others such as alu-
mina particles or iron oxide pellets do not due to the different wetting characteristics with the slag.
The fundamentals of slag foaming have been applied to a variety of processes such as oxygen and
EAF steelmaking, bath smelting and ladle processing. The fundamental principles allow for rea-
sonable predictions of foaming in these processes.
20
CaO-SiO2–FeO–(Al2O3)
FeO=30%, Al2O3 = 3~5%
100
10
Predicted foam index (sec)
S (s)
5 10
1573 K
2 1673 K 1
1
0 0.5 1.0 1.5 2.0
0.1
CaO 0.1 1 10 100
(wt %)
SiO2 + Al2O3 Measured foam index (sec)
Fig. 2.85 The foam index of CaO–SiO2–FeO– Fig. 2.86 Comparison between the measured
Al2O3 slags illustrating the effect of second phase foam index and that predicted by the correlation
particles. From Ref. 126. equation 2.5.48. From Ref 126.
2.5.12 Slag Models and Empirical Correlations for Thermodynamic Properties

There have been several attempts to develop models and correlations for slags, from which thermo-
dynamic properties such as activities and capacities can be predicted. Simple models such as a regu-
lar solution have been used.129 However because of the ionic nature and structure of slags they are
generally inadequate. Recently Gaye and Lehmann reviewed the current status of structural models.130
The most successful models have been developed by Gaye and coworkers131 at IRSID and Blander
et al.132 In the IRSID model the slag is represented using an oxygen sublattice and a cationic sub-
lattice. The structure is defined in terms of cells composed of a central oxygen surrounded by two
cations. Two parameters are used to calculate the energy of the system; the energies of formation
of asymmetric cells and the interaction energies limited to one parameter per couple of cations.
These parameters are computed from the known thermodynamics of the binary systems. It is
beyond the scope of this chapter to describe the model in greater detail. The model has been used
successfully to estimate the phase diagrams and thermodynamic activities in multi-component sys-
tems as well as sulfide capacities. Whereas this model is complex and requires binary data, to date
it is the best model available for estimating thermodynamic properties of slags.
In recent years various capacities such as the sulfide and phosphate capacities have been correlated
to a term called the optical basicity. This concept has been recently reviewed by Summerville et
al.133 Briefly each oxide is assigned a value for its optical basicity which is related to the Pauling
electronegativity. By definition the optical basicity of CaO is taken as unity. Components such as
Na2O have optical basicities greater than one while acidic oxides such as SiO2 have values less than
one. The optical basicity of a slag is then computed from a weighted average of the components in
the slag. The optical basicity has been used to correlate a large number of studies on sulfide and
other capacities.133, 134 Whereas the optical basicity is an interesting concept, the correlations are
not sufficiently accurate to predict capacities for industrial purposes. For example the deviations
between the correlation and actual data is 25–50% in some cases. This is not accurate enough for
computing sulfur equilibrium in industrial applications. However the optical basicity correlation
maybe useful for making crude estimates of capacities for which no data exists or to predict trends
when changing slag compositions slightly.
2.6 Fundamentals of Ironmaking Reactions

Reference may be made to a book entitled “The Iron Blast Furnace”, authored by Peacey and Daven-
port.135 In this book, the theoretical, experimental and operational data have been put together about the
blast furnace with particular emphasis on the interrelations between countercurrent heat and mass
transfer between gases and solids in the blast furnace stack. Reference may also be made to the earlier
studies of Rist and co-workers136–140 on the heat and mass transfer and reduction reactions in the stack.
In this section, the following aspects of the blast furnace reactions are considered: oxygen poten-
tial diagram, role of vapor species in blast furnace reactions, slag-metal equilibrium, reaction
mechanisms and operational slag-metal data.
2.6.1 Oxygen Potential Diagram

The reducibility of a metal oxide relative to other oxides, or the oxidisability of a metal relative to
other metals, can readily be assessed from the free energy data.
For the oxidation reaction involving pure metals and metal oxides in their standard states,
i.e. aM = 1, aMO = 1
2M + O2 = 2MO (2.6.1)
the isothermal equilibrium constant is
1
K= (2.6.2)
pO 2
where pO is the equilibrium oxygen partial pressure for which the standard state is 1 atm at the
2
temperature under consideration. The standard free energy change is
DG° = –RT ln K = RT 1n pO2 (2.6.3)
which is also called the oxygen potential.
In an earlier study, Richardson and Jeffes141 compiled the free energy data on metal oxides which
were then available, and presented the data, Fig. 2.87, as an oxygen potential diagram.
The oxides for which the oxygen potential lines are above that of CO may be reduced by carbon.
As the affinity of the metals for oxygen increases, i.e. DG° decreases, the temperature of reduction
of the oxides by carbon increases.
For easy conversion of oxygen potentials to the corresponding values of pO2 or to the equilibrium
ratios of H2/ H2O and CO/CO2, appropriate scales are included in the enclosed oxygen potential
diagram for various oxides of metallurgical interest.
Scale for pO2:
Lines drawn from the point O on the ordinate for the absolute zero temperature through
the points marked on the right hand side of the diagram give the isobars. For example, for
the Fe–FeO equilibrium the oxygen potential is –82 kcal (–343 kJ) at 1200°C. By drawing
a line passing through this point and the point O, the oxygen partial pressure of about 7 ´
10–13 atm is read off the log pO2 scale.
Scale for CO/CO2:
Draw the line from point C to a point on the oxygen potential; the extension of this line
gives the corresponding equilibrium log CO/CO2 ratio.
Scale for H2/H2O:
Same as above by using point H.
These diagrams can be used to predict CO/CO2 or H2/H2O ratios as well as the pressure of oxygen
in equilibrium with numerous metals and their oxides. They are particularly useful for predicting
the CO/CO2 ratio in a blast furnace or the H2/H2O ratio in direct reduction with natural gas. It also
gives an indication of the relative reduceability of the oxides. It should be remembered that the dia-
grams are for unit activity of the metal. If the metal is in solution, it is easier to reduce. For exam-
ple, if the activity of Si is 10–3 relative to pure Si in liquid iron, the equilibrium oxygen pressure is
103 times higher. Similarly, if the activity of SiO2 is lowered in a slag it is more difficult to reduce.
2.6.2 Role of Vapor Species in Blast Furnace Reactions

Because of the high temperatures involved, vaporization plays a significant role in pyrometallurgi-
cal reactions. In a reactor such as a blast furnace, the formation and condensation of various vapor
species is responsible for the recycle of some of the elements between the high and low tempera-
ture regions of the furnace.
The countercurrent flow in the blast furnace of solids from low- to high-temperature zones and of
gases from high- to low-temperature zones brings about a cyclic process of vaporization and con-
densation which has a decisive influence on the overall operation of the furnace.
2.6.2.1 Vapor Species SiO, SiS and CS

It is now well established that the silicon and sulfur in the coke are transferred to the slag and iron
in the blast furnace via the vapor species SiO, SiS, CS, and other minor sulfur-bearing species. The
earlier and recent studies of quenched blast furnaces have revealed that while the sulfur content of
the slag increases with the descent of the burden in the bosh, the sulfur and silicon contents of the
metal droplets reach maxima and then decrease as they pass through the slag layer. This phenom-
enon of vapor-phase mass transfer has been substantiated by the experimental work of Tsuchiya et
al,142 Turkdogan et al 143 and Ozturk and Fruehan144 under conditions partly simulating the bosh
region of the blast furnace.
In the presence of carbon and depending on temperature and activity of silica in the slag or coke,
the SiO vapor is generated by one of the following reactions
SiO2 (in coke ash or slag) + C ® SiO + CO (2.6.4)
SiC + CO ® SiO + 2C (2.6.5)
log H2/H2O
log pO2
log CO/CO2 –8 –7 –6 –5 –4 –3
–8 –7 –6 –5 –4 –3 –2 –2
–1
O 0 0
0 –1
–20 )
–1
(B B
O 3
B
Fe 2 –2
=6
+O
2
(A)
– 40 M (B) 2CO
2
e 3O
4
NiO M =
4F M =2 ) + O2 –3
O2 O (C 2CO
0
2N
i+ M 2Co
O2 = ) 1
(B) M 2Co + 2H 2O
(A
(A) O M M =
– 60 Pb + O2
u 2O =2 2H 2
= 2C O2
+ O2 b+ –4
4Cu 2P
1
= SO 2 (B
) O (A)
•S 2 + O 2 O 4
O2 = 2Fe 2
Fe 3 Fe +
– 80 M =2 M 2
+O
2
O M
6F
e –6
2SnO + O2 = 2SnO2
–100 C + O2 = CO2 2
) M 3
O 2 (B
O2 = Sn
Sn + B
(B)
(A) O
e 3O 4 Zn B –8
–120 =F M O 5 (C
) =2 (B)
M
O2
+O
2
H +2 O3 2C
3Fe = 2 P2
2Z
n ) M 2Cr 2 +O 4
3
(D
) (B B 3O 2 = =2 M
+ 5O 2
O O r +
2
CO
Na
2
2P 2 K2 4C (A) M
=2 =2 (B)
RT log PO2 (kcal)
C –140 O 2
O2 nO T
4K+ M a+ = 2M M
4 N + O2 M –10
2M n
B
) 4
)
(C 5
(B
V 2O 3
–160 M =2
2
iO
O2
2S
+3 B)
4V (
=
iO 2
=S
2
O
O2
+
Si + –12
iO
–180
2S
(B) 6 5
TiO 2 O2
= =U
O2 O2
Ti +
B U+
S
– 200
B –14
7 6
(B)
O 3
M Suggested accuracies
Al 2
– 220 =2 B B
+ 3O 2 (A) ± 1 kcal
4 Al
M (B) ± 3 kcal –16
8
– 240 ) (C) ± 10 kcal
(C
O 7
Li 2 (C
) (D) ± 10 kcal
=2 O
O 2 Mg
i+ =2 M
4L C) Changes of state Element Oxide
+O O(
2
– 260 g Ca M
–18
M 2M =2 Melting point M M 9
O2
a+
2C
Boiling Point B B 8
– 280 Sublimation pt. S S
10 –20
Transition pt. T T
9
– 300
0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400
Absolute zero Temperature (°C) 11
10 –22
14 13 12
13 12 11 –24
–200 –150 –100 –90 –80 –70 –60 –50 –44 –38 –34 –30 –28 –26
Fig. 2.87 Oxygen potential diagram. From Ref. 141.
At 1 bar pressure of CO, aSiO2 = 1 and above about 1500°C, SiO2 is converted to SiC and SiO is gen-
erated by reaction 2.6.5; for aSiO2 = 0.1, reaction 2.6.5 applies only at temperatures above about
1610°C. In the experiments cited,143 the SiO generated by passage of CO through a coke bed at tem-
peratures of 1600 to 1900°C, were found to be close to the equilibrium values for reaction 2.6.5.
Volatile SiS is generated by reactions of the type with the coke ash:
CaS (in coke ash) + SiO ® SiS + CaO (2.6.6)
FeS (in coke ash) + SiO + C ® SiS + CO + Fe (2.6.7)
The reaction of these vapor species with metal and slag may be represented by the equations:
[ ] [ ]
ìSiO + C = Si + CO
ï
Metal í (2.6.8)
ïSiS = Si + S
î [ ] []
ì 2- æ | ö
ïSiO + 12 O + 12 C = ç Si O - ÷ + 1
2 CO
ïï è | ø
Slag í (2.6.9)
ï æ ö
( )
|
2-
ïSiS + 2 O + 2 C = ç
3 1 Si O - ÷ + S2 - + 1
2 CO
ïî è | ø
Ozturk and Fruehan,144 in carefully controlled experiments, demonstrated that the rate of the SiO
reactions with metal and slag are controlled by gas phase mass transfer and are rapid.
Coke samples taken from the tuyere zone of the blast furnace usually contain 12 to 16% ash which
has a basicity (CaO + MgO)/SiO2 of about 0.6; for a typical blast furnace slag, the basicity is about
1.5. Therefore, the partial pressures of SiO and sulfur-bearing species in the gas near the coke sur-
face are expected to differ from those interacting with the slag. For the purpose of comparison, cal-
culations are made for assumed local equilibrium at 1500°C for the systems gas-slag and gas-coke
ash; the results are given in Table 2.12. The amount of ash in the tuyere coke and its composition
suggests that, although most of the silica and sulfur in the coke ash are removed from the furnace
by the slag and liquid iron, there is significant recycle of silicon and sulfur in the bosh region by
the vaporization and condensation of these species on the coke particles.
Table 2.12 Calculated Equilibrium Vapor Pressures in bar for the Systems Gas–Slag
and Gas–Coke Ash at 1500°C and 1 bar Pressure of CO.
B = 1.5, aSiO2 = 0.05 B = 0.6, aSiO2 = 0.3

2%(S), CS = 3 x 10–4, 5%(S), CS = 10–5,
Vapor (wt%) * (wt%)*
SiO 2.6 ´ 10–4 1.5 ´ 10–3
SiS 3.9 ´ 10–5 1.7 ´ 10–2
S2 8.6 ´ 10–9 4.9 ´ 10–5
CS 5.5 ´ 10–5 4.3 ´ 10–3
COS 5.5 ´ 10–6 1.2 ´ 10–4
* CS(wt%) is the sulfide capacity of the slag.
2.6.2.2 Alkali Recycle and Removal by Slag

The alkalies constitute another dominant vapor species in the bosh and stack regions of the blast fur-
nace. Many studies have been made of the alkali recycle in the blast furnace via the process of vapor-
ization and condensation; the subject is well covered in the references cited. 145–147 General
consensus evolved from elementary thermodynamic calculations is that the alkali silicates in the ore
and coke ash decompose at elevated temperatures in the lower part of the bosh and the combustion
zone. In part the alkali vapors carried away with the ascending gas react with the slag and can there-
fore be removed from the furnace; part are converted to alkali cyanides and carbonates and
deposited on the burden and the refractory lining of the furnace stack. Some new thoughts are pre-
sented in the following discussion on the sequence of reactions and vapor species that may govern
the alkali recycle in the blast furnace.
As an example, let us calculate the equilibrium vapor pressure of potassium for the reduction reaction:
(K2O) + C = 2K(g) + CO(g) (2.6.10)
The computed equilibrium values in Table 2.13 are for a blast furnace slag containing 0.5% K2O
and having a basicity of B = 1.5: K is the equilibrium constant for reaction 2.6.10 from the ther-
mochemical data and aK2O is the activity of K2O in the slag, relative to the hypothetical solid K2O,
from the experimental data of Steiler.148
If the potassium input with the blast furnace burden is 4 kg K/t–HM and 80% of it is removed by
the slag, about 0.8 kg K/t–HM will be in the vapor phase carried away by the furnace gas. For a
typical total gas volume of 2300 nm3/t–HM, the partial pressure of the total potassium vapor
species would be 7.2 x 10–4 bar (for a total gas pressure of 3.6 bar in the bosh region). Compari-
son of this total potassium pressure in the gas with those in Table 2.13 for the gas-slag equilibrium
indicates that the slag particles will pick up potassium from the gas only at temperatures below
1200°C. The ash in the coke samples taken from the tuyere zone contains 2 to 4% (Na2O + K2O)
which is several fold of that in the original coke ash. Because of the low basicity of coke ash, the
amount of alkali therein is expected to be 5 to 10 times greater than that found in the slag. In the
upper part of the bosh, the slag and coke will pick up alkalies from the gas. As the temperature of
the slag and coke increases during descent in the bosh, they will emit alkali vapors to the gas phase.
However, the partial pressure of potassium vapor in the gas would probably be below that given in
Table 2.13 for the gas-slag equilibrium. Nevertheless, the numerical examples cited suggest that
much of the alkalies accumulated in the bosh and hearth zone of the blast furnace, via recycling,
are in the gas phase.
Table 2.13 Equilibrium Vapor Pressure of Potassium for Reaction 2.6.10 with Slag
having 0.5% K2O and B = 1.5.
Temperature (°C) K, bar aK2O pK, bar

1200 3.10 ´105 2.1 ´ 10–12 7.1 ´ 10–4
1300 2.69 ´ 106 5.5 ´ 10–12 3.4 ´ 10–3
1400 1.81 ´ 107 1.4 ´ 10–11 1.4 ´ 10–2
1500 9.78 ´ 107 3.5 ´ 10–11 2.6 ´ 10–1
According to a mass spectrometric study of vaporization of potassium cyanide by Simmons et.

