3rd AMIREG International Conference (2009): Assessing the Footprint of 433

Resource Utilization and Hazardous Waste Management, Athens, Greece

Pelletization and reductive smelting of Greek bauxite residues for iron

production

A. Xenidis, C. Zografidis and I. Kotsis

School of Mining and Metallurgical Engineering, National Technical University of Athens, Greece

D. Boufounos
Aluminium of Greece SA, Agios Nikolaos, Viotia, Greece

ABSTRACT stituents has been a matter of scientific interest

The pelletization and reductive smelting of
Greek bauxite residues to produce iron was in-
vestigated. Bauxite residues pelletization was
performed without the addition of any binding
agent or using Ordinary Portland Cement or
bentonite as binders. The pellets produced were
subjected to green drop, green strength and dry
strength tests, in order to determine the compli-
ance of their mechanical properties with the iron
ore industry requirements. Moreover, mixtures
of fine-grained Greek bauxite residue and lig-
nite were subjected to smelting reduction. The
effect of parameters such as the smelting tem-
perature, the amount of the reducing agent in the
mixture, the retention time and the addition of
fluxes on the quality of the metallic product, as
well as the basicity and the desufulfurization
capability of the slag, were investigated. The
obtained results regarding both the mechanical
characteristics of the pellets and the chemical
properties of the metallic iron produced were
very promising providing input for further re-
search on the optimization of the proposed
method.
1. INTRODUCTION
Bauxite residue, also known as alumina red
mud, is the by-product of the Bayer process
which is universally applied for the refining of
bauxite to produce smelting grade alumina and
then aluminium. In general, each tonne of alu-
mina results in the production of approximately
0.5-2 tonnes (dry wt.) of bauxite residue. The
application of various processing methods for
the effective recovery of its major metallic con-
by numerous researchers (Paramguru et al.,
2005). The iron content of bauxite residue de-
pends on the quality of the bauxite ore treated,
and if expressed as iron (III) oxide it ranges
from 30 to 50%. Thus, much attention has been
attended towards its recovery. The current work
constitutes a part of an integrated research effort
which aims to investigate alternative ways of
utilizing bauxite residue as a raw material in the
iron ore industry, by applying pyrometallurgical
treatment methods. Within this framework,
roasting reduction and magnetic separation for
iron enrichment of the Greek bauxite residue
have already been investigated (Xenidis et al.,
2009). This study is focused on the pelletization
and reductive smelting of the same raw material
for iron production.
Reduction smelting constitutes the most sig-
nificant pyrometallurgical process for the pro-
duction of cast iron. It is mainly conducted in
blast furnaces or electrical furnaces. Important
parameters affecting the economics and produc-
tivity of such a process are the electrical energy
needs as well as the reducing agent consump-
tion, which is usually a solid fuel (coal, coke or
graphite). Many efforts of smelting reduction of
bauxite residue material have been reported
(Visnyovsky, 1971; Eremin, 1969; Zazubin
et al., 1967). The most often reducing agent util-
ized is coke while the finally obtained iron re-
covery is always over 90%.
Pelletization is a typical process for the ag-
glomeration of fine grained raw materials in
iron metallurgy, in order to obtain constant ore
grain size of the feed, absence of very fine
grained material resulting in dust production,
and sufficient strength in order to withstand the
3rd AMIREG International Conference (2009): Assessing the Footprint of 434
Resource Utilization and Hazardous Waste Management, Athens, Greece

