Kho Swinbourne10 Mpm188 EAF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230816525

Understanding stainless steelmaking through computational


thermodynamics Part 1: electric arc furnace melting

Article  in  Mineral Processing and Extractive Metallurgy IMM Transactions section C · March 2010
DOI: 10.1179/174328509X431454

CITATIONS READS

14 1,481

5 authors, including:

Bart Blanpain Sander Arnout


KU Leuven InsPyro
420 PUBLICATIONS   7,702 CITATIONS    44 PUBLICATIONS   328 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Geopolymerisation: a generic solution for the valorisation of high-temperature residues View project

Bauxite mineralogy View project

All content following this page was uploaded by Sander Arnout on 30 May 2014.

The user has requested enhancement of the downloaded file.


Understanding stainless steelmaking through
computational thermodynamics
Part 1: electric arc furnace melting
T. S. Kho1, D. R. Swinbourne*1, B. Blanpain2, S. Arnout2 and D. Langberg1
Stainless steel alloys are widely used in many important applications but their production presents
difficulties because they contain expensive chromium, which can be extensively oxidised during
decarburisation to the very low carbon levels required. Modern stainless steelmaking largely
avoids this problem by having two distinct stages and is therefore described as duplex practice.
Molten high carbon stainless steel is produced in an electric arc furnace and then the melt is
decarburised in an argon–oxygen converter or a vacuum oxygen decarburising converter. In this
work, computational thermodynamics has been used to examine the major reactions occurring in
the electric arc furnace and to show the effect of various process variables on chromium recovery.
It was shown that significant oxidation of the scrap must occur during melting, and that
subsequent carbon/oxygen injection initially oxidises some chromium, but then mostly oxidises
the added carbon. Chromium was predicted to exist in the slag as CrO and CrO1?5 in almost equal
proportions. Increasing the temperature should improve chromium recovery but results in less
benefit than expected due to the decreasing activity coefficients of CrO and CrO1?5 and the
increasing oxygen partial pressure. Ferrosilicon additions reduce chromium oxides from the slag,
but much of the silicon simply dissolves into the steel. Computational thermodynamics is seen to
be a very effective educational tool for gaining an understanding of smelting processes.
Keywords: Electric arc furnace, Stainless steel, Thermodynamics, Computational model

Introduction necessary to ensure that grain boundary corrosion will


not take place.3
Introduction to stainless steels Carbon is removed from steels by oxidation, carbon
Stainless steels are iron based alloys containing a being much more readily oxidised than iron, as can be
minimum of 10 wt-%Cr and varying amounts of other seen on the Ellingham diagram given as Fig. 1. The
alloying elements. Table 1 shows the composition of vertical axis will be DGu, the standard Gibbs free
several widely used grades of stainless steels. The energy change, for all lines except those for a reduced
minimum chromium content is determined by the need carbon activity, for which it will be the general Gibbs
for a passive film of chromium oxide to form on the free energy change DG. Chromium is clearly much
surface, which prevents aqueous corrosion. Corrosion more readily oxidised than iron. At y1200uC the
and staining resistance, low maintenance and familiar oxidation of carbon and chromium at unit activity are
lustre make stainless steels ideal materials for a host of equally likely, but as the temperature is increased
commercial applications including cookware, cutlery, carbon becomes increasingly favoured. As carbon is
hardware, surgical instruments, industrial equipment oxidised to very low levels its activity drops rapidly and
and structural applications in the automotive and the C/O2/CO line on the Ellingham diagram rotates
aerospace industries. Importantly, stainless steel is upwards, which therefore makes chromium oxidation
recyclable, any stainless steel object typically containing increasingly favourable. The main issue with producing
60% of recycled material – 25% from end-of-life stainless steel is the restriction imposed by the carbon–
consumer products and 35% being scrap from manu- chromium equilibrium,4 i.e. the need to oxidise carbon
facturing processes.1 Very low carbon contents are without also oxidising much chromium. Failure to
develop a strategy to achieve this would lead to
1
excessive loss of chromium and necessitate the addi-
School of Civil, Environmental and Chemical Engineering, RMIT
University, 124 Latrobe Street, Melbourne 3000, Australia
tion of expensive low carbon ferrochromium to achi-
2
Departement Metaalkunde en Toegepaste Materiaalkunde, Katholieke eve the required grade of stainless steel, without
Universiteit Leuven, Kasteelpark Arenberg 44, B 3001 Leuven, Belgium simultaneously increasing the carbon content beyond
*Corresponding author, email [email protected] specifications.