al,149 the equilibrium pressures of vapor species are in the order p(KCN)2 > p(KCN) > p(KCN)3. For
atmospheric pressure of nitrogen and at graphite saturation, the following are the equilibrium ratios
of the vapor species: at the melting point of KCN (635°C), (KCN)2: KCN : K = 2 : l : 0.76 and at
the boiling point (1132°C), (KCN)2: KCN : K = 4.6: l : 0.19. It is all too clear that (KCN)2 is the
dominant alkali-bearing species in the blast furnace. The sodium in the burden undergoes cyclic
reactions similarly to potassium; however, the concentration of sodium-bearing species are about
one-tenth of the potassium-bearing species.
One method of minimizing the ascent of alkali vapors to the stack, hence reducing the alkali recy-
cle, is by operating the furnace with a slag of low basicity, particularly when the alkali input is high.
The slag of low basicity, however, will have an adverse effect on the composition of the hot metal
produced; it will result in low manganese, high silicon and high sulfur in the metal. For a given
preferred basicity, the slag mass per ton of hot metal may have to be increased when there is an
increase in the alkali input to the furnace. An increase in amount of slag will increase the coke rate,
hence will increase the alkali input. Obviously, with a burden of high alkali input, a compromise
has to be made in adjusting slag basicity and slag mass so that the alkali recycle would be low.
Even with a low alkali input some alkali buildup will occur. To prevent excessive accumulation of
alkalies, which lead to scaffolding, gas channeling, furnace upsets and so on, periodic additions of
gravel (silica) or olivine (magnesium silicate) are made to the blast furnace charge. The silica reacts
with alkali carbonates and cyanides, thus facilitates the discharge of the accumulated alkalies from
the furnace by the slag. Practical experience also shows that periodic addition of calcium chloride
to the burden also facilitates the removal of alkalies by the slag.
2.6.2.3 Ammonia and Hydrogen Cyanide in Blast Furnace Stack

The contaminants in the wash water for the iron blast-furnace off-gas are in the ranges 10–300 mg
NH3/l, 1–30 mg CNt (total)/l, and 0.3–15 mg phenols/l. In terms of ppm by volume in the stack
gas, the concentrations are usually in the ranges 3000–4000 ppm NH3 and 150–250 ppm CNt. As
observed by Pociecha and Biczysko,150 for example, most of the cyanide in the de-dusted furnace
off gas is in the form of hydrogen cyanide. They also found that the blast furnace dust contains only
a small amount of solid cyanides (presumably alkali cyanides) in comparison with the amount of
hydrogen cyanide present in the de-dusted top gas. The same contaminants in the ferromanganese
blast furnace off gas are about an order of magnitude greater than those in the iron blast furnace.
The experimental work of Turkdogan and Josephic,151 under conditions partly simulating the blast
furnace stack, substantiated the validity of this mechanism of generation of ammonia and hydro-
gen cyanide in the stack, by the following reactions.
2K(Na)CN + 3H2O = (K, Na)2CO3 + 2NH3 + C (2.6.11)
NH3 + CO = HCN + H2O (2.6.12)
The amounts of ammonia and hydrogen cyanide generated will increase with an increase in the
concentration of water vapor in the stack gas. Other reactions will occur with these volatile species
generating some amino-nitro phenols, e.g. NH2(NO2)C6H3OH. The equilibrium concentrations of
these volatile organic species become greater at lower temperatures.
2.6.3 Slag-Metal Reactions in the Blast Furnace

Slag-metal reactions in the hearth of the blast furnace control the chemistry of the hot metal. In
particular, the silicon, sulfur and manganese contents depend upon these reactions. In this section
the equilibrium and reaction mechanisms are presented.
2.6.3.1 Slag-Metal Equilibrium for Blast Furnace Reactions

From the thermodynamics of the slag systems and of the various solutes in iron such as sulfur, sil-
icon and manganese, it is possible to predict the equilibrium relationship for blast furnace reactions.
The three-phase reactions to be considered are for graphite-saturated melts at 1 atm pressure of car-
bon monoxide, which is close to the partial pressure of CO in the blast furnace bosh and hearth zones.
2.6.3.1.1 Silicon Reaction The reaction equilibrium to be considered is
(SiO2) + 2[C] = [Si] + 2CO(g) (2.6.13)
K Si =
[% Si] f Si æ p CO ö
2
ç ÷ (2.6.14)
(a )
SiO 2 è aC ø
where the SiO2 and C activities are with respect to pure solid silica and graphite, respectively.
The temperature dependence of the equilibrium constant is from the free energy data
30 , 935 (2.6.15)
log K Si = - + 20.455
T
Substituting aSiO2 = (%SiO2)gSiO2/60 ´ 1.65 and for the 10

activity coefficient fSi = 15 in graphite-saturated iron
containing Si < 2%, the equilibrium metal/slag silicon
distribution ratio is represented by
1600°C
[ % Si ] p 2CO = 6.73 ´ 10 -4 g SiO 2 K Si
1
(% SiO2 ) (2.6.16)
[%Si] r2 (atm2)
The equilibrium silicon distribution ratios computed 1500°C
co
previously, using the gSiO2 values for the blast furnace
0.1
type slags (CaO, 10% MgO, 10% Al2O3, SiO2), are
(%SiO2)
given in Fig. 2.88. The dotted curves are derived from
other data discussed later. For the blast furnace type
slags, the basicity is usually defined by the mass ratio 1400°C
(%CaO + %Mg)/%SiO2. 0.01
2.6.3.1.2 Manganese Reaction For the reaction

(MnO) + [C] = [Mn] + CO(g) (2.6.17)
K Mn =
[% Mn] f Mn pCO
(2.6.18)
0.001
0.8 1.0 1.2 1.4 1.6
(a )
MnO aC % CaO + %MgO
% SiO2
15 , 090
log K Mn = - + 10.970 (2.6.19) Fig. 2.88 Silicon distribution ratio for
T graphite–saturated melts for blast furnace
type slags containing about 10% MgO and
For the graphite-saturated iron, the activity coefficient 10% Al2O3. Lines – – – are derived from
fMn = 0.8, and inserting aMnO = (%MnO) gMnO/71 ´ the data in Fig. 2.90 and equation (2.6.23).
1.65, the equilibrium manganese distribution ratio is
represented by
[% Mn ] p = 1.07 ´ 10 -2 g MnO K Mn (2.6.20)

(% MnO) CO
Abraham et al.152 measured the MnO activity in CaO–Al2O3–SiO2– MnO melts containing Al2O3
< 20% and MnO < 8%. The values of gMnO derived from their data are seen in Fig. 2.89 to
increase with an increasing slag basicity. The equilibrium manganese distribution ratio for
graphite saturated melts thus evaluated are given in Fig. 2.90.
2.4
1.6
Fig. 2.89 Activity coefficient of MnO, rela- %Al2O3
20
g MnO
tive to pure solid oxide, in CaO–Al2O3–

SiO2–MnO slags, with MnO < 8%, at 1500 10
to 1650°C, derived from the data of Abra- 0
0.8
ham, Davies and Richardson152.
0
0.4 0.8 1.2 1.6 2.0
%CaO
%SiO2
16
12
r (atm)
(%Mn) co
1600°C
8
[%Mn]
Fig. 2.90 Manganese distribution ratio for graphite-

saturated melts derived for blast furnace type slags
containing about 10% MgO and 10% Al2O3.
4
1500°C
1400°C
0
0.4 0.8 1.2 1.6
%CaO + %MgO
%SiO2
2.6.3.1.3 Si–Mn Coupled Reaction

The silicon-manganese coupled reaction
2(MnO) + [Si] = (SiO2) + 2[Mn] (2.6.21)
between graphite-saturated iron and blast furnace type slags has been investigated by several exper-
imental studies.143,153,154
[ ] (% SiO2 ) = 0.17 K 2Mn

2
æ % Mn ö g 2MnO
K=ç ÷
è ( % MnO) ø [% Si] K Si g SiO 2
(2.6.22)
The combination of equations 2.6.16 and 2.6.20 gives the following equilibrium constant for reac-
tion (2.6.21)
The variation of the equilibrium relation KMnSi with slag basicity is shown in Fig. 2.91. The dotted
line for 1400 to 1600°C is calculated from the thermochemical data, i.e. from the computed equi-
librium data in Figs. 2.88 and 2.90. The difference of the calculated equilibrium dotted line from
the average experimental values by a factor of 2.7 (º ~15 kJ) is due to the accumulated uncertain-
ties in the free energy data on MnO and SiO2 as well as DGs for solutions of Mn and Si in
graphite-saturated iron.
The experimental data in Fig. 2.91 for graphite saturated melts at 1400 – 1600°C may be repre-
sented by the following equation.
æ %CaO + % MgOö
log K MnSi = 2.8 ç ÷ - 1.16 (2.6.23)
è % SiO2 ø
To be consistent with the experimental data in Fig. 2.91, the dotted curves are derived from the reli-
able equilibrium data in Fig. 2.90 and the experimental data represented by equation (2.6.23) for
the Si–Mn coupled reaction in graphite saturated melts.
In a study of slag-metal equilibria between CaO–MnO–SiO2 slags and graphite-saturated man-
ganese-base alloys, Turkdogan and Hancock155 found that the ratios Si/SiO2 and Mn/MnO varied
with slag basicity in a manner similar to that for iron base alloys. Their results for 1400°C are
shown in Fig. 2.92; the dot-dash line is for the iron-base alloys, reproduced from Fig. 2.91.
2000
104 Temp CaO-SiO2-10% Al2O310% MgO
°C CaO-SiO2-10% Al2O3
1000
1400 800 Thermodynamic data
1550
600
1400
103 1500 400
1600
calculated
200
100
102 80
60
kMnSi
40
10 20
10
8
6
1 4
0.5 0.7 0.9 1.1 1.3 1.5 1.7 0.6 1.0 1.4 1.8
(a) (b)
%CaO + %MgO
%SiO2
Fig. 2.91 Experimental data showing the equilibrium relation KMnSi with basicity of the blast furnace type slags
at graphite saturation: (a) reproduced from Ref. 143 and (b) from Ref. 154; lines---derived from equation
2.6.23 and Fig. 2.90.
2.6.3.1.4 Si–S and Mn–S Coupled Reactions The silicon-sulfur coupled reaction is represented by
[S] + 1/2[Si] + (CaO) = (CaS) + 1/2(SiO2) (2.6.24)
for which the isothermal equilibrium constant is
(% S) ìï (% SiO2 ) üï
1
2
1
K SiS = í ý
[% S] ïî [% Si] ïþ (%CaO) (2.6.25)
From an analysis of the most available experimental data, Turkdogan et al.143 developed a relation
for KSiS shown in Fig. 2.93. For the range 1400 to 1600°C, the temperature dependence of KSiS is
represented by the equation
6327 æ %CaO + % MgOö
log K SiS = - 4.43 + 1.4 ç ÷ (2.6.26)
T è % SiO2 ø
From the experimental data represented by equations 2.6.23 and 2.6.26, the following equation is
obtained for the temperature dependence of the equilibrium relation for the manganese-sulfur cou-
pled reaction in graphite-saturated melts.
log K MnS = log

(% S) (% MnO) 1 = 6327 - 3.85
[% S] [% Mn ] (%CaO) T (2.6.27)
The equilibrium sulfur distribution ratio as a function of slag basicity in graphite-saturated melts
shown in Fig. 2.94.
1000 %Fe 15-24

800 25-44 45-62
600
400
~95
%Mn 2 (SiO2)
[%Si]
200
%MnO
100
80
60
40
20
10
0 0.4 0.8 1.2
%CaO
%SiO2
Fig. 2.92 Product of manganese and silicon ratios for graphite–saturated ferromanganese melts, of indicated
iron contents, equilibrated with CaO–MnO–SiO2 slags at 1400°C and 1 atm CO, derived from the data of Turk-
dogan and Hancock.155
3.0 3.5
(%SiO2) (%SiO2) ⁄2
(%SiO2) (%SiO2) ⁄2
1
1
2
2
}}
}}
[Si]
[Si]
2.5 1400°C 3.0

1
1
Log }}
Log }}
[%S]
[%S]
2.0 2.5
1500°C
(a) (b)
1.5 2.0
log(%CaO) + 1.4B log(%CaO) + 1.4B
3.0
(%SiO2) (%SiO2) ⁄2
1600°C
1
2 }}
2.5 1550°C
[Si]
Fig. 2.93 (a) 1400°C; (b) 1500°C; (c) 1550°C and

1
2.0 1600°C. Variation of equilibrium relation for the sil-

Log }}
icon-sulfur coupled reaction with slag basicity B for

[%S]
unit activity of carbon. Reference cited in Ref. 143.

1.5
(c)
1.0
2.0 2.5 3.0 3.5 4.0 4.5
log(%CaO) + 1.4B
300
30%
50%
200
pCO (atm)
Fig. 2.94 Sulfur distribution ratio for
graphite-saturated melts as a function 1600°C
(%S)
[%S]
of slag basicity. From Ref. 143.
100
1500°C
1400°C
0
0.4 0.8 1.2 1.6
B
2.6.3.1.5 Si–Ti Coupled Reaction Delve et al.156 have studied the silicon-titanium coupled reaction in
graphite-saturated melts using blast furnace type slags:
(TiO2) + [Si] = (SiO2) + [Ti] (2.6.28)
Their experimental results, for temperatures of 1500 and 1600°C and basicities of 1 to 2, may be
summarized by the following equation.
K SiTi =
[% Ti] (%SiO ) 2
(% TiO ) [%Si] 2
(2.6.29)
%CaO
log K SiTi = 0.46 + 0.39 (2.6.30)
%SiO2
2.6.3.2 Mechanisms of Slag-Metal Reactions

2.6.3.2.1 Experimental Work Experiments were made by the authors 143 to simulate the events in
the upper part of the blast furnace hearth where metal droplets pass through the slag layer.
Solid pieces (4 to 5 mm dia.) of graphite-saturated iron containing Si and S were dropped on
the surface of a pool of remelted blast furnace slag (65 g) contained in a graphite tube (20
mm dia.) at 1480°C; an argon atmosphere was maintained over the melt. Twenty pieces of
metal (7 g total) were dropped one at a time at 10–20 s intervals, within 7–10 min total time
of the experiment. The metal pieces melted rapidly and descended a 10 cm deep column of
molten slag as individual metal droplets. After the last piece was dropped the melt was
rapidly cooled. The residence time of the droplets in the slag column is estimated to be about
1–2 s, if there is no appreciable buoyancy effect due to gas evolution. The metal droplets col-
lected in a well at the bottom of the slag column; metal and slag were subsequently analyzed
for manganese, silicon and sulfur.
With the addition of MnO to the slag, there is silicon transfer from metal to slag. As is seen from
the experimental results in Fig. 2.95, the extent of silicon oxidation increases with increasing
%MnO in the slag via the Si–Mn coupled reaction
2.0
1.59 [%Si]i
1.16 [%Si]i
0.93[%Si]i
0.63 [%Si]i
1.6
Silicon in metal (wt %)
1.2
Fig. 2.95 Change in silicon content of metal
droplets passing through a 10 cm deep slag
column. From Ref. 143.
0.8
0.4
0 4 8 12 16
(MnO)i (wt %)
2(MnO) + [Si] ® 2[Mn] + (SiO2) (2.6.31)

In addition, there is also MnO reduction by carbon in the metal droplets.
(MnO) + [C] ® [Mn] + CO (2.6.32)
With the slags containing MnO > 3%, the metal droplets desiliconized down to 0.1 – 0.3% Si.
Within this region, the silicon and manganese distribution ratios are within the ranges
(%SiO2)/[%Si] = 120 to 350 and [%Mn]/(%MnO) = 0.7 to 1.8. These ratios give KMnSi values
below the equilibrium value of 1096 derived from equation 2.6.23 for the slag basicity of 1.5.
The effect of MnO content of the slag on the extent of desulfurization of metal droplets during
descent in the slag column is shown in Fig. 2.96. The shaded area within the range 0.010 to
0.035% S corresponds to the state of slag-metal equilibrium represented by equation 2.6.26, for
metal droplets containing 0.1 to 0.3% Si and the slag of basicity 1.5 containing 41% CaO, 35%
SiO2 and 1.5% S.
The extremely fast rates of reactions observed in the foregoing experiments may be attributed to
the presence of a thin gas film (consisting of the vapor species CO, SiO, SiS and Mn) around the
metal droplets, which provides the reaction path for the transfer of reactants to and from slag and
metal via the gas phase.
2.6.3.3 Blast Furnace Tap Chemistry Data