transportation and handling steps during the given in Table 1. The total iron (Fetot) content in
treatment in blast furnaces or rotary kilns with- the bauxite residue is about 33.5%, which is in-
out degradation. Bentonite and Ordinary Port- dicative of the fact that it constitutes an impor-
land Cement (OPC) are the most often industri- tant potential iron source for the iron ore indus-
ally used binding agents. The binding mecha- try. Its main mineral constituents, as determined
nism of bentonite has been ascribed either to the by mineralogical analysis conducted by X-Ray
dispersion mechanism, i.e. dispersing into indi- diffraction, were hematite, gibbsite, diaspore,
vidual platelets and coating of the surface of the calcite and hydroxysodalite.
ore particles like a gel (Kawatra and Ripke, Pellets were produced in a concrete mixer.
2002, 2003; Forsmo et al., 2006), or to the fiber The methodology followed for pellets produc-
mechanism, i.e. under proper conditions, a fi- tion include:
brous matrix is formed through the bentonite i) mixing of the dried bauxite residue with the
addition that ties the iron ore particles together. binding agent to produce a feed of approxi-
The binding efficiency of cement is ascribed to mately 2-3 kg for the production of pellets.
the formation of constant hydrated mineralogi- Portland cement (10% by weight on a dry ba-
cal phases by the addition of water, almost in- sis) and bentonite (0.3, 0.6 and 1.0% by
soluble in the water, providing significant co- weight on a dry basis) were used as binding
herence between the ore particles. Pelletization agents, the chemical analysis of which can be
of fine materials has been an object of investiga- also seen in Table 1. The determination of
tion by quite many researchers (Logomerac, ore to binding agent ratio was based on the
1963, 1974; Srb and Ruzickova, 1988). industrial pelletizing practice.
ii) production of pellet seeds (pellets with a di-
2. EXPERIMENTAL ameter of -6.3+3.327 mm), by scattering
about 0.5 kg of the pulverized mixture inside
2.1 Pelletization the rotating apparatus. About 200-300 g of
The sample of bauxite residue utilized in the the produced seeds were put back in the ap-
present study was provided by the Greek alu- paratus.
mina refinery owned by the Aluminum of iii)forming of green pellets, through mixing of
Greece S.A. It is a fine grained material, given the raw materials and continuous addition of
that a percentage of 40% by weight is less than water in the same apparatus.
2 μm. Chemical analysis of bauxite residue is The fraction of the green pellets used for the
determination of its physicomechanical proper-
Table 1: Chemical analysis of Greek bauxite residue and
binding agents used for pelletization. ties was -12.5 + 9.5 mm, based on the require-
Bauxite Portland ments of the standard test method (ISO4700:
Component Bentonite 1996(E)). Apart from testing the mechanical
residue Cement
Fe2O3 48 4.44 29.09 strength of the pellets formed by the addition of
Na2O 3.26 2.83 2.67 binding agent, the same properties of pellets
SiO2 6.96 53.81 formed without any binding agent addition, was
CaO 14.84 4.91 0.5 also tested, due to the fact that the content of
MgO 0.24 3.39 2.9 clay minerals in bauxite residue can be regarded
Al2O3 15.85 19.11 as high enough to justify such a series of tests.
K2O 0.07 0.69 The tests conducted were:
TiO2 7.06 0.79 i) Green drop number: Each single freshly
CO2 2.24 3.78 made (undried) pellet is dropped repeatedly
SO2 0.78 from a height of 18 in. onto a steel plate. The
L.O.I. 11.37 number of drops required for fracture is re-
3CaΟ.SiO2 45
corded. This is repeated for 20 pellets, and
2CaO.SiO2 27
the results are averaged. This test measures
3CaO.Al2O3 11
the ability of the green pellet to remain intact
4CaO.Al2O3.Fe2O3 8
CaSO4 3.1
during handling.
3rd AMIREG International Conference (2009): Assessing the Footprint of 435
Resource Utilization and Hazardous Waste Management, Athens, Greece

ii) Dry compression strength: Pellets are dried at
105 oC for at least 24 h, and single pellets are
then compressed using a compression test
machine, with an almost constant rate of 10
lb/sec. The load required to fracture the pellet
is recorded. This test measures the ability of
dried pellets to survive handling during the
firing process.
The results are presented in Figures 1 and 2.
It is noted that the moisture content of the green
pellets in every case was fluctuating among 19.6
and 21.2%, while the moisture content of the
pellets of iron ores is usually 8-10%. This is
60
50
40
Drop number
probably ascribed to the high content of the hy-
droscopic Al2O3 (16% approximately) and clay
minerals in bauxite residue.
Taking into consideration that the plasticity
and therefore the green drop strength of the pel-
lets increase linearly with the moisture content
of the pellets (Forsmo et al., 2006), it is deduced
from Figure 2 that the green drop strength of the
bauxite residue pellets is very satisfactory. Al-
though the green drop number regarding iron
ore pellets with bentonite, rarely exceeds 15-20,
it comes up to 50 when 1% by weight of ben-
tonite is added. It is also noted that increase of
the bentonite addition from 0.3 to 1% favors the
strength of the pellets. Portland cement addition
has the most negative effect on the obtained
green drop number, although the quantity added
(10% by weight) was significantly higher than
that of bentonite (0.3-1% by weight).