ß 2010 Institute of Materials, Minerals and Mining and The AusIMM


Published by Maney on behalf of the Institute and The AusIMM
Received 18 March 2009; accepted 16 August 2009 Mineral Processing and Extractive
DOI 10.1179/174328509X431454 Metallurgy (Trans. Inst. Min Metall. C) 2010 VOL 119 NO 1 1
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

2 Electric arc furnace8

1 Ellingham diagram for relevant iron, chromium and car- et al.6 and was intended to understand the chemistry of
bon oxidation reactions: range of steelmaking tempera- the process as well as to be used as an aid to control
tures is shaded stainless steelmaking operations. This work has an
educational focus and will utilise the HSC Chemistry
Stainless steelmaking technology for Windows software package as a tool for gaining an
understanding of the chemical changes taking place
In the early days of stainless steel production carbon within the EAF. A subsequent paper will take the same
steel scrap, iron ore and lime (CaO) were charged into approach to examine decarburisation of molten stainless
an electric arc furnace (EAF). When the charge was steel in AOD and VOD converters.
molten, ferrosilicon, lime and fluorspar (CaF2) were The duty of the EAF is to efficiently and economically
added and the temperature of the bath was increased so deliver molten metal of the required composition and
that a large amount of low carbon ferrochromium could temperature to a converter. The EAF is the most widely
be added.5 With the advent of tonnage oxygen, oxygen employed melter due to its ability to process a wide
injection into the molten steel using a lance greatly variety of raw materials.5 An EAF, as shown schema-
improved the rate of decarburisation. However, the tically in Fig. 2, consists of a refractory lined hearth,
more oxidising conditions had the expected adverse vertical cylindrical sidewalls, and a removable roof.7
effect of extensive oxidation of chromium to the slag. A Graphite electrodes pass through holes in the roof and
reduction period using ferrosilicon to reduce the are clamped to arms which move vertically on masts
oxidised chromium from the slag became necessary mounted to the furnace assembly. This allows the
before low carbon ferrochromium could be added. electrodes and roof to be raised and swung to one side
Modern stainless steelmaking practice does not to allow furnace charging. Conventional furnaces have a
attempt to make stainless steel in an EAF alone. The horizontal tapping spout where the entire furnace can be
EAF is used to provide a molten steel charge for tilted for tapping, while some prefer eccentric bottom
converters, without any significant decarburisation of tapping furnace because it allows slag free tapping and
the melt and with minimum chromium oxidation to the decreased steel temperature losses during tapping.8
slag. The melt is then decarburised in a variety of Power is supplied from a three-phase, multivoltage tap
converters, with argon oxygen decarburisation (AOD) transformer to three carbon electrodes, though some
and vacuum oxygen decarburisation (VOD) being the furnaces use dc power and have only one or two
most widely used. As a result, these practices are electrodes.
classified as ‘duplex’ in that there are two stages in the Many modern EAFs are equipped with oxy fuel
production of stainless steel. Sometimes further refining burners and lances for injecting oxygen, lime and carbon
takes place under vacuum to produce stainless steels to increase the energy input into the furnace and reduce
which have very low carbon and nitrogen requirements melting cycle times. In such furnaces y60% of the
and this is known as ‘triplex’ practice.5 energy input is electrical with the remainder being
In this work, only the chemical reactions taking place chemical energy from oxy fuel combustion, oxidation of
during EAF melting will be discussed and modelled silicon and carbon and other chemical reactions.7
using computational thermodynamics. A mathematical Carbon and oxygen injection also generates a large
model designed to predict the compositions of slag and volume of carbon monoxide gas and so can cause the
steel as a function of time has been described by Arnout slag to foam around the electrodes, as will be discussed
later.
Table 1 Composition ranges for typical grades of A typical heat commences with the charging of a total
stainless steels,2 wt-% of y10 t carbon steel scrap, 90 t stainless steel scrap and
Grade Classification C Mn Cr Ni Mo 10 t high carbon ferrochromium (50 Cr, 7 C, 4 Si,
balance Fe, wt-%) through the open roof of the furnace.
304 Austenitic 0.05 1.5 18.5 9 … The roof is then swung into place, the electrodes are
310 Austenitic 0.08 1.5 25 20 … lowered and current flow is initiated so that the
316 Austenitic 0.05 1 17 11 2 electrodes bore through the scrap to form a pool of
410 Martensitic 0.1 0.5 12 … …
liquid metal. About 5 t fluxes containing mostly lime
430 Ferritic 0.03 0.4 16.5 … …
(CaO) and some magnesia (MgO) are added soon after.

