VSEBINA – CONTENTS
IZVIRNI ZNANSTVENI ^LANKI – ORIGINAL SCIENTIFIC ARTICLES
The mechanical properties of two-phase Fe-NiCrMo alloys at room temperature and 290 °C after ageing in the temperature
range 290–350 °C
Mehanske lastnosti dvofaznih zlitin Fe-NiCrMo pri sobni temperaturi in pri 290 °C po staranju v razponu temperature 290 °C do 350 °C
J. Vojvodi~ Tuma, B. [u{tar{i~, R. Celin, F. Vodopivec. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Mechanisms of HF bonding in dry scrubber in aluminium electrolysis
Mehanizmi vezave HF v ~istilnem sistemu pri elektrolizi aluminija
I. Paulin, ^. Donik, M. Jenko . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
The corrosion behaviour of duplex stainless steel in chloride solutions studied by XPS
XPS raziskave korozijskega vedenja dupleksnega nerjavnega jekla v kloridnih raztopinah
A. Kocijan, ^. Donik, M. Jenko . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
The application of an artificial neural network for determining the influence of the parameters for the deposition of a zinc
coating on steel tubes
Uporaba umetnih nevronskih mre` za dolo~itev debeline cinkove plasti na jeklenih ceveh
S. Re{kovi}, Z. Glava{ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
The application of the program QFORM 2D in the stamping of wheels for railway vehicles
Uporaba programa QFORM 2D pri kovanju koles za `elezni{ka vozila
A. Shramko, I. Mamuzi}, V. Danchenko . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
STROKOVNI ^LANKI – PROFESSIONAL ARTICLES
Influence of the working technology on Al-alloys in semi-solid state
Vpliv tehnologije preoblikovanja Al-zlitin v testastem stanju
M. Torkar, M. Dober{ek, I. Nagli~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Steel refining in a vacuum unit with chemical boosting
Rafinacija jekla v vakuumski napravi z vpihovanjem legirnih dodatkov
Z. Adolf, M. Dostál, Z. [ána . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Historical survey of iron and steel production in Bosnia and Herzegovina
Zgodovinski pregled proizvodnje `eleza in jekla v Bosni in Hercegovini
S. Muhamedagi}, M. Oru~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
17. MEDNARODNA KONFERENCA O MATERIALIH IN TEHNOLOGIJAH, 16. – 18. november, 2009, Portoro`, Slovenija
17th INTERNATIONAL CONFERENCE ON MATERIALS AND TECHNOLOGY, 16–18 November, 2009, Portoro`, Slovenia . . . . . 175
ISSN 1580-2949
UDK 669+666+678+53 MTAEC9, 43(4)177–229(2009)
MATER.
TEHNOL.
LETNIK
VOLUME 43
[TEV.
NO. 4
STR.
P. 177–229
LJUBLJANA
SLOVENIJA
JULY–AUG.
2009

J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
THE MECHANICAL PROPERTIES OF TWO-PHASE
Fe-NiCrMo ALLOYS AT ROOM TEMPERATURE AND
290 °C AFTER AGEING IN THE TEMPERATURE
RANGE 290–350 °C
MEHANSKE LASTNOSTI DVOFAZNIH ZLITIN Fe-NiCrMo PRI
SOBNI TEMPERATURI IN PRI 290 °C PO STARANJU V RAZPONU
TEMPERATURE 290 °C DO 350 °C
Jelena Vojvodi~ Tuma, Borivoj [u{tar{i~, Roman Celin, Franc Vodopivec
Institute of Metals and Technology, Lepi pot 11, 1000 Ljubljana, Slovenia
jelena.tuma imt.si
Prejem rokopisa – received: 2009-02-19; sprejem za objavo – accepted for publication: 2009-03-09
Fe-NiCrMo alloys were aged for up to 17 520 h in the temperature range 290-350 °C to achieve the spinodal decomposition of
δ-ferrite. The tensile properties and the Charpy notch toughness were determined at room temperature and at 290 °C. The
ageing affected the tensile properties only a little, but decreased very strongly the notch toughness and to a greater extent the
content of α-ferrite. The tensile properties were lower at 290 °C than at room temperature and the difference was virtually
independent of the content of ferrite. The notch toughness was significantly higher at 290 °C, and the difference amounted to
three times the lowest toughness of the aged alloy with the maximum content of δ-ferrite. The different effects of ageing are
explained in terms of the ferrite microhardness, the strain hardening of the austenite and the mechanisms of deformation and
fracturing.
Key words: Fe-NiCrMo alloys, microstructure, spinodal decomposition, tensile properties, notch toughness, testing temperature
Zlitine Fe-NiCrMo so bile starane do 17 520 h pri temperaturah med 290 °C in 350 °C zaradi spinodalne razgradnje trdne
raztopine v δ-feritu. Dolo~ene so bile raztr`ne lastnosti in zarezna `ilavost po Charpyju pri sobni temperaturi in pri 290 °C.
Staranje malo vpliva na raztr`ne lastnosti, a mo~no zmanj{a zarezno `ilavost, in to tem bolj, ~im ve~ je v zlitini ferita in ~im bolj
je napredoval spinodalen proces. Raztr`ne lastnosti so bile ni`je pri temperaturi 290 °C kot pri sobni temperaturi, razlika pa je
bila prakti~no neodvisna od vsebnosti ferita in od stopnje spinodalnega procesa. Zarezna `ilavost je bila ve~ja pri 290 °C kot pri
sobni temperaturi in razlika je bila trikratna pri najni`ji izmerjeni `ilavosti starane zlitine z najve~ ferita. Razli~en vpliv staranja
na lastnosti je razlo`en z upo{tevanjem trdote δ-ferita, deformacijske utrditve avstenita ter mehanizmov plasti~ne deformacije in
preloma.
Klju~ne besede: zlitine Fe-NiCrMo, spinodalni razpad, raztr`ne lastnosti, zarezna `ilavost, temperatura preizkusa
1 INTRODUCTION
In alloys with a high content of chromium the solid
solution of chromium and nickel or cobalt in α-iron is
not stable. In the temperature range up to approximately
750 °C the solid solution is decomposed with a spinodal
process into two constituents: one enriched in chromium
and the other in nickel or cobalt1. Both phases retain the
initial α-iron lattice but have a different lattice parameter
that depends on the composition of the solid solution.