In a previous study157 an assessment was made of the state of slag-metal reactions in the blast fur-
nace hearth from an analysis of the daily average compositions of metal and slag samples as nor-
mally recorded in the plants. The plant data used were those reported by Okabe et al.158 from Chiba
Works of Kawasaki Steel and the daily average data for a period of a month from U.S. Steel plants.
In the plant data used, the melt temperatures at tap are mostly in the range 1500 ± 25°C. The aver-
age slag compositions in mass percent are in the range 38–42% CaO, 8–10% MgO, 34–38% SiO2,
10–12% Al2O3, 0.5–1.0% MnO, 2% S and minor amounts of other oxides. The slag basicity is about
0.16
[%Si]i [%S]i
1.59 0.14
1.16 0.084
0.93 0.10
0.63 0.14
0.12
Sulfur in metal (wt. %)
0.08
0.04
0 4 8 12 16
(MnO)i (wt. %)
Fig. 2.96 Change in sulfur content of metal droplets passing through a 10 cm deep slag column; shaded
region is for slag-metal equilibrium at 1480°C. From Ref. 143.
1.4 ± 0.15. The hot metal compositions in the tap stream are in the range: 0.4–0.8% Mn, 0.5–1.5%
Si, 0.02–0.05% S, 5% C and other usual impurities.
The silicon, manganese and sulfur distribution ratios from plant data are scattered within the
shaded areas in Fig. 2.97. The dotted lines are for the slag-metal equilibria at 1 atm pressure of
CO, which is close to the CO partial pressure in the hearth zone of the blast furnace. It should be
noted that the equilibrium line for the Si/SiO 2 ratio is based on equation 2.6.16 representing the
experimental data and the equilibrium relation in Fig. 2.88, i.e. dotted curves. It is only the sili-
con distribution ratio that is scattered about the gas-slag-metal equilibrium line for the three-
phase reaction
(SiO2) + 2[C] = [Si] + 2CO(g) (2.6.33)
Values of KMnSi scattered within the shaded area in Fig. 2.98 are below the equilibrium line (dot-
ted). It should be pointed out that in the upper part of the slag layer, the iron oxide content will be
higher than manganese oxide. Therefore, the silicon in metal droplets will be oxidized more read-
ily by iron oxide in the upper part of the slag layer, resulting in the observed non-equilibrium state
for the Si–Mn coupled reaction in equation 2.6.31.
With iron ore of low basicity in the blast furnace burden, the slag basicity will be low and fur-
thermore, the iron oxide content in the upper part of the slag layer is expected to be higher. It is
0.08
1550°C
(%SiO2) 0.06
[%Si]
0.04 1500°C
0.02
1400°C
3
(%MnO)
1500°C
[%Mn]
1
1400°C
80
1550°C
60
(%S)
[%S]
1400°C
40
Plant data
Equilibrium
20
1.2 1.3 1.4 1.5 1.6
%CaO + %MgO
%SiO2
Fig. 2.97 Si, Mn and S distribution ratios are compared with the equilibrium data for graphite-saturated melts
at 1 atm CO. From Ref. 159.
presumably for this reason that a low slag basicity, the K MnSi values are much lower than the equi-
librium values for the Si–Mn couple reaction.
In an earlier study of plant data from a ferromanganese blast furnace in the UK,160 it was found
that there were departures from equilibrium for the Si oxidation by MnO in the slag in a manner
similar to that observed for the iron blast furnace.
The plant data in Fig. 2.99 for the sulfur distribution ratio are scattered about the equilibrium line
for the Mn–S coupled reaction at 1500°C, represented by equation 2.6.27. However, the ratios
1000
600
400
[%Mn] 2 (%SiO2)
[%Si]
200
Fig. 2.98 KMnSi values for plant 100
(%MnO)
data are compared with the equi- 80
librium values (dotted line). From 60
Ref. 159.
40
20
10
1.2 1.3 1.4 1.5
%CaO + %MgO
% SiO2
120
1400°C
1500°C
80 1600°C
(%S)
[%S]
40
Plant data
Equilibrium
0 2 4 6 8
[%Mn]
(%MnO)
Fig. 2.99 The blast furnace data are compared with the slag–metal equilibrium values for the Mn–S coupled
reaction. From Ref. 159.
(%S)/[%S] in the plant data are lower than the equilibrium values for the Si–S coupled reaction
represented by equation 2.6.26. Nevertheless, despite departures from equilibrium, the silicon and
sulfur contents of the blast furnace iron change in a systematic manner, i.e. high silicon/low sulfur
and low silicon/high sulfur.
It should be borne in mind that because of fluctuations in the operating conditions in the blast fur-
nace, there will be variations in temperature and the extent of reaction of metal droplets during
descent through the slag layer. Consequently, the metal droplets of varying composition collecting
in the stagnant hearth will bring about the composition and temperature stratification that is com-
mon to all blast furnaces.
2.7 Fundamentals of Steelmaking Reactions

The oxygen steelmaking and electric-arc furnace steelmaking processes are described in detail
in Chapters 9 and 10 respectively. The practical and technical aspects of the steel refining in the
ladle furnace and vacuum degassing are described in Chapter 11. In this section of Chapter 2, the
discussion of steelmaking reactions will be confined to an assessment of the reaction mechanisms
and the state of slag-metal reactions at the time of furnace tapping.
2.7.1 Slag-Metal Equilibrium in Steelmaking

The reaction equilibria in the liquid steel-slag systems have been extensively studied, both experi-
mentally and theoretically by applying the principles of thermodynamics and physical chemistry.
In a recent reassessment of the available experimental data on steel-slag reactions103, it became evi-
dent that the equilibrium constants of slag-metal reactions vary with the slag composition in dif-
ferent ways, depending on the type of reaction. For some reactions the slag basicity is the key
parameter to be considered; for another reaction the key parameter could be the mass concentra-
tion of either the acidic or basic oxide components of the slag.
2.7.1.1 Oxidation of Iron

In steelmaking slags, the total number of g-mols of oxides per 100 g of slag is within the range
1.65 ± 0.05. Therefore, the analysis of the slag-metal equilibrium data, in terms of the activity and
mol fraction of iron oxide, can be transposed to a simple relation between the mass ratio
[ppm O]/(%FeO) and the sum of the acidic oxides %SiO2 + 0.84 ´ %P2O5 as depicted in Fig.
2.100(a). The experimental data used in the diagram are those cited in Ref. 103. There is of course
a corollary relation between the ratio [ppm O]/(%FeO) and the slag basicity as shown in Fig.
2.100(b).
80
1600°C
xx
x
(%FeO)
[ppm O]
x x
40 x
x
x xx x x
x
xx x
x x
0
0 10 20 30
(a) %SiO2 + 0.84 ´ %P2O5
Fig. 2.100 Equilibrium ratio

100 [ppm O]/(%FeO) related to (a)
SiO2 and P2O5 contents and (b)
1700°C slag basicity; experimental data
are those cited in Ref. 103.
80
1600°C
60
[ppm O]
(%FeO)
1500°C
40
20
0
0 1 2 3 4 5 6
(b) B
2.7.1.2 Oxidation of Manganese

For the FeO and MnO exchange reaction involving the oxidation of manganese in steel, formulated
below,
(FeO) + [Mn] = (MnO) + [Fe] (2.7.1)
the equilibrium relation may be described in terms of the mass concentrations of oxides
K ¢FeMn =
(% MnO)
(% FeO)[% Mn ]
(2.7.2)
where the equilibrium relation K¢FeMn

8
depends on temperature and slag
composition.
The values of K¢FeMn derived from
the equilibrium constant for reaction
2.7.1, given in Ref. 27 and the activ-
ity coefficient ratios gFeO/gMnO in Fig. 6
2.67, are plotted in Fig. 2.101 against
the slag basicity. In BOF, OBM(Q-
BOP) and EAF steelmaking, the slag
basicities are usually in the range 2.5
to 4.0 and the melt temperature in
K¢FeMn
the vessel at the time of furnace tap- 4

ping in most practices is between
1590 and 1630°C for which the equi-
librium K¢FeMn is 1.9 ± 0.3. The plant
analytical data for tap samples give
K¢FeMn values that are scattered about
the indicated slag-metal equilibrium 2 1800 C
values.
1650 C
Morales and Fruehan161 have recently
determined experimentally the equi- 1700 C
librium constant K¢FeMn for reaction
2.7.1 using MgO-saturated calcium
silicate melts. Their values of K¢FeMn 0
are plotted in Fig. 2.102 together with 1.5 2.5 3.5 4.5
some data from the studies of Chip- B
man et al.162 and Suito et al.163. The
broken line curve is reproduced from Fig. 2.101 Equilibrium relation in equation 2.7.2 related to slag
basicity. From Ref. 103.
Fig. 2.101. Resolution of observed
differences in the values of K¢FeMn
awaits future studies.
2.7.1.3 Oxidation of Carbon

With respect to the slag-metal reaction, the equilibrium relation for carbon oxidation would be
(FeO) + [C] = CO + [Fe] (2.7.3)
p CO ( atm )
K FC = (2.7.4)
[%C ] a FeO
5730 (2.7.5)
log K FC = - + 5.096
T
2.0
Morales and Fruelhan 161
Chipman et al162
1.5 Suito et al163
From Fig. 2.101
log K'Mn-Fe
1.0
0.5
0.0
0 1 2 3 4 5
(%CaO + %MgO)
Basicity, B =
%SiO2
Fig. 2.102 Experimental values of K¢FeMn measured recently by Morales and Fruehan161.
For 1600°C, g FeO = 1.3 at slag basicity of B = 3.2 and pCO = 1.5 atm (average CO pressure in the
vessel), we obtain the following equilibrium relation between the carbon content of steel and the
iron oxide content of slag.
K FC = 108.8
a FeO = 1.3 N FeO »

1.3
72 ´ 1.65
(% FeO) = 0.011 ´ (% FeO) (2.7.6)
(%FeO)[%CO] = 1.25
2.7.1.4 Oxidation of Chromium
There are two valencies of chromium (Cr2+ and Cr3+) dissolved in the slag. The ratio Cr2+/Cr3+
increases with an increasing temperature, decreasing oxygen potential and decreasing slag basic-
ity. Under steelmaking conditions, i.e. in the basic slags and at high oxygen potentials, the triva-
lent chromium predominates in the slag. The equilibrium distribution of chromium between slag
and metal for basic steelmaking slags, determined by various investigators, is shown in Fig. 2.103;
slope of the line represents an average of these data.
(%Cr ) = 0.3 ± 0.1 ´ % FeO
( ) ( ) (2.7.7)
[%Cr ]
2.7.1.5 Oxidation of Phosphorus
It was in the late 1960s that the correct formulation of the phosphorus reaction was at last realized, thus
[P] + 5/2[O] + 3/2(O2–) = (PO43–) (2.7.8)
10
6
(%Cr)
[%Cr]
0 4 8 12 16 20 24 28
FeO (wt %)
Fig. 2.103 Variation of chromium distribution ratio with the iron oxide content of slag, in the (D) open hearth164
and (o) electric arc furnace165 at tap is compared with the results of laboratory experiments166(·).
At low concentrations of [P] and [O], as in most of the experimental melts, their activity coeffi-
cients are close to unity, therefore mass concentrations can be used in formulating the equilibrium
relation KPO for the above reaction.
(% P) %O -
[% P] [ ]
5
K PO = 2
(2.7.9)
The equilibrium relation KPO, known as the phosphate capacity of the slag, depends on tempera-
ture and slag composition.
From a reassessment of all the available experimental data, discussed in detail in Ref. 103, it was con-
cluded that CaO and MgO components of the slag, had the strongest effect on the phosphate capac-
ity of the slag. Over a wide range of slag composition and for temperatures of 1550 to 1700°C, the
steel-slag equilibrium with respect to the phosphorus reaction may be represented by the equation
21,740
log K PO = - 9.87 + 0.071 ´ BO (2.7.10)
T
where BO = %CaO + 0.3 (%MgO).
2.7.1.6 Reduction of Sulfur

The sulfur transfer from metal to slag is a reduction process as represented by this equation
[S] + (O2–) = (S2–) + [O] (2.7.11)
for which the state of slag-metal equilibrium is represented by
(% S) %O
[% S] [ ]
K SO = (2.7.12)
As is seen from the plots in Fig. 2.104, the sulfide capacities of slags, KSO, measured in three inde-
pendent studies are in general accord. The effect of temperature on KSO is masked by the scatter in
the data. The concentration of acidic oxides, e.g. %SiO2 + 0.84 ´ %P2O5, rather than the slag basic-
ity seems to be better representation of the dependence of KSO on the slag composition.
2.7.2 State of Reactions in Steelmaking

There are numerous versions of oxygen steelmaking such as top blowing (BOF, BOP, LD, etc.),
bottom blowing (OBM, Q-BOP, etc.) and combined blowing (K-OBM, LBE, etc.); these are
described in detail in Chapter 10. The state of the refining reactions depend to some degree on the
process. Bottom blowing in general provides better slag-metal mixing and reactions are closer to
equilibrium. In the following section the state of reactions for top blowing (BOF) and bottom blow-
ing (OBM) is given. The state of reactions for combined or mixed blowing processes would be
between these two limiting cases. The analytical plant data used in this study were on samples
taken from the vessel at first turndown from the BOP and Q-BOP shops of U.S. Steel. These plant
data were acquired through the kind collaboration of the USS research personnel at the Technical
Center.
2.7.2.1 Decarburization and FeO Formation

The most important reaction in steelmaking is decarburization. It not only determines the process
time but also the FeO content of the slag, affecting yield and refining. When oxygen is injected into
an oxygen steelmaking furnace a tremendous quantity of gas is evolved, forming a gas-metal-slag
emulsion which is three to four times greater in volume than the non-emulsified slag and metal.
The chemical reactions take place between the metal droplets, the slag and gas in the emulsion.
These reactions have been observed in the laboratory using x-ray techniques indicating in many
cases, the gas phase (primarily CO) separates the slag and metal, and that gaseous intermediates
play a role in decarburization.
1.2
°C
1600
1600
1.0
1623-1715
1600
0.8
[%O]
(%S)
[%S]
0.6
kSO =
0.4
0.2
0
0 10 20 30 40 1 2 3 4 5
%SiO2 + 0.84 ´ %P2O5 B
Fig. 2.104 Sulfide capacities of slags. From Ref. 103.
When oxygen first contacts a liquid iron-carbon alloy it initially reacts with iron according to reaction
2.7.13, even though thermodynamically it favors its reaction with carbon. This is due to the relative
abundance of iron in comparison to carbon. Carbon in the liquid metal then diffuses to the interface
reducing the FeO by reaction 2.7.14. The net reaction is the oxidation of carbon, reaction 2.7.15.
Fe + 1/2O2 = FeO (2.7.13)
FeO + C = CO + Fe (2.7.14)
C + /2O2 = CO
1
(2.7.15)
However, carbon is only oxidized as fast as it can be transferred to the surface.
At high carbon contents the rate of mass transfer is high such that most of the FeO formed is
reduced and the rate of decarburization is controlled by the rate of oxygen supply:
N O 2 M C 100
d %C
dt
=-
W
(f + 1) (2.7.16)
where
·
NO2 = the flow rate of oxygen in moles,
MC = the molecular weight of carbon (12),
W = the weight of steel,
f = the fraction of the product gas which is CO; the remainder is CO2 and f is
close to unity (0.8 to 1).
Below a critical carbon content the rate of mass transfer is insufficient to react with all the injected
oxygen. In this case the rate of decarburization is given by168:
r
d %C
dt
=-
W
(%C - %C e ) å mi Ai (2.7.17)
i
where
r = density or steel,
%Ce = the equlibrium carbon with the slag for reaction 2.7.14 and is close to zero,
mi = the mass transfer coefficient for the specific reaction site,
Ai = the metal-FeO surface area for the specific reaction site.
For top blown processes the reaction takes place between the metal droplets ejected into the emul-
sion and the FeO in the slag. For bottom blowing there is less of an emulsion and the reaction takes
place at the interface of the metal bath and the rising bubbles which have FeO associated with
them. The critical carbon is when the rates given by 2.7.16 and 2.7.17 are equal and is typically
about 0.3% C.
The actual values of mi and Aiare not known, consequently an overall decarburization constant (kC)
can be defined and the rate below the critical carbon content is given by 2.7.19 and 2.7.20:
r
kC =
W i
å m i Ai (2.7.18)
= - k C ( %C - %C e )
d %C (2.7.19)
dt
or,
(%C - %C e ) = - k t - t
ln C( C) (2.7.20)
(%C C - %C e )
where tc is the time at which the critical carbon content is obtained. The equilibrium carbon con-
tent (%Ce) is close to zero. However in actual steelmaking there is a practical limit of about 0.01
to 0.03 for %Ce.
The value of kC increases with the blowing rate since the amount of ejected droplets and bubbles
increases. Also, kc decreases with the amount of steel, (W), as indicated by 2.7.19. In actual
processes, the blowing rate is proportional to the weight of steel and , therefore, kC has similar val-
ues in most oxygen steelmaking operations. For top blowing, kC is about 0.015 s–1 and for bottom
blowing 0.017 s–1. The values of the critical carbon content are also similar ranging from 0.2 to
0.4% C. The rate of decarburization for a typical steelmaking process is shown in Fig. 2.105. At
the initial stage when silicon is being oxidized, which will be discussed later, the rate of decarbur-
ization is low. The value of kC is higher for bottowm blowing because mixing is more intensive and
the reaction occurs at the interface of the rising bubbles as well as the slag-metal emulsion.
Below the critical carbon content, when the rate of mass transfer of carbon is insufficient to reduce
all of the FeO formed and, therefore, the FeO content of the slag increases rapidly. In actual
processes, initially, some FeO forms because it has a low activity in the slag resulting in about 5-
10 % FeO in the slag. The FeO remains constant until the critical carbon content and then the FeO
increases rapidly. The amount of FeO can be computed from a mass balance for oxygen. Specifi-
cally the oxgyen not used for carbon, silicon or manganese oxidizes iron to FeO. The moles of FeO
in the slag at anytime (NFeO) is given by:
d %C W(1 + f ) ù
t
é
ò
N FeO = 2 ê N˙ O 2 -
êë dt M C 100
ú dt - N O 2 + N O 2
ûú
Si Mn
( ) (2.7.21)
0
where NOSi2 and NMn