2 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

When the scrap is fully melted, carbon in the form of


coke (y300 kg) is added to the steel bath together with
y250 Nm3 of oxygen which is injected through a lance
over a period of y10 min.
After the oxygen blow, y900 kg of 75 wt-%Si
ferrosilicon is added to reduce chromium oxides from
the slag according to the reaction
3Siz4CrO1:5 ~4Crz3SiO2
DG 0 at 16000 C~{422 kJ
DH 0 at 16000 C~{588 kJ (1)
In order to improve slag/metal contact, argon or
nitrogen is often injected through a bottom plug. Guo
et al.9 reported that this bottom stirring is largely
ineffective in promoting the reduction of chromium
oxides by silicon. Most reduction was found to take
place in a spout-tapped EAF during the tapping of slag
and molten steel into the ladle. Strong mixing of slag
and metal is achieved in a spout-tapped furnace because
slag is tapped first, then steel pours onto the slag.
Reduction was minimal in a bottom tapped furnace
where the steel is tapped first, followed by the slag, and 3 Equilibrium carbon and chromium contents as function
mixing is therefore much less vigorous. of temperature
The silica content of the slag increases during
reduction and the slag becomes more viscous, so before Dennis and Richardson11 argued that CrO1?5 had such a
tapping y100 kg fluorspar (CaF2) and 150 kg borax very low solubility in slag that the slag would be
(Na2O.2B2O3) is usually added to reduce the viscosity. saturated with CrO1?5, so its activity could be taken as
The tap-to-tap time is y1 h and produces y100 t unity. The partial pressure of CO could also be taken to
molten metal and 15 t slag. be 1 atm. Taking the mole fractions of carbon and
chromium in steel to be proportional to their weight per
The carbon–chromium equilibrium cent concentrations
An understanding of the competitive oxidation equili- K’~(cC wt-%C)3 =(cCr wt-%Cr)2 (4)
brium of carbon and chromium is the key to stainless
steelmaking. While carbon simply oxidises to carbon The chromium contents of stainless steel in equilibrium
monoxide gas, there is some dispute regarding the form of with various carbon contents, as a function of tempera-
chromium oxide in the slag. Hilty10 suggested that the ture, can then be graphed as in Fig. 3.
stable oxide of chromium in the decarburisation reaction Dennis and Richardson11 did not state the values they
above 1711uC is CrO. Dennis and Richardson11 stated used for the activity coefficients, but their results are
that it had been established that Cr2O3 is the stable oxide consistent with the assumptions that chromium forms an
of chromium below 1400uC but did not discuss which ideal solution and that the activity coefficient of carbon
oxide was stable at higher temperatures. Similarly, is y0?5. The isotherms in Fig. 3 show that as the carbon
Fruehan12 assumed that chromium oxidizes to Cr2O3 content decreases, so does the chromium content. At a
and did not mention CrO as a possible species, although temperature of 1600uC, an alloy containing 18 wt-%
in a subsequent publication Kerr and Fruehan13 stated chromium has an equilibrium carbon content of
that CrO was the predominant form of chromium oxide y0?42 wt-%. Decarburising below this carbon content
in slag. Voj et al.14 took both oxidation states of leads to a loss of chromium to the slag. If the melt were
chromium into consideration by using the oxidation decarburised to ,0?08 wt-% carbon as required by most
reactions for both CrO and Cr2O3 in their analysis, while stainless steel grades, the chromium content would be
Sun et al.15 considered the carbon–chromium equilibria reduced to ,5 wt-% which is well below that required. It
assuming that Cr3O4 was the stable oxide, i.e. that there can also be seen that increasing the temperature allows
were equal numbers of moles of CrO and Cr2O3 formed. lower carbon contents at any given chromium content.
Assuming for simplicity that the carbon–chromium For the 18 wt-% chromium alloy, the equilibrium
equilibrium can be expressed satisfactorily as follows, as carbon content can be decreased to 0?1 wt-% if the
was adopted by several authors8,11 temperature is raised to 1800uC. While higher tempera-
tures permit greater extents of decarburisation there are
2Crz3CO(g)~2CrO1:5 z3C a number disadvantages, including higher energy costs
DG 0 at 16000 C~173 kJ and higher refractory consumption.
In practice EAF slags are not necessarily saturated
DH 0 at 16000 C~{77 kJ (2) with chromium oxides so the values of the activity
The equilibrium expression can be expressed as coefficients of CrO and CrO1?5 in slag are significant,
and depend on the composition of the slag. Maximising
(aCrO1:5 )2 (ac )3 the values of these activity coefficients is desired because
K~ (3) this minimises the concentration of chromium oxides in
(aCr )2 (pCO )3

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1 3
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

