Feni: Formation and Reversal of Silicon: Egil J.M. Jahnsen July 4, 2007
Feni: Formation and Reversal of Silicon: Egil J.M. Jahnsen July 4, 2007
Feni: Formation and Reversal of Silicon: Egil J.M. Jahnsen July 4, 2007
Title
FeNi : Formation and Reversal of Silicon
Author Date
Egil J.M. Jahnsen July 4, 2007
TABLE OF CONTENTS
6. REFERENCES...................................................................................................................................2
FeNi_states.DOC 1/21
1. FeNi Smelting in Electric Furnaces
Temp. °C
Solid calcine entering furnace. Direct &
indirect reduction. Gasification of
800 - 1300 Solid Calcine carbon by CO2 and remaining H2O is
active, and will accelerate when temp.
+ exceeds 950°C. Formation of solid
FeNi, solid solution metal prills with a diameter of
approx. 1 micron.
FeNi
liquid Formation of cohesive layers of partially
molten calcine and molten slag. Initial
dissolution of carbon in liquid metal can
take place.
There are many questions with respect to where, when and how the reduction of the various
metal oxides are taking place in the furnace. Observations from various FeNi furnaces indicate
that there may be several different mechanisms at work for the reduction in the various zones
inside the furnace, and that the conditions may change depending on the power settings, furnace
geometry, type of reductants and the overall composition of the calcine.
FeNi_states.DOC 2/21
1.2 General Aspects of Mixed Reduction in RKEF Operation
The calcine entering the electric furnace is partly pre-reduced in the sense that the better part of
the Fe is present as Fe2+, but some portion of iron will still be present as Fe3+, probably in a form
that can be regarded as Fe3O4. A minor part of the nickel may also be present as metallic Ni0.
Carbo-thermic reduction reactions in a high-powered smelting furnace do not occur at only a
single temperature. The furnace itself will represent a number of various zones with large
temperature differences and retention times. Direct and indirect reduction by carbon will be
predominant in various zones inside the furnace, and their preconditions will vary with the
mode of furnace operation. Some reduction reactions also need temperatures well above their
respective equilibrium temperature in order to obtain reasonable reduction rates.
The carbon content in the calcine must always be thoroughly balanced to cover for the
remaining reduction of Ni and Fe in order to produce the required grade of the crude FeNi, and
also for the eventual amount of dissolved carbon and reduction of Si, Cr and Mn. The extent of
Si reduction and C dissolution is closely related to the Ni-grade, in the sense that it will depend
on the total amount of carbon available in the charge.
heat transfer
calcine flow
gas flow
furnace
centre
electrodes
electrodes
slag
height
slag bath
metal
height above metal bath
taphole
Figure 2 above depicts the case of resistance heating, where the electrode tips are dipped into
the slag. The yellow arrows depict the material flow in the calcine bed, while the red ones
FeNi_states.DOC 3/21
indicate the heat flux from the superheated slag. The black arrows depict the dissipation of the
percolating reaction gases.
The calcine is mainly melted by the heat transferred from the slag, where the thermal energy is
created by the resistance heating in the slag bath. There will, however, be a higher energy
concentration around the electrodes, but a distinct melting crater is not always clearly observed.
Due to the absence of such a melting crater, one cannot expect the reduction reactions to pick up
speed and come to a completion until the major part of the calcine has reached the high
temperature zones relatively deep into the furnace and close to the calcine-slag interface. A
considerable amount of the carbon may therefore not be consumed until the calcine reaches the
zone at which it is liquefied, resulting in a considerable degree of carbon dissolution in the
reduced metal prills. As will be discussed later, this is also closely connected with high contents
of silicon in the metal. FeNi melting by resistance heating is therefore suitable mainly for low
grade FeNi (15 – 25% Ni) and when high carbon and silicon can be accepted.
By increasing the impedance on the furnace above the bath resistance then the electrode tip will
be forced out of the slag bath, and the furnace will operate with a small arc under each of the
electrode tips. This will increase the energy concentration in this confined area, and increase
the melting rate and material flow close to the electrodes. One will then observe a more
confined melting crater, from which the flow of escaping gases will noticeably increase.
heat transfer
calcine flow
gas flow
furnace
centre
electrodes
brush arc
slag
height
slag bath
metal
height above metal bath
taphole
Figure 3 depicts the situation when the melting operation is carried out with a “brush arc”.
From an operator’s point of view, it will become necessary to increase the calcine feeding rate
to the center of the furnace, and typically some 65 – 70% of the calcine will be consumed in the
central area of the furnace and less than 35% in the lateral part.
The very hot gases escaping the arcing zones will improve the heat transfer and reducing
potential conditions for the solid-state reductions to be completed in the calcine bed inside the
melting crater, and thus increase the rate at which the carbon is consumed, and less carbon will
reach the liquid phase. This will lead to conditions for generally lower carbon contents in the
metal, while a high degree of reduction can be obtained for production of low grade FeNi with
the associated low Ni losses to the slag.