O2
are the moles of oxygen consumed in oxidizing Si and Mn. Since the rate of
decarburizaiton is slightly higher for bottom and mixed blowing these processes have lower FeO
contents.
At all levels of turndown carbon, the iron oxide content of BOF slag is about twice that of OBM
(Q-BOP) slag (Fig. 2.106). The dotted line depicts the slag-metal equilibrium value of
(%FeO)[%C] » 1.25.
As depicted in Fig. 2.107, the square root correlation also applies to the product (%FeO)[%C] for
low carbon contents, with marked departure form this empirical correlation at higher carbon con-
tents. The slopes of the lines for low carbon contents are given below.
5 30
% Silicon
4 Temperature: 1610 ± 20°C
C and Si content (%)
% Carbon
BOF
3 OBM
Equilibrium
20
2
FeO (%)
10
0
0 5 10 15
Time (min)
Fig. 2.105 Computed carbon and silicon contents of 0

steel in oxygen steelmaking processes. 0 0.2 0.4 0.6
C (%)
Fig. 2.106 Iron oxide-carbon relations in BOF and

OBM (Q-BOP) are compared with the average equi-
librium relation (-------). From Ref. 103.
%C %C
0.02 0.05 0.1 0.2 0.3 0.4 0.5 0.6 0.02 0.05 0.1 0.2 0.3
4
Temperature: 1610 ± 20°C
32
BOF
3 OBM
(FeO)[%C]
24
2
[ppm O][%C]
1
16
0
0 0.2 0.4 0.6 0.8
…%C 8 Temperature: 1610 ± 20°C
BOF
Fig. 2.107 Variation of product (FeO)[%C] with car-
bon content of steel at first turndown. From Ref. 103. OBM
0
0 0.2 0.4 0.6
Ö%C
Fig. 2.108 Variation of product [ppm O][%C] with
carbon content of steel at first turndown. From Ref.
103.
BOF with C < 0.10%,

(2.7.22)
(% FeO) %C = 4.2 ± 0.3
OBM (Q - BOP) with C < 0.10%,

(2.7.23)
(% FeO) %C = 2.6 ± 0.3
2.7.2.2 Oxygen-Carbon Relation

Noting from the plot in Fig. 2.108, there is an interesting correlation between the product [ppm O]
[%C] and the carbon content of the steel. The square root correlation holds up to about 0.05% C
in BOF and up to 0.08% C in OBM(Q-BOP) steelmaking. At low carbon contents, the oxygen con-
tent of steel in the BOF practice is higher than that in OBM steelmaking. For carbon contents above
0.15% the product [ppm O] [%C] is essentially constant at about 30 ± 2, which is the equilibrium
value for an average gas (CO) bubble pressure of about 1.5 atmosphere in the steel bath.
At low carbon levels in the melt near the end of the blow, much of the oxygen is consumed by the
oxidation of iron, manganese and phosphorus, resulting in a lower volume of CO generation. With
the bottom injection of argon in the BOF combined-blowing practice, and the presence of hydro-
gen in the gas bubbles in OBM, the partial pressure of CO in the gas bubbles will be lowered in
both processes when the rate of CO generation decreases. A decrease in the CO partial pressure
at low carbon contents will be greater in OBM than in BOF practices, because the hydrogen con-
tent of gas bubbles in OBM is greater than the argon content of gas bubbles in BOF. It is pre-
sumably for this reason that the concentration product [O][C] in OBM steelmaking is lower than
in the BOF combined-blowing practice, particularly at low carbon levels.
The non-equilibrium states of the carbon-oxygen reaction at low carbon contents in BOF and
OBM(Q-BOP) are represented by the folowing empirical relations.
BOF with C < 0.05%,
(2.7.24)
[ppm O] %C = 135 ± 5

(2.7.25)
[ppm O] %C = 80 ± 5
2.7.2.3 Desiliconization
Silicon is oxidized out of hot metal early in the process. The oxidation reaction supplies heat and
the SiO2 reacts with the CaO to form the slag.
Si + 1/2O2 = (SiO2) (2.7.26)
The thermodynamics of the reaction indicate virtually all of the Si is oxidized. The rate is con-
trolled by liquid phase mass transfer, represented by the following equation:
A r m Si
[ ]
d % Si
=- % Si - % Si e (2.7.27)
dt W
where
mSi = mass transfer coefficient for silicon,
%Sie = the silicon content in the metal in equilibrium with the slag and is
close to zero.
As with decarburization, an overall rate parameter for Si can be defined, kSi, similar to kC. Within
our ability to estimate kSi it is about equal to kC since A, W, and are the same and mass transfer
coefficients vary only as the diffusivity to the one half power. Therefore 2.7.27 simplifies to:
% Si
ln = - k Si t (2.7.28)
% Si o
where %Si° is the initial silicon and kSi is the overall constant for Si and is approximately equal to
kC. The rate for Si oxidadtion is shown in Fig. 2.105.
2.7.2.4 Manganese Oxide–Carbon Relation

At low carbon contents, the ratio [%Mn]/(%MnO) is also found to be proportional to Ï%
wC
w, rep-
resented as given below.
BOF with C < 0.10%,
[% Mn ] 1
= 0.1 ± 0.02 (2.7.29)
(% MnO) %C

[% Mn ] 1
= 0.2 ± 0.02 (2.7.30)
(% MnO) %C
2.7.2.5 FeO–MnO–Mn–O Relations

From the foregoing empirical correlations for the non-equilibrium states of reactions involving the
carbon content of steel, the relations obtained for the reaction of oxygen with iron and manganese
are compared in the table below with the equilibrium values for temperatures of 1610 ± 20°C and
slag basicities of B = 3.2 ± 0.6.
Values for
BOF OBM(Q-BOP) slag-metal
C < 0.05%
__________ C < 0.08%
____________ equilibrium
__________
[ppm O]
_______ 32 ± 4 32 ± 5 26 ± 9
(%FeO)
[%Mn][ppm O]
_____________ 13.6 ± 3.2 16.1 ± 2.6 18 ± 6
(%MnO)
(MnO)
_____________ 2.6 ± 0.5 2.2 ± 0.2 1.9 ± 0.2
(%FeO)[%Mn]
It is seen that the concentration ratios of the reactants in low carbon steel describing the states of
oxidation of iron and manganese, are scattered about the values for the slag-metal equilibrium.
However, as indicated by the plant data in Fig. 2.108, the oxidation of iron and manganese are in
the non-equilibrium states for high carbon contents in the steel at turndown. Although the concen-
tration product [O][C] is close to the equilibrium value for an average CO pressure of about 1.5
atm in the steel bath, the concentrations of iron oxide and manganese oxide in the slag are above
the equilibrium values for high carbon contents in the melt.
It is concluded from these observations that:
(i) at low carbon contents the equilibrium state of iron and manganese oxidation con-
trols the concentration of dissolved oxygen
(ii) at high carbon contents it is the CO–C–O equilibrium which controls the oxygen
content of the steel.
60
Turndown C: 0.1-0.3%
BOF
40 1630°C
[ppm O]
(%FeO)
OBM
Slag-metal equilibrium
20 1590°C
30
1630°C
[%Mn] [ppm O]
20
(%MnO)
1590°C
10
0
2.4 2.8 3.2 3.6 4.0
B
Fig. 2.108 Non-equilibrium states of oxidation of iron and manganese at high carbon contents in steel. From
Ref. 103.
160 2.7.2.6 State of Phosphorus Reaction

The slag/metal phosophorus distribution
BOF
ratios at first turndown in BOF and OBM
OBM
Equilibrium for
(Q-BOP) steelmaking are plotted in Fig.
120 2.109 against the carbon content of the steel.
(O) contents in OBM
The plant data are for the turndown temper-
atures of 1610 ± 20°C and slags containing
50 ± 2% CaO and 6 ± 2% MgO. The shaded
(%P)
[%P]
80 area is for the slag-metal equilibrium for the

above stated conditions in OBM(Q-BOP),
based on the empirical [O][C] relations, i.e.
for C < 0.08%, [ppm O] Ïw %Cw = 80 and for
C > 0.15%, [ppm O][%C] = 30.
40
Noting that the oxygen contents of the steel
at low carbon levels are greater in BOF
steelmaking, the equilibrium phosphorus
%CaO: 48 52
distribution ratios will likewise be greater
0 in BOF than in (OBM) Q-BOP steelmaking.
0 0.1 0.2 0.3 For example, at 0.05% C and about 600 ppm
%C O in BOF at turndown the average equilib-
Fig. 2.109 Slag/metal phosphorus distribution ratios at rium value of (%P)/[%P] is about 200 at
first turndown in BOF and OBM(Q–BOP) practices; 1610°C, as compared to the average value
slag–metal equilibrium values are within the hatched of 60 in OBM(Q-BOP) at 0.05% C with
area for Q–BOP. From Ref. 103.
about 360 ppm O.
Below 0.04% C the phosphorus distribution ratios in OBM(Q-BOP) are in general accord with the
values for slag-metal equilibrium. However, at higher carbon contents the ratios (%P)/[%P] are
well above the equilibrium values. In the case of BOF steelmaking below 0.1% C at turndown, the
ratios (%P)/[%P] are much lower than the equilibrium values. On the other hand, at higher carbon
contents the phosphorus distribution ratios are
higher than the equilibrium values as in the case
200
of OBM(Q-BOP).
The effect of temperature on the phosphorus dis- Equilibrium for BO = 52% and
tribution ratio in OBM(Q-BOP) is shown in Fig. [O] contents at
0.014% C
2.110 for melts containing 0.014% to 0.022% C 160 0.022% C
with BO = 52 ± 2% in the slag.
2.7.2.7 State of Sulfur Reaction

(%P)
[%P]
A highly reducing condition that is required for 120

extensive desulfurization of steel is opposite to the
oxidizing condition necessary for steel making.
However, some desulfurization is achieved during
oxygen blowing for decarburization and dephos- 80
phorization. As seen from typical examples of the
BOF and OBM(Q-BOP) plant data in Fig. 2.111,
the state of steel desulfurization at turndown,
40
described by the expression [%O](%S)/[%S], is 1600 1640 1680 1720
related to the SiO2 and P2O5 contents of the slag. Temperature °C
Most of the points for OBM(Q-BOP) are within
the hatched area for the slag-metal equilibrium Fig. 2.110 Effect of turndown temperature on the
slag/metal phosphorus distribution ratios in OBM
reproduced from Fig. 2.104. However, in the case (Q-BOP) for turndown carbon contents of 0.014 to
of BOF steelmaking the slag/metal sulfur distrib- 0.022% C; curves are for slag–metal equilibrium.
ution ratios at turndown are about one-third or From Ref. 103.
1.2 one-half of the slag-metal equilibrium values.

BOF 1590-1630°C, C < 0.05% At higher carbon contents, e.g. C > 0.1%, the
OBM 1590-1700°C, C < 0.1% sulfur distribution ratios in both processes are
Slag metal equilibrium below the slag-metal equilibrium values.
0.8 2.7.2.8 Hydrogen in BOF and OBM(Q-BOP)
[%O]
Steelmaking
(%S)
[%S]
Hydrogen and nitrogen are removed from the

steel bath during oxygen blowing by CO car-
0.4
rying off H2 and N2.
2H(in steel) ® H2(g) (2.7.31)
2N(in steel) ® N2(g) (2.7.32)
1.2
0 10 20 30 There is however continuous entry of both
%SiO2 + 0.84 ´ %P2O5 hydrogen and nitrogen into the bath by various
Fig. 2.111 Equilibrium and non–equilibrium states of means. In both BOF and OBM(Q-BOP) there is
sulphur reaction in OBM(Q-BOP) and BOF steel- invariably some leakage of water from the water
making. From Ref. 103. cooling system of the hood into the vessel. In
the case of OBM(Q-BOP), the natural gas
(CH4) used as a tuyere coolant is a major source
of hydrogen.
The hydrogen content of steel in the tap ladle, measured by a probe called HYDRIS, is less than 5
ppm in BOF steelmaking and 6 to 10 ppm H in OBM(Q-BOP) steelmaking. Because natural gas
(CH4) is used as a tuyere coolant in the OBM, the hydrogen content of steel made in this vessel is
always higher than that in the BOF vessel. In both practices the re-blow will always increase the
hydrogen content of the steel. A relatively small volume of CO evolved during the re-blow cannot
overcome the hydrogen pickup from various sources.
2.7.2.9 The Nitrogen Reaction

As demonstrated, the decarburization reaction is reasonably well understood as well as the slag-
metal re-fining reactions which approach equilibrium. The removal of nitrogen is complex and
depends on numerous operating variables. With the advent of the production of high purity inter-
stitial free (IF) steels the control of nitrogen has become of great importance. For years there have
been “rules of thumb” for achieving low nitrogen. For example, using more hot metal and less
scrap in the charge or with combined blowing processes, switching the stirring gas from nitrogen
to argon have resulted in lower nitrogen contents.
The sources of nitrogen in oxygen steelmaking include the hot metal, scrap, impurity nitrogen in
the oxygen, and nitrogen in the stirring gas. Nitrogen from the atmosphere is not a major factor
unless at first turndown a correction or reblow is required, in which case the furnace fills up with
air which is entrained into the metal when the oxygen blow restarts, resulting in a significant nitro-
gen pickup of 5 to 10 ppm.
Goldstein and Fruehan169 developed a comprehensive model to predict the nitrogen reaction in
steelmaking. Nitrogen is removed by diffusing to the CO bubbles in the emulsion,or to the bubbles
in the bath in bottom blowing. The nitrogen atoms combine to form N2, the rate of which is con-
trolled by chemical kinetics and the nitrogen gas is removed with the CO bubbles. Both mass trans-
fer and chemical kinetics contribute to the overall rate of removal.
N(metal) = N(surface) (2.7.33)
2N(surface) = N2 (2.7.34)
The mixed control model presented previously is the basis for the model. There are, however, sev-
eral complications. For example, all of the rate parameters are functions of temperature. The sulfur
content of the metal changes with time and the oxygen content at the surface is considerably dif-
ferent from the bulk concentration. Consequently the temperature and surface contents of oxygen
and sulfur must be known as functions of time. Also, scrap is melting during the process changing
the composition of the metal. These details are dealt with and discussed in detail elsewhere.169
An example of the results of the model calculations are shown in Fig. 2.112 and Fig. 2.113. In Fig.
2.112 the nitrogen contents are shown for two hypothetical cases for a 200 ton oxygen steelmaking con-
verter blowing oxygen at 800 Nm3/min. In one case the charge contains 80% heavy scrap, 16 cm thick
containing 50 ppm nitrogen; in the second case there is no scrap but cooling is achieved by the use of
a 15% addition of DRI containing 20 ppm nitrogen. For the case of the heavy scrap, it melts late in the
process releasing its nitrogen after most of the CO is generated, causing the nitrogen content to increase
to 30 ppm at the end of the blow. With no scrap the final nitrogen is significantly lower at 20 ppm. The
model also indicates if ore, which has no nitrogen, is used for cooling, levels of 10 ppm nitrogen can
be obtained. In comparing the bulk and surface nitrogen contents, it is found that the surface concen-
tration is only slightly less than the bulk, indicating that the nitrogen reaction is primarily controlled by
chemical kinetics. In particular, towards the end of the blow the oxygen content at the surface is high,
retarding the rate. This model is useful for optimizing the process for nitrogen control. For example,
the effect of oxygen purity, the time of switching the stirring gas from N2 to argon, the metal sulfur
content and the amount and size of scrap can be determined. The effect of switching the stirring gas
from N2 to Ar is shown in Fig. 2.113.
2.7.2.10 General Considerations

The analyses of plant data on slag and metal samples taken at first turndown have revealed that
there are indeed equilibrium and non-equilibrium states of reactions in oxygen steelmaking. In all
the reactions considered, the carbon content of steel is found to have a decisive effect on the state
of slag-metal reactions.
0.0090
0.0080 DRI 15% with 20 ppm N

Scrap 20% with 50 ppm N
0.0070
Nigtrogen content (%)
0.0060
0.0050
0.0040
0.0030
0.0020
0.0010
0 2 4 6 8 10 12 14
Time (min)
Fig. 2.112 Computed nitrogen content in a 200 ton oxygen steelmaking converter for 20% heavy scrap – 80%
light scrap and DRI at a blowing rate of 800 Nm3/min O2. From Ref. 167.
100
Si Oxidation
80
Decarburization Stirring gas

Nitrogen (ppm)
60
N2
40
N2–Ar
20
Ar
0
0 5 10 15
Time (min)
Fig. 2.113 The effect of bottom stirring practice on nitrogen removal in the BOF with N2 during the entire blow,
Ar during the entire blow, and with a switch from N2 to Ar at 50% through the blow. From Ref. 167.
Considering the highly dynamic nature of steelmaking with oxygen blowing and the completion of
the process in less than 20 minutes, it is not surprising that the slag-metal reactions are in the non-
equilibrium states in heats with high carbon contents at turndown. With regard to low carbon heats
all the slag-metal reactions are close to the equilibrium states in bottom blown processes. In the
case of BOF steelmaking however, the states of steel dephosphorization and desulfurization are
below the expected levels for slag-metal equilibrium. As we all recognize it is of course the bottom
injection of lime, together with oxygen, that brings about a closer approach to the slag-metal equi-
librium in OBM(Q-BOP) as compared to the BOF practice, particularly in low carbon heats.
2.8 Fundamentals of Reactions in Electric Furnace Steelmaking