further, but then decreases until at the end of injection it


is only a little lower than it was prior to injection.
Following ferrosilicon addition the CrOx content of the
slag drops rapidly, as described earlier.
Making slags which foam and surround the electrodes
in the EAF is standard practice in carbon steelmaking.
Foam shields the refractories from the arcs so longer,
more powerful arcs can be used and productivity is
therefore increased. Slag foaming in stainless steel
practice is utilised, but is difficult to achieve, partly
because of the presence of solid particles in the slag but
especially because the rate of generation of carbon
monoxide from carbon reduction of chromium oxides in
the slag is much too low to sustain a foam.13 However,
Gőrnerup and Jacobsson20 claim that this is not true in
practice and that significant carbon reduction of
chromium oxides does take place.
The carbon monoxide partial pressure pCO is also a
very significant variable affecting the carbon–chromium
equilibrium, being a cubic term in equation (3). The
lower pCO the greater the equilibrium chromium content
of the steel at any given carbon content. However, in the
4 Liquid region at 1600uC in CaO–SiO2–MgO–5 wt-%Al2O3 EAF pCO cannot be manipulated because the furnace is
system16 open to the atmosphere. Reducing the carbon monoxide
partial pressure is technically possible in converters and
slag and therefore minimises chromium losses. Stainless this is the basis of modern stainless steel production. The
steelmaking slags contain mostly CaO, SiO2, MgO and pCO can be reduced by either dilution of the input
Al2O3 and the relevant phase diagram at 5 wt-% Al2O3 oxygen with argon as in the AOD converter, or by
is given as Fig. 4.16 reducing the total pressure in the system as in the VOD
Silica (SiO2) is formed by the oxidation of silicon in converter.
the scrap steel charge, lime (CaO) is needed to flux the In summary, the processing conditions favourable for
silica and produce a fluid and basic slag and magnesia minimising chromium losses during decarburisation are:
(MgO) is required to reduce the rate of attack of the slag (i) operating at as high a temperature as possible
on the chrome-magnesite furnace refractories.17 The (ii) choosing a slag with the maximum possible
most important chemical property of the slag is its basicity
basicity. Pretorius and Muan18 and Pei and Wijk19 (iii) minimising the slag volume
found that the activity coefficients of CrO and CrO1?5 in (iv) reducing the input partial pressure of oxygen to
the slag increase with increasing basicity, although both reduce pCO
defined basicity a little differently. Pretorius and Muan (v) lowering the oxygen partial pressure in the slag
defined basicity as wt-%CaO/wt-%SiO2 while Pei and and steel with ferrosilicon additions.
Wijk defined it as (NCaO zNMgO )=NSiO2 where Ni is the Higher temperatures involve more severe attack on
mole fraction of ‘i’ in the slag. Gőrnerup and furnace refractories, while operating with a reduced
Jacobsson20 also confirmed the relationship between oxygen partial pressure increases the time required to
slag basicity and chromium recovery to the steel using decarburise the steel to the necessary low carbon
industrial data. Figure 4 shows that at 1600uC the contents. If replacement of oxidised chromium is
maximum basicity of the slag is obtained at the point required, expensive low carbon ferrochrome must be
where the slag is double saturated with Ca2SiO4 and employed. The evolution of stainless steelmaking
MgO. In practice, the slag may also be saturated with technology has been guided by the need to overcome
MgO.Cr2O3. Guo et al.9 examined quenched slag or minimise these process constraints.
samples and mainly observed idiomorphic crystals of
MgO.Cr2O3, although they stated that crystals of HSC simulation of the EAF
(Ca,Mg)2SiO4 were sometimes also seen. The presence The ‘equilibrium module’ in the HSC Chemistry for
of MgO.(Cr,Al)2O3 spinel in slag was confirmed by Guo Windows v.5?1 software package21 is a convenient way
et al.9 early during a heat but in some heats it had to determine the equilibrium state of a multicomponent
disappeared in the final slag. multiphase system. It has been widely used to examine a
Guo et al.9 reported on the evolution of the chromium variety of complex metallurgical processes.22–25 The
oxide content of the EAF slag throughout the melting software requires as inputs the specification of all likely
process. They showed that at ‘flat bath’, where the steel phases in the system; the species which are in, or likely to
has charge has fully melted but before oxygen injection be created in, each phase; the activity coefficients of all
commences, the slag already contains a large concentra- species in each phase; the masses of all input species, and
tion of CrOx. This clearly shows that a significant the temperature and pressure of the system. The
amount of oxidation of the charge takes place during software uses a Gibbs free energy minimisation routine
melting as a result of air ingress into the furnace and to output the quantities of all species in all phases at
possibly the presence of some rust on steel scrap. During equilibrium. It cannot take into account rates of
oxygen injection the CrOx content of the slag rises reactions or heat and mass transfer effects, but does