The lattices of both phases accommodate with elastic
internal stresses that increase the hardness and brittle-
ness, and after magnetisation give the alloy hard-magne-
tic properties2. The kinetics of decomposition depends
on the diffusional transport of atoms in a substitutional
solid solution. The rate of diffusion depends strongly on
the temperature and it is very slow in the temperature
range in which alloys of this type are operating in
nuclear power plants. In the same conditions of tempe-
rature and time, the solid solution in austenite is stable; it
may change only with the precipitation of carbides if the
content of carbon is above the solubility limit that
depends on the temperature. At higher temperatures the
spinodal decomposition is replaced by the formation of σ
phases that also decrease the mechanical properties at
room temperature3. In the process of spinodal decom-
position, the properties of alloys with a two-phase
microstructure of austenite and δ-ferrite are changed,
depending on the volume share of ferrite and the extent
of the decomposition, respectively, and the difference in
the chemical composition and the lattice parameters
between both spinodal constituents.
In some nuclear power plants essential parts of the
equipment are manufactured from Fe-Ni-Cr-Mo alloys
and may have, depending of the chemical content of
alloying elements and impurities as well as the eventual
thermal history, a different content of δ-ferrite formed
with equilibrium or non-equilibrium solidification. The
operation temperature of nuclear power plants is rela-
tively low, in many plants it is around 300 °C, and the
process of spinodal decomposition is very slow, as is the
rate of change of the properties. For this reason, in this
work the ageing was performed at a relatively low
temperature and for a time that should suffice for a
Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187 179
UDK 669.01:620.18:539.42 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 43(4)179(2009)
reliable evaluation of the kinetics and mechanisms of the
changes of the properties of built-in alloys.
In the investigation of the instability of the properties
of Fe-Ni-Cr-Mo alloys with a two-phase microstructure4
it was established that the change of properties was
greater for an alloy with 14.5 % of δ-ferrite than for an
alloy with 8.5 % of δ-ferrite when ageing in the tempera-
ture range 300–400 °C, while at higher temperatures the
ageing effect was smaller. In5 these findings we con-
firmed and found that the content of δ-ferrite determined
from the Schaeffler diagram was unreliable and that its
distribution in as-cast alloys is inhomogeneous and can
vary in the range 1.5 to 22.5 for the same cast piece. The
ageing effect on Charpy toughness was very strong in
temperature range 303 °C to 325 °C and the initial
toughness was achieved again after annealing at 550 °C.
Of several processes that could affect the alloys’ pro-
perties, the main embrittlement process is the spinodal
decomposition6. A correlation was developed7 for the
assessment of the thermal embritlement and the
prediction of changes in the fracture and Charpy and
tensile properties of as-cast two-phase alloys. The use of
small specimens for the investigation of in-service-aged
elbows gives reliable values for the J- a results in
conditions where strongly deviating specimens are
rejected8. The low-cycle fatigue increases rapidly with
the increase of the ageing time9. An essential difference
in the properties obtained at room temperature and at
300 °C for aged alloys with a different content of
δ-ferrite is observed in ref. 10.
2 EXPERIMENTAL WORK
Three alloys with the compositions in Table 1 were
prepared by melting in a laboratory induction furnace
from the same raw materials and cast into square ingots
of 100 mm thickness. The average contents of δ ferrite
were 2 %, 11 % and 27 %. In the two alloys with lower
contents of ferrite, this phase was formed mostly during
the solidification of grain boundaries, forming a network
that was more closed with a higher content of ferrite
(Figure 1 and 2). It is concluded that δ-ferrite solidified
from the residual melt, enriched in elements that
improve the stability of ferrite in the equilibrium binary
and ternary systems. In the alloy with the highest content
of ferrite the microstructure is explained by the start of
solidification with the formation of ferrite and the
subsequent solidification with the formation of austenite
in a peritectic reaction between the solid ferrite phase
and the melt enriched in elements stabilising the γ-phase
(Figure 3). All the alloys were assumed to be suited for
the investigation because in all these the δ-ferrite
constituent of the microstructure were embedded in
austenite.
All the alloys were aged for a time up 4 320 h for the
tensile tests and up to 17 520 h for the Charpy notch tests
and hardness at three temperatures: 290 °C, 320 °C and
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
180 Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187
Figure 1: Microstructure of the alloy with 2 % of δ-ferrite
Slika 1: Mikrostruktura zlitine z 2 % δ-ferita
Figure 2: Microstructure of the alloy with 11 % of δ-ferrite
Slika 2: Mikrostruktura zlitine z 11 % δ-ferita
Figure 3: Microstructure of the alloy with 27 % of δ-ferrite
Slika 3: Mikrostruktura zlitine s 27 % δ-ferita
350 °C. It was assumed that the rate of spinodal decom-
position depended on the exchange of atoms between
both phases with volume diffusion. The diffusion rate is
about 3 times greater at 350 °C than at 290 °C. It was
thus expected that the ratio of the rates of spinodal
decomposition was equal and that after 17 520 h of
ageing at 350 °C the decomposition of the solid solution
would be achieved at 290 °C, after a much longer ageing
time at 290 °C. This assumption does not imply that the
effect of ageing on properties is equal for both tempera-
tures. The answer to these questions, observations of
spinodal constituents with transmission electron micro-
scopy would be necessary.
Tensile properties and Charpy notch toughness of
aged specimens were determined at room temperature
and at 290 °C, operating temperature of parts in a
nuclear power plant. The hardness and microhardness
were determined only at room temperature. Austenite
does not fracture with cleavage, thus according to8, the
very wide spread of Charpy results were rejected as a
result of the local, greatly increased content of ferrite in
the specimen with a small section. Most of the
specimens strongly deviating from the parallels and
clearly deviating in the curve depicting the effect of
ageing time were found for the Charpy specimens.