The EAF process is described in detail in Chapter 10. The slag-metal equilibrium for EAF refining
reactions are similar to those for oxygen steelmaking discussed in detail in Section 2.7. The state of
reactions in EAF steelmaking are similar but, in general, are slightly further from equilibrium due to
less stirring and slag-metal mixing as compared to oxygen steelmaking. In this section, the slags and
the state of the carbon, sulfur and phosphorous reactions are briefly discussed. The nitrogen reaction
is considerably different and , along with control of residuals, is more critical and consequently is dis-
cussed in more detail.
2.8.1 Slag Chemistry and the Carbon, Manganese, Sulfur and Phosphorus
Reactions in the EAF
The state of slag-metal reactions at the time of furnace tapping discussed here is based on about fifty
heats acquired from EAF steelmaking plants. The data considered are for the grades of low alloy
steels containing 0.05 to 0.20% C and 0.1 to 0.3% Mn plus small amounts of alloying elements.
50
40
BOF
FeO (%)
30
BOF
20
10
0 10 20 30 20 30 40 50
SiO2 (%) CaO (%)
Fig. 2.114 FeO, SiO2 and CaO contents of EAF slags at tap are compared with the BOF slags. From Ref. 27.
The iron oxide, silica and calcium oxide contents of slags are within the hatched areas shown in Fig.
2.114. The silica contents are slightly higher than those in oxygen steelmaking slags. On the other hand
the CaO contents of EAF slags are about 10% lower than those in the BOF slags. This difference is due
to the higher concentrations of Al2O3, Cr2O3 and TiO2 in the EAF slags; 4 to 12% Al2O3, 1 to 4% Cr2O3
and 0.2 to 1.0% TiO2. In these slags the basicity ratio B is about 2.5 at 10% FeO, increasing to about 4
at 40% FeO.
40 As is seen from the plant data within the
hatched area in Fig. 2.115, the iron oxide con-
tents of EAF slags are much higher than those
of the BOF slags for the same carbon content
30 at tap. However, the product [%C][ppm O] at
FeO (%)
tap is about 26 ± 2 which is slightly lower than

that for oxygen steelmaking.
20 EAF For the slag-metal reaction involving [Mn],
BOF
(FeO) and (MnO), the equilibrium relation
K¢FeMn = (%MnO)/(FeO)[%Mn] is 1.8 ± 0.2 at
OBM 1625 ± 15°C and the basicity B > 2.5. The
10 EAF data are scattered about K¢FeMn = 1.8 ±
0 0.05 0.10 0.15 0.20
0.4, which is near enough to the slag-metal
C (%)
equilibrium similar to those in oxygen steel-
Fig. 2.115 Relation between %FeO in slag and %C in making.
steel at tap in EAF steelmaking is compared with the
relations in BOF and OBM(Q–BOP) steelmaking. For steels containing 0.07 to 0.09% C in the
From Ref. 27. furnace at tap, the estimated dissolved oxy-
gen contents will be about 370 to 290 ppm O.
For slags containing 38–42% CaO and 5–7% MgO, the equilibrium constant KPO for the phospho-
rus reaction, given in equations 2.7.10, is in the range 2.57 ´ 104 to 4.94 ´ 104 at 1625°C. For the
tap carbon contents and slag composition ratios will be in the range 4 to 13. The EAF plant data
show ratios (%P)/[%P] from 15 to 30. Similar to the behavior in BOF steelmaking, the state of
phosphorus oxidation to the slag in EAF steelmaking is greater than would be anticipated from the
equilibrium considerations for carbon, hence oxygen contents of the steel at tap.
The state of steel desulfurization, repre- 300

sented by the product {(%S)/[%S]} ´
(%FeO), decreases with increasing con-
tents of SiO2 and P2O5 in the slag, as
shown in Fig. 2.116. The EAF plant data
200
(% FeO)
within the shaded area are below the val- Slag-metal equilibrium
ues for the slag-metal equilibrium. This
non-equilibrium state of the sulfur reac-
tion in EAF steelmaking is similar to that
(% S)
[% S]
observed in BOF steelmaking. 100
General indications are that the states of EAF plant
slag-metal reactions at tap in EAF steel- data at tap
making are similar to those noted in oxy-
gen steelmaking at carbon contents above
0
0.05% C. That is, the slag/metal distribu-
0 10 20 30
tion ratio of manganese and chromium are
scattered about the equilibrium values; % SiO2 + 0.84 ´ % P2O5
the phosphorus and sulfur reactions being
in non-equilibrium states. Also, the iron Fig. 2.116 Slag/metal sulfur distribution ratios at tap in
oxide content of the slag is well above the EAF steelmaking are compared with the slag-metal equi-
slag-metal equilibrium values for carbon librium values. From Ref. 27.
contents of steel at tap. On the other hand,
the product [%C][ppm O] = 26 ± 2 approximately corresponds to the C–O equilibrium value for
gas bubble pressures of 1.3 ± 0.2 atm in the EAF steel bath.
2.8.2 Control of Residuals in EAF Steelmaking

The growth of EAF steelmaking in the 1990’s was in the production of higher grades of steel,
including flat rolled and special bar quality steels. These steels require lower residuals (Cu, Ni, Sn,
Mo, etc.) than merchant long products normally produced from scrap in an EAF. It is possible to
use high quality scrap such as prompt industrial scrap, but this is expensive and not always avail-
able. The solution to the problem is the use of direct reduced iron such as DRI, HBI, iron carbide
or other scrap substitutes such as pig iron.
The control of residuals in EAF steelmaking is based on a simple mass balance. Residual elements
such as copper will report almost entirely to the steel. The control of residuals can be demonstrated
in a very simple example. Consider the copper content of steel produced from two grades of scrap,
Number 1 bundles and shredded scrap, and DRI/HBI. It is critical to know the yield of each charge
material and its copper content, which is not always easy to assess accurately. There are also sec-
ondary effects such as increased slag levels. Nevertheless, this simple example demonstrates the gen-
eral concepts.
Yield Charge Fe Yield Cu Content Cu

Material
_________ (%)
____ (kg)
_____ (kg)
______ (wt. %)
_________ (kg)
_____
#1 Bundles 95 363 345 0.05 0.17
Shredded 95 500 475 0.25 1.19
DRI/HBI 90 200
_____ 180
_____ 0.0 0
____
Totals 1063 1000 1.36
Cu content in the steel will be 0.136%
2.8.3 Nitrogen Control in EAF Steelmaking

The nitrogen content of steels produced in the EAF is generally higher than in oxygen steelmak-
ing primarily because there is considerably less CO evolution, which removes nitrogen from steel.
The usual methods of controlling nitrogen are by carbon oxidation and the use of direct reduced
iron or pig iron. The control of nitrogen in the EAF is discussed in recent publications by Gold-
stein and Fruehan.170,171 They developed a model to predict the rate of nitrogen removal similar to
the one presented earlier for oxygen steelmaking. The removal of nitrogen is shown for a 100 ton
EAF as a function of oxygen usage in Fig. 2.117. Nitrogen removal decreases once the carbon con-
tent falls below approximately 0.3%C, as most of the oxygen is then reacting with Fe and is there-
fore not producing CO. Starting at a higher initial carbon allows for more CO evolution and
reduces the activity of oxygen, which retards the rate of the nitrogen reaction.
The other method of reducing nitrogen is the use of DRI or HBI. These materials reduce nitrogen
primarily through dilution. There was a belief that the CO evolved from these products also
removed nitrogen. However it has been demonstrated that the CO evolved from the reaction of C
and FeO in the HBI/DRI is evolved at lower temperatures (1000°C) and is released while heating
or in the slag phase. Since it does not pass through the metal it does not remove nitrogen. The CO
does dilute any N2 in the furnace atmosphere.
The nitrogen removal using 25% and 50% DRI/HBI is shown in Fig. 2.118. The primary effect is
simple dilution.
100
80 Meltdown carbon = 0.5%

Nitrogen (ppm)
60
Fig. 2.117 Change in bulk nitrogen
concentrations during EAF steelmak-
40 ing. From Ref. 171.
Meltdown carbon = 1.0%
20
0
0 2 4 6 8 10 12 14
.
V O2 (Nm3/ton)
100
No DRI
Nitrogen content (ppm)
80
60 25% DRI charged
Fig. 2.118 Effect of DRI on nitrogen

content during EAF steelmaking.
40 From From Ref. 171.
50% DRI charged
20
0
0 10 20 30 40
Time (min)
2.9 Fundamentals of Stainless Steel Production

There are numerous processes to produce stainless steel including the AOD (argon oxygen decar-
burization), VOD (vacuum oxygen decarburization), and other variations of these processes. All of
the processes are based on the reduction of the CO pressure to promote the oxidation of carbon in
preference to chromium. The processes and reactions are described in detail in Chapter 12. This
section will be limited to the reaction mechanisms and fundamentals of decarburization, nitrogen
control and slag reduction.
2.9.1 Decarburization of Stainless Steel

Stainless steels can not be easily produced in an EAF or oxygen steelmaking converter as under
normal conditions Cr will be readily oxidized in preference to C at low carbon contents. This leads
to excessive Cr yield loss and necessitates the use of high cost low carbon ferrochrome rather than
lower cost high carbon ferrochrome.
The critical carbon content at which Cr is oxidized rather than carbon can be computed based on
the following reaction:
Cr2O3 + 3C = 3CO + 2Cr (2.9.1)
The equilibrium constant is given by:
3
pCO aCr
2
K=
[ ] (2.9.2)
3
aCr2O 3 f C3 %C
where fC is the activity coefficient of carbon 0.5

and ai is the activity of species i. The reac-
tion could be written in terms of Cr3O4 and
the results would be similar. In Fig. 2.119
0.4
the equilibrium for reaction 2.9.1 is given as
a function of pCO for an 18% Cr steel.
The critical carbon content is defined as 0.3
C (%)
the carbon content below which Cr is 1600°C

oxidized. The critical carbon increases
with chromium content or activity (aCr2/3)
0.2
and decreasing temperature.
However decarburization in stainless
steelmaking is not controlled by equilib- 0.1
rium but rather reaction kinetics. It was
found that for the AOD process to be 1700°C
effective the gas had to be injected deep
0
in the steel bath. Fruehan172 demon- 0 0.2 0.4 0.6 0.8 1.0
strated in laboratory experiments that PCO (atm)
when oxygen is injected into an Fe–Cr–C
bath the oxygen initially oxidizes Cr. Fig. 2.119 Critical carbon content for an 18% Cr steel as a
This lead to the following reaction mech- function of CO pressure.
anism and model for decarburization.
When oxygen initially contacts an Fe–Cr–C bath it primarily oxidizes Cr to Cr2O3. As the chrome
oxide particles rise through the bath with the gas bubbles, carbon diffuses to the surface and reduces
the oxide according to the reaction 2.9.1. The rate of the reaction is controlled by mass transfer of
carbon. The reaction also takes place with the top slag. The rate equation can be expressed as:
d %C
dt
r
= - å m i Ai %C - %C ie
W i
[ ] (2.9.3)
where r is the density of steel, W is the weight of the metal, mi are the mass transfer coefficients,
Ai are the surface areas, %Cie is the equilibrium carbon content, and the subscript i refers to the
individual reaction sites such as the rising bubble and the top slag.
The equilibrium carbon content is determined from the equilibrium for reaction 2.9.1 which is a
function of temperature, chromium content and local CO pressure. The CO pressure in turn
depends on the rate of decarburization and the Ar or N2 in the gas and the total pressure. The details
of the calculation are given elsewhere.
Equation 2.9.3 is valid when mass transfer of carbon is limiting. At high carbon contents mass
transfer of carbon is sufficient to consume all the oxygen and the rate of decarburization is simply
given by the mass balance based on oxygen flow rate.
The Cr loss to the slag can be computed from the mass balance for oxygen and is given by
10 -2 W
[ ] [ ]
4 M Cr ˙
D %Cr = N t - D %C (2.9.4)
3 W 10 -2
O 2
2MC
1.4 where
MCr, MC = molecular weights of
1.2
O2:Ar = 3:1 Cr and C,
1.0 O2:Ar = 1:1 %C <0.4 .W = weight of steel,
NO2 = molar flow rate of
C (wt %)
0.8 oxygen,
D[%C] = change in carbon
0.6
content.
0.4 Simplified calculations for the rates
0.2
of decarburization and Cr oxidation
of an 18–8 stainless steel are pre-
sented in Fig. 2.120 and Fig. 2.121.
0 10 20 30 40
Time (min)
These calculations indicate that
switching the O2/Ar ratio at 0.4% C
Fig. 2.120 Calculated rate of decarburization for an 18-8 stain- slightly increases the rate of decar-
less steel. From Ref. 172. burization but significantly reduces
Cr oxidation.
This model is the basis of process
control models for the AOD, VOD
and other similar processes. These
4 are discussed in Chapter 12 in more
(0.15% C) detail.
3 O2:Ar = 3:1
DCr (wt %)
2 (0.15% C)
Fig. 2.121 Calculated rate of Cr oxida-

tion for an 18–8 type stainless steel.
1 O2:Ar = 1:1 From Ref. 172.
(0.40% C)
0 10 20 30 40 50
Time (min)
2.9.2 Nitrogen Control in the AOD

It is desirable to use nitrogen (N2) in the AOD in place of Ar because of its lower cost. Also, for
some steels it is desirable to alloy nitrogen by blowing N2 gas. A model for nitrogen control was
developed by one of the authors based on fundamental principles and is briefly described below.
The nitrogen control model presented here is based on the mixed control model for chemical kinet-
ics and mass transfer in series given in section 2.2.4. Equation 2.9.5 requires knowledge of the
pressure of N2 which is given by
W R T PT d % N N˙ N 2
pN2 = + PT (2.9.5)
100 M N 2 dt N˙ CO + N˙ N 2 + N˙ Ar
where .
Ni = molar flow rate of species i,
W = weight of steel,
PT = total pressure,
MN2 = molecular weight of N2 (28).
The rate of CO evolution is determined by the rate of decarburization, equations 2.9.3 and 2.9.4, and
consequently it is necessary to include the decarburization model given in the previous section.
The rate of dissociation of N2 depends on the Cr and S contents. Glaws and Fruehan173 measured
the rate for Fe–Cr–Ni–S alloys and found that Cr increased the rate while sulfur decreased the rate.
The mass transfer parameter can be estimated from basic principles but requires knowledge of the
bubble size; alternatively, it can be determined with a limited quantity of plant data by determin-
ing the best value of mA, the mass transfer coefficient times the surface area. Basic principles can
be used to estimate m reasonably accurately using equations 2.2.74 and 2.2.75, thereby allowing
for an estimation of A. Results of the model are presented in Fig. 2.122 and 2.123.
0.16
0.14
0.12
0.10
Nitrogen (%)
Fig. 2.122 Nitrogen pickup O2

in 75 metric tons of 18–8 =1
N2
stainless steel containing 0.08
0.03% S using O2/N2 = 3
and O2/N2 = 1 gas mix-
tures, and with total flow of
0.06
0.85 m3/s at 1600°C. From
Ref. 50.
0.04
O2
=3
0.02 N2
0.00
0 5 10 15 20 25 30 35 40
Time (min)
0.20
0.16
0.03 %S
Nitrogen (%)
0.12
Fig. 2.123 Rate of nitrogen removal
from 75 metric tons of 18–8 stainless
steel containing 0.005 and 0.03% S
using 0.85 m3/s of O2 and Ar at
0.08 1600°C. From Ref. 50.
0.008 %S
0.04
0.00
0 5 10 15 20 25 30 35 40
Time (min)
2.9.3 Reduction of Cr from Slag

In stainless steel production some chromium is oxidized to slag, approximately 3% for an 18% Cr
steel. It is necessary to recover this Cr by adding a reductant, usually Si, (as ferrosilicon) or alu-
minum, which reacts with Cr2O3. For example, for Si the reaction is given by
3Si + 2(Cr2O3) = 3(SiO2) + 4Cr (2.9.6)
The equilibrium distribution of chromium between slag and metal is given in Fig. 2.124 as a func-
tion of Si content and slag chemistry. From Fig. 2.124 and a mass balance from reaction 2.9.6, it
is possible to compute the Si required for the desired Cr reduction.
The typical compositions of AOD slags after decarburization and after silicon reduction are given
in Table 2.14. As is seen from the experimental data in Fig. 2.124, the higher the slag basicity and
higher the silicon content of steel, the lower is the equilibrium slag/metal distribution of chromium,
i.e. the greater the chromium recovery from the slag.
Table 2.14 Ranges of AOD Slag Composition after Decarburization and after Silicon
Reduction.
Composition, wt.%
After decarburization
____________________ After silicon reduction
____________________
FeO 4–6 1–2
MnO 4–8 1–3
SiO2 12–18 30–40
Al2O3 18–22 3–8
CaO 8–15 33–43
MgO 7–15 10–20
Cr2O3 20–30 1–3
Aluminum is a stronger reductant and nearly all will react to reduce the Cr2O3.
1
SiO2-saturated
CaO
SiO2
0.7
Fig. 2.124 Equilibrium

0
slag/metal chromium distrib- 1.3
log (% Cr)
ution varying with the con-
[% Cr]
centration of silicon in iron 1.5
coexisting with chromium
oxide containing CaO–Al2O3–
SiO2 slags at temperatures of
1600 to 1690°C. See Ref. 3 -1
for references to experimental
data.
-2
-3 -2 -1 0 1
log [% Si]
2.10 Fundamentals of Ladle Metallurgical Reactions