The conclusion deduced by the dry compres-

30
sion strength of the pellets (Fig. 2) is that the
obtained values are satisfactory as well, since
20
the minimum dry compression strength value of
pellets required by the iron ore industry is
10
5 lb/pellet. Even regarding the worst result ob-
tained for the addition of 0.3% by weight ben-
0
tonite, the dry compression strength value was
No binding Bentonite Bentonite Bentonite OPC
agent (0.3%) (0.6%) (1%) (10%)
approximately 6 lb/pellet. Increase of bentonite
addition favors the obtained strength values,
Addition of Binding Agent (% by weight of the though its effect is almost negligible among 0.6
mixture)
and 1% addition by weight.
Figure 1: Effect of the type and the quantity of the bind- Moreover, the result obtained by the addition
ing agent on Green Drop strength of the bauxite residue
pellets. of Portland cement was the best compared to
10 any addition of bentonite, contrary to the inves-
9
tigation of green drop number, where the oppo-
site conclusion was deduced. This is ascribed to
8
the fact that a different mechanism is responsi-
Compressive strength (lb)

7 ble for deformation of the pellet that takes place

6 in each case. Thus, in case of wet drop number
5 determination, the strength of the pellet is main-
4
ly dependent on the force of the liquid capillary
‘liquid bridge’ formed between the particles.
3
This means that the addition of bentonite, which
2 has higher ability than cement to absorb water,
1 leads to a more considerable increase of the
0 number of liquid bridges between particles and
No binding Bentonite Bentonite Bentonite OPC thus the dropping damage resistance is in-
agent (0.3%) (0.6%) (1%) (10%) creased. On the contrary, the property of dry
Addition of Binding Agent (% by weight of the
compression strength also depends on the pres-
mixture) ence of very fine particles which restrict the
Figure 2: Effect of the type and the quantity of the bind- layer movement inside the pellet and as a result
ing agent on Dry Strength of the bauxite residue pellets. of it the resistance to compression force is in-
3rd AMIREG International Conference (2009): Assessing the Footprint of
Resource Utilization and Hazardous Waste Management, Athens, Greece
creased. Probably, the increased amount of very
fine particles contained in cement in comparison
with bentonite, enhances the compression
strength of the pellets.
In any case it is very important the fact that
even without adding any binding agent, the ob-
tained physicomechanical properties of the
bauxite residue pellets are very satisfactory.
This is probably ascribed to the high content of
clay minerals in bauxite residue. Thus, taking
for granted the high cost mainly of bentonite, it
would be very interesting for the economics of
an industrial agglomeration process to be able to
obtain pellets without addition of binders, hav-
ing almost the same physicomechanical proper-
ties with those obtained after the addition of
binders.
2.2 Smelting reduction
A second series of experiments was conducted
including smelting reduction of the Greek baux-
ite residue with solid reducing agents. The ob-
jective was to directly produce a metallic prod-
uct, the chemical properties of which are as
close as possible to those of cast iron. Lignite
(Cfix = 30.1%, ash = 20.1% and volatile matter =
49.75%) and coke (Cfix = 81.7%, ash = 5.0%
and volatile matter = 13.3%) were used as re-
ducing agents. Moreover, a single experiment
was conducted regarding smelting reduction of
bauxite residue in form of pellets with lignite
(bauxite residue/lignite: 1/3 by weight and ben-
tonite: 0.6% by weight as a binding agent).
The melting temperature of the bauxite resi-
due was initially determined with the LECO AF
600 apparatus, by employing the Seger cones
technique, based on the determination of four
characteristic temperatures: start of softening,
end of softening, start of melting and end of
melting points. The melting point of bauxite re-
sidue was determined to be 1400 oC, as a mean
value of five values. The determined value is in
agreement with the theoretically calculated val-
ue by the ternary phase diagram of FeO - CaO -
Al2O3 (the Fe2O3 content of the bauxite residue
is stoichiometrically converted to FeO content).
The smelting reduction experiments were
conducted in two types of laboratory furnaces:
i) Electric resistance heating furnace Tamman
(60 KVA, Tmax: 2500 oC): its main part is a
cylindrical heating tube made of carbon ma-
436
terial, surrounded by a thermal insulating ma-
