Sei sulla pagina 1di 10

Effect

Ore

By

of Alkalis

on Reduction

Behavior

Iron

Pellets*

Taymour

ELKASABGY**

Synopsis
Under conditions relevant to the process of an iron blast furnace with
alkali cycling, acid iron ore pellets were tested. The alkalis were introducedin three different ways: as part of thepellets gangue phase, as fine
crystals of alkali carbonatesnear the pellets surface, and, as metal vapor
mixed with the reducinggas.
During reductionalkalis flux thesilicate binding bridges and react with
ferrous iron oxideforming low melting point slag which weakensthepellets
and causes them to degrade. A mechanism of degradation will be given
which relates the behavior of the pellets to theformation of this primary
slag. Also, certain measuresto control the reductionwill be discussed.

I.

Introduction
Alkalis are known to cycle and sometimes to accumulate in the iron blast furnace.)-4) This phenomenon is related to the counter current character of the
furnace and the existence of two gradients of temperature and oxygen potential between its top and the
tuyeres level.5'6~ By sampling and direct observation, during relining of the furnace refractories, it is
known the alkali contents in the burden materials increase as they move down towards the melting zone
in the blast furnace.3'5,6> The higher the alkalis in
the raw materials the easier it is for them to circulate
and accumulate and to build up to dangerously high
levels.
The adverse effects of alkalis are higher coke rate,
poor quality of hot metal and deterioration of the furnace lining. Also, they cause irregularities in the
operation process and decrease the production. However, the determinate effects of alkalis are more pronounced when the stability of the coke and/or the
iron ore is relatively low.
In the present work it is intended to simulate to a
certain extent reduction conditions relevant to the
process of an iron blast furnace with alkali cycling and
to investigate the behavior of a typical acid iron ore
pellets before the formation of a strong metallic shell.
II.

Behavior of Alkalis in the Blast Furnace


This paper is not designed to investigate in detail
the stability of various alkali compounds and the
mechanism of alkalis cycling in the iron blast furnace,
they are given elsewhere.)-6~ The most important
facts regarding the behavior of alkalis may be summarized as follows :
(1) Alkalis are known to cycle and sometimes to
accumulate in the blast furnace.
(2) The movement of alkali compounds may be
represented by two open loops. The first describes
their movements as vapors (either metallic : Na and
* ReceivedJune 28, 1983. 1984ISIJ
**

of Acid

K, or in the form of cyanides KCN or NaCN), as fine


droplets of cyanides and/or carbonates : Na2C03 and
K2C03, or fine crystals of alkali cyanides, carbonates,
or silicates, with the ascending gas stream. The second loop shows their movement down with the charge
as condensed phases. When these two loops are combined they complete one cycle.
(3) At the lower parts of the blast furnace where:
the temperature is higher, the reducing power of the
gas is stronger, and the blast is almost free from alkalis,
it is thermodynamically possible that alkalis, in the
condensed phases are gasified to form vapours.
(4) Along the height of the furnace, as the gaseous
stream moves upward, there is thermodynamic possibilities of reactions which either produce or consume
certain alkali compounds.
(5) Inside the iron ore pellets, the oxygen potential is relatively much higher than in the bulk gas
stream, due to iron oxide reduction. At the gas/iron
oxide interface it is expected, based on thermodynamics, that wustite would oxidize the alkali metal
vapors, cyanides, and, carbonates to form oxides,
provided silica as stabilizer is available. Complex
alkali silicate compounds would result from this interaction between reducing gases, iron oxide, alkalis and
silica of the gangue phases. They have the highest
chemical stability, among all the alkali compounds
under blast furnace conditions. The only way to
reduce alkali silicates is to get them down, with the
charge, to the lower parts of the furnace to the higher
temperature and reducing power zones.
(6) Based on the ternary phase diagrams, between silica ferrous iron oxide and alkali oxides : alkalis drastically reduce the melting point of silica and
iron silicate gangues and slags.7'8)
From the above-mentioned points it may be concluded:
(1) The alkali contents of the charge can be much
higher than originally in the raw materials.
(2) Alkalis influence both the iron oxides and the
gangue phases, which are of silicate origin, in most of
iron ore deposits.
(3) Alkalis may be introduced during reduction
in the blast furnace and interact with silica and iron
oxides to form low melting point slag.
Most of the experimental work which had been
carried out concerned almost exclusively with the influence of alkalis on the iron oxides.9-11> It is well
known that high grades iron ore pellets, which contain alkalis, swell abnormally due to iron whiskers

Horas Investments (Kingston), Kingston, Ontario K7L 3C7, Canada. On leave from Department of Metallurgy & Materials Science,
The Catholic University of Rio de Janeiro, Rio de Janeiro, CEP 22453, Brazil.