Several books and conferences have been devoted to ladle or secondary metallurgy. The processes
and fundamentals are described in detail in Chapter 11. In this section only the fundamentals of
deoxidation, desulfurization and inclusion modification are discussed.
2.10.1 Deoxidation Equilibrium and Kinetics

There are primarily three elements used in steel deoxidation:
(l) Mn as low or high C ferro alloy,
(2) Si as low or high C ferro alloy or as silico manganese alloy,
(3) Al of approximately 98% purity.
2.10.1.1 Deoxidation with Fe/Mn

When the steel is partially deoxidized with Mn, the iron also participates in the reaction, forming
liquid or solid Mn(Fe)O as the deoxidation product.
[Mn ] + [O] ® MnOüï

ý liquid or solid Mn ( Fe)O (2.10.1)
Fe + O ® FeO ïþ
[ ] []
The state of equilibrium of steel with the deoxidation product Mn(Fe)O is shown in Fig. 2.215.
2.10.1.2 Deoxidation with Si/Mn

Depending on the concentrations of Si and Mn added to steel in the tap ladle, the deoxidation prod-
uct will be either molten manganese silicate or solid silica.
[Si] + 2 [O] ® SiO

ü
ï
2
ý molten xMnO × SiO2 or solid SiO2 (2.10.2)

Mn + O ® MnOïþ
[ ] []
0.10
1650 C
Oxygen (wt%)
0.05 1600 C Fig. 2.125 Manganese and oxygen contents of

iron in equilibrium with solid FeO–MnO deoxida-
tion product.
1550 C
0
0 0.5 1.0 1.5
Manganese (wt%)
From the experimental work of various investigators174–176 the following equilibrium relation is
obtained for the Si/Mn deoxidation reaction.
[Si] + 2(MnO) = 2[Mn] + (SiO2) (2.10.3)
[ ]
2
æ % Mn ö aSiO 2
K MnSi =ç ÷
è a MnO ø %Si [ ] (2.10.4)
1510
log K = + 1.27 (2.10.5)
T
where the oxide activities are relative to pure solid oxides. For high concentrations of silicon (>
0.4%) the activity coefficient fSi should be used in the above equation, thus log fSi = 0.11 ´ %Si.
The activities of MnO in manganese silicate melts have been measured by Rao and Gaskell.177 Their
results are in substantial agreement with the results of the earlier work by Abraham et al.152 The
activity of the oxides (relative to solid oxides) are plotted in Fig. 2.126. For liquid steel containing
Mn < 0.4% the deoxidation product is a MnO-rich silicate with FeO < 8%; therefore the activity data
in Fig. 2.126 can be used together with equation 2.10.4 in computing the equilibrium state of the
Si/Mn deoxidation as given in Fig. 2.127. The deoxidation product being either solid silica or molten
manganese silicate depends on temperature, Si and Mn contents, as shown in Fig. 2.127.
1.0
0.8
0.6
aMnO
1600°C
1500°C
Fig. 2.126 Activities in MnO–SiO2
0.4
melts with respect to solid oxides.
From Ref. 27.
0.2
0
0.01 0.1 1.0 10 100
aSiO2
2
aMnO
0.03
(a)
[0.0% Mn]
[0.4% Mn]
[0.6% Mn]
0.02
[0.8% Mn]
[1.0% Mn]
Oxygen (wt%)
saturated with solid SiO2
0.01
molten silicate
0
0 0.1 0.2 0.3 0.4
Silicon (wt%)
0.15
(b)
Temp, °C 1650°C 1600°C 1550°C 1500°C
Silicon (wt%)
0.10
deoxidation
product
solid SiO2
0.05 deoxidation
product
molten
manganese
silicate
0
0 0.25 0.50 0.75 1.0
Manganese (wt%)
Fig. 2.127 Equilibrium relations for deoxidation of steel with silicon and manganese at 1600°C. From Ref. 27.
At silica saturation, the deoxidation is by silicon alone, for which the equilibrium relation for unit
SiO2 activity is reduced to
[ ][ ]
2 30 , 410 (2.10.6)
log % Si ppm O =- + 19.59
T
2.10.1.3 Deoxidation with Si/Mn/Al

Semi-killed steels with residual dissolved oxygen in the range 40 to 25 ppm are made by deoxidiz-
ing steel in the tap ladle with the addition of a small amount of aluminum together with silicoman-
ganese, or a combination of ferrosilicon and ferromanganese. In this case, the deoxidation product is
molten manganese aluminosilicate having a composition similar to 3MnO•Al2O3•3SiO2. With a small
addition of aluminum, e.g. about 35 kg for a 220 to 240 ton heat together with Si/Mn, almost all the
aluminum is consumed in this combined deoxidation with Si and Mn. The residual dissolved alu-
minum in the steel will be less than 10 ppm. For the deoxidation product 3MnO•Al2O3•3SiO2 satu-
rated with Al2O3, the silica activities are 0.27 at 1650°C, 0.17 at 1550°C and decreasing probably to
about 0.12 at 1500°C. Using these activity data the deoxidation equilibria are calculated for Al/Si/Mn;
these are compared in Fig. 2.128 with the residual ppm O for the Si/Mn deoxidation at the same con-
centrations of Mn and Si.
160
% Mn % Si
Si/Mn
0.4 0.05
0.4 0.50
120
Al/Si/Mn
O (ppm)
80
Si/Mn
40 Al/Si/Mn
0
1500 1525 1550 1575 1600
Temperature (°C)
Fig. 2.128 Deoxidation equilibria with Si/Mn compared with Al/Si/Mn for the deoxidation product saturated
with Al2O3. From Ref. 27.
0.10
0.05
0.02 Al2O3
Al (wt%)
0.01
0.005
liquid
inclusion
0.002
SiO2
0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 10
Si (wt%)
Fig. 2.129 Equilibrium inclusions for an Fe–Al–Si–1.0% Mn steel at 1600°C. From Ref. 178.
When intentionally using some Al for deoxidation or unintentionally with Al being in the ferrosili-
con the equilibrium deoxidation product can be solid Al2O3, molten manganese silicate or solid SiO2.
To avoid clogging of continuous casting nozzles it is desirable to have liquid inclusions. The equi-
librium deoxidation product is given as a function of Al and Si contents in Figure 2.129. As seen in
this figure even small amount of Al in solution will lead to Al2O3 inclusions which can clog nozzles.
2.10.1.4 Deoxidation with Al

Numerous laboratory experiments have been made on the aluminum deoxidation of liquid iron
using the EMF technique for measuring the oxygen activity in the melt. The equilibrium constants
obtained from independent experimental studies, cited in Ref. 179 agree within about a factor of
two. An average value for the equilibrium constant is given below.
Al2O3(s) = 2[Al] + 3[O] (2.10.7)
[% Al] [ ppm O ´ f ]
2 3
O
k= (2.10.8)
a Al 2O 3
62 , 680 (2.10.9)
log K = - + 31.85
T
The alumina activity is with respect to pure solid Al2O3. The effect of aluminum on the activity
coefficient of oxygen dissolved in liquid steel is given by log fO = 3.9 ´ [%Al]. At low concentra-
tion of aluminum, fAl ~ 1.0.
It should be noted that in the commercial oxygen sensors the electrolyte tip is MgO-stabilized zir-
conia. At low oxygen potentials as with aluminum deoxidation, there is some electronic conduc-
tion in the MgO-stabilized zirconia which gives an EMF reading that is somewhat higher than Y2O3
or ThO2 stabilized zirconia where the electronic conduction is negligibly small. In other words, for
a given concentration of Al in the steel the commercial oxygen sensor, without correction for par-
tial electronic conduction, registers an oxygen activity that is higher than the true equilibrium
value. To be consistent with the commercial oxygen sensor readings, the following apparent equi-
librium constant may be used for reaction 2.10.7 for pure Al2O3 as the reaction product.
62 , 680
log K a = -
+ 32.54 (2.10.10)
T
Another point to be clarified is that in the commercial oxygen sensor system, the EMF reading of
the oxygen activity is displayed on the instrument panel in terms of ppm O, as though the activity
coefficient fO = 1.0 in the Al-killed steel. If the deoxidized steel contains 0.05% Al, the apparent
oxygen activity using equation 2.10.10 will be (ppm O ´ fO) = 3.62; noting that at 0.05% Al, fO =
0.64, the apparent concentration of dissolved oxygen will be 3.62/0.64 = 5.65 ppm 0.
When the Al-killed steel is treated with Ca–Si the alumina inclusions are converted to molten cal-
cium aluminate. For the ratio %CaO/Al2O3 = 1:1, the activity of Al2O3 is 0.064 with respect to pure
Al2O3 at temperatures in the range 1500–1700°C. The apparent equilibrium relations, consistent
with the readings of commercial oxygen sensors, are shown in Fig. 2.130 for the deoxidation prod-
ucts: pure Al2O3 and molten calcium aluminate with %CaO/%Al2O3 = 1:1.
2.10.1.5 Silicon and Titanium Equilibrium in Aluminum Deoxidized Steel

When steels are deoxidized with aluminum and also contain silicon and titanium slag metal equi-
librium is established for these elements according to the reactions.
3SiO2 + 4Al = 2(Al2O3) + 3Si (2.10.11)
3TiO2 + 4Al = 2(Al2O3) + 4Ti (2.10.12)
In principle the equilibrium distribution ratio %Si/(%SiO2) and %Ti/(%TiO2) can be computed from
basic thermodynamics. The equilibrium ratios observed in practice are shown in Figure 2.131 and 2.132.
16 80
Deoxidation product ~1580°C tank degasser180
Pure Al2O3
Molten CaO-AlO3 1600°C equilibrium data181
12 CaO/Al2O3 = 1:1
60
O (ppm)
1625°C
(% SiO2)
[%Si]
40
4 1575°C
1625°C
1575°C
0
0 0.02 0.04 0.06 0.08
Al (%) 20
Fig. 2.130 Deoxidation with aluminum in equilib-
rium with Al2O3 or molten calcium aluminate with
CaO/ Al2O3 = 1:1. From Ref. 27.
0
0 0.02 0.04 0.06 0.08 0.10 0.12
Al (%)
Fig. 2.131 Tank degasser data are compared with

the experimental equilibrium data for aluminum
reduction of silica from lime-saturated calcium alu-
minate melts containing SiO2 < 5%.From Ref. 27.
250
(%TiO2) 0.55
=
[%Ti] [%Al]4/3
Titanium distribution (TiO2) 200
[Ti]
150
100
50
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Al content after vacuum, [%]
Fig. 2.132 Silicon and titanium distribution after vacuum. From Ref. 178.
2.10.1.6 Rate Phenomena in Deoxidation

The equilibrium states for the common deoxidation reactions pertaining to steelmaking conditions
have been established reasonably well. However, the rate phenomena concerning the deoxidation
reactions is complex and is discussed in detail in a review paper by Turkdogan.182 A brief mention
of the state of our knowledge is adequate for the present purpose.
There are three basic consecutive steps involved in the deoxidation reaction; namely, formation of
critical nuclei of the deoxidation product in a homogeneous medium, progress of deoxidation
resulting in growth of the reaction products and their flotation from the melt.
Turkdogan has shown that because of the high interfacial tension between liquid iron and oxide and
silicate inclusions, a high supersaturation of the reactants in the metal is needed for spontaneous
nucleation of the deoxidation products as predicted from the theory of homogeneous nucleation. In
estimating the supersaturation ratio likely to be achieved under practical conditions, homogeneous
solution of deoxidizers in the steel was assumed. However, the dissolution of added deoxidizers in
liquid steel takes a finite time during which certain regions of the melt are expected to be very rich
in solute concentration:; in these regions the solution is sufficiently supersaturated locally for
homogeneous nucleation of the deoxidation product. Owing to the agitation in the ladle, the nuclei
thus formed are considered to be distributed in the melt, soon after the addition of deoxidizers.
Another source of nuclei is, of course, the thin oxide layer on the surface of particles of solid deox-
idizers added to steel.
A generally accepted view is that the deoxidation reactions at steelmaking temperatures are fast rela-
tive to other rate-controlling processes responsible for the growth and ultimate flotation of inclusions.
The rate phenomena is deoxidation is complex because of the side effects caused by the interplay
of several variables which cannot readily be accounted for in mathematical simulations of the deox-
idation process. However, certain important deductions can be made from the results of several
conceptual analyses based on simplified models and those of experimental observations.
1. The number of nuclei (z) formed at the time of addition of deoxidizers is of the
order of z =107/cm3 or higher.
2. The diffusion-controlled deoxidation reaction is essentially complete within a few

seconds when z > 106/cm3.
3. The deoxidation reaction may cease prematurely in parts of the melt depleted of
nuclei or oxide inclusions.
4. The inclusion size during deoxidation is in the range 1 to 40 mm.
5. In laboratory experiments with inductively stirred melts (~5 cm deep) most of the
oxide inclusions float out of the melt in 5 to 10 minutes.
6. The growth by collision and coalescence of ascending inclusions does not seem pos-
sible under the conditions of laboratory experiments with unstirred or moderately
stirred melts.
These observations are not mutually consistent. One possible explanation perhaps is that the nuclei
formed at the time of dissolution of deoxidizers are unevenly distributed in molten steel. In parts
of the melt where the number of nuclei is small, e.g. 104/cm3 to 105/cm3, the inclusions 20 to 40
mm in size rapidly float out of the melt, presumably prior to the deoxidation reaction. In parts of
the melt containing about 108 nuclei/cm3, the inclusions grow only to a micron size and ascend in
the melt with a creeping velocity. Convection currents or other means of stirring eventually bring
about more uniform distribution of these small inclusions. The particles thus brought to the parts
of the melt where the deoxidation reaction was incomplete, due to early depletion of inclusions,
bring about further deoxidation, growth and flotation.
Under practical conditions of deoxidation during filling of the ladle, there is sufficient stirring that
some inclusion growth may take place by collision and coalescence; also stirring brings about a
motion in the melt such that the inclusions could get attached to the surface of the ladle lining and
caught by the slag layer. Controlled gas stirring at low flow rates for 4–8 minutes is common prac-
tice to enhance inclusion removal.
2.10.2 Ladle Desulfurization

It is possible to desulfurize aluminum killed steels in the ladle or ladle furnace, using CaO based
slags, to less than 20 ppm S. The chemical reaction can be written as
3(CaO) + 2Al + 3S = 3(CaS) + Al2O3 (2.10.13)
The equilibrium sulfur distribution ratio can be calculated from the thermodynamics of reaction
2.10.13. In terms of the ionic reaction the reaction is
/3Al + S + (O2–-) = (S2–) + 1/3(Al2O3)
2
(2.10.14)
for which the equilibrium ratio is
(% S) % Al -
[% S] [ ]
2
K SA = 3
(2.10.15)
The value of KSA for CaO–Al2O3 and CaO–Al2O3–SiO2–MgO slags is shown in Figure 2.133.
The sulfur distribution ratios are shown in Fig. 2.134 and Fig. 2.135 for the stated Al contents.
From the data presented and a mass balance for sulfur it is possible to compute the final equi-
librium sulfur contents.
The slags will absorb sulfur until CaS forms decreasing the amount of dissolved CaO and decreas-
ing KSA. The solubility of CaS is given in Fig. 2.136.
Steels deoxidized with Si are difficult to desulfurize because the oxygen potential is significantly
higher than for Al killed steels. For these steels desulfurization is limited to 10 to 20%. Calcium
carbide is an effective desulfurizer because it decreases the oxygen potential and the resulting CaO
then is able to desulfurize.
4.8
CaO CaO-Al2O3-SiO2
saturation CaO-Al2O3-SiO2-6% MgO
Temperature °C
4.4 4.0 1600°C
1500
1550 % SiO2
Iog KSA
1600
4.0 3.6 0
1650 2
4
3.6 3.2
6
8
3.2 2.8
0.8 1.0 1.2 1.4 1.6 1.1 1.2 1.3 1.4 1.5
% CaO % CaO
% Al2O3 % Al2O3
Fig. 2.133 Effects of temperature and slag composition on the equilibrium relation KSA for the calcium-mag-
nesium aluminosilicate melts. From Ref. 27.
103 1500°C
1650°C
Fig. 2.134 Sulfur

distribution for
CaO–Al2O3 slags
at 1600°C. From
Ref. 187. % Al
102
0.05
(%S)
[%S]
0.05
0.01
0.005
101 0.01
0.005
100
30 40 50 60
CaO (wt%)
148
Copyright © 1998, The AISE Steel Foundation, Pittsburgh, PA. All rights reserved.
% CaO
60 50 40
20
Liquidus 10
449
O
275 166
Ca
387
50
70
670
2
%
%S10
iO
180
500
200
656
355
400
560
500 180
20 30 40 50
%Al2O3
Fig. 2.135 Sulfur distribution for CaO–Al2O3–SiO2 slags at 1600°C steel containing 0.03% Al. From Ref. 178.
Desulfurization is controlled by liquid phase mass