It must, however, be emphasized that operation with an arc to decrease the carbon content in the
crude metal will require a better control of the carbon content in the calcine in order to maintain
the correct carbon balance. Too much carbon, or a carbon material with a low reactivity, can
promote carbon accumulation in, or around, the arc zones, which will give conditions for
FeNi_states.DOC 4/21
excessive silicon reduction and formation of SiO-gases at very high pressures under these very
hot, highly reducing conditions.
By increasing the electrode voltage and bring impedance further above the bath resistance the
arc length will increase, and one will reach the conditions referred to as ‘shielded arc smelting’.
Figure 4 depicts the extreme case of arc smelting in the rectangular furnaces at Xstrata
Dominicana, where the major part of the total energy input (>75%) is in the form of arc power.
To a lesser degree, Cerro Matoso have also operated their circular furnaces in this mode of high
voltage smelting.
heat transfer
calcine flow
electrodes
arc bubble
slag
height
slag bath
metal
height above metal bath
taphole
Typical for shielded arc melting is that most of the calcine (~90%) is fed either directly towards
the electrodes (Xstrata) or into the central area of the furnace (Cerro Matoso). Due to the very
high melting rate inside a small area, this mode of smelting is only suitable for high Ni-grade
metal where the carbon content in the calcine is relatively low. The heat transfer from the arc
zones to the surrounding burden is extremely high, and the high flow of hot gases dissipating
from the arc bubbles ensure the solid state reduction reactions to be completed at a very high
rate and minimizing any accumulation of carbon below the electrodes. Shielded arc smelting,
combined with highly reactive carbon source therefore makes it possible to produce a carbon
and silicon free high grade FeNi with acceptable Ni-losses to the slag.
FeNi_states.DOC 5/21
ii) creation of volatile gases such as Mg and Si vapors and SiO gas may escape from the
arc zones, resulting in high off-gas temperatures, low carbon utilization, under-
reduction and high specific power consumption;
With respect to point i) above it should be kept in mind that if the melting rate of the feed
material becomes higher than the carbon consumption rate then there is a possibility for an
accumulation of carbon material around the electrodes, such as observed in other types of
ferroalloy furnaces (FeSi, FeSiMn etc).
The power densities when smelting with arcs, brush arcs or shielded arcs, may also be restricted
by the volume of reaction gases leaving the melting crater. The volume of the reaction gases is
proportional to the amount of carbon being consumed by the reduction reactions at a given
melting rate of the calcine. It may therefore be of some interest to make a comparison between
some of the various FeNi producers known to operate their furnaces in this manner. First, a few
simple definitions can be made. Although they are not of a very scientific nature, they can help
to get a feeling of what power level and carbon content where problems may arise. The two
parameters called Active Crater and Crater Gas Flow Rate can be defined as follows:
"Active crater":
For circular furnaces one can measure the distance from the center of the furnace to the outside
face of one of the electrodes. Add 1 m to this distance, and one now has the radius of what can
be considered to be the "active melting crater". One can assume that about 80% of the calcine
will be melted inside this area; hence also 80% of the reduction off-gases will be created within
this area, when operating a furnace in shielded arc mode.
For a rectangular furnace we can calculate the "active crater" as the rectangle we get when we
go 1 m out on either side of the electrodes. Again, we assume that 80% of the feed material is
melted (and reduced) within this rectangle.
"Crater gas flow rate":
Actual off-gas flow rate at 1000°C per square unit of "active crater" per hour.
FeNi_states.DOC 6/21
2)
The data from FeNi II at Antam are taken from Hatch’s commissioning during October
2005, which included operating for a number of days at >24 MW. At 24 MW and an
estimated arc length of 50 mm serious problems occurred with arc flares and “burners”
around the electrodes.
3)
Assumed PCD = furnace inside diameter / 3.1 (typical Elkem design)
It is quite interesting to notice that in all cases one gets close to the "magic" number of 165
m3/m2/hr for the crater gas flow rate. However, no considerations have been made with respect
to eventual differences in the gas permeability in the charge material.
Unit reaction iii), representing the Boudouard’s reaction, can be quite slow at low temperatures,
but is known to ‘ignite’ at about 900°C and at temperatures above 1000°C, it starts to run very
quickly. On the other hand, unit reaction i) may be very slow, because it includes a mechanism
for diffusion by either carbon or oxygen in solid particles. Solid state reduction with a gaseous
reductant at 900°C and above is at least one, if not two, orders of magnitude faster than with
solid reductants, so we can therefore ignore the effect of direct reduction with solid carbon
according reaction i).
Figure 5: Effect of temperature on the initial rate of reduction for hematite– carbon pellets, for
standard conditions (solid line), and where the rate constants are increased by an order of
magnitude (broken line). The equilibrium temperature for reduction of hematite to magnetite is
indicated. For standard conditions, the reduction rate only becomes significant if the
temperature is higher than 800°C. (Ref. 1)
Solid state reduction therefore depends on the combination of reactions ii) and iii), such that:
CO + FeO = Fe + CO2 ii)
└→ CO2 + C = 2CO iii)
└→ 2CO + 2FeO = 2Fe + 2CO2 ii)
└→ etc iii)
FeNi_states.DOC 7/21
It should be noticed that unit reaction ii) can run from left to right in the calcine bed if the
temperature is sufficiently high, but it will come to a halt in the forward direction, if the CO
content in the gas should drop below a critical level at the given temperature. (See Figure 6)
0.4
<Fe3O4> + CO = 3<FeO> + CO2
0.3
C + CO2 = 2CO
0.2
0.1 <NiO> + CO = <Ni> + CO2
0.0
400 500 600 700 800 900 1000 1100 1200 1300
Temperature, °C
For each mole of CO consumed by the reduction, the Boudouard’s reaction will be able to
regenerate the formed CO2 into two new moles of new CO. However, while unit reaction ii) is
slightly exothermic, the Boudouard’s reaction is highly endothermic. At 1000°C, each mole of
Fe reduced will require as much as 152 kJ, which is equivalent to 0.76 kWh per kg reduced Fe.
For a calcine containing 18 wt-% FeO and a FeNi smelting including 60% Fe reduction (Pamco,
Antam) this amounts to nearly 64 kWh/ton calcine, - or about 12% of the total furnace power.
If sufficient thermal energy is not supplied to the calcine bed, then the solid state reduction of Fe
in the upper part will not be able to proceed, but will be delayed until the calcine has moved
deeper down towards the calcine-slag interface. As a consequence, the concentration of CO gas
in the upper calcine level will diminish.
The furnace charge, which consists of calcine and reduction material, will, on its descent toward
the melting crater, be heated and reduced to some extent by the rising CO gas. The heating
effect by the percolating gas is, however, in itself rather small. In the case of calcine smelting
by pure resistance heating, for the sake of simplicity, the following extreme conditions can be
assumed:
- the reaction gases rise evenly distributed over the cross section of the charge bed;
- the calcine is charged into the furnace at a temperature of 850°C;
- the calcine contains 20% FeO, 30% MgO and 50% SiO2;
- the extent of reduction is equivalent to 72% of the FeO reduced to Fe;
- the initial temperature of the rising reaction gases is 1600°C.;
- the composition is 75 vol-% CO and 25 vol-% CO2.
The equation for the gas heating upper calcine layers can then be written as:
0.287FeO850°C + 0.744MgO850°C + 0.832SiO2850°C + 0.16CO1600°C+ 0.04CO21600°C + Q
= 0.287FeOTemp + 0.744MgOTemp + 0.832SiO2Temp + 0.16COTemp + 0.04CO2Temp
The specific heat capacity of the CO gas is in the range of 34 – 36 J/mol/K between 900°C and
1600°C, and the respective values for CO2 are 56 – 60 J/mol/K. The ΔH for reaction above
becomes zero already at Temp ≈ 900°C. In other words, heating of the calcine by the rising
gases is very small. For a typical production of low grade FeNi it represents less than 2 kJ per
mole of gas from the reduction, or less than 1.5% of the energy required for sustaining the solid
state reduction reactions in the calcine bed.
FeNi_states.DOC 8/21
Other mechanisms for heat transfer to the overlying calcine bed would either include conduction
or combustion of CO gas by infiltration air. The thermal conductivity of loosely packed grain
material, such as calcine, is low, and the specific heat capacity is quite high. When considering
the overall retention time for the calcine, the probabilities for conduction alone to bring
sufficient thermal heat to the upper layers in the calcine bed are rather small. Combustion of
carbon and CO gas in the bed by infiltration air will, on the other hand, release large amount of
heat:
C + 2.381air = CO + 1.88N2 ΔH1000°C = -113 kJ/mol iv)
CO + 2.381air = CO2 + 1.88N2 ΔH1000°C = -280 kJ/mol v)
The equation v) above indicates that combustion of approx. 50% of the formed CO gas is
sufficient to cover the energy required to sustain the solid state reduction reactions in the
calcine. This, however, will bring the CO/CO2 ratio in the reaction gases down to a level where
the extent of solid state Fe-reduction will become quite limited. Both reactions iv) and v) may
take place in the upper calcine layers. Gas sampling has indicated that the ratio between the
volumes of reaction gases (CO + CO2) and infiltration air may be about 3:1 in the upper calcine
layer.
The importance of sufficient transport of thermal energy to the reaction site in connection with
solid state reduction is well known from production of DRI, and is documented in published
literature. The Figure 7 below shows the modeled reduction times at various heat transfer rates
for hematite-carbon pellets. The graphs indicate a nearly linear correlation between reduction
rate and heat transfer rate.
FeNi_states.DOC 9/21
Arc smelting, on the other hand, with the electrodes above the slag level, will increase the
melting rate of calcine in a certain area around the electrodes. The conditions for heat transfer
in this melting crater are excellent due to the radiation from the arc and the flow of exited gases
and vapors escaping from the arc bubble. The combination of increased temperature and highly
efficient energy transfer in the melting crater will provide optimum conditions for the solid state
reduction reactions to run to completion at very high rates. However, the power released in the
arcs (arc power, Pa) must be balanced according to the carbon content in the calcine and the
degree of reduction required. High impedance combined with highly reducing conditions can
easily result in excessive arc flare that will penetrate the calcine bed and escape to the freeboard
and lead to excessive fuming and deposition of white magnesia and silica products. Arc
smelting is therefore most suitable for smelting of calcine containing less than about 1.5 - 2 wt%
carbon, and when using reductant materials associated with high reactivity.
Thermodynamic calculations cannot be expected to be very precise, but may be useful for
comparison of the possible conditions for carbon dissolution and silicon reduction under the
various modes of operation. Extreme temperatures may exist inside the arc bubble itself, and
large temperature gradients will exist in the melting crater. Not all the calcine is reduced in the
high temperature zone, but also in the colder lateral zones in the furnace. For a simplified
example one may assume an average effective temperature of the reduction gases in and around
the melting crater to be about 1 800ºC. The resulting oxygen pressure can then be estimated
from the resulting slag and metal analysis. For this calculation the typical data from the current
operation at Cerro Matoso can be used:
%Ni ≈ 38
αFe ≈ 0.6
αFeO ≈ 0.14
αSiO2 ≈ 0.8
FeNi_states.DOC 10/21
[Si]FeNi + ½O2 = SiO2
With a SiO2 activity of about 0.6 the activity of the dissolved Si can then be estimated to:
a[Si]FeNi ≈ 0.04 (1 wt-% ref.)
By iterative calculations using thermodynamic functions for the activity coefficients in FeNi
containing 38% Ni, the equilibrium carbon and silicon contents are estimated to be:
wt-%[C]FeNi ≈ 0.013 wt-%
wt-%[Si]FeNi ≈ 0.025 wt-%
These numbers compare well with the FeNi produced at Falcondo and Cerro Matoso, who both
produce high grade FeNi from a calcine containing less than 1.5% carbon. The calculations
above indicate that the carbon and silicon contents in their metals are the result of a predominant
solid state reduction of the calcine inside their furnaces.
FeNi_states.DOC 11/21
of carbon in the liquid metal. The degree of carbon dissolution will depend on factors such as
metal composition, time for exposure, temperature, type of carbon material etc, and is widely
studied and reported in published papers. The results from a pilot plant test smelting of FeNi,
reported by I. Candy and A. Matyas [Ref. v)] clearly indicated that the use of low reactive
carbon materials, such as anthracite, would increase the carbon content in the metal. The reason
was assumed to be that less carbon was consumed by the solid state reduction reactions, and
would result in deeper coke beds and increased contact time between carbon and liquid FeNi
prills. A study published by Sheau Tsuey Cham & al [Ref. iv)] concluded that non-graphitic
carbon dissolution is much slower than dissolution of graphitic materials. In addition to the
large apparent activation energies, the dissolution rates were also attributed to the effect of
sulfur and ash in the non-graphitic carbon source. Table 1 shows the apparent dissolution
constants found for synthetic graphite and two different Australian coke types at various
temperatures. It is most interesting to notice that the dissolution rate for the graphite material is
quite constant with temperature.
FeNi_states.DOC 12/21
Due to the high SiO2-activity in the slag (and SiO2-content in the ashes from the reductant) the
Si-content in the FeNi may, theoretically, reach its saturation level even at temperatures below
1500°C. The rate of silicon reduction is, however, known to be slow at low temperatures, and
the Si content in the produced metal can therefore be expected to be proportional to the
concentration of dissolved C for a given process temperature and amount of reductants in the
calcine. The C/Si ratio in the metal can also be expected to increase by increasing the amount
of anthracite in the reductant mix, and to decrease by a higher temperature in the melting zones.
The energy required for silicon reduction and carbon dissolution should be taken into
consideration. At 1600°C each kg of Si reduced requires nearly 7.5 kWh, and about 1.3 kWh
for each kg of C dissolved. For a low grade FeNi containing some 2% C and 2.5% Si this
additional energy requirement will then be equivalent to about 30% of the energy requirement
for the total Fe-reduction.
In connection with production of low grade FeNi, for which low impedance operation is most
suitable, it is most tempting to try increasing the electrode voltage to increase the productivity.
By introducing additional arc smelting under such conditions, one must take into consideration
that the newly reduced metal produced in the two different reduction zones may have different
composition. The metal produced in the melting crater around the electrode may be low in
carbon but have high silicon content, whereas the metal reduced in the area outside the central
crater may be high in both carbon and silicon. Reducing the carbon content in the calcine in
order to reduce the Si-content, or for the purpose to avoid arc flare, may decrease the C/Si ratio
in the metal prills until the point where a considerable degree of Si-reversion can take place.
The reasoning behind the statements above is based on the assumption that an operating a FeNi
furnace with arcs of some lengths, in combination with excessively high carbon content in the
calcine, can promote direct reduction inside the melting crater. A considerable part of the
reduction reactions can then occur at very high temperatures by solid carbon reacting directly
with the liquefied slag.
C + FeO = Fe + CO
The equilibrium oxygen pressure for the carbon oxidation can be estimated by adding the two
reactions x) and xi) below:
C + ½O2 = CO x)
Fe+ ½O2 = FeO xi)
FeNi_states.DOC 13/21
By adding the equations we then get:
Fe + C + O2 = FeO + CO xii) ( K ≈ exp[1.42 + 19308/T] )
As a calculation example one may again assume the activities for Fe and FeO to be 0.75 and
0.08, respectively, and an effective temperature in (and around) the melting crater of 1 800°C.
The oxygen pressure for the reduction reactions is then estimated as:
pO2 = 0.08 * 1.3 / (0.75 * exp[1.42 + 19 308/T]) = 2.5E-12 atm.
The equilibrium activity of dissolved carbon in metal particles being reduced from the slag
stream under such oxygen pressure and temperature is only about 1.5, which represents about
0.5 wt-% carbon in a pure Fe-Ni-C system at 25 wt-% Ni. The silicon, however, reaches a value
close to its saturation level and can reach very high levels in the metal phase, depending on
amount of carbon available, time exposure and rate of reduction reaction rates. The formed
metal prills having a high content of Si can, during their decent through the cooler slag bath
result a high degree of Si-reversion due a shift in the C – Si – O equilibrium. The released
thermal reaction energy will increase the metal temperature.
[Si]FeNi + 2{FeO} = 2Fe + {SiO2} xiii) ΔH1550°C = -3.08 kWh/kg Si
Furnace
center
electrodes
arc bubble
slag
height
slag bath
metal
height above metal bath
taphole
It must also be taken into account that SiO-gas, Si and Mg vapors at relatively high pressures
can easily be created in the arc zones from the molten products when getting into contact with
solid carbon particles. The electrode tip itself will supply vaporized carbon into the arc bubble,
but under very reducing conditions where pieces of coal or anthracite can accumulate inside the
melting crater, and problems with strong flaring and erratic silicon reduction can take place.
Even if extrapolation of some thermodynamic data may not be quite precise, the Table 1 below
shows the predicted activities of the various slag components at 2000°C for a typical slag for
low grade FeNi smelting.
Table 2 : Slag component activities at 2000°C
Slag composition Slag composition, wt-% Oxide activities at 2000°C
SiO2 55 0.6
MgO 32 0.1
Al2O3 3 0.003
CaO 2 0.0003
FeO 8 0.065
FeNi_states.DOC 14/21
MexOy + yC = yCO(1 atm.) + xMe(g)
If we apply the estimated activities for SiO2 and MgO in the liquid slag and set the CO pressure
to 1 atm. then we can calculate the gas and vapor pressures at various temperatures.
Table 3: Vapor pressures at various temperatures
Comp./Temp.(C°) 1800°C 2000°C 2200°C 2400°C
Si(g) 0.001 atm. 0.03 atm. 2.6 atm. 1240 atm.
Mg(g) 0.05 atm. 1.1 atm. 13.9 atm. 120 atm.
SiO 1.4 atm. 38 atm. 590 atm. 5860 atm.
Proper arc operation includes constant feeding of new charge material around the electrodes to
maintain the arcs covered. The above vapors (and gases) will, on contact with the colder
calcine, act as effective reducing agents:
Si(g) + 2FeO = SiO2 + 2Fe
Mg(g) + FeO = MgO + Fe
SiO(g) + FeO = SiO2 + Fe
All the reactions above are highly exothermic, and will release large amounts of thermal energy
to the surrounding calcine. However, if the arc bubble is not sufficiently shielded, or if a
“chimney” is created, then these gases will escape to the freeboard where they will oxidize with
the atmospheric oxygen. This will result in very high off-gas temperatures and deposits in the
ducts. Considerable losses of Mg vapors and SiO gases to the freeboard will thus represent a
loss of reducing agent for the process, which will lead to under-reduction of Fe and increasing
Ni content in the final FeNi product.
The calculation example thus demonstrates that shielded arc operation with high carbon
contents in the calcine can lead to extremely erratic results. Very high silicon levels can be
created in the produced metal, which can result in:
- increased bath temperatures, and small delta-T between slag and metal;
- erratic silicon contents in the tapped metal;
- low and erratic carbon utility;
- increased specific power consumption;
- erratic nickel grade in the tapped metal.
3. Silicon Reversion
Silicon reversion reveals itself by a gradual or sudden loss of silicon in the crude FeNi, and
results in a considerable increase of the temperature in the metal bath. The Si content in the
metal bath will drop from a relatively high level and down to nearly zero. In some cases the
substantial increase in temperature will also lead to a carbon boil, which further aggravates the
situation. Most often, an occurrence of Si-reversion is preceded by a slight in carbon content,
but there may also be other factors that can trigger the incident.
To the knowledge of this author, silicon reversion is only known to be a common problem in
furnaces operating in either brush-arc mode or shielded arc mode. The preconditions for a
reversion to start is that there exists a chemical potential between the Si in metal and FeO in the
slag. Once the reversion reaction is triggered, then it tends to run until equilibrium condition is
established.
Experience from Antam has indicated that operation in “brush-arc” mode is fully possible, but
that the impedance applied must be limited in order to avoid problems with “burners” and arc
flare. Similar to previous experience at Cerro Matoso in the early 90’s it was also observed that
relatively small carbon deficits in the calcine would decrease the C-content in the metal, and
when reaching a critical level then a Si-reversion would easily occur.
FeNi_states.DOC 15/21
3.1 Reduced Metal at Cerro Matoso
When Cerro Matoso initially operated at low power (20 – 35 MW), producing a high grade
FeNi that contained 40% Ni, 0.3% C and about 2% Si, then the problem with occasional Si-
reversions was never an issue. When switching to high voltage operation and increasing the
power level, then Si-reversion became more and more frequent until they finally had to change
their production to an oxygenated metal with nearly any carbon and silicon at all. The change-
over was carried out by a reduction of carbon in the calcine by approx. 0.2%. (See Table 4)
It may seem strange that tiny metal droplets containing 0.28% C and 1.6% Si can sink through a
slag bath with 18% FeO without being completely oxidized according to reaction scheme:
{FeO} + [C]FeNi = Fe + CO xiii) ΔG° = 106918 - 97.1T
{FeO} + ½[Si]FeNi = Fe + ½SiO2 ivx) ΔG° = -159166 + 52.2T
However, by careful calculation of the activities for Fe, C and Si in the FeNi, and a best possible
estimation of the activities for FeO and SiO2 in the slag, we find that:
αFe = 0.59 a[Si] = 4.86 a[C] = 1.26
αFeO = 0.17 αSiO2 = 0.87 pCO is assumed to be 1.3 atm.
The equilibrium oxygen activity for carbon and silicon can now be calculated:
a[O]equil.C = pCO / (a[C] * KC-O = 20 ppm
a[O]equil.Si = αSiO2 / (a[Si] * KSi-O = 20 ppm
In this case it can be seen that C and Si are both at the same chemical potential with respect to
oxygen. Carbon’s affinity to oxygen decreases with decreasing temperature, while silicon’s
oxygen affinity increases. That means that Si is locked by C at the actual temperature. The
only way for Si to react with FeO is either by a decrease in temperature or by an oxidation of
carbon.
Studies from the CMSA operation in 1991 indicated that Si in the metal prills formed in the
melting crater would readily re-oxidize and get into equilibrium with C. The Figure 9 below
shows the daily averages of metal and slag temperatures as well as the Si and C in FeNi as
function of the furnace impedance. The fact that the graphs show a clear increase of the tapped
metal temperature with increasing impedance may seem a bit odd at first glance. However, the
graphs also show a slight decrease of the slag temperature, while the carbon content seems to
remain rather insensitive to changes in the impedance.
A slight drop in temperature will allow some Si to oxidize with FeO according to equation ivx).
The very strong interaction between C and Si will, in turn, lead to a reduction of the C-activity
and extend the Si-oxidation. By thorough recalculations of the activities for the solutes in metal
and the slag components one will find that a drop in slag temperature from 1600°C to 1580°C
FeNi_states.DOC 16/21
will, in fact, lead to a drop in the silicon content from 1.6% down to 1.0% when carbon is kept
constant at 0.28%. In spite of the variation in the data points, the trend lines seem to support the
validity of the thermodynamic calculations in the range of furnace impedances between 45 and
65 mΩ.
A higher degree of silicon reduction in the melting crater, as a result of a higher temperature
imposed by the increasing impedance, is also a possible explanation. In such case one would
expect to see a decrease in the carbon content, which is not directly evident from the collected
furnace data. From a thermodynamic point of view, a considerable drop in slag temperature in
combination with lower carbon content in the metal prills could trigger a dramatic degree of Si-
reversion due to the spiraling effect caused by the strong interaction between Si and C. If Si
oxidizes then the C-activity drops, which, in turn, allows more Si-oxidation and so on. This
scenario may explain why high impedance operation in combination with a carbon deficit in the
calcine would easily lead to a complete reversion of Si and C.
Daily averages CMSA, Jan.1 - Dec.31, 1991 Daily averages CMSA, Jan.1 - Dec.31, 1991
1520 3.50
1500 3.00
1480
2.50
1460
Metal temp. °C
%Si in FeNi
2.00
1440
1.50
1420
1.00
1400
0.50
1380
1360 0.00
30.0 40.0 50.0 60.0 70.0 80.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0
Daily averages CMSA, Jan.1 - Dec.31, 1991 Daily averages CMSA, Jan.1 - Dec.31, 1991
1720 0.40
1700 0.35
1680
0.30
1660
0.25
Slag temp. °C
%C in FeNi
1640
0.20
1620
0.15
1600
0.10
1580
1560 0.05
1540 0.00
30.0 40.0 50.0 60.0 70.0 80.0 30.0 35.0 40.0 45.0 50.0 55.0 60.0 65.0 70.0 75.0
The question then remains: Why does not the carbon dissolved in the metal prills, or in the
metal bath, react with FeO in the slag? The answer to this question seems very complex, and is
probably associated with the possibilities for nucleation of CO bubbles within the metal
droplets. Distin & al [Ref. viii)] carried out an investigation on decarburizing of droplets where
various phenomena connected to the surface-active oxygen could play a major role. Their
experiments proved that CO-pressures in equilibrium with the bulk carbon and oxygen
concentrations could reach up to 10 – 40 atm. Anyhow, experiences from CMSA indicated
clearly that when carbon eventually starts to oxidize then it would oxidize quite rapidly down to
a very low level.
FeNi_states.DOC 17/21
3.2 Silicon Reversion in Low Grade FeNi
Reversion of silicon was observed in a number of cases during the period of the HATCH
Operational Support for FeNi II and Antam. A stable “brush-arc” operation at 25MW at the
impedance in the range of 20-25 mΩ was achieved. Over a two month period, the mean slag
temperature was 1550°C, with the average composition of 12.5% FeO, 55% SiO2, 30% MgO,
3% CaO and 2% Al2O3. The corresponding crude ferronickel metal was tapped at 1450°C and
contained 26% Ni, 0.3% Co, 1.5% C, 2.8% Si, 0.6% Cr, 0.03% Mn, 0.03% P and 0.3% S.
Contrary to the case of CMSA, the C and Si in metal would not be at the same oxygen potential
at the measured slag temperature of 1550°C, but rather at a temperature close to that of the
metal. By thermodynamic calculation of the activities for solutes and slag components from the
analysis above it can be found that
a[O]equil.C = a[O]equil.Si = 2 ppm at T = 1470°C
Thermodynamic calculations of this kind can never be very precise, but if trustworthy enough to
be compared with those made for the CMSA case, a higher degree of Si-reversion should be
expected in the newly reduced metal prills during their decent through the slag layer. This
assumption is supported by the fact that the delta-T between slag and metal at Antam was only
in the range of 135-150°C (sometimes even much lower), while it was generally in the range of
200 - 220°C at CMSA during the years with production of reduced FeNi.
The reports from the Operational Support engineers at Antam gave the strong impression that
the occasional incidents of Si-reversion from the metal bath occurred as a consequence of
carbon deficit in the calcine while operating the furnace in the “brush-arc” mode. Low slag
temperature may also be an important factor that can trigger such an incident, and the influence
of the waffle coolers cannot be ruled out. As mentioned before, a slight drop in temperature
will allow Si to oxidize with FeO in the slag and the C-activity will drop, allowing Si to oxidize
further. An eventual additional cooling effect by the sidewall coolers will be fuelling this chain
of reactions, and a nasty spiraling effect may start.
FeNi_states.DOC 18/21
4. Conclusions:
This study includes considerations on why high silicon contents in FeNi will be formed and how
it is closely connected with the carbon content. From what has been described through
calculations and references, the following conclusions can be made:
1. Reduction of silicon is not necessarily connected to the extent of Fe-reduction from the
calcine. It will also take place in production of high-grade FeNi where the FeO content
is high, provided there is an excess amount of carbon available in the charge. Silicon
reduction in the melting zone can also take place in connection with production of
oxygenated metal, but if will readily re-oxidize with FeO in the slag due to the absence
of dissolved carbon in the metal prills.
2. Whenever silicon reduction takes place in the melting zones, the Si content in the crude
FeNi in the furnace is most often proportional to the amount of carbon dissolved in the
newly reduced metal prills. However, the type of reductant material used may have
strong influence on the C/Si ratio in the metal. The Si content in the crude metal will
also depend on the slag temperature, such that a lower slag temperature is liable to
decrease the Si analysis. While the Si content in the newly reduced metal prills may be
considerably higher than in the tapped metal, a continuous re-oxidation before reaching
the slag-metal interface can lead to increased transport of heat towards the metal bath.
3. A certain degree of silicon reduction seems inevitable in FeNi smelting, and may not
represent a problem rather than it steals a lot of energy from the melting process. The
real problem, however, is the possibilities for Si-reversions in the metal bath, which will
lead to a serious superheat of the furnace. The exact reasons and mechanisms for a
sudden Si-reversion and oxidation of the metal bath are not clear, but seem to be
associated with high voltage furnace operation combined with a lasting deficit of carbon
in the calcine.
4. The exclusion of periodic incidents with Si-reversions and oxidations of the metal bath
can only be fully guaranteed by production of oxygenated metal, such as currently done
by Xstrata Dominicana and Cerro Matoso. Oxygenated metal also allows for high
impedance furnace operation at very high power densities. A second option, but a much
less attractive one, would be production of a low grade FeNi with sufficiently high
carbon content in the metal and low FeO in the slag to suppress any Si-reversion. This
would include melting at low impedance (resistance heating) and restricted power
density. A careful selection of a mix of reductants with high and low reactivity may
also be required in order to ensure a sufficiently high C/Si ratio in the reduced metal.
FeNi_states.DOC 19/21
5. Calculation Tools Applied
Table 5: Interaction parameters used in calculations.
a/T + b
i j k e[i, j] r[i, j, k] Tmeasure, K
a b
C C 0.1425 1873 158 0.0581
C C C 0.00737 1873 8.94 0.0026
C Cr -0.0230 1873
C Ni 0.012 1873
C S 0.0428 1873
C Si 0.0785 1873 162 -0.008
C Si Si 0.00074 1873 1.94 -0.0003
Si C 0.18 1873 380 -0.023
Si Cr -0.0003 1873
Si Ni 0.0050 1873
Si Si 0.103 1873 34.5 0.089
Si Si Si -0.00203 1873 6.5 -0.0055
O C -0.421 1873
O Cr -0.0519 1873 -380 0.151
O Cr Cr 0.0006 1873 10.2 -0.00487
O Ni 0.006 1873
O O -0.1743 1873 -1750 0.76
O S -0.133 1873
O Si -0.066 1873
The data in the table above is mainly taken from G. K. Sigworth and J. F. Elliott: Metal Sci., 8
(1974)
Activity coefficient fi of i in a diluted Fe-i-j-k multi-component alloy is normally calculated as:
log f i = ∑ e ij [% j] + ∑∑ ri j,k [% j][% k ]
j j k
Each term may be replaced with a temperature dependent function on the form such as:
logfi = a*T + b
In those cases where the temperature dependent functions are reliably determined, the values for
the parameters a and b are listed in the table.
For alloys such as FeNi better precisions can be obtained if the activities for Fe and solutes are
calculated acc. to the modified interaction parameter formalism as described by A.D. Pelton &
C.W. Bale for non-dilute solutions. This method includes recalculating the wt-% related values
for e[i,j] and r[i,j,k] into their mole fraction related equivalents ε[i,j] and ρ[i,j,k], and the activity
for the base metal, Fe, is then determined as:
ln γFe = -0.5*Σ(Xi*εij) - 0.67*Σ(Xi*ρijk)
This expression is then used for calculation of the modified interaction parameter for each of the
solutes according to the expression:
ln γi / γ0i = ln fi = ln γFe + Σ(εij * Xj)+ Σ(ρijk * Xj * Xk)
The activities for the slag components have been calculated by using my programs SlagAn.
Other thermodynamic calculations use data from HSC data base, v. 5.1
All the calculation programs applied for this study are available for the Hatch personnel upon
request to the author at the email address: [email protected]
FeNi_states.DOC 20/21
6. References
i) P. Chris Pistorius, “Kinetics of Carbothermic Reduction Reactions under Heat Transfer
Control: Modelling Results”, Scandinavian Journal of Metallurgy 2005; 34: 122–130.
ii) L.T. Loe, “Aspects of Smelting Pig Iron in Electric Furnaces”, CIM Bulletin, August
1970
iii) H. Sun, “Analysis of Reaction Rate between Solid Carbon and Molten Iron by
Mathematical Models”, ISIJ International, vol. 45, No. 10
iv) Sheau Tsuey Cham & al, “Influence of Temperature on Carbon Dissolution of Cokes
inMolten Iron”, ISIJ International, Vol. 46, No. 5
v) I. Candy & A. Matyas, Hatch internal Project Report, PR10625.001, February 28, 2001
vi) E.T. Turkdogan, “Fundamentals of Iron and Steelmaking”, The Making, Shaping and
Treating of Steel, Edition 10.
vii) F.D. Richardson, “Physical Chemistry of Melts in Metallurgy”, ISBN: 0-12-587901-6
viii) P.A. Distin, G.D. Hallet and F.D. Richardson, Journal of Iron and Steel Inst., 206, 1968.
ix) L.R. Nelson, J.M.A. Geldenhuis1, T. Miraza2, T. Badrujaman, A. Taofik Hidyat, I.
Jauhari3, F.A. Stober, N. Voermann,B.O. Wasmund4 and E.J.M. Jahnsen, “Role Of
Operational Support In Ramp-Up Of The Feni-Ii Furnace At Pt Antam In Pomalaa”,
Infacon XI – 18-21 February 2007
FeNi_states.DOC 21/21