terial, so that the inside of the heating tube
acts as a heat treatment chamber, in which
the graphite sample holder is disposed. The
temperature was recorded by a Raytek opti-
cal sensor pyrometer (laser).
ii) Induction smelting furnace: This consists of
an induction heating system, a water cooled
induction coil, a supporting body having a
suitable circumference for supporting a gra-
phite crucible. The temperature during the
experimental procedure was measured by the
same Raytek optical sensor pyrometer.
The sample fed in both furnaces was 70-80 g.
It is noted that two types of furnaces were used
for the conduction of smelting reduction of ex-
periments, due to the fact that each one has sev-
eral advantages and disadvantages. It is noted
that the determination of temperature was more
precise in the experiments conducted in the
Tamman furnace, however, the duration of each
experiment was more than 4 hours in Tamman
and only half an hour in the induction smelting
furnace. Nevertheless, due to the intense eddy-
ing of the material resulting from the Foucault
current, there was loss of fine material and
therefore, it was difficult to conduct mass bal-
ance calculations.
The effect of the following parameters on the
final result of smelting reduction was investi-
gated: temperature, retention time of the smelted
material, bauxite residue/lignite ratio and the
addition of fluxes. The determination of the
bauxite residue/lignite ratio was determined
based on the stoichiometrically required quan-
tity of carbon required for the complete reduc-
tion of Fe2O3 to metallic iron. Moreover, the
addition of fluxes (CaO and SiO2 reagents), was
based on the theoretical determination of the de-
sirable lower melting point, according to the
ternary phase diagram of Al2O3-CaO-SiO2, tak-
ing into consideration that these oxides are the
three main constituents of the bauxite residue
slag. Thus, the addition of three different pairs
of CaO and SiO2 was investigated per 100 g of
the mixture bauxite residue/lignite (4/1 by
weight): i) SiO2-CaO: 20-0 g, ii) SiO2-CaO: 35-
5 g, iii) SiO2-CaO: 62-38 g.
The experimental conditions of reduction
smelting as well as the chemical analysis of the
metallic products, are presented in Table 2. The
3rd AMIREG International Conference (2009): Assessing the Footprint of 437
Resource Utilization and Hazardous Waste Management, Athens, Greece

Table 2: Experimental conditions of reduction smelting experiments and chemical analysis of the metallic product
Raw Reducing BR/Reducing Fluxes Retention (%) Content in the Metal
Material Agent Agent Ratio SiO2 (g) CaO (g) Time (min) Fe S C Rest Admixtures
Pellets
- -
B.R. * Lignite 4/1 15 94.8 0.1 3.4 1.7
B.R. Lignite 4/1 35 5 15 91.3 0.7 5.4 2.6
B.R. Coke 11/1 - - 15 94.2 0.2 4.6 1.0
B.R. Lignite 4/1 20 - 15 91.7 0.4 5.3 2.6
B.R. Lignite 4/1 62 38 15 91.9 0.4 5.2 2.5
B.R. Lignite 4/1 - - 30 94.5 0.3 3.5 1.7
B.R. Lignite 4/1 - - 45 95.3 0.2 3.0 1.5
B.R. Coke 11/1 - - 30 95.0 0.1 3.2 1.6
B.R. Coke 11/1 - - 45 95.8 0.5 2.4 1.2
B.R. Lignite 3/1 - - 15 95.6 0.3 2.7 1.3
B.R. Lignite 1.5/1 - - 15 95.8 0.4 2.8 1.0
* BR: Bauxite Residue
most important conclusion deduced is that re- ducing agent and reaction time on carbon con-
gardless of the conditions employed, a metallic tent in the pig iron is given. It is seen that in-
product with iron and carbon content ranging crease of the retention time of the bauxite resi-
from 91.3-95.8% and 2.4-5.4% respectively, due melt results in a significant decrease of the
was produced. The quality of iron produced is carbon content.
satisfactory, since according to a typical analy- Moreover, after 15 minutes of smelting re-
sis of cast iron, the iron content is approxi- duction, the carbon content of the metallic
mately 94% and carbon content fluctuates be- product fluctuates between 4.6% and 5.1%,
tween 3.5% and 4.5%. In Figure 3, the effect of while after 45 minutes the respective values are
the ratio bauxite residue/lignite on mixture melt- 2.04-3.01%, in compliance with the require-
ing point is presented. It is noted that increase of ments of commercial cast iron. The use of a sol-
the aforementioned ratio results in a consider- id fuel like coke, which is much more reactive
able increase of the melting temperature, which at higher temperatures (>1000 oC) favors the re-
can be attributed to the formation of refractory duction contrary to lignite, which is much more
iron carbides. reactive and evolves its thermal energy at lower
In Figure 4 the effect of the type of the re- temperatures.
6
1800
5
1750
C content in iron (%)