Table 1: Chemical composition of the examined alloys
Tabela 1: Kemi~na sestava zlitin
Alloy Elements, w/%
C Si Mn P S Ni Cu Mo
I 0.06 0.43 1.59 0.03 0.01 11.9 18 1.84
II 0.07 0.67 1.04 0.03 0.01 11 21.7 2.03
II 0.06 1.68 0.67 0.03 0.01 9 20.8 2.46
3 HARDNESS AND MICROHARDNESS
The ageing time at 350 °C does not affect the hard-
ness of the alloy with 2 % of δ-ferrite; only very slightly
does it affect the hardness of the 11 % ferrite alloy, and
slightly the hardness of the 27 % ferrite alloy (Figure 4).
At a lower ageing temperature the changes of hardness
were also small for the 27 % ferrite alloy. The ageing
time does not affect the austenite microhardness and
increases the microhardness of the ferrite (Figure 5).
The increase is larger and faster for the ageing tempe-
rature of 350 °C. For this alloy, the initial microhardness
of the ferrite was increased by 82 % after 4 320 h of
ageing, doubled after 8 760 h and increased by 2.2 times
after 17 520 h of ageing. The chemical composition of
the δ-ferrite is similar in all the alloys. It can thus be
assumed that the hardness of this phase increased with
the ageing time in a similar way also for the alloys with a
lower content of α-ferrite. The minor share of ferrite in
the alloys explains why the ageing affected the alloys’
hardness to a much smaller extent than the micro-
hardness of the ferrite.
4 TENSILE PROPERTIES
In Figures 6, 7 and 8 the effect of ageing time is
shown for the yield stress at room temperature and at
290 °C. In the 2 % ferrite alloy, the yield stress at room
temperature is slightly decreased during the ageing at all
three tested temperatures. For the alloy with 11 % of
ferrite a similar yield-stress decrease is found for the
ageing temperatures of 290 °C and 320 °C, while when
ageing at 350 °C the yield stress is slightly increased.
For the alloy with 27 % of ferrite the yield stress is
increased at all ageing temperatures. Compared to the
increase of the ferrite microhardness, which amounts to
82 % after 4 320 h of ageing, the increase of the yield
stress is much lower; it amounts to only approximately 4
%. It is assumed that the small decrease of the yield
stress is due to the relaxation of the cooling stresses in
both low-amount ferrite alloys.
At 290 °C the yield stress is lower for all alloys than
at room temperature, and the effect of the ageing time is
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187 181
Figure 5: Microhardness of δ-ferrite and austenite in the 27 % ferrite
alloy and the dependence on the ageing time at 350 °C
Slika 5: Mikrotrdota δ-ferita in avstenita v zlitini s 27 % ferita v
odvisnosti od ~asa staranja pri 350 °C
Figure 4: Hardness of all three alloys with the dependence on the
ageing time
Slika 4: Trdota vseh treh zlitin v odvisnosti od ~asa staranja
similar for the alloy 1. The yield stress is not affected by
the ageing for the alloy with 11 % of ferrite. It is
constant for the alloy with 27 % of ferrite during ageing
at 290 °C, slightly increased during ageing at 320 °C,
and increased a great deal at both test temperatures after
ageing at 350 °C.
Ageing at 290 °C and 320 °C has no effect on the
tensile strength at room temperature and at 290 °C for
the 2 % and 11 % ferrites (Figure 9, 10 and 11). For the
alloys with 11 % and 27 % ferrite the tensile strength is
increased only at room temperature. At both higher
ageing temperatures the tensile strength increases with
the ageing. The increase is greater at 350 °C, when the
strength at room temperature and at 290 °C, compared to
the initial value it is higher by 12 % and 10 %,
respectively. Again the increase of the tensile strength is
much smaller than the increase of the ferrite hardness.
The elongation is very slightly affected by the ageing
of all alloys at 290 °C and 320 °C, while after ageing at
350 °C it is lower due to the higher content of ferrite
(Table 2). It is not affected significantly by the ageing
time and temperature at both testing temperatures and for
all the alloys it is lower during testing at 290 °C.
For all the alloys the reduction of the area is virtually
unaffected by the content of ferrite and the ageing time
at 290 °C and 320 °C. At room temperature it is virtually
equal for all the alloys and ageing conditions (Table 3).
For all the alloys it is significantly lower at 290 °C and
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
182 Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187
Figure 8: Effect of ageing time at 350 °C on the yield stress at room
temperature and at 290 °C for all alloys
Slika 8: Vpliv trajanja staranja pri 350 °C na mejo plasti~nosti vseh
zlitin pri sobni temperaturi in pri 290 °C
Figure 6: Effect of ageing time at 290 °C on the yield stress for all
alloys at room temperature and at 290 °C
Slika 6: Vpliv trajanja staranja pri 290 °C na mejo plasti~nosti vseh
zlitin pri sobni temperaturi in pri 290 °C
Figure 9: Effect of ageing time at 290 °C on the tensile strength of all
alloys at room temperature and at 290 °C
Slika 9: Vpliv ~asa staranja pri 290 oC na raztr`no trdnost vseh zlitin
pri sobni temperaturi in pri 290 °C
Figure 7: Effect of ageing time at 320 °C on the yield stress for all
alloys at room temperature and at 290 °C
Slika 7: Vpliv trajanja staranja pri 320 °C na mejo plasti~nosti vseh
zlitin pri sobni temperaturi in pri 290 °C
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187 183
Figure 11: Effect of ageing time at 350 °C on the tensile strength at
room temperature and at 290 °C for all alloys
Slika 11: Vpliv ~asa staranja pri 350 °C na raztr`no trdnost vseh zlitin
pri sobni temperaturi in pri 290 °C
Table 2: Effect of ageing time at 350 °C on the elongation at room
temperature and at 290 °C
Tabela 2: Vpliv ~asa staranja na raztezek pri sobni temperaturi in pri
290 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 51 35 45 36 40 28
24 h 54 39 46 40 37 30
168 h 54 34 52 36 37 28
720 h 48 38 48 36 37 27
4320 h 51 34 44 33 33 26
Figure 10: Effect of ageing time at 320 °C on the tensile strength at