4 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

provide valuable information on the effects of process infinite dilution can be calculated from the Gibbs free
variables on the amounts and compositions of molten energy conversion from the pure substance to the wt-%
phases as a function of temperature or the amount of standard states by the following equation
one or more of the inputs.  : 
One problem associated with modelling stainless 0 5585c0i
DGr?wt ~RT ln (8)
steelmaking is that while the software thermodynamic Mi
database is very large, it does not contain data for the where Mi is the atomic weight of species i in g mol21.
species CrO(l). However it is possible to add the required The Gibbs free energies for converting from the pure
data from information in the literature. Toker et al.26 substance to the wt-% standard states were taken from
gave the Gibbs standard free energy change for the the literature.32 All activity coefficients used for species
formation of CrO(l) as follows in the slag and molten steel are given in Table 2. The
3 activity coefficients are given to only two significant
Cr(s)z O2 (g)~CrO(l) figures because they were calculated for representative
2
0 compositions of both phases, so greater precision cannot
DG ~{334218z63:18T J at 1938 to 2023K (5) be justified. Slag was assumed to be 50 CaO–25 SiO2–
Data in the form given in equation (5) is easily entered by 7 CrOx–7 AlO1?5–5 MgO–5 MnO–1 FeO. Steel was assumed
selecting ‘Database’ then ‘Convert data to HSC format’ and to be Fe–18 Cr–8 Ni–1 Mn–0?8 C–0?3 Si (wt-%). It can
following the prompts. The data entry for CrO(l) in the be seen that most of the activity coefficients are not very
standard HSC format has the following parameters: H5 sensitive to temperature, with the notable exceptions
2334?218 kJ mol21, S562?400 J mol21 K21, A532?430, being CrO(l) and CrO1?5(s) in the slag and silicon in the
B520?958, C520?381, D522?623. molten steel.
The most difficult issue in modelling is the selection of During the EAF cycle there are two periods during
appropriate values for the activity coefficients of all which chemical changes take place in the melt. The first
species in all phases. Activity coefficients are a function period is when oxygen is blown into the molten steel,
of the concentration of the species in a solution phase, and the second is when ferrosilicon is added to return
but are acceptably close to constant if the species are chromium to the molten steel. These two periods were
expressed in the monocation form.27 The gas phase can modelled separately.
be considered to be an ideal solution at high tempera-
tures so the activity coefficients of all gas species are Oxygen blowing
unity. The activity coefficients of species in the slag The masses of input species were calculated on a per
phase were estimated using the regular solution model,28 tonne of scrap charged basis from the mass balance data
in which the activity coefficients are given by the for a typical heat from ArcelorMittal Stainless Genk in
following equation Belgium. Elements present in small amounts, such as
X molybdenum, were omitted. Fluorspar and borax fluxes
RT ln ci ~ aij Xj2 were also inconsequential to the modelling and so were
XX  omitted. The phases and species considered, together
z aij zaik {ajk Xj Xk zDGconv (6) with their input masses and activity coefficients, are
where ci is the activity coefficient of species i, aij is the listed in Table 3. The temperature was taken to be
interaction coefficient between species i and j, Xi is the 1600uC throughout this period. Carbon and oxygen
cation fraction of species i and DGconv is the Gibbs free were considered to be injected simultaneously, with the
energy for conversion from the hypothetical regular oxygen being raised incrementally by 0?035 Nm3 steps
solution standard state to the pure solid or pure liquid and the carbon by 0?05 kg, for 80 steps. Oxidation of
standard states. The interaction energies and conversion scrap by air ingress during melting was simulated by
Gibbs free energies were taken from the literature.28,29 initially adding 3 kg/tonne (2?1 Nm3) of oxygen, this
Interaction energies for CrO–MgO and CrO1?5–MgO small amount being found to satisfactorily reproduce the
were estimated by multiple regression of equation (6) CrOx evolution in the slag observed by Guo et al.9
with the data of Pei and Wijk.19 Activity coefficients during oxygen injection.
were not calculated for NiO because it is expected that
those of FeO could be used for NiO, nor for AlO1?5 Table 2 Calculated activity coefficients for species in
because in the modelling it was not considered to slag and molten steel
partition to any other phase.
Temperature,
Activity coefficients in the molten iron phase were
uC 1550 1600 1650 1700 1750 1800
estimated using the dilute solution model described by
Sigworth and Elliott,30 using the tabulation of Lupis31 CrO (l) 12 10 8.6 7.3 6.3 5.5
for the interaction coefficients. The activity coefficients CrO1?5 (s) 27 22 18 15 12 10
are given by the following equation FeO (l) 1.8 1.8 1.8 1.8 1.8 1.9
X j MnO 0.21 0.24 0.27 0.31 0.34 0.38
ln ci ~ ln c0i z eX (7) CaO (s) 0.048 0.048 0.048 0.049 0.049 0.049
j i j MgO (s) 0.074 0.075 0.076 0.078 0.079 0.080
SiO2 (s) 0.14 0.14 0.15 0.15 0.15 0.15
where ci is the Raoultian activity coefficient, c0i is the Cr (l) 0.83 0.83 0.83 0.83 0.83 0.83
Raoultian activity coefficient at infinite dilution, eji is the Ni (l) 0.72 0.74 0.77 0.79 0.82 0.85
binary first order interaction coefficient between species i Mn (l) 1.19 1.18 1.17 1.17 1.16 1.15
and j, and Xj is the mole fraction of species j. The Si (l) 0.0017 0.0021 0.0026 0.0032 0.0039 0.0048
C (g) 0.37 0.36 0.35 0.33 0.32 0.31
temperature variation of the activity coefficient at

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1 5
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