1700 4
Melting point (OC)

1650
3
1600
2
1550

1500 1 Lignite
Coke
1450 0
1400 0 4020 60
Time (min)
4:1 3:1 1.5:1
Figure 4: Effect of the reducing agent and retention time
Bauxite residue/lignite ratio (wt/wt) on carbon content of the metallic product (reducing agent
Figure 3: Effect of the bauxite residue/lignite ratio on the in 50% excess of the stoichiometrically required quan-
melting point. tity).
3rd AMIREG International Conference (2009): Assessing the Footprint of 438
Resource Utilization and Hazardous Waste Management, Athens, Greece

Figure 5: Melting temperature of slag samples (A,B,C) produced by the addition of: A) SiO2-CaO: 20-0 g, B) SiO2-CaO:
35-5 g, C) SiO2-CaO: 62-38 g per 100 g of the mixture bauxite residue /lignite (4/1 by weight).

The addition of fluxes results in the alteration
of the basicity index of the slag from 4.5 to 0.6,
at an addition rate of 35 g SiO2 - 5 g CaO per
100 g of bauxite residue/lignite mixture. Due to
the reduction of slag basicity, the % sulfur con-
tent in the metallic product increases to ap-
proximately 0.7%. It is noted that the sulfur
content of a typical cast iron should not exceed
0.04% and this constitutes a crucial parameter
of its quality.
The addition of fluxes in the current work
can be characterized as successful, since the
melting temperature measured by the optical py-
rometer during the three experiments conducted
with the addition of CaO and SiO2, fluctuated
between 1400 and 1450oC. This is in agreement
with the theoretically calculated melting tem-
peratures using the ternary phase diagram seen
in Figure 5. Moreover, the slag produced by the
addition of 35 g SiO2 and 5 g CaO (Point B) per
100 g of mixture (bauxite residue/lignite), is in
the eutectic region of the slag produced by the
blast furnace operation.
The smelting reduction of pellets of Greek
bauxite residue with lignite (diameter of pellet: -
9.5+6.3 mm, bauxite residue /lignite: 4/1 by
weight) resulted in the production of a metallic
product of good quality (94.8% iron and 3.1%
carbon), something which is ascribed to the bet-
ter contact of the bauxite residue with the solid
reducing agent, favoring in such a way the iron
oxide reduction.
A metallographical section of grains of the
metallic product form the smelting reduction of
bauxite residue with lignite (4/1 by weight) with
the addition of 35 g SiO2 and 5 g CaO per 100 g
of the mixture, was studied by SEM/EDS
(JEOL® JSM-6380LV). A back-scattered elec-
tron image of the metallic product is presented
in Figure 6. This is a typical feature, where the
metal matrix and the lamellar structure of iron
carbide flakes can be seen.
3. CONCLUSIONS
Treatment of Greek bauxite residue including
3rd AMIREG International Conference (2009): Assessing the Footprint of
Resource Utilization and Hazardous Waste Management, Athens, Greece
Figure 6: Back-scattered electron image of the metallic
product from the smelting reduction of Greek bauxite re-
sidue with lignite.
pelletization and smelting reduction revealed
that it can be used successfully either for direct
cast iron production or as a raw material in the
form of pellets with improved properties, as a
potential feed for sponge or cast iron produc-
tion. Both the ability of the green pellets of
bauxite residue to remain intact during handling
and the ability of dried pellets to survive han-
dling during the firing process, determined in
terns of the green drop number and dry com-
pression strength respectively, were within the