room temperature and at 290 °C for all alloys
Slika 10: Vpliv ~asa staranja pri 320 °C na raztr`no trdnost vseh zlitin
pri sobni temperaturi in pri 290 °C
Table 3: Effect of ageing time at 350 °C on the reduction of area at
room temperature and at 290 °C
Tabela 3: Vpliv ~asa staranja na kontrakcijo pri sobni temperaturi in
pri 290 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 64 61 68 62 64 47
24 h 68 58 62 58 68 51
168 h 67 56 63 58 67 50
720 h 68 56 61 60 64 50
4320 h 66 50 60 56 60 46
Table 4: Yield stress reported for the 2 % ferrite alloy at room tem-
perature and 290 °C; ageing temperature 350 °C
Tabela 4: Razmerje proti meji plasti~nosti zlitine z 2 % ferita pri
sobni temperaturi in pri 290 °C; temperatura staranja 350 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 1 1 1.14 1.24 1.52 1.71
24 h 0.94 0.96 1.18 1.24 1.52 1.68
168 h 0.95 0.98 1.15 1.29 1.55 1.74
720 h 0.91 0.92 1.19 1.19 1.57 1.70
4320 h 0.92 0.89 1.18 1.22 1.59 1.79
Table 5: Tensile strength reported for the 2 % ferrite alloy at room
temperature and 290 °C; ageing temperature 350 °C
Tabela 5: Razmerje proti raztr`ni trdnosti zlitine z 2 % ferita pri sobni
temperaturi in pri 290 °C; temperatura staranja 350 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 1 1 1.15 1.19 1.41 1.49
24 h 1.0 1.05 1.15 1.18 1.43 1.52
168 h 1.0 1.01 1.16 1.28 1.47 1.55
720 h 1.0 1.01 1.18 1.28 1.51 1.53
4320 h 1.0 1.04 1.22 1.27 1.58 1.59
Table 6: Elongation for the 11 % and 27 % ferrite alloys reported for
the 2 % ferrite alloy at room temperature and 290 °C; ageing tempe-
rature 350 °C
Tabela 6: Razmerje proti razteznosti zlitine z 2 % ferita pri sobni tem-
peraturi in pri 290 °C; temperatura staranja 350 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 1 1 0.88 1.02 0.78 0.80
24 h 0.98 1.11 0.90 1.03 0.72 0.85
168 h 0.89 0.97 1.05 1.0 0.72 0.80
720 h 0.87 1.08 1.02 1.0 0.74 0.77
4320 h 0.94 0.97 0.97 0.94 0.66 0.74
all ageing conditions and diminished the most after the
ageing of the 27 % ferrite alloy at 350 °C.
The effect of the content of ferrite and ageing time on
the change of the yield stress and the tensile strength at
both testing temperatures after different ageing times at
350 °C is shown in Tables 4 and 5. Both properties are
higher with an increased content of ferrite and are
increased slightly more when testing at 290 °C than at
room temperature. The data in Tables 6 and 7 show that
the initial elongation is less with a higher content of
ferrite. For the 2 % and 11 % ferrite alloys the elongation
is not affected by the ageing time at both testing
temperatures, while it is slightly decreased with longer
ageing. For the 27 % ferrite alloy the elongation is
slightly greater when testing at 290 °C. The scatter of the
test results was great for the reduction of area and, for
this reason, the effect of ageing time at 350 °C is not
clear; it is, however, very small. During testing at 290 °C
the difference in the properties is greater. The yield
stress and the tensile properties of the non-aged and aged
alloys increase with the content of ferrite in a non-linear
dependence, which is similar for the yield stress and the
tensile strength.
5 NOTCH TOUGHNESS
In Figures 12, 13 and 14 the effect of time for
different ageing temperatures on the Charpy notch
toughness is shown for both testing temperatures and the
2 %, 11 % and 27 % δ-ferrite alloys. With 2 % of ferrite
and ageing at 290 °C and 320 °C the notch toughness
starts to decrease gradually with an intermediate ageing
time, and it is decreased to a value of about 80 % of the
initial level after the longest ageing time of 17 520 h (2
years). With ageing at 350 °C the notch toughness drops
after ageing for approximately 4 320 h when testing at
room temperature and 290 °C to approximately 2/3 of
the initial value. In all cases the notch toughness is
higher when testing at 290 °C than at room temperature.
The difference is about 30 J (16 %) for all the tested
specimens.
The effect of ageing is stronger for the 11 % ferrite
alloy. At room temperature the notch toughness starts to
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
184 Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187
Table 7: Reduction of area for the 11 % and 27 % ferrite alloys
reported for the 2 % ferrite alloy at room temperature and 290 °C;
ageing temperature 350 °C
Tabela 7: Razmerje proti kontrakciji zlitine z 2 % ferita pri sobni
temperaturi in pri 290 °C; temperatura staranja 350 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
0 h 1 1 1.06 1.01 1.0 0.87
24 h 1.06 0.95 0.97 0.85 1.06 0.84
168 h 1.05 0.92 0.98 0.95 1.05 0.82
720 h 1.06 1.0 0.95 0.98 1.0 0.92
4320 h 1.03 0.82 0.94 0.92 0.94 0.81
Figure 14: Alloy with 27 % of δ-ferrite. Effect of the ageing time at
different ageing temperatures on the notch toughness at room
temperature and at 290 °C10.
Slika 14: Zlitina s 27 % δ-ferita. Vpliv trajanja staranja pri razli~nih
temperaturah na zarezno `ilavost pri sobni temperaturi in pri 290 °C
Figure 13: Alloy with 11 % of δ-ferrite. Effect of the ageing time at
different ageing temperature on the notch toughness at room
temperature and at 290 °C10
Slika 13: Zlitina z 11 % δ-ferita. Vpliv trajanja staranja pri razli~nih
temperaturah na zarezno `ilavost pri sobni temperaturi in pri 290 °C
Figure 12: Alloy with 2 % of δ-ferrite. Effect of the ageing time at
different ageing temperatures on the notch toughness at room
temperature and at 290 °C10
Slika 12: Zlitina z 2 % δ-ferita. Vpliv trajanja staranja pri razli~nih
temperaturah na zarezno `ilavost pri sobni temperaturi in pri 290 °C
decrease already after 168 h of ageing at 350 °C and 720
h at 290 °C. With longer annealing the notch toughness
decreases faster, to approximately 35 J, thus to 30 % of
the initial value. At the test temperature of 290 °C the
notch toughness is higher by 30 J to 40 J (approximately
25 %) than at room temperature. The effect of ageing
time is similar; it is, however, smaller and the minimum
test value of 80 J is 2.3 times greater at 290 °C than at
room temperature after the same ageing.