6 Masses of chromium oxides in slag as function of tem-


5 Equilibrium amount of carbon, manganese and silicon perature for 3 kg of injected oxygen
in steel, chromium oxides in slag and CO in gas during
simultaneous oxygen and carbon injection, at 1600uC
considered, i.e. CrO1?5, MgCr2O4 and Ca2SiO4, were
and with 3 kg/tonne of ingress oxygen
predicted by the HSC model to form over the range of
oxygen addition examined. This is contrary to the
The evolution of the masses of chromium oxides in findings of Arnout et al.6 who predicted the formation of
slag, carbon, manganese and silicon in the steel melt and significant amounts of calcium aluminium silicates and
carbon monoxide in the gas are given in Fig. 5. MgCr2O4.
Initially manganese, silicon and chromium in the steel The HSC module ‘heat and material balances’ was
melt are being oxidised and injected carbon simply used to determine the temperature rise caused by the
dissolves in the steel. Both chromium oxide species are exothermic oxidation reactions occurring during the
formed in approximately equal amounts, although CrO carbon/oxygen injection period. The temperature of the
is predominant. This supports the simplification of Sun slag and steel prior to injection was assumed to be
et al.15 that the carbon–chromium equilibria can be 1600uC while the carbon and oxygen being injected were
considered to produce Cr3O4, i.e. CrO.Cr2O3. No assumed to be at 25uC. The adiabatic temperature rise
carbon is oxidised until about 0?75 kg/tonne of injected was calculated to be 38uC, a significant increase arising
oxygen (3?75 kg of total oxygen). Once carbon oxidation from chemical reactions rather than from more expen-
commences only very minor reduction of chromium sive electrical energy.
oxides from the slag takes place. Typically, 2?5 to 3 kg/ Increasing the furnace temperature is expected to
tonne of oxygen are injected in industrial practice so greatly reduce the loss of chromium to the slag, as
significant formation of CO (g) is to be expected. These indicated by Fig. 3, so this was modelled by constructing
observations support some of the findings from the an HSC input file in which the temperature varied from
modelling performed by Arnout et al.6 i.e. that carbon 1550 to 1800uC and the injected oxygen input was held
injection raises the carbon content of the steel, leads to at 3 kg/tonne. One difficulty with this modelling was
the formation of CO (g) and reduces some chromium that the activity coefficients of the chromium oxides are
from the slag. None of the potentially saturating species strong functions of temperature, as seen in Table 2. In
one case the arithmetic mean values of the activity
coefficients over the temperature range were used, i.e.
Table 3 Input data for the HSC file used to simulate 8?8 for CrO (l) and 19 for CrO1?5 (s). In the second, the
oxygen blowing at 1600uC activity coefficients were given their true values from
Table 2, this requiring six separate input files to be
Species Amount, kg Species Amount, kg
executed. The results are seen in Fig. 6.
Gas Metal It can be seen that the effect of temperature on the
O2 (g) 2.1 Nm3 Fe 718 loss of chromium to the slag is actually quite weak,
CO (g) Cr 190 especially when the activity coefficients of CrO(l) and
CO2 (g) Ni 72 CrO1?5 are allowed to vary with temperature. The
Mn (g) Mn 12
MnO (g) C 8
simplified case using fixed values of the activity
Si coefficient falsely exaggerates the effect of temperature
on chromium loss. The other reason why the chromium
Slag Pure loss does not decrease as quickly as expected is that the
CaO 30 CrO1?5 curves in Fig. 3 were calculated on the assumption that
SiO2 20 MgCr2O4
the carbon monoxide partial pressure remained fixed at
MgO 15 Ca2SiO4
AlO1?5 5 1 atm. HSC can calculate the oxygen partial pressure in
CrO (l) the system and the values were found to be
CrO1?5 1?79610212 atm at 1550uC and 4?08610211 atm at
FeO 1800uC, i.e. the oxidising potential increased by over an
MnO order of magnitude, leading to further stabilisation of
NiO
the chromium oxides.