4.8
1600°C transfer. Since the sulfur distribution ratio is high
1550°C sulfur mass transfer in the metal is the primary rate
Lime controlling process. For this case the rate of desul-
4.0 saturated furization is given by the following178.
(% S) % Al -
1500°C
[% S] [ ]
2
K SA = 3
(2.10.16)
CaS (mass %)
3.2
(% S t )
2.4
[ %Se = ] LS
(2.10.17)
(%St ) = W
W
m
( [%So ] - [%St ])
s
(2.10.18)
1.6
Aluminate saturated where
m = mass transfer coefficient,
0.8 A = surface area,
0.4 0.8 1.2 1.6 2.0 Wm = weight of metal,
Ws = weight of steel,
Mass ratio % CaO r = density of steel,
% Al2O3
LS = sulfur distribution ratio,
Fig. 2.136 Solubility of CaS in calcium aluminate
melts related to the mass ratio %CaO/%Al2O3. (%St) = sulfur content of the slag,
From Ref. 27. [%St] = sulfur content of the metal,
[%Se] = equilibrium sulfur content of the metal,
[%So] = initial sulfur content of the metal.
Equations 2.10.16 through 2.10.18 can be solved and the rate is given by
ïì 1 æ Wm ö é %S t ù 1 æ Wm ö ïü
ln í1 + ç ÷ê ú- ç ÷ý
ïî LS è Ws ø ë %So û LS è Ws ø ïþ m A r
= t (2.10.19)
1 æ Wm ö Ws
1+ ç ÷
LS è Ws ø
Desulfurization increases with stirring rate, which increases m and more importantly A by provid-
ing slag-metal mixing, increased values of Ls, and higher slag volumes. It has been shown that 90%
desulfurization can be achieved in 10–15 minutes of intense stirring.
2.10.3 Calcium Treatment of Steel

Calcium is usually employed in ladle metallurgy as Ca or CaSi cored wire or by injecting CaSi
powder. Calcium is highly reactive and it could deoxidize, desulfurize, modify oxide inclusions or
modify sulfide inclusions. It should only be used in deoxidized steels because it is too expensive
to be used as a deoxidizer.
Since often the primary purpose of a calcium injection into the steel bath is to convert solid Al2O3
inclusions to liquid calcium aluminates to prevent Al2O3 from clogging casting nozzles, it is nec-
essary to know under what conditions it will react with the inclusions or simply react with sulfur.
There have been numerous thermodynamic calculations to predict the conditions for Al2O3 inclu-
sion modification. Several of these required knowledge of the thermodynamics of Ca in steel,
which is not accurately known. The conditions can be computed based on the thermodynamic prop-
erties of the inclusions themselves.
Consider the following reaction equilibrium in the calcium treated steel,
3(CaS) + (Al2O3) = 3(CaO) + 2Al + 3S (2.10.20)
where CaS is that formed by the reaction of calcium and sulfur, (Al2O3) is an alumina rich inclu-
sion such as Al2O3, CaO•Al2O3, etc and (CaO) represents an inclusion richer in CaO. The results of
these calculations are given in Fig. 2.137 in which the inclusion stability is given as a function of
Al and S contents183. Below
the CA curve CaO•Al2O3 is the 0.05
stabler oxide and below the
C12A7 liquid curve, liquid cal-
1550°C
cium aluminates form. For
example if Ca is added to a
steel containing 0.04% Al and 0.04
1500°C
0.015% S the alumina inclu-
sion will be converted to solid
CaO•Al2O3 and calcium there
will react to form CaS. These
Sulfur (wt%)
inclusions will clog casting 0.03

nozzles. For effective inclu- CA
sion modification the sulfur
content should be below
0.01% for a 0.04% Al steel.
0.02
Calcium also helps eliminate
MnS inclusions which form CA
during solidification. For a
steel low in sulfur which con-
tains liquid calcium aluminate 0.01
inclusions, much of the remain-
ing sulfur will be absorbed dur-
ing cooling and solidification C12A7 (liquid)
by these inclusions as they are C12A7 (liquid)
excellent desulfurizers. Also, 0.00
if any dissolved Ca is remain- 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.0
ing, although the authors Aluminum (wt%)
believe this will be very small, Fig. 2.137 Univariant equilibrium for CaS (aCaS = 0.74) and C12A7 or
it could react with sulfur on CA as a function of %Al and %S at 1550 and 1500° C.
the inclusions. The net result is a duplex inclusion consisting of a liquid calcium aluminate with a
CaS (MnS) rim. This is preferable to stringer MnS which can cause brittle fracture.
Effective calcium treatment is a critical issue and is discussed in Chapter 12 in more detail.
2.11 Fundamentals of Degassing

During the past two decades there has been an increase in vacuum degassing to reduce hydrogen
and nitrogen contents. However the major increase in vacuum degassing is for decarburization to
low carbon contents ( 40 ppm) which are required for good formability such as for interstitial free
(IF) steels. Vacuum degassing is discussed in detail in Chapter 11. Briefly there are two major types
of degassers: circulating, such as RH and RH–OB, and non-recirculating such as ladle and tank
degassers. In some cases oxygen is used to enhance the reactions, examples of these are RH–OB
(oxygen blowing) and VOD (vacuum oxygen decarburization).
Details on the types of degassers and the processes are given in Chapter 11. In this section only the
fundamental aspects of the reactions will be discussed.
2.11.1 Fundamental Thermodynamics

In vacuum degassing hydrogen, nitrogen, carbon, and oxygen can be removed by the following reactions.
H = 1/2H2 (2.11.1)
N = 1/2N2 (2.11.2)
C + O (FeO, 1/2O2) = CO + (Fe) (2.11.3)
Hydrogen and nitrogen are dissolved in
steel and removed by forming diatomic
1500
molecules. The thermodynamics are given
in Section 2.4.
Carbon is removed by reaction with oxy-
gen as dissolved oxygen, FeO in slag or
gaseous oxygen (O2) to form CO. Due to
the reduced pressure the formation of CO
is favored. This is shown schematically in
1000 Figure 2.138 as a plot of the equilibrium
carbon and oxygen contents at 1 and 0.2
Oxygen (ppm)
atmosphere pressure. In RH and other

processes with oxygen the carbon reacts
with oxygen and follows the indicated
reaction path. Theoretically one atom of
oxygen requires one atom of dissolved
RH-OB oxygen. However in actual processes
500 there are other sources of oxygen such as
leaks and unstable oxides in the slag (FeO
pCO=1
and MnO). In oxygen assisted processes
the reaction path is indicated as RH–OB
RH
(or VOD). Initially gaseous oxygen
pCO=0.2 removes carbon, the oxygen blow is ter-
minated, and dissolved oxygen reacts
0 with carbon.
0.05 0.10 0.15 0.20
Carbon (%)
Fig. 2.138 Carbon–oxygen equilibrium as a function of
CO pressure. From Ref. 184.
2.11.2 Vacuum Degassing Kinetics

In recirculating systems H2, N2 and CO are formed at two reaction sites: the rising argon bubbles
and the metal-vacuum interface above the melt. The model and plant data given by Bannenberg185
et al indicates that the Ar bubbles are the major reaction surface area for reactions at high Ar rates.
Recent work by Uljohn and Fruehan186 indicate at lower Ar flow rates both sites should be consid-
ered. As the argon bubbles burst at the surface they generate a large number of metal droplets and
surface area. In recirculating systems the reactions include the rising argon bubbles, the metal
droplets and homogenous or heterogeneous nucleated bubbles in the bath.
The hydrogen reaction is simply controlled by liquid phase mass transfer of hydrogen. Hydrogen
diffusivity is high and consequently the reaction is fast. For non recirculating systems the reactions
increase with argon stirring in a complex manner. Due to the expansion of the gas bubbles and the
dependence of bubble size on operating parameters, the rate equations are complex and the reader
is referred to Ref. 184 for details.
It should be noted that the reaction is limited not by H2 gas in the vessel but rather H2O. Even at
very low H2O pressures the equilibrium hydrogen content for reaction 2.11.4 is relatively high.
3H2O + 2Al = Al2O3 + 6H (2.11.4)
For example, for pH2O = 2 ´ 10–6 atm and an aluminum content of 0.03%, the equilibrium hydro-
gen content is approximately 1 ppm.
The nitrogen reaction is controlled by mass transfer of nitrogen and chemical kinetics in series and
the mixed control model given in Section 2.2.4 can be applied. Again the reactions are complex
and the details are given elsewhere. However, the rate depends on the argon flow rate and sulfur
content; sulfur is surface active and retards the chemical reaction as discussed in Section 2.2.3.
Examples of calculated rates of nitrogen are given in Figs. 2.139 and 2.140. Nitrogen can only be
removed effectively at low sulfur contents and high Ar bubbling rates.
80
70
Nitrogen content [ppm]
60
Stirring rate : 1.8 m3/min
50 Pressure: 1 mbar
40
30
20
10
0 5 10 15 20 25 30 35 40
Volume of argon stirring gas (m3)
Fig. 2.139 Decrease of nitrogen content during vacuum for a sulfur content of 10 ppm. From Ref. 185.
60
50
Nitrogen content (ppm)

40 Sulfur content (ppm)
200
100
30 40
10
20
10
0 5 10 15 20 25 30 35 40
Volume of argon stirring gas (m3)
Fig. 2.140 Influence of sulfur content on the denitrogenisation. From Ref. 185.
The rates in recirculating systems (RH) are complicated by the fact they also depend on the circu-
lation rate into the RH unit. For example the rate of decarburization in the ladle is given by
[ ]
d %C Q (2.11.5)
=- %C - %C RH
dt W
where
Q = metal recirculation rate,
W = the weight of steel,
%CRH = the carbon content in the RH.
The change of carbon in the RH is given by
[ ] [ ]
d %C RH Q
= - k %C RH - %C e + %C - %C RH (2.11.6)
dt WRH
where k is the decarburization rate constant, %Ce is the equilibrium carbon content and WRH is the
weight of steel in the RH. An approximate solution to equation 2.11.6 is given by
æ %C - %C e ö
ln ç ÷ = -K t (2.11.7)
è %C o - %C e ø
where
%Co = the initial carbon content,
K = the overall reaction rate constant given by
Q æ k WRH ö
K=
W çè k WRH + Q÷ø
(2.11.8)
Therefore the rate can be increased by increasing Q, WRH and k. For low circulation rates the rate
is controlled by the circulation rates. Consequently newer RH units have large snorkels and high
circulation rates.
The hydrogen reaction in RH units are given by similar expressions. The major difference is that
kRH for hydrogen is larger because of faster mass transfer of hydrogen. Nitrogen removal in RH
units is generally slow and generally less than 5 ppm is removed.
References
1. L. Onsager and R. M. Fuoss, J. Phys. Chem. 36 (1932): 2687.
2. A. Einstein, Ann. Phys (Leipzig) 17 (1905): 54; Z. Elektrochem (1908): 14,235.
3. E. T. Turkdogan, “Physicochemical Properties of Molten Slags and Glasses,” The Metals
Society (now The Institute of Materials), London, 1983.
4. J. Szekely, Fluid Flow Phenomena in Metals Processing (New York: Academic Press, 1979).
5. J. Szekely and N. J. Themelis, Rate Phenomena in Process Metallurgy (New York: Wiley,
1971).
6. G. H. Geiger and D. R. Poirier, Transport Phenomena in Metallurgy (Reading, MA: Addi-
son-Wesley, 1973).
7. J. Szekely, J. W. Evans, and J. K. Brimacombe, The Mathematical and Physical Modelling of
Primary Metals Processing Operations New York: Wiley, 1988).
8. J. Szekely, G. Carlsson, and L. Helle, Ladle Metallurgy (Berlin: Springer, 1989).
9. J. Szekely and O. J. Ilegbusi, The Physical and Mathematical Modelling of Tundish Opera-
tions (Berlin: Springer, 1989).
10. J. S. Sheasby, W. E. Boggs, and E. T. Turkdogan, Metal Science 18 (1984): 127.
11. C. Wagner, Z. Phys. Chem., Abt B 21 (1933), 25 (in “Atom Movements,” ASM (1950): 153.
12. L. Himmel, R. F. Mehl, and C. E. Birchenall, Trans. AIME 197 (1953): 827.
13. P. Hembree and J. B. Wagner, Trans. AIME 245, (1969): 1547.
14. P. L. Walker, F. Rusinko, and L. G. Austin, Advan. Catalysis 21 (1959): 33.
15. E. T. Turkdogan et al., Carbon 6 (1968): 467.
16. E. T. Turkdogan and J. V. Vinters, Carbon 7 (1969): 101; 8 (1970): 39.
17. E. T. Turkdogan, R. G. Olsson, and J. V. Vinters, Carbon 8 (1970): 545.
18. R. H. Tien and E. T. Turkdogan, Carbon 8 (1970): 607.
19. D. A. Aderibigbe and J. Szekely, Ironmaking and Steelmaking 9 (1982): 32.
20. W .M. McKewan, Trans. TMS-AIME 218 (1960): 2.
21. J. M. Quets, M. E. Wadsworth, and J. R. Lewis, Trans. TMS-AIME 218 (1960): 545: 221
(1961): 1186.
22. N. J. Themelis and W. H. Gauvin, A.E. Ch. E.J. 8, (1962): 437; Trans. TMS-AIME 227 (1963):
290.
23. N. A. Warner, Trans. TMS-AIME 230 (1964): 163.
24. R .H. Spitzer, F. S. Manning, and W. O. Philbrook, Trans. TMS-AIME 236 (1966): 726.
25. E. T. Turkdogan and J. V. Vinters, Metall. Trans. 2 (1971): 3175; 3 (1972): 1561.
26. P. H. Calderbank, Chem. Eng. (London) 212 (1967): CE 209.
27. E. T. Turkdogan, “Fundamentals of Steelmaking,” The Institute of Materials, London, 1996.
28. I. Leibson et al., A.I.Ch.E.J. 2, (1956): 296.
29. V. G. Levich, Physicochemical Hydrodynamics (Englewood Cliffs: Prentice-Hall, 1962).
30. S. Glasstone, K. J. Laidler, and H. Eyring, The Theory of Rate Processes (New York:
McGraw-Hill, 1941).
31. C. N. Hinshelwood, Kinetics of Chemical Change, (London: Oxford University Press, 1942).
32. T. Fuwa, S. Ban-ya, and T. Shinohara, Tetsu-to-Hagane 53 (1967): S328.
33. M. Inouye and T. Choh, Trans. Iron Steel Inst. Japan 8 (1968): 134.
34. K. Mori and K. Suzuki, Trans. Iron Steel Inst. Japan 10 (1970): 232.
35. R. J. Fruehan and L. J. Martonik, Metall. Trans. B 11B (1980): 615.
36. M. Byrne and G. R. Belton, Metall. Trans. B 14B (1983): 441.
37. P. Glaws and R. J. Fruehan, Metall Trans. B 16B (1985): 551.
38. S. Ban-ya et al., Tetus-to-Hagane 60 (1974): 1443.
39. G. Sain and G. R. Belton, Metall Trans. B 5B (1974): 1027–32; 7B (1976): 403–7.
40. F. Manmion and R. J Fruehan, Metall Trans. B, 20B (1989): 853.
41. T. Nagasaka and R. J. Fruehan, Metall Trans. B, 25B (1994): 245.
42. G. R. Belton, Metall Trans. B, 7B (1976): 35–42.
43. E. A. Gulbransen, K. F. Andrew, and F. A. Brassort, Carbon 2 (1965): 421.
44. J. D. Blackwood, Aust. J. Appl. Sci. 13 (1962): 199.
45. S. Ergun, J. Phys. Chem. 60 (1956): 480.