The initial notch toughness is similar for the alloys
with 27 % and 11 % ferrite; however, the effect of
ageing is enhanced for the 27 % ferrite alloy. The
toughness starts to decrease after a similar ageing time as
for the 11 % ferrite alloy; however, it decreases faster by
increasing the ageing time, since by ageing at 350 °C the
level of 35 J (approximately 1/3 of the initial value) is
achieved already after 720 h and the toughness of
approximately 20 J already after 4 320 h of ageing at 350
°C, and this remains unchanged also after the longest
ageing time of 17 520 h. At the lower ageing tempe-
rature of 320 °C the decrease in the notch toughness is
slower and then the final value of 33 J (32 % of the
initial value) is achieved after 8760 h of ageing. During
ageing at 290 °C the rate of decrease of the toughness is
slower and the minimum value is approximately three
times higher than for the 27 % ferrite alloy after the
same ageing time at 350 °C.
For the 27 % ferrite alloy the notch toughness is
higher when testing at 290 °C by about 25 J, and the
minimum value after the longest ageing time at 350 °C is
approximately three times greater than when testing at
room temperature. The effect of the ageing temperature
is similar as at room temperature; however, the
differences after the longest annealing are smaller than
those at room temperature.
An easier comparison of the effects of the content of
ferrite and ageing temperature is possible from the
relative values in Table 8. The data show that the effects
of the content of ferrite and ageing temperature are much
greater for the notch toughness than for the tensile
properties. The notch toughness at room temperature is
lower for all the alloys after the longest ageing at 290
°C; however, the decrease is stronger and virtually equal
for the 11 % and 27 % ferrite alloys. A similar extent of
notch-toughness lowering is found for the 2 % ferrite
alloy after ageing at 320 °C, whereas it is much stronger
and again virtually equal for the 11 % and 27 % ferrite
alloys. After ageing at 350 °C the notch toughness is
slightly diminished for the 2 % ferrite alloy, while it is
diminished more for the 11 % ferrite alloy and even
more for the 27 % ferrite alloy. In all cases the decrease
of the notch toughness is lower when testing at 290 °C
than at room temperature.
6 ANALYSIS OF THE EXPERIMENTAL
FINDINGS
The effect of ageing temperature and time is
explained by the extent of the completion of the spinodal
decomposition of the solid solution in the δ-ferrite. For
the tensile properties, the change is relatively small and,
when detected, it increases continuously with the ageing
time and it is greater with a higher ageing temperature.
The Charpy notch toughness is decreased slowly in the
first period and very fast after the level of the hardness
of ferrite was achieved. The ageing time of the initial
rapid decrease of the notch toughness is shorter with a
higher ageing temperature. The lowest toughness was
achieved at an ageing temperature of 350 °C. The hard-
ness of the ferrite increased continuously at all the
ageing temperatures and more rapidly at the higher
temperature, and it was not finished even after the
longest applied ageing at 350 °C, when the initial
hardness was increased by 2.2 times. Thus, the increase
of the ferrite hardness affects differently the tensile
properties and the notch toughness.
It is assumed that the explanation is in the differences
of the behaviour of the ferrite inserts by the axial
deformation and the localised flexion deformation. With
the axial deformation the aged ferrite inserts start to
deform plastically at a sufficient level of strain hardening
of the austenite, inducing only small changes in the
elongation and a reduction of the area for different
contents and microhardnesses of the aged ferrite and the
different width of the austenite ligaments. For this
reason, up to a microhardness of ferrite of HV 537
obtained after ageing for 4 320 h at 350 °C, the fracture
surface of the tensile specimens is entirely ductile and
dimpled, confirming that the ferrite and the austenite
matrix are fractured after the cold deformation in the
ductile mode.
The explanation of the initial slow decrease of the
notch toughness is probably the same, while the phase of
the rapid decrease of notch the toughness begins when
the aged ferrite starts to fracture with clevage2. The
difference in the behaviour of the ferrite of the same
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187 185
Table 8: Charpy notch toughness for all the alloys reported for the 2
% ferrite alloy at room temperature and 290 °C; ageing temperature
350 °C
Tabela 8: Razmerje proti zarezni `ilavosti zlitine z 2 % ferita pri
sobni temperaturi in pri 290 °C; temperatura staranja 350 °C
δ-ferrite
(w/%) 2 11 27
Ageing
time
Test temperature
22 °C 290 °C 22 °C 290 °C 22 °C 290 °C
Ageing temperature 290 °C
0 h 1 1 0.76 0.82 0.80 0.80
17 520 h 0.61 0.79 0.39 0.66 0.37 0.62
Ageing temperature 320 °C
0 h 0.96 1.03 0.75 0.78 0.69 0.73
17 520 h 0.58 0.82 0.22 0.60 0.21 0.50
Ageing temperature 350 °C
0 h 0.94 0.97 0.70 0.78 0.69 0.73
17 520 h 0.53 0.77 0.21 0.43 0.13 0.38
hardness with the flexion tests of the notched specimens
is related to the difference in the mechanism and the rate
of deformation and fracture. The deformation and
fracturing of the Charpy specimens in the ductile range
occurs in approximately 10–2 s, while the tensile test time
is several orders of magnitude longer. The presence of
the notch limits the volume of the plastic deformation
and from the moment of the crack propagation from the
notch the plastic deformation is limited to a layer of
metal with a thickness of less than 50 µm on both sides
of the crack lips11, while the width of the plastic
deformation in the phase of the increase of the reduction
of the area of tensile specimens is much greater. The
rapid plastic deformation with the Charpy test generates
adiabatic heating of the deformed metal. For this reason,
with the Charpy fracturing the temperature in the layer
of the plastic deformation at the crack tip and both sides
of the crack is increased, the more this happens the
greater is the extent of plastic deformation in structural
steels, even by several hundreds of °C11. These differen-
ces make the comparison of tensile and Charpy tough-
ness values unreliable for the same alloy.