limit values required by the iron ore industry.
Increase of the bentonite addition as a binding
agent favors the strength of the pellets in all
cases examined. On the contrary, Ordinary Port-
land Cement addition has a more complex be-
havior as a binding agent, since its effect is neg-
ative regarding the obtained green drop number,
while it is positive regarding the compression
strength, in comparison to the addition of ben-
tonite. In any case, even without adding any
binding agent, the obtained physicomechanical
properties of the bauxite residue pellets are very
satisfactory, which is very essential for the eco-
nomics of the process.
The reduction smelting of bauxite residue
with solid reducing agents resulted in the pro-
duction of a metallic product with an iron con-
tent fluctuating among 91-95.8%, which can be
characterized as satisfactory considering that a
typical cast iron has an approximate analysis of
91-96 % in iron. The addition of solid reducing
agents results in a significant increase of the
melting point of the bauxite residue, from
1400oC up to 1740 oC using a great excess of
439
the reducing agent, due to the formation of re-
fractory iron carbides. The addition of CaO and
SiO2 reagents as fluxes favored the decrease of
the melting point, in agreement with the respec-
tive ternary phase diagrams. The type of the sol-
id reducing agent used, proved to be a critical
parameter affecting the quality of the metallic
product, in terms of the desired sulfur content,
the basicity of the slag and the kinetics of the
reductive procedure. The results render the fu-
ture investigation for optimization of the pro-
posed processing method very promising.
ACKNOWLEDGEMENTS
The authors would like to acknowledge the fi-
nancial support of the General Secretariat for
Research and Technology of Greece, Pro-
gramme PAVET 2005, No. 05PAB102.
REFERENCES
Eremin, N.I., 1969. Complex processing of bauxites,
Proc. 2nd Int. Symp. ICSOBA Bauxites, Alumina,
Aluminium, Leningrad.
Forsmo, S.P.E, A.J. Apelqvist, B.M.T. Björkman and P.-
O. Samskog, 2006. Binding mechanisms in wet iron
ore green pellets with a bentonite binder. Powder
Technology, 169, pp. 147-158.
Kawatra, S.K. and S.J. Ripke, 2002. Effects of bentonite
fiber formation in iron ore pelletization. International
Journal of Mineral Processing, 65, pp. 141-149.
Kawatra, S.K. and S.J. Ripke, 2003. Laboratory studies
for improving green ball strength in bentonite-bonded
magnetite concentrate pellets. International Journal of
Mineral Processing, 72, pp. 429-441.
Logomerac, V.G., 1963. Metallurgical Processing of red
mud to produce useful elements. Symp. Bauxites, Ox-
ides, Hydroxides, Aluminium, Zagreb.
Logomerac, V.G., 1974. Complex processing of red mud.
Neue Hutte, 20.
Paramguru, R.K., P.C. Rath and V.N. Misra, 2005. Trends
in Red Mud Utilization-A Review, Mineral Processing
and Extractive Metallurgy, 26, pp. 1-29.
Srb, J. and Z. Ruzickova, 1988. Pelletization of fines, De-
velopments in mineral processing, 7.
Visnyovszky, L., 1971. Complex utilization of Hungarian
bauxites. Femip. Kut. Intez. Kozlem, 9.
Xenidis, A., C. Zografidis, I. Kotsis and D. Boufounos,
2009. Reductive Roasting and Magnetic Separation of
Greek Bauxite Residue for its Utilization in Iron Ore
Industry, Proc. Light Metals 2009, 138th TMS Annual
Meeting, San Francisco, California, pp. 63-67.
Zazubin, A.I., A.N. Barshchevskaya and G.M. Potapova,
1967. Complex processing of red mud. Tr. Inst. Met.
Obogashch., Akad. Nauk. Kaz. SSR, 25.