46. K. Hedden and A. Lowe, Carbon 5 (1967): 339.
47. H. J. Grabke, Ber. Bunsenges. Physik Chem. 70 (1966): 664.
48. D. A. Aderibigbe and J. Szekely, Ironmaking & Steelmaking 8 (1981): 11.
49. S. Story and R. J. Fruehan, To be submitted to Metall Trans., 1997
50. R. J. Fruehan, B. Lally, and P. C. Glaws, Trans. ISS, I&SM, (April 1987), 31.
51. E. T. Turkdogan and J. V. Vinters, Metall. Trans. 5 (1974): 11.
52. E. T. Turkdogan, Physical Chemistry of High Temperature Technology (New York: Academic
Press, 1980).
53. H. J. Donald, “An Annotated Bibliography,” Mellon Inst. Ind. Res. Pittsburgh, PA. 1956.
54. J. V. Kehrer and H. Leidheiser, J. Phys. Chem. 58, (1954): 550.
55. W .R. Walker, J. F. Rakszawski, and G. R. Imperial, J. Phys. Chem. 63 (1959): 133, 140.
56. W. R. Ruston et al., Carbon 7 (1969): 47.
57. R. G. Olsson and E. T. Turkdogan, Metall. Trans. 5 (1974): 21.
58. J. Macak, P. Knizek, and J. Malecha, Carbon 16 (1978): 111.
59. E. R. Gilliland and P. Harriott, Ind. Eng. Chem. 46 (1954): 2195.
60. T. H. Byrom, J. Iron Steel Inst. 92 (1915): 106.
61. H. C. H. Carpenter and C. C. Smith, J. Iron Steel Inst. 96 (1918): 139.
62. O. Stelling, J. Met. 10 (1958): 290.
63. H. Schenck and W. Maachlanka, Arch. Eisenhuttenwes.31 (1960): 271.
64. R. J. Fruehan and L. J. Martonik, Metall Trans. 4 (1973): 2129.
65. S. Chapman and T. G. Cowling, The Mathematical Theory of Non-uniform Gases, 3rd ed.
(London: Cambridge University Press, 1970).
66. J. O. Hirschfelder, C. F. Curtiss, and R. B. Bird, Molecular Theory of Gases and Liquids,
(New York: Wiley, 1954).
67. A. Eucken, Physik.Z. 14 (1913): 324.
68. M. Knudsen, Ann. Phys. (Leipzig) 28 (1909): 75.
69. C. H. Bosanquet, Br. TA Rep. BR–507, September 1944.
70. E. T. Turkdogan, R. G. Olsson, and J. V. Vinters, Metall. Trans. 2 (1971): 3189.
71. G. K. Sigworth and J. F. Elliott, Met. Sci. 8 (1974): 298.
72. E. T. Turkdogan and L. E. Leake, J. Iron and Steel Inst. 179 (1955): 39.
73. E. T. Turkdogan and R. A. Hancock, 179 (1955): 155; 183 (1956): 69.
74. J. Chipman et al., Trans. ASM. 44 (1952): 1215.
75. J. Chipman and T. P. Floridis, Acta Met 3 (1955): 456.
76. N. R. Griffing, W. D. Forgeng, and G. W. Healy, Trans. AIME 224 (1962): 148.
77. J. F. Elliott, M. Gleiser, and V. Ramakrishna, Thermochemistry for Steelmaking, vol. II
(Reading, MA: Addison-Wesley, 1963).
78. S. N. Singh and K. E. Blazek, J. Metals (AIME) 26(10) (1974): 17.
79. M. M. Wolf and W. Kurz, Metall.Transaction B 12B (1981) 85.
80. A. A. Howe, Applied Scientific Research 44 (1987): 51.
81. H. Yamada, T. Sakurai, and T. Takenouchi, Tetsu-to-Hagane 76 (1990): 438.
82. M. M. Wolf, in “1st European Conference on Continuous Casting”, p. 2.489. Florence, Italy,
Sept. 23–25, 1991.
83. C. R. Taylor and J. Chipman, Trans. AIME 154 (1943): 228.
84. P .A. Distin, S. G. Whiteway, and C. R. Masson, Can. Metall. Q. 10 (1971): 13.
85. A. E. Lord and N. A. Parlee, Trans. Met. Soc. AIME 218 (1960): 644.
86. D. L. Sponseller and R. A. Flinn, Trans. Met. Soc. AIME 230 (1964): 876.
87. P. K. Trojan and R. A. Flinn, Trans. ASM 54 (1961): 549.
88. P. J. Guichelaar et al., Metall. Trans. 2 (1971): 3305.
89. B. J. Keene, A survey of extant data for the surface tension of iron and its binary alloys, NPL
Rep. DMA(A) 67, June 1983.
90. E. T. Turkdogan, in Foundry Processes—Their Chemistry and Physics, ed. S. Katz and C. F.
Landefeld (New York: Plenum Press, 1988), 53.
91. P. Kozakevitch, in Surface Phenomena of Metal (London: Chem. Ind., 1968), 223.
92. E. Selcuk and D. H. Kirkwood, J. Iron and Steel Inst. 211 (1973): 134.
93. I. Jimbo and A. W. Cramb, Iron Steel Inst. Japan International 32 (1992): 26.
94. A. Kasama et al., Can. Met. Quart. 22 (1983): 9.
95. L. D. Lucas, Compt. Rend. 248 (1959): 2336.
96. R. N. Barfield and J. A. Kitchener, J. Iron and Steel Inst. 180 (1955): 324.
97. G. Tromel et al., Arch. Eisenhuttenwes. 40 (1969): 969.
98. C. R. Taylor and J. Chipman, Trans. AIME 159 (1943): 228.
99. M. Timucin and A. E. Morris, Metall. Trans 1 (1970): 3193.
100. S. H. Liu, A. Morales, and R. J. Fruehan, To be submitted to Metall Trans. B, 1997.
101. R. H. Rein and J. Chipman, Trans. Met. Soc. AIME 233 (1965): 415.
102. T. Fujisawa and H. Sakao, Tetsu-to-Hagane 63 (1977): 1494, 1504.
103. “Ethem T. Turkdogan Symposium”, pp. 253–269. The Iron and Steel Society of AIME, May
15–17, 1994.
104. Y. Iguchi, S. Ban-ya and T. Fuwa, Trans. Iron and Steel Inst. Japan 9 (1969): 189.
105. K. Ito and R. J. Fruehan, Metall. Trans. B 19B (1988): 419.
106. D. J. Min and R. J. Fruehan, Metall Trans. B 21B (1990): 1025.
107. H. Suito and R. Inoue, Tetsu-to-Hagane 73 (1987): S246.
108. K. Nomura, B. Ozturk, and R. J. Fruehan, Metall Trans. B 22B (1991): 783.
109. S. Ban-ya and J. D. Shim, Can. Metall. Q. 21 (1982): 319.
110. P. Kozakevitch, Rev. Metall. 46 (1949): 505, 572.
111. K. C. Mills and B. J. Keene, Inter. Met. Rev. 26 (1981): 21.
112. H. Gaye et al., Can. Metall. Q. 23 (1984): 179.
113. K. Ogino et al., Trans. Iron and Steel Inst. Japan 24 (1984): 522.
114. I. Jimbo and A. W. Cranb, The Sixth International Iron and Steel Congress, Naguya, ISIJ, 1990, 499.
115. D. R. Gaskell, A. McLean, and R. G. Ward, Trans. Faraday Soc. 65 (1969): 1498.
116. Y. Skiraishi et al., Trans. Japan Inst. Met. 19 (1978): 264.
117. Yu M. et al., Soob A Gruz. SSR 32(1) (1963): 117.
118. L. Segers, A. Fontana, and R. Winand, Electrochem. Acta. 23 (1978): 1275.
119. J. W. Tomlinson, M. S. R. Heynes, and J. O’M. Bockris, Trans. Faraday Soc. 54 (1958): 1822.
120. L. R. Barrett and A. G. Thomas, J. Glas Technol. (1959): 179s.
121. H. Winterhager, L. Greiner, and R. Kammel, Forschungsberichte des Landes (Nordrhein-
Westfalen, Nr. 1630: Westdeutscher Verlag, 1966).
122. E. V. Krinochkin, K. T. Kurochin, and P. V. Umrikhin, Fiz. Khim. Poverkh. (Kiev: Naukova
Dumba, Yavlenii Rasp., 1971) ,179–83.
123. J. S. Machin and T. B. Yee, J. Am. Cerm. Soc. 32 (1948): 200; 37 (1954) 177.
124. K. Ito and R. J. Fruehan, Metall Trans. B 20B (1989): 509.
125. K. Ito and R.J. Fruehan, Metall Trans. B 20B (1989): 515.
126. Y. Zhang and R. J. Fruehan, Metall Trans. B 26B (1995): 811.
127. T. Tokumetsu et al., Process Technology Conference Proceedings, Toronto, ISS, 1989.
128. Y. Zhang and R. J. Fruehan, Metall Trans. B. 26B (1995): 803.
129. S. Ban-ya, ISIJ International 33 (1993): 2–11.
130. H. Gaye and J. Lehmann, Molten Slags, Fluxes, and Salts Conference, Sydney, 1997, IS War-
rendale, PA, 27–34.
131. H. Gaye and J. Welfinger, Proc. Second International Symposium on Slags and Fluxes, 1984,
TMS Warrendale, PA 357–375.
132. M. Blander, A Pelton and G. Erikssen, 4th Internation Conference on Slags and Fluxes, 1992,
ISIJ, 56–60.
133. I. D. Sommerville, A. McLean and Y. D. Yang, Molten Slags, Fluxes, and Salts Conference,
Sydney, 1997, ISS Warrendale PA, 375–383.
134. D. J. Sosinsky and I. D. Sommerville, Metall Trans. B. 17B (1986): 331–7.
135. J. G. Peacey and W. G. Davenport, The Iron Blast Furnace (Oxford: Pergamon Press, 1979)>
136. A. Rist and G. Bonnivard, Rev. Met. 60 (1963): 23.
137. A. Rist and N. Meysson, Rev. Met. 61 (1964): 121.
138. N. Meysson, J. Weber, and A. Rist, Rev. Met. 61 (1964): 623
139. A. Rist and N. Meysson, Rev. Met. 62 (1965): 995.
140. N. Meysson, A. Maaref, and A. Rist, Rev. Met. 62 (1965): 1161.
141. F. D. Richardson and J. H. E. Jeffes, J. Iron and Steel Inst. 160 (1948): 261.
142. N. Tsuchiya, M. Tokuda, and M. Ohtani, Metall. Trans B7 (1976): 315.
143. E. T. Turkdogan, G. J. W. Kor, and R. J. Fruehan, Ironmaking and Steelmaking 7 (1980): 268.
144. B. Ozturk and R. J. Fruehan, Metall Trans. B 16B (1985): 121.
145. N. Standish and W.-K. Lu, eds., Alkalies in Blast Furnaces (Ontario: McMaster University
Press, 1973).
146. K. P. Abraham and L. J. Staffansson, Scand. J. Metall. 4 (1975): 193.
147. J. Davies, J. T. Moon and F. B. Traice, Ironmaking and Steelmaking 5 (1978): 151.
148. J. M. Steiler, “Etude Thermodynamique des Laitiers Liquides des Systems K2O–SiO2 et
K2O–CaO–SiO2–Al2O3,” IRSID, PCM–RE 646, July 1979.
149. L. L. Simmons, L. R. Lowden and T. C. Ehlert, J. Phys. Chem. 81 (1977): 709.
150. Z. Pociecha and J. Biczysko, in Seminar on Problems of Air and Water Pollution Arising in
the Iron and Steel Industry, Leningrad, Aug. 1971, CI.
151. E. T. Turkdogan and P .H. Josephic, Ironmaking and Steelmaking 11 (1984): 192.
152. K. L. Abraham, M. W. Davies, and F. D. Richardson, J. Iron and Steel Inst. 19 (1960): 82.
153. E. W. Filer and L. S. Darken, Trans. AIME 194 (1952): 253.
154. W. Oelsen, Arch. Eisenhuttenwes 35 (1964): 699, 713, 1039, 1115.
155. E. T. Turkdogan and R. A. Hancock, Trans. Inst. Min. Metall. 67 (1957–58): 573.
156. F. D. Delve, H. W. Meyer, and H. N. Lander, in Phys. Chem. Proc. Metallurgy, ed. G. R. St.
Pierre (New York: Interscience, 1961).
157. E. T. Turkdogan, Metall. Trans. B 9B (1978): 163.
158. K. Okabe et al., Kawasaki Steel Tech. Res. Lab. Rept, pp. 1–43, May 4, 1974 (English Trans-
lation BISI 13657); pp. 1–11 November 27, 1974 (English Translation 13658).
159. E. T. Turkdogan, Trans. Iron and Steel Inst. Japan 24 (1984): 591.
160. E. T. Turkdogan, J. Iron and Steel Inst. 182 (1956): 74.
161. A. Morales and R. J. Fruehan, Metall Trans. B, To be published 1997.
162. J. Chipman et al., Trans. AIME 188 (1950): 341.
163. H. Suito and R. Inoue, Trans. ISIJ 24 (1984): 301.
164. P. Bremer, Stahl u. Eisen 71 (1951): 575.
165. E. Aukrust, P. J. Koros, and H. W. Meyer, J. Met. 18 (1966): 433.
166. R. V. Pathy and R. G. Ward, J. Iron and Steel Inst. 202 (1964): 995.
167. R. J. Fruehan, in Advanced Physical Chemistry for Process Metallurgy, ed. N. Sano, wKLW
and P. Ribaud (London: Academic Press, 1977).
168. R. J. Fruehan, Ironmaking and Steelmaking 1 (1976): 33.
169. D. Goldstien and R. J. Fruehan, Submitted to Metall Trans. B, 1997.
170. R. J. Fruehan and D. Goldstien, 5th European Electric Steel Congress, 1995, Paris, Revue de
Metallurgic Paris.
171. D. Goldstien and R. F. Fruehan, Submitted to Metall Trans. B, 1997.
172. R. J Fruehan, Ironmaking and Steelmaking 3 (1976): 158.
173. P. C. Glaws and R. J. Fruehan, Metall Trans. B 17B (1986): 317.
174. F. Korber and W. Oelsen, Mitt. Kaiser-Wilhelm Inst. Eisenforsch 17 (1935): 231.
175. S. Matoba, K. Gunji and T. Kuwana, Tetsu-to-Hagane 45 (1959): 229.
176. J. Chipman and T. C. Pillay, Trans. Met. Soc. AIME 221 (1961): 1277
177. B. K. D. P. Rao and D. R. Gaskell, Metall. Trans. B. 12B (1981): 311.
178. R. J. Fruehan, “Ladle Metallurgy Principles and Practices, ISS Warrendale PA, 1985.
179. D. Janke and W. A. Fischer, Arch. Eisenhuttenwes, 47 (1976): 195.
180. N. Bannenberg and B. Bergmann, Stahl u. Eisen, 112(2) (1992): 57.
181. B. Ozturk and E. T. Turkdogan, Metal Science, 18 (1984): 306.
182. E. T. Turkdogan, J. Iron and Steel Inst. 210 (1972): 21.
183. K. Larsen and R. J. Fruehan, Trans. ISS, I&SM, July 1990, 45.
184. R. J. Fruehan, “Vacuum Degassing of Steel”, ISS Warrendale PA, 1990.
185. N. Bannenberg, B. Bergmann, H. Wagner and H. Gaye, Proceeding of 6th International Iron
and Steel Congress Nagoya, 1990, ISIJ, 3, 603.
186. H. Uljohn and R. J. Fruehan, To be submitted into Trans ISS, 1997.
187. E. T. Turkdogan, Arc. Eisenhuttenwes., 54 (1983) :1.

Chapter 2 PDF

Uploaded by

Copyright:

Available Formats

Chapter 2 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chapter 2 PDF

Uploaded by

Copyright:

Available Formats

Chapter 2

Fundamentals of Iron and

1.01325 ´ 105 ´ 22.414 ´ 10 ±3

2.1.2 Thermodynamic Laws

Above 298 K, the temperature dependence of Cp is represented by:

For an isobaric and isothermal reaction,

ò [SC p ( products) - SC p æè reactantsöø ] dT

2.1.2.2 The Second Law

2.1.2.3 The Third Law

The entropy of reaction is

2.1.2.4 Gibbs Free Energy

2.1.2.4.1 Generalization of Entropy Change of Reaction

2.1.3 Thermodynamic Activity

In terms of the relative partial molar enthalpy and entropy of solution

DG° = DH° – DS°T

{K} = (K) –84.5 82.0 0.5 63–759b

<Mn> = {Mn} –14.6 9.6 1 1244m

Table 2.1 (continued)

{Na} = (Na) –101.3 87.9 1 98–883b

1/2(S ) + (O2) = (SO2) 361.7 72.7 0.5 25–1700

<Si> = {Si} –49.3 30.0 2 1412m

<Ti> = {Ti} –18.6 9.6 – 1660m

<V> + {V} –22.8 10.4 – 1920m

{Zn} = (Zn) –118.1 100.2 1 420–907b

<Zr> = {Zr} –20.9 9.8 – 1850m

2.1.3.2 Activity Coefficient

2.1.3.2.1 Henry’s Law for Dilute Solutions In infinitely 0.8

2.1.3.3 Conversion from One Standard State to Another

2.1.4 Reaction Equilibrium Constant

2.2 Rate Phenomena

2.2.1.2 Chemical and Self-Diffusivities

2.2.2 Mass Transfer

2.2.2.1 Parabolic Rate of Oxidation of Iron

Hembree & Wagner13

0 0.05 0.10 0.15

Fig. 2.4 Self-diffusivity of iron as a function of (1 – NFe/NO) fraction of iron deficiency.

2.2.2.2 Oxidation of Carbon in CO2–CO Mixtures

2.2.2.3 Reduction of Iron Oxides

pellet surface and particularly the resistance to dif-

Fig. 2.10 Initial rate in hydrogen 600°C

ticle radius at indicated temper-

1000°C 900°C 800°C 95

2.2.2.4 Mass Transfer with Gas Bubbles in Oxygen Steelmaking

hand, the surface active solutes decrease the bub-

[ppm X ] d ( ppm X ) 22.414 ´ 10 –3

2.2.3 Chemical Kinetics

classical text books by Glasstone et al.30 and by

In terms of a single rate constant, the net reaction rate is formulated as

where Fo is a pre-exponential constant and

reaction at a given site fillage q.

positive, the enthalpy of activation for the DH*

2.2.3.1 Examples of Experiments on Rates of Interfacial Reactions

With this substitution the isothermal rate equation is

d[%N ] 100 A ì [%N]2 ü

Fig. 2.19 Rates of dissociation

Rate, fractional mass loss per minute

1/T x 103 (K–1)

2.2.4 Mixed Control

R = k (pis – pei ) (2.2.72)