It is assumed that with localised plastic deformation
the much stronger effect of ageing on the notch tough-
ness than on the tensile properties may be explained by
assuming that the ferrite with the hardness above a
determined level started to fracture with cleavage ahead
of the tip of the propagating crack. Ferrite inserts with
very narrow cracks would also increase the local stress
concentration and generate crack initials in austenite
ligaments ahead of the tip of the propagating crack.
Most of the energy consumed for the brittle
fracturing of structural steels is consumed for the elastic
deformation that generates the stress concentration
necessary to start the cleavage of ferrite at the notch
tip11. With ductile fracturing of structural steels, 6 J to 7
J of energy is consumed for the elastic deformation
before the plastic deformation is initiated. Generally, for
the cleavage fracturing of structural steels below the
brittle-fracture threshold, a similar energy is consumed.
The content of δ-ferrite in the alloy with the lowest
Charpy notch toughness of 20 J was 27 %. Assuming
that the energy consumed for the fracturing of ferrite is
proportional to its content in the alloy, it can be deduced
that at the lowest level of Charpy toughens of 20 J, only
approximately (8 × 0.27) = 2.1 J are consumed for the
fracture of the aged ferrite. Thus, with a level of 20 J,
approximately 10 % of the energy is consumed for
fracturing the aged ferrite and 90 % for fracturing the
austenite ligaments. The total content and shape of the
ferrite inserts could modify significantly the energy
deduced for the Charpy cleavage of the ferrite. It can be
reliably concluded that in reality the energy consumed
for the cleavage of the ferrite could be only lower than
the deduced value, while it is reliable to assume that the
formation of the initial crack in the austenite ahead the
fracture tip would also decrease the energy consumed for
the fracture of the austenite.
Generally, it is found for low carbon iron alloys that
by increasing the test temperature up to a level depend-
ing on the chemical composition and microstructure, the
yield stress and the tensile strength are decreased and the
elongation and the reduction of the area remain constant
or are even slightly increased. Logically, the yield stress
and the tensile strength are lower at 290 °C. However, a
smaller elongation and a reduction of the area are found
independently of the content and the microhardness of
ferrite. The conclusion is that the lower tensile properties
at 290 °C than at room temperature are very probably
due to the lower intensity of the strain hardening by the
axial tensile deformation at higher temperatures. The
same explanation could also be applied for the greater
Charpy notch toughness at 290 °C, since, it was found
that the volume of plastic deformation before the crack is
started at the notch tip is greater with lower strain
hardening, while the energy spent in the crack propa-
gation is virtually identical in both cases11.
7 CONCLUSIONS
The following conclusions are suggested on the basis
of the experimental findings and their explanation:
• the effect of the content of ferrite on the properties of
non-aged two-phase Fe-Ni-Cr-Mo alloys is explained
by the differences in the chemical composition and
the linear grain size;
• the ageing affects the properties more with an
advanced process of spinodal decomposition of the
δ-ferrite. For this reason, the properties are changed
more after equal ageing times at 350 °C than at 290
°C;
• the effect of ageing is relatively small for tensile
properties. It is much greater for the Charpy tough-
ness, which is diminished down to approximately 1/5
of the initial value after the longest ageing at the
highest applied temperature;
• the tensile properties are lower at 290 °C than at
room temperature, independent of the content of
ferrite and the ageing. In contrast, the notch tough-
ness is higher at 290 °C than at room temperature for
all the ferrite contents and ageing temperatures;
• the difference in the properties at room temperature
and 290 °C is explained in terms of the hardness of
the ferrite, the strain hardening of the austenite and
the difference in the mechanisms of plastic defor-
mation and fracture propagation by the axial tensile
test and the Charpy notch flexion test.
The authors are indebted to the Slovenian Research
Agency and NPP Kr{ko for their financial support of the
project, to Mr. B. Breskvar for the preparation of alloys,
Mr. D. Kmeti~ for the thermal treatments and Mr. B.
Arzen{ek for the mechanical tests.
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
186 Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187
8 REFERENCES
1 F. Vodopivec, M. Pristavec, J. @vokelj, D. Gnidovec, F. Gre{ovnik:
Zaitschrift für Metallkunde 79 (1988), 648–653
2 F. Vodopivec, D. Gnidovec, B. Arzen{ek, M. Torkar, B. Breskvar:
@elezarski Zbornik 23 (1989), 73–78
3 F. Tehovnik, Izlo~anje  -faze v avstenitnih nerjavnih jeklih, Mate-
riali in tehnologije, will be printed
4 G. Slama, P. Petrequin, T. Mager: Assuring structural integrity of
steel reactor pressure boundary components: SMIRT post-conference
seminar, 1983, Monterey, USA, 211
5 C. Jannson: Degradation of cast stainless steel elbows after 15 years
of service: Fontenraud III-Internationa Symposium, Royal Abbey of
Fontenraud-Franc, September, 1990, 1/8–8/8
6 H. M. Chung, T.R. Leax: Mater. Sci. Techn. 6 (1990), 249
7 O. K. Chopra: Estimation pof mechanical properties of cast stainless
steels during thermal ageing in LWR; Trans. 13th Intern. Conf. On
structural mechanics and reactor technology, Porto Allegre, Brazil,
August 1995
8 S. Jayet-Gendrot, P. Ould, T. Meylogan: Nucl. Eng. Des. 184 (1998),
3–11
9 J. Kwom, S. Woo, Y. Lee, J-C Park, Y-W Park: Nucl. Eng. Des. 206
(2001), 35–44
10 J. Vojvodi~ Tuma, B. [u{tar{i~, F. Vodopivec: Nucl. Eng. Des. 238
(2008), 1511–1517
11 F. Vodopivec, B. Arzen{ek, J. Vojvodi~ Tuma, B. Celin: Metalurgija
47 (2008), 173–179
J. VOJVODI^ TUMA ET AL.: THE MECHANICAL PROPERTIES OF TWO-PHASE Fe-NiCrMo ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 179–187 187

I. PAULIN ET AL.: MECHANISMS OF HF BONDING IN DRY SCRUBBER IN ALUMINIUM ELECTROLYSIS
MECHANISMS OF HF BONDING IN DRY SCRUBBER IN
ALUMINIUM ELECTROLYSIS
MEHANIZMI VEZAVE HF V ^ISTILNEM SISTEMU PRI
ELEKTROLIZI ALUMINIJA
Irena Paulin1,2, ^rtomir Donik2, Monika Jenko2
1 TALUM, d. d., Kidri~evo, Tovarni{ka cesta 10, 2325 Kidri~evo, Slovenia
2 Institute of Metals and Technology, Lepi pot 11, SI-1000 Ljubljana, Slovenia
irena.paulin imt.si
Prejem rokopisa – received: 2009-04-21; sprejem za objavo – accepted for publication: 2009-05-26
Modern dry scrubbers in electrolytic winning of aluminium operate with very high efficiency, typically above 99 %, and all but
the fugitive fluoride losses are returned into electrolytic cell. The principle of purification of exhaust gases from electrolytic
cells is adsorption of HF onto primary alumina in dry scrubber reactors, where gases from cells and flow of active (primary)
alumina run into each other and react. It is very important how the reaction between gases and primary alumina takes place and
which chemical mechanisms are taking part. Nevertheless, also industrial conditions are playing an important role in efficiency
of the scrubbing reaction.