6 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

chromium reduction from the slag at high silicon addi-


tions. Gőrnerup and Jacobsson20 state that there is no
significant benefit in having .0?4 wt-%Si in steel, which
in this example corresponds to an addition of 6 kg/tonne
of silicon, while Guo et al.9 recommend a maximum of
0?3 wt-%Si in steel. Equation (1) also shows that
chromium losses to slag will be reduced by reducing
the activity of silica in slag and Guo et al.9 comment that
this is another reason why the basicity of the slag must
be maximised during stainless steel melting.
The silicon reduction of chromium oxides is exother-
mic, as shown in equation (1), but a heat balance
revealed that the temperature of the steel was unchanged
after the reduction process. The heat liberated by the
chemical reaction is almost exactly balanced by that
required to heat the ferrosilicon addition from 25uC to
7 Masses of chromium oxides in slag, and manganese the steel melt temperature.
and silicon in steel, as function of mass of silicon
reductant added, at 1638uC
Conclusions
Ferrosilicon reduction Computational thermodynamics modelling of reactions
The effect of adding ferrosilicon to the EAF was taking place during the melting of steel scrap in an EAF
modelled by creating another HSC input file, where to produce a melt suitable for further refining into
the masses of inputs were those applicable to an oxygen stainless steel provides valuable insights into the effect of
addition of 2?5 kg and where the amount of ferrosilicon operational variables on the efficiency of the process. It
added was varied. Steelmakers typically use 75 wt-%Si has been shown that significant oxidation of the steel
ferrosilicon so it was inappropriate to use the species scrap occurs during scrap melting, and that subsequent
FeSi from the HSC database in the modelling. Pure carbon/oxygen injection in the EAF initially oxidises
silicon was used as the input species because the iron some chromium to the slag, but then mostly forms
content of ferrosilicon plays no part in the reduction carbon monoxide gas. Carbon/oxygen injection also
reactions. In practice approximately 9–10 kg of ferrosi- results in a significant temperature increase of the steel
licon are used per tonne of steel, which contains 6?75 to bath. Chromium exists in the slag in almost equal
7?5 kg of silicon, so the silicon input was varied from proportions of CrO and CrO1?5, with CrO predominat-
zero to 10 kg. The temperature was assumed to be ing. Increasing the temperature of the steel results in less
1638uC, the temperature of the steel after it has been chromium loss to slag, but the effect is not as
heated by carbon/oxygen injection as discussed above. pronounced as expected due to the decreasing activity
Silicon (as ferrosilicon) can be seen to effectively coefficients of CrO and CrO1?5 and the increasing
reduce chromium oxides from the slag, the mass of CrOx oxygen partial pressure in the furnace. Ferrosilicon
in the slag after ferrosilicon addition being only 30% of additions reduce chromium oxides from the slag and so
that before the addition when 7 kg/tonne of silicon is enhance the recovery of chromium to the steel melt, but
much of the silicon simply dissolves into the steel and
used. However, much of the added silicon is not used for
this presents problems in the subsequent converter
reducing chromium oxides. Some is used to reduce other
refining stage.
metal oxides in the slag, e.g. MnO, but much of it simply
dissolves into the steel. In this example, when 7 kg/tonne
of silicon is used the residual concentration of silicon in Acknowledgement
the steel is 0?5 wt-% and y64% of the added silicon has The authors wish to thank the staff of ArcelorMittal
dissolved into the steel. This is unfortunate because it Stainless Genk, Belgium, for providing data and for
causes problems in the subsequent converter refining helpful discussion of their steelmaking practices.
stage. More oxygen in the refining process must be
consumed in oxidising excess silicon and the additional
slag formed may impede efficient oxygen transfer to the References
steel. Most importantly, silicon oxidises very exother- 1. Available at: http://www.worldstainless.org/ISSF/Files/Recycling/
mally so the steel in the converter becomes overheated Flash.html (accessed 12 December 2008).
2. Available at: http://www.azom.com (accessed 5 January 2009).
and this damages the refractories.
3. R. G. Ward: ‘An introduction to the physical chemistry of iron and
The results in Fig. 7 explain various comments in the steel making’, 143; 1962, London, Edward Arnold Ltd.
literature regarding the role of silicon in steel in 4. C. Moore and R. I. Marshall: ‘Modern steelmaking methods’, 74;
‘protecting’ chromium from oxidation,20 implying that 1980, London, Institution of Metallurgists.
silicon is sacrificially oxidised in place of chromium. 5. B. V. Patil, A. H. Chan and R. J. Choulet: ‘Refining of stainless
steels’, in ‘The making, shaping and treating of steel’, 11th edn,
Equation (1) for the silicon reduction of chromium 715–741; 1998, Pittsburgh, PA, Association of Iron and Steel
oxide shows that low levels of dissolved chromium in Engineers.
slag must be accompanied by high silicon levels in steel, 6. S. Arnout, F. Verhaeghe, B. Blanpain, P. Wollants, R. Hendrickx
as seen in Fig. 7. It has also been found33 that at very and G. Heylen: ‘A thermodynamic model of the EAF process for
stainless steel’, Steel Res. Int., 2006, 77, (5), 317–323.
high levels of residual silicon in steel the chromium 7. R. J. Schmitt: ‘Introduction to electric arc furnace steelmaking’, 1–
recovery to the steel improves very little, as can be seen 4; 1997, TechCommentary, EPRI Centre for Materials Production,
in Fig. 7 from the rapidly decreasing extent of Carnegie Mellon Research Institute.

Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1 7
Kho et al. Understanding stainless steelmaking through computational thermodynamics: Part 1