612 ) Research

Article

Transactions

formation, the practical solution is to decrease the alkali content, and/or, the alkali to silica ratio in the
used ore.9'10~ On testing ordinarily acid pellets, which
contains about 5 wt% silicate gangue phases, it was
found that they swell and crack without evidence of
iron whiskers formation.12,13)
In designing the experiments alkalis were introduced in three different ways corresponding to an
iron blast furnace with alkalis cycling:
(1) As part of the gangue phases : representing
pellets originally of high alkali content.
(2) As alkali carbonate fine crystals near the pellets outer surface : representing pellets with precipitated alkali compounds, as a result of condensation
and chemical reactions involving alkali vapors in the
blast furnace stack. Alkali cyanides cannot be used
for safety reasons; however, they are assumed to behave much like alkali carbonates.5,s~
(3) As alkali metal vapors mixed with the reducing gas : representing the conditions of reduction by a
blast furnace gas contains alkali vapors.
III.

Experimental

Chemical

analysis

of iron

24,

1984

(613)

2.

ExperimentalApparatusand Procedures
The experimental apparatus consisted, mainly, of
a Marshall resistance heated tubular furnace, which
contained a movable quartz tube of 40 mm inside diameter. The components of the reducing gas mixture : CO, CO2 and N2 were supplied by separate
cylinders, with flow rates monitored by individual flow
meters, and, mixed before entering the reaction tube.
The specimens were placed in the middle zone of the
tube on top of ceramic or platinum boats, during the
reduction tests. They were moved together with
the quartz tube to the hot zone of the furnace, and
left under N0 flow for about 10 min until the isothermal temperature level was reached, which was checked using a separate thermocouple placed just above
the specimens. Then, the reducing gas mixture was
allowed into the tube and the reduction time was
counted using an automatic stopwatch. Once the reduction is over, the gas mixture was switched off, and
purified N2 was allowed to flow, while the partially
reduced pellets were taken together with the tube to
the cooler end of the Marshall furnace.
To simplify the comparison of the results, most of
the reduction test were carried out under the same
standard conditions for 1 hr using a gas mixture of
CO, CO2, and, N2 at a total flow rate of 500 cc (STP)/
min (the individual flow rates were : N2 300, CO
160, and CO2 40 cc (STP)/min, respectively), at a
temperature of 900 C, which corresponds to that of

1. Sample and Its Preparation


Most of the specimens used were commercial iron
ore pellets which had about 5 wt% silicate gangues.
The same iron ore concentrates, which was used in
manufacturing those pellets was used to prepare pellet specimens doped with alkalis. The chemical composition of the pellets and ore, as received from the
plant, is given in Table 1. It represents magnetite
iron ore deposit, and, hematite pellets, respectively,
To examine the sole effect of alkalis on the iron oxides,
the silicate gangue phases was removed from an iron
ore concentrate sample, by washing in warm concentrated hydrofluoric acid solution, followed by distilled
water, several times. The treated sample was dried
and fired in air at 1200 C for about one day, to
remove the traces of the fluoric acid and to decompose
the ferrous and ferric fluorides. After this treatment
the sample had about 0.025 wt% SiO2, according to
chemical analysis.
The specimens used in the reduction and swelling
tests can be grouped in the following way:
(1) Commercial acid iron ore pellets: were used
to test the sole effect of acid gangue phases, also were
dipped in alkali carbonate aqueous solutions followed
by drying at 150 C to result in specimens with a layer
near their outer surface which contained fine alkali
carbonate crystals, called impregnated, and, were reduced with gas mixture contained alkali metal vapors.
1.

Vol.

(2) Modified pellets: same like in (1) but were


pre-reduced to magnetite or wustite at a temperature
of 900 C or pre-reduced followed by oxidizing in air
at 800 C. This treatment was done to modify their
structure and made the pellets' body more open so
that the alkali carbonate can be distributed more uniformly when they were dipped in its solution.
(3) Laboratory pellets made of iron ore concentrate : they were made from the same iron ore concentrate; however, alkali carbonates were introduced
prior to pelletizing and firing in air for 2 hr called
doped pellets.
(4) Laboratory pellets and briquettes of high purity iron oxides : the pellets were made of chemical
reagent hematite (with a purity of 99.8 % Fe2O3) but
no alkalis were introduced before pelletizing and, firing in air, then were dipped in alkali carbonate solutions followed by drying at 150 C to result in high
purity pellets with a layer near their outer surface,
which contained alkalis. The briquettes were made
from the iron ore sample after gangue removal, and
mixing with alkali carbonate, followed by briquetting
and firing in air for 2 hr.

Work

Table

ISIJ,

ore

concentrate

and

commercial

pellets.

Research

Article

( 614)

Transactions

ISIJ, Vol. 24,

1984

the thermal
and chemical
reserves zone of some blast
furnaces.14)
Weighing
and volume measurement
by a mercury
displacement
were carried
out before and after the
reduction
experiments
to obtain the extents of reduction and swelling.
The degree of reduction
was calculated based on the initial oxygen content with the
iron oxides, from chemical
analysis, and, the amount
of oxygen taken during
reduction.
The swelling index is defined
as the percentage
increase
of the apparent volume of the specimen
after reduction.
some occasions, due to the effect of alkalis, cracks
fissures were developed
during
reduction.
The

On
and
ap-

parent volumes of the reduced specimens


measured
by
mercury
displacement
did not include
that of the fissures with a width larger than about 10 to 12 ,u, due
to the effect of the mercury
surface tension.
Accordingly, the given swelling index may be smaller than
that if we include the fissures' volume.
Unfortunately, it was rather
difficult
to measure
the volume
of
those macro-fissures
and cracks; hence the results will
be presented
without
adding
their volume.
It was
found that their size and distribution
is a function
of
the amount and distribution
of alkali in the partially
reduced
pellets which contained
silicate ,gangues.
For the reduction
tests performed
with reducing
gas contained
alkali vapors,
they were generated
by
passing
the gas mixture
over alkali carbonate
melt,
which was placed in a separate
platinum
crucible.
It
was located
prior to where the specimen
was positioned in the upstream
in the direction of the gas flow.
The partial
pressure
of the alklai metal
vapor was
measured
indirectly,
by weighing
the alkali carbonate's crucible
before and after gasification,
and assuming the rate of generating
the alklai vapors was
constant
during the test.
To protect the quartz tube
from the alkali vapors,
a mild steel tube, of about
38 mm outside diameter,
and 35 mm inside, was used.
It was placed inside the quartz
tube and carried the
tested materials,
during the reduction
experiments.
A scanning electron microscope
SEM, and, an electron probe microanalyzer
were used to examine
the
topography
of the grains,
and, the formation
and
movement
of the liquid slag during reduction.

Table

Research

Article

2.

Standard

reduction

Iv.

Experimental Results
As received commercial acid pellets, which had a
porosity of about 25 %, swelled normally (about
18 %) after the standard reduction test. SEM observation indicated no iron whiskers formation, also,
the pellets did not show a tendency to form macrocracks or fissures during reduction. Of course, care
was exercised not to choose pellets with defects orr irregularities, which may arise during their industrial
making, and, effect the results, for any of the tests in
this work.
To study the effect of gangue, the pellets made
from chemical reagent hematite were tested. Those
of a porosity of about 20 % disintegrated during reduction, but more compact specimens of porosity less
than 10 % swelled normally, after the standard reduction test.
1.

The SoleEffect of Alkalis


The results of the standard reduction tests are given
in Table 2. The two groups of specimens represents
high purity hematite samples which were almost free
from the silicate gangue phases. However, the method of introducing the alkalis was not the same in each
group.
Alkalis did not seem to effect the reduction behavior
of the high purity impregnated pellets. Neither the
swelling index nor the degree of reduction were influence, in a noticeable way, by the alkalis, which
were introduced after making the pellets, without subsequent firing.
The alkali doped briquettes, which had a porosity
comparable to the high purity pellets (about 10 %),
behaved in a different way. They swell abnormally
and SEM examination indicated the formation of iron
whiskers. The swelling index and the degree of reduction were also influenced by the firing temperature, prior to reduction, and increased with elevating
the induration temperature from 1 200 to 1 300 C,
under otherwise identical conditions.
It was more appropriate to use the same material
(the sample of iron ore concentrate, which was treated
for gangue removal) in making both types of the high
purity specimens. However, the amount of the mateof gangue

free

specimens.

Transactions

rials, left after the treatment, was not enough to make


the necessary number of specimens.
2.

The CombinedEffectof Acid Gangueand Alkalis


In Fig. 1, there is two groups of laboratory pellets,
which are represented by two curves. Those made
without adding alkali carbonate prior to pelletizing
and firing seemed to have better resistance to swelling
which was increased by elevating the firing and induration temperature in air, prior to reduction. On
the other hand, alkali doped pellets which were fired
at 1 200 and 1 300 C, seemed to swell much more,
also, they developed hairline cracks. With the pellets
fired at 1 100 C, the alkali doped specimens had
slightly lower swelling index; however, both types
showed a tendency of cracking and sometimes split
into pieces, after reduction. Figure 2 illustrates the
effect of alkalis on increasing the swelling index and
the degree of reduction of alkali doped pellets which
was fired and indurate in air at 1 200 C before reduction. The extent of swelling and reduction increased
with the alkalis.
The results of the standard reduction test using impregnated commercial pellets are given in Fig. 3. To
illustrate that the observed swelling and crack formation were independent of the step of metallization,
which may cause abnormal swelling due to iron
whiskers formation (as in the case of alkali doped

Effect
tory

Fig . 2.

of firing

and

alkalis

on the

swelling

Fig.

3.

alkalis

and
prior

reduction
to standard

of acid
reduction.

pellets

with

and

reduction

dipped

drying

24,

(615)

1984

in alkali

and

standard

Alkali impregnated

of commercial
carbonate

pellets

solution

which

followed

by

reduction.

commercial

pellets reduced

with

gas mixtures of different


reducing
power which
shows the swelling is almost independent
of the
metallization
of wustite and the formation of iron.

of 1abora-

doped

Swelling
were

pellets.

Swelling

Vol.

briquettes in SectionIV. 1), impregnated pellets with


almost the same amount of alkalis were examined and
reduced with reducing gas mixtures with different reducing power, but under otherwise similar conditions.
The results are given in Fig. 4. The swelling index
seemed to be independent of reducing power, as far
as the gas mixture was able to reduce the hematite to
magnetite or wustite.
The experimental results given in Fig. 5, shows the

Fig. 4.
Fig . 1.

ISIJ,

Fig.

5.

Swelling
duced

and
with

reduction
gas

mixture

of commercial
containing

pellets,

re-

alkali vapors.

Research

Article

(616)

Transactions

ISIJ, Vol.

24,

1964

effect of alkali metal vapors. The extent of swelling


and reduction increased with increasing the partial
pressure of the potassium with the reducing gas mixture, under similar conditions. The Marshall furnace
used to carry the tests had a three zone windings to
adjust the temperature profile, which were employed
so that the gasification zone (where the alkali vapors
were generated) can be adjusted to have a temperature level which may be equal or higher than that of
the standard reduction of 900 C. The three gasification temperatures, which corresponded to the three
partial pressures of potassium given in Fig. 5, were
900, 1000, and, 1 100 C, respectively.
The partially reduced pellets, in the three mentioned cases, were examined by the SEM, and, there
was no evidence that the influence of alkalis on the
reduction behavior of commercial pellets is directly
related to iron whiskers formation. Photograph 1 is
given to compare the appearance of the specimens
under SEM after standard reduction, they represent
commercial pellets as well as high purity specimens.
Photographs 1(A) and (B) were taken after standard reduction of alkali impregnated commercial
pellet (with 1.32 wt% K2O, before reduction). They
represent two different regions of the partially reduced
specimen. Photograph 1(A) represents an area on
the outer surface zone where one of the hairline cracks
started, while Photo. 1(B) represents another interior
zone where the same observed crack was about to end.
The pellet's grains near or on the surface were bigger
in size and it could be observed, from their shape and

Photo.

Research

Article

1.

Some

SEM

photographs

of partially

topography, that they were split, most probably due


to the movement of slag, which was seen to concentrate in certain areas and formed small pockets within
the grains. In the second interior zone, the grains
were much smaller, with rounded edges, and, almost
with a continuous layer of slag around them, in a pattern which may indicate that in this region the slag
phase and the iron oxide particles were agglomerated
together in a manner which made it more difficult for
the interior grains to be reduced further.
Photographs 1(C) and (D) represent the appearance of the partially reduced high purity specimens
under SEM (Table 2). The examination showed no
indication of liquid slag formation. Photograph 1(C)
represents a typical view of partially reduced alkali
impregnated pellets, there was no strong evidence of
iron whiskers formation, while Photo. 1(D) shows the
only case in this work where iron whiskers were observed under SEM with alkali doped briquettes.
The two regions of the partially reduced impregnated pellets which are represented by Photos. 1(A)
and (B) were examined further by the X-ray unit of
the SEM. In both cases it was observed that the slag
areas were high in Si and K. Photograph 2 gives the
X-ray mapping for K, Si, and Fe. The only difference between the two regions was in the distribution
of the slag which was more uniformly distributed between the grains in the interior zone.
In order to understand better the mechanism of the
slag formation and its movement the following two
experiments were carried out using commercial pel-

reduced

s pecimens

after

standard

reduction.

Transactions

lets:
(1) A cylindrical specimen of about 1 g was made
by cutting and shaping a commercial pellet. It was
placed on top of about 0.1 g potassium carbonate, in
the upright position. The specimen was then reduced
using the standard reduction procedures.
(2) A commercial pellet was left, immersed in
molten potassium carbonate, under flowing N2 for
1 hr at 900 C.
The two specimens were examined using an electron probe microanalyzer (with electron beam of
about 30 pm in diameter), and the distributions of K,
Si, and Fe were plotted. The results given in Fig. 6
indicate that although the pellet was completely immersed in molten potassium carbonate under flowing
N2 for 1 hr there was relatively poor penetration.
Near the surface the silica was moved inward with the
potassium to a distance of about 200 dim (region AB
in Fig. 6). Beyond this region where silica was depreciated there was another region BC, of about 900
im where Si and K, appeared to be in phase. Further inward, the penetration of potassium nearly stopped, and larger peaks of Si, representing the silicate
gangue phases was observed in the pellets interior.
The reproducibility of these results was checked in
three different locations which showed almost the same
profiles for K, Si, and Fe. It was confirmed that the
potassium carbonate was in its molten state by its appearance in the platinum crucible after cooling the
specimen to room temperature.
In the experiment with the specimen which was
left to be reduced on top of potassium carbonate melt,
all the potassium carbonate was taken up by the speci-

Photo.

2.

SEM
shows

photographs
clearly

and
that

the

X-ray
slag

mapping
areas

are

of Fe,
high

ISIJ,

Vol. 24,

1984

(617)

men's body, as shown in Fig. 7, which also illustrates


that the distributions
of K and Si were almost identical ` and uniformly
dispersed
throughout
the specimen's body from the bottom to the top.
No peaks of
Si could be observed,
which may give an indication
that almost all the gangue phases were reacted with
potassium
to form the slag.
Also, it seems that the
movement
of the slag occurs against the gravity force,
most probably
due to capillary
force, which results
from the liquid slag surface tension.
Modified
commercial
pellets, which were restructured by prereduction
or prereduction
followed
by
oxidation
at 800 C and were relatively
more porous
than as received pellets, showed a weaker tendency
to
swell when reduced
using the standard
reduction
procedures, as given in Table 3.
To demonstrate
that
the observed
swelling
and
cracking did not occur during the subsequent
cooling,
after reduction.
An alkali impregnated
commercial
pellet was photographed,
while was : heated,
reduced,
and, during cooling.
The results are given in Photo.
3 which shows clearly that the swelling and cracking
occurred
during the standard
reduction
test, and remained during cooling or heating
under N2 flow.
V.

Discussion

After firing
their strength

and induration
the iron ore pellets owe
to the formation
of hematite,
and,

gangue bridges which find the iron oxide grains together.


During
reduction
once the hematite
bridges
are transferred
to magnetite
the structure
as a whole
is relaxed
and the pellets
swell.
Normal
swelling
occurs
almost
to all kinds of hematite
pellets and

Si, and

in both

Si and

after

reduction

of

alkali

impregnated

pellets.

It

K.

Research

Article

(618)

Transactions

Fig.

Is",

6.

Fig . 7.

gue

which

Distribution

pellets

Research

Article

was

contains

phases.15-17)
The experimental

oratory

24,

1984

Distribution of K, Si, and Fe near the pellet surface after heated under N2
no reduction) but was immersed totally in molten K,C03 during testing.

which

compacts,

Vol.

made

results
without

of K,
placed

Si, and
on

reasonable
given
alkali

the

Fe

across

top of K2C03

amount
in Fig.
addition,

the

center
melt

of gan-

1 with

lab-

showed

line

during

after

standard

reduction

(300 CC STP/min

flow

of

acid

pellet

and

specimen

reduction.

that the swelling


firing temperature.
pellets which
to the acid

index decreased
with elevating
the
On the other hand, high purity

had structure
and overall porosity closer
pellets
tended
to disintegrate
as ex-

Transactions

Table

Photo.

3.

taking

Photographs
us, prior
duction

to standard
and

3.

Standard

reduction

of

modified

to the cooling

pected.15,16) It may be claimed that high purity


specimens, which were prepared without alkali addition, and were more consolidated with a lower porosity (less than 10 %) swelled almost normally, because
there was still a hard core of unreduced hematite within the specimens after reduction for 1 hr.
I t seems the silicate gangue bridges between the
iron oxide grains can bear with the stresses which may
arise during magnetite to hematite transformation.
Of course their effect in helping the pellets to resist
degradation depends on their relative amount and
distribution inside the pellets as well as the firing
conditions. The presence of about 5 wt% gangues is
said to be sufficient for ordinary ore pellets.l5>
The results with high purity specimens, which was
almost free from silica, suggest high temperature firing
for sufficient time is needed to cause iron whiskers
formation as illustrated in Table 2 and Photos. 1(C)
and (D). It must be pointed out, the effect of high
temperature firing on iron whiskers formation is opposite compare to the case when lime was added to
high purity iron oxides.18 It may be assumed, the
decomposition of alkali ferrites, which may form during induration and firing, and the incorporation of
alkalis, as solid solution in wustite, in irregular pattern, is the reason for iron whiskers formation.

and heating

Vol.

24,

1984

(619)

pellets.

during heating, reduction,


and cooling of commercial
reduction
test.
It is clear the swelling and cracking

is not very sensitive

ISIJ,

procedures

pellet, impregnated
occur during early

under

with alkastage of re-

N2.

The experiments
carried out to simulate
the influence of alkali on the reduction
behavior
of acid pellets
showed, clearly, that the formation
of iron whiskers is
not the cause of their degradation.
Visual observation, and, SEM examination
of the partially
reduced
specimens,
showed the pellets to swell, and to form
macrocracks
and fissures as result of alkalis attack.
The results given in Figs. 1 to 3, and 5 suggest that
the extent of swelling and reduction
to increase with
alkalis.
Also, Fig. 4 supports
the idea that the observed swelling
is independent
of the metallization
step.
The correlation
between
the existence of liquid
slag high in both alkalis and silica and the formation
of cracks is supported
by SEM and electron
probe
microanalyzer
examinations.
The first step of reduction
from hematite
to magnetite seems to be necessary
for the movement
and
progress of the alkali contained
slag inside the pellets.
This important
piece of information
can be seen if the
results in Figs. 6 and 7 are compared.
Reduction
and liquid formation
are necessary
conditions,
however, not sufficient.
Uneven
distribution
of alkali
liquid slag inside the pellets is needed.
The results
given in Table 3, with alkali impregnated
modified
pellets, indicated
no big change in swelling and cracking, when alkalis were relatively
more uniformly
dis-

Research

Article

(620)

Transactions

ISIJ,

Vol.

24,

1984

tributed.
During drying of green wares in ceramic industry
different shrinkage rates, due to uneven moisture content between the surface, and the interior parts can
cause cracking. The case under investigation may be
close but in the reverse direction. During reduction
liquid alkali silicate slag is formed near the surface,
and moves inward most probably due to capillary
forces. However, at any moment its distribution is
not uniform. The uneven distribution can be understood in the following way : the fine crystals of alkali
carbonates were concentrated near the surface when
reduction started (alkali impregnated pellets) ; when
they were added prior to pelletizing and firing, alkalis
were more or less distributed uniformly, however, reduction started near the surface; therefore, the distribution of liquid slag was also not uniform (alkali
doped pellets) ; with reduction using alkali contained
gas, both the reduction and the slagging processes occurred at the gas/solid contacts near the surface, in a
way very similar to the case with impregnated pellets.
The slagging process near the surface started the cracks
(Photo. 1(A)), while its movement, and, in later stage,
its aggregation with the iron oxide grains, facilitated
their propagation (Photo. 1(B)).
Testing commercial acid pellets in the laboratory
showed they swell and crack due to alkalis, the important question is whether it happens inside the blast
furnace. If alkali contents are high in the ore, which
is of ordinary composition, at the moment when ferrous oxide becomes available, cracking and swelling
are expected. Also, when the first stages of reduction
are taking place using blast furnace gas containing
alkalis : ferrous oxide and alkali oxides are formed
simultaneously at the gas/solid interface creating favorable conditions for cracking and swelling. However, these two cases are not quite normal. It is reasonable to assume the conversion of hematite to
magnetite, for the locations near the pellets surface, is
completed before they arrive to the higher alkali zones
in the blast furnace. Swelling and cracking are not
expected to cause serious problems if the alkalis are
introduced after prereduction as illustrated in Table
3. However, since the pellets contain relatively large
quantities of liquid phase due to the formation of alkali
iron silicate slag, and are subject to compression load,
they are expected to soften during reduction inside the
blast furnace.
In most of the experiments carried out, the alkalis
to silica ratio was varied; however, the initial gangue
was kept the same. Figure 8 is given to illustrate the
case when the silica content of the ore is modulated.
It was calculated based on equilibrium between wustite, silica, and potassium oxide at 900 C. Of course
the formation of liquid phase and its relative amount
depend on kinetic factors; however, the informations
given in Fig. 8 suggest the amount of liquid slag to be
sensitive to the absolute amount of the acid gangues.
Therefore acid pellets subject to the same level of
alkali cycling are expected to have better resistance
when the relative amount of gangues exceeds certain
limit. The silica rich gangues may have a dual effect.
Research

Article

Fig.

8.

Formation
based
Fe0

on

of
the

in presence

alkali

iron

equilibrium

silicate
between

slag

at

900

Si0 2, K20,

C,
and

of iron.

Besides its effect on liquid slag formation,


the remaining silicate binding bridges, which are formed by adequate
firing, are roughly
proportional
to the silica
content
of the pellets and would
be responsible
for
absorbing
the stresses which cause degradation.
The result of this work showed that alkalis increase
the reducibility
of acid pellets.
It may be attributed
to their influence
on swelling,
and crack formation,
which enhance
the mass transfer
of the reducing
gas
between the iron oxide grains.
Also, it had been suggested alkalis and lime to help the metallization
step,
due to their effect on the nucleation
and growth
of
iron, when they are incorporated,
as solid solutions,
with wustite.10,11,18-20)
The second claim may not
have a strong effect in our case, since SEM, and, electron probe microanalyzer
examinations
showed
the
alkali to be more concentrated
with the slag.
However, a small amount
of alkali, which was not detected,
may exist with the iron oxides and effect the metallization
process.
VI.

Conclusions

As a general mechanism
formation
of iron whiskers
cannot
adequately
be accepted
to explain
all the
causes of degradation
of iron ore pellets due to circulation
and accumulation
of alkalis in the blast furnace.
It seems the alkalis react with the silica rich
gangue and ferrous iron oxide to form primary
liquid
slag, at relatively
lower temperature
levels.
Its formation and movement
between
the iron oxide grains
cause the binding phase to disintegrate
during reduction.
When acid pellets were tested in the laboratory,
and the alkalis were introduced
in three ways relevant
to alkalis' cycling in the blast furnace,
they showed a
tendency
to swell and crack due to formation
of unevenly distributed
liquid slag coupled by hematite
to

Transactions

magnetite
phase transformation.
Formation
of liquid
alkali iron silicates lowers the pellets resistance
and
the phase transformation
is likely to create stresses
which
may act as a driving force for cracking
and
swelling.
Inside
the blast furnace
the reduction
process
is
accompanied
by compression
load.
Therefore,
the
formation
of liquid alkali iron silicate slgas is expected
to weaken the pellets.
However,
they are not likely
to swell, and crack because the first reduction
steps to
magnetite
usually takes place in the upper parts of the
furnace
stack, where the extent of alkalis is a minimum.
Softening
may be the real cause of degradation when relatively
large
quantity
of liquids
are
formed due to alkalis.
Of course, the pellets resistance is a function of both the alkalis and gangue quantities, under
similar reduction
conditions.
Minimization of alkalis, which are cycling in the blast furnace,
is always desirable.
Also, there is a minimum
amount
of gangue
contents
to ensure smooth
operation
and
optimum
behavior,
which in turns is a function
of
alkalis to silica ratio during reduction.
Acknowledgements

3)
4)
5)
6)
7)

8)
9)
10)
11)
12)

13)
14)
15)

The author thanks Prof. W-K Lu, Stelco


Department
of Metallurgy
and Materials
McMaster
University,
Hamilton,
Ontario,
for his useful suggestions
and discussion.

Professor,
Science,
Canada,

REFERENCES
1) F. D. Richardson and J.H.E. Jeffes: JISI,163 (1949), 397.
2) "Alkalis in Blast Furnaces ", 1st McMaster Symposium on

16)
17)
18)
19)
20)

ISIJ,

Vol.

24,

1984

(621)

Iron and Steelmaking, edd by N. Standish and W-K. Lu,


McMaster University Press, Hamilton, (1973).
"A session on Blast Furnace Operating Problems with High
Alkali Burdens ", RIME Ironmaking Proc, 35 (1976).
J. Davies, J. T. Moon and F. B. Traice: Ironmaking Steelma/ring, 5 (1978), 151.
K. P. May and W-K Lu: AIME Ironmaking Proc, 37
(1978), 582.
T. ElKasabgy : Metallurgia ABM, 36 (1980), 228.
J. M. Steiler: "A Thermodynamic Study of Liquid K20Si02 and K20-CaO-Si02-A1,03
MgO Blast Furnace Type
Slags ", Proc. of PCS, Versailles, (1978).
Phase Diagrams for Ceramists, ed, by American Ceram.
Soc., (1969), supplement.
T. U. Fuwa and S. Ban-ya:
Trans. ISIJ, 9 (1969), 137.
R. Nicolle and A. Rist: Trans. AIME, lOB (1979), 429.
W-K. Lu : Scand. J. Met., 2 (1973), 65 and 169.
D.W.R. George and J. A. Peart: "The Influence of Alkalis
on Blast Furnace Performance," 1st McMaster Symposium
on Iron and Steelmaking, McMaster University Press,
Hamilton, (1973). Ref. 2) paper 4.
T. Elkasabgy and W-K. Lu : AIME Ironmaking Proc., 36
(1977), 2.
Blast Furnace Theory and Practice, ed, by J. H. Strassburger, 2 (1969), 737.
R. D. Walker, N. S. Ford and D. L. Carpenter:
Trans.
ISIJ, 11 (1971), 473.
H. B. Edwards, HEN. Stone and B. L. Daniell: JISI, 207
(1969), 1565.
R. L. Bleifuss : Trans. ISIJ, 11 (1971), 51.
T. Elkasabgy and W-K. Lu: Trans. AIME,11B (1980), 409.
S. E. Khalafalla and P. L. Weston:
Trans. Met. Soc. AIME,
239 (1967), 1494.
R. Piepenbrock, K. Koch and G. Tromel:
Arch. Eisenhuttenw., 47 (1976), 141.

Research

Article

Potrebbero piacerti anche