In references on dealing with related systems there can be found that most research was made to improve dry scrubbing systems
and their efficiency, but not much was done to explain chemical mechanisms of fluoride bonding on active alumina that is
essential for the cleaning process of exhaust gases. Chemical and micro-chemical analyses enabled to determine the fraction of
bonded fluoride on the surface and in the interior of the fluorinated alumina grains. Also the correlation between the most
efficient fluoride bonding on active alumina and real industrial conditions was determined.
Key words: chemical adsorption, HF-alumina interaction, fluoride bonding mechanism, EDS, dry scrubber
Ve~ina ~istilnih naprav pri proizvodnji primarnega aluminija ima u~inkovitost ve~jo od 99 % in ves ujeti fluor se vra~a v proces
elektrolize. ^i{~enje plinov HF iz elektroliznih celic deluje po principu vezave fluora na primarno glinico, ki se dozira v reaktor
~istilne naprave, skozi katero te~ejo plini. Zato je zelo pomembno, kako poteka vezava fluora na aktivno glinico in kak{ni so
najugodnej{i pogoji za to.
V prej{njih raziskavah je bilo veliko narejeno na podro~ju u~inkovitosti ~istilnih sistemov, mehanizmi vezave fluora na aktivno
glinico pa {e niso bili podrobno raziskani. S kemijsko in EDS-analizo elementov smo ugotovili dele` vezanega fluora na
povr{ini in v notranjosti zrn glinice. Poiskali smo povezavo med u~inkovitostjo mehanizma kemijske vezave fluora na aktivno
glinico in industrijskimi pogoji, pri katerih poteka ~i{~enje plinov, ki nastajajo med procesom elektrolize.
Klju~ne besede: kemijska adsorpcija, interakcija glinica-HF, mehanizmi vezave fluora, EDS, ~istilni sistem
1 INTRODUCTION
Aluminium is one of the most important and widely
used metals in industry and in everyday life in modern
world, also. Pure aluminium is extracted by electrolysis
of alumina (Al2O3) prepared from bauxite. Primary
aluminium is extracted from alumina with electrolytic
process according to the 2 Al2O3 + 3 C (graphite)  4 Al + 3
CO2 (g) reaction. The process, with additives of cryolite,
AlF3, CaF2 etc. into electrolyte, proceeds at temperatures
below 1000 °C and at constant DC, most frequently 180
kA to 190 kA. Electrolytic cells are connected in series
in order to keep constant electrical current in all the cells
according to the Ohm’s law.
Like in all the other industries, the major driving
force in the primary aluminium industry over the last 20
years has been to produce more metal of better quality at
lower costs. But production of greater quantities of metal
results in significant increase of emissions. Hence,
aluminium industry has developed unique industrial
emission control and recovery systems. Fluoride
emissions of the highest concern during the electrolytic
process are those related to cryolite, (Na3AlF6), chiolite,
(Na5Al3F14), and atmolite, (NaAlF4) particulates, and to
HF, CF4, and C2F6 gases1. The focus in this paper was
directed to fluorides that were caught and bonded onto
the alumina surface during the cleaning process.
Minimizing the fluoride emission in aluminium
electrolytic cells has many positive effects; reduction of
local and global environmental impacts of the process,
improved working environment as well as economic
effects and advantages in cell operation.
Efficiency of removal of fluorides from flue gases in
modern dry scrubbing systems is now higher than 99 %,
and fluoride emissions from smelters are typically close
to 0.5 kg F/ton Al 2,3. Nevertheless, further reduction of
fluoride losses through emissions can lead to chemically
more stable electrolyte and the level of fugitive fluoride
emissions (emissions through the pot-room roof) is thus
reduced, too. Fugitive emissions are estimated to be 4–5
times higher than the stack ones3 of which up to 80 %
may be represented by standard background emissions
4–6.
In the past there have been some significant but
limited studies on exact contributions of various sources
Materiali in tehnologije / Materials and technology 43 (2009) 4, 189–193 189
UDK 669.71:546.16:620.18 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 43(4)189(2009)
to the total fluoride emissions from cells. 7,8 In 1963
Henry 7 measured the influence of a number of variables
in 10 kA experimental aluminium reduction cells on
particulate and gaseous emissions. In 1972 Grjotheim et
al. 8 prepared a review of available data for careful
assessment of contributions of important parameters.
Mathematical models to predict total fluoride losses
were based on these data. 9,10 Over 30 years of research
brought to improved efficiency of existent cleaning
systems and to an examination of fluoride bonding
mechanism on active alumina in dry scrubbing reactors.
Principle of the dry scrubbing system is simple.
Scrubbers operate by direct extraction of flue gases from
electrolytic cells into reactors where gases react with
active alumina. At first, poisonous fluoride and HF are
adsorbed from the gas mixture onto active alumina
surface. Then gases with active alumina are transported
through bag filters where solid sandy alumina with
adsorbed gases is removed. The fluoride-enriched
alumina is after filtration returned back into electrolytic
process. Effectiveness of the cleaning process depends
on details of the system and the process. Continuous
sustaining "equal parts of alumina" in the process gas
stream represents the opportunity of modern dry
scrubbers to reduce fluoride emissions11.
Most of dry methods essentially consist of adsorption
and chemisorption of HF from flue gases coming from
aluminium cells by active alumina. Research projects
mainly deal with various technologies and their
improvements, rarely with exact mechanisms how HF is
actually bonded onto raw alumina. Dando13 made review
of experiments that were performed from 1970 till now.
However, it is well known that more than one mecha-
nism of HF bonding onto primary alumina was proposed
but none of existent hypotheses was undoubtedly
confirmed with analyzing techniques, such as XRD,
XPS, TGA, NMR etc., were used in various experiments
and analyses.
Primary focus in previous examinations of alumina –
HF systems was directed to various types of alumina and
their capacity for HF adsorption. The results of these
studies have been used to propose mechanisms of HF
adsorption on alumina. Two basic models emerged:
1) physical adsorption of HF on the alumina surface
(hydroxyl groups bonded to surface and/or physi-
sorbed by water) to form alternating layers of HF and
H2O 14, or
2) direct chemisorption of HF at reactive sites on the
alumina surface with formation of Na-F and/or Al-F
species and with additional monolayer being
physically adsorbed on top of chemisorbed layer 15.
However, Wagener et. al. observed that alumina
readily adsorbs HF under either hydrous or anhydrous
conditions16. Some authors17,18 believe that certain num-
ber of H2O molecules is needed to bind HF molecules to
the surface of alumina. Others19 believe that an
Al2O3·nHF type of compound is formed which is trans-
formed into AlF3 when heated above 300 °C. Several
research works have shown that both mechanisms took
place depending on process conditions. However,
Wagener et. al.16 concluded that H2O did not play an
essential role in the alumina-HF interaction. HF was not
only adsorbed physically, but it reacted with the exposed
alumina surface to form an AlOF or Al(F,OH)3 type of
compound. If a large excess of HF has passed over
alumina also AlF3 was formed (and alumina particles
became soft and friable).
Our research represented a more theoretical analysis
of chemical bonding reactions. There exist two
possibilities for HF bonding on primary alumina. The
first one is the Lewis acid-base complex and the other
one is the theory of hydrogen bonding. Lewis theory
claims that acid can take electron pair and base can
donate it. Al3+ is by definition Lewis hard base and F- is
Lewis hard acid. Due to their chemical properties they
form together most stable complex.
The other theory is the theory of hydrogen bonding.
This theory was already well described in some
papers14,20,21. Basically, it is a surface process where
humidity is present. The reaction mechanism involves
few steps: H2O adsorption forming an aqueous layer on
the alumina surface; HF adsorption to acidify the surface
water layer; dissolution of the alumina surface to form
AlO2– and AlOH2–, and precipitation of AlFx(OH)3–x ·
nH2O. The overall adsorption rate is controlled by the
rate of the surface chemical reactions rather than by
transport of fluid phase, mass transfer, and intraparticle
diffusion.
2 EXPERIMENTAL WORK
Smelting grade alumina samples, both primary and
reacted, were analyzed as-received from TALUM
company, Kidri~evo. The reacted alumina was prepared
in industrial conditions. Primary alumina passed through
the dry scrubber system where HF from electrolytic cells
was bonded in controlled reactor conditions to vapour-
phase HF. The temperature in the dry scrubber system
was about 90 °C (363 K). The flow rate of air sucked
from electrolytic cells into dry scrubber system was of
500 000 m3/h and all the air passed through six reactors
into which new (primary) alumina was dosed.
Electron microscope was used to analyze bonding of
fluoride onto primary alumina. Sandy alumina before
and after the dry scrubbing was examined in scanning
electron microscope (SEM – JEOL-JSM 6500F). There
were made also micro-chemical analyses with energy
dispersive X-ray spectroscopy (EDS Oxford Instruments
INCA x-sight ENERGY 400). This technique is an
analytical technique used for the analysis or of a
sample, limited to detection of about 0.1 % of chemical
elements and to 1–3 µm of analyzed volume, depending
on analyzed materials.
I. PAULIN ET AL.: MECHANISMS OF HF BONDING IN DRY SCRUBBER IN ALUMINIUM ELECTROLYSIS
190 Materiali in tehnologije / Materials and technology 43 (2009) 4, 189–193
Cross-section micro-chemical EDS analyses of
fluorinated alumina grains were made with larger grains
to find differences in fluoride concentrations on the
surface and in the interior of grains. Alumina sample was
prepared as metallographic sample with grinding and
polishing.
Chemical analyses of fluoride in the fluorinated
alumina were made by standard Pechiney method with
ion-selected electrode (Thermo ORION EA940, elec-
trode: Thermo 9409 BM).
3 RESULTS AND DISCUSSION
Alumina grains had large surface area because of
their morphology (Figure 1). Larger was the surface
more of vapour-phase HF could be bonded onto alumina
in reactors, and less HF escaped into environment
through the chimney after dry scrubbers.
Some authors22 used TGA, XRD and NMR and
proposed that fluorides were adsorbed onto alumina in at
least 2 ways, i.e. as mobile and as rigid species. Though
no real structures were identified, it was hypothetically
proposed that adsorbed fluoride existed as surface-
bonded species or as fluoride incorporated into the
alumina structure network (Al-F-Al bonds), respectively.
We have used EDS technique to analyze fluorinated
alumina grain to find differences in fluoride concen-
trations on the surface of alumina grain and in its
interior. Figure 2 shows fluorinated alumina grain with a
line of discrete spots analyzed with the EDS.
Results of EDS analyses are presented in Figure 3.
More fluoride was detected on the grain surface than in
the interior. This indicated that adsorption as well as
diffusion progressed from surface into interior. However,
there was difference in intensity of fluoride bonding;
greater amount of fluoride was found on the surface.
Reduced oxygen content and increased fluoride content
were determined on the grain surface. The ratio of
elements detected with the EDS enabled to propose that
AlF3 and Al2O3 were the most probable compounds on
the grain surface. Alumina represented substratum on top
of which aluminium fluoride was bonded.
Results confirmed the theory of Lewis acid-base
complex since by definition Lewis hard acid Al3+ and
Lewis hard base F- most likely formed strong complex.
Reaction: Al2O3 