8. J. A. T. Jones, B. Bowman, and P. A. LeFrank: ‘Electric furnace 21. A. Roine: ‘HSC Chemistry 3?0 in metallurgical applications’, in
steelmaking’, in ‘The making, shaping and treating of steel’, 11th ‘Sulfide smelting ’98: current and future practices’, (ed. J. A.
edn, 525–660; 1998, Pittsburgh, PA, Association of Iron and Steel Asteljoki and R. L. Stephens), 49–58; 1998, Warrendale, PA, TMS.
Engineers. 22. D. R. Swinbourne: ‘Computational thermodynamics modelling of
9. M. Guo, D. Durinck, P. T. Jones, G. Heylen, R. Hendrickx, R. indium behaviour during KIVCET smelting’, Proc. EMC 2007,
Baeten, B. Blanpain and P. Wollants: ‘EAF stainless steel refining. Vol. 3; 1241–1254; 2007, Dusseldorf, GDMB Gesellschaft für
Part I: observational study on chromium recovery in an eccentric Bergbau, Metallurgie, Rohstoff- und Umwelttechnik.
bottom tapping furnace and a spout tapping furnace’, Steel Res. 23. D. R. Swinbourne, S. Yan and A. Yazawa: ‘Behaviour of thallium
Int., 2007, 78, (2), 117–124. during lead smelting’, Proc. Int. Symp. on ‘Metallurgical and
10. D. C. Hilty: ‘The relation between chromium and carbon in materials processing: principles and technologies’, (ed. F. Kongoli
chromium steel refining’, Trans. AIME, 1949, 185, 91–95. et al.), 1111–1122; 2003, Warrendale, PA, TMS.
11. W. E. Dennis and F. D. Richardson: ‘The equilibrium controlling 24. K. Yamaguchi, D. R. Swinbourne and A. Yazawa: ‘Behaviour of
the decarburisation of iron-chromium-carbon melts’, J. Iron Steel minor species during oxidation smelting of lead concentrate’, in
Inst., 1953, 175, 264–266. ‘Lead–zinc 2005’, (ed. T. Fujisawa), 1231–1246; 2005, Kyoto,
12. E. T. Turkdogan and R. J. Fruehan: ‘Fundamentals of iron and MMIJ.
steelmaking’, in ‘The making, shaping and treating of steel’, 11th 25. E. C. Vanderstaay, D. R. Swinbourne and M. Monteiro: ‘A
edn, 13–157; 1998, Pittsburgh, PA, Association of Iron and Steel computational thermodynamics model of submerged arc electric
Engineers. furnace ferromanganese smelting’, Miner. Process. Extr. Metall.,
2004, 113, (1), C38–C44.
13. J. J. Kerr and R. J. Fruehan: ‘Additions to generate foam in
26. N. Y. Toker, L. S. Darken and A. Muan: ‘Equilibrium phase
stainless steelmaking’, Metall. Mater. Trans. B, 2004, 35B, 643–
relations and thermodynamics of the Cr–O system in the
650.
temperature range of 1500uC to 1825uC’, Metall. Mater. Trans.
14. L. Voj, V. Risonarta, I. Wagenknecht and H. Pfeifer: ‘Control of
B, 1991, 22B, 225–232.
chromium slagging in stainless steel production at EAF’, Proc.
27. A. Yazawa, S. Nakazawa and Y. Takeda: ‘Distribution behaviour
AISTech 2007, Indianapolis, IN, USA, May 2007, ISS, 1057–
of various elements in copper smelting systems’, in ‘Advances in
1063.
sulfide smelting’, (ed. H.Y. Sohn et al.), 99–117; 1983, Warrendale,
15. S. Sun, P. Uguccioni, M. Bryant and M. Ackroyd: ‘Chromium PA, TMS.
control in the EAF during stainless steelmaking’, Proc. Conf. on 28. S. Ban-Ya: ‘Mathematical expression of slag-metal reactions in
‘Electric furnace’, 297–302; 1997, Warrendale, PA, TMS. steelmaking process by quadratic formalism based on the regular
16. D. Springorum (ed.): ‘Slag Atlas’, 2nd edn, 156; 1995, Dusseldorf, solution model’, ISIJ Int., 1993, 33, 2–11.
Verlag Stahleisen GmbH. 29. Y. Xiao, L. Holappa and M. A. Reuter: ‘Oxidation state and
17. E. B. Pretorius and R. C. Nunnington: ‘Stainless steel slag activities of chromium oxides in CaO–SiO2–CrOx slag system’,
fundamentals: from furnace to tundish’, Ironmaking Steelmaking, Metall. Mater. Trans. B, 2002, 33B, 595–603.
2002, 29, (2), 133–139. 30. G. K. Sigworth and J. F. Elliott: ‘The thermodynamics of dilute
18. E. B. Pretorius and A. Muan: ‘Activity-composition relations of iron alloys’, Met. Sci., 1974, 8, 298–310.
chromium oxide in silicate melts at 1500uC under strongly reducing 31. C. H. P Lupis: ‘Chemical thermodynamics of materials’, 524–527;
conditions’, J. Am. Ceram. Soc., 1992, 75, (6), 1364–1377. 1983, Englewood Cliffs, NJ, Prentice Hall.
19. W. Pei and O. Wijk: ‘Experimental study on the activity of 32. Y. K. Rao: ‘Stoichiometry and thermodynamics of metallurgical
chromium oxide in the CaO–SiO2–Al2O3–MgOsat–CrOx slag’, processes’, 892; 1985, Cambridge, Cambridge University Press.
Scand. J. Metall., 1994, 23, 228–235. 33. C. W. McCoy and F. C. Langenburg: ‘A key to greater chromium
20. M. Gőrnerup and H. Jacobsson: ‘Foaming slag practice in electric recovery: slag-metal equilibrium during stainless steel melting’,
stainless steelmaking’, Iron Steelmaker, 1998, 25, (6), 59–66. J. Met., 1964, 421–424.

8 Mineral Processing and Extractive Metallurgy (Trans. Inst. Min. Metall. C) 2010 VOL 119 NO 1

View publication stats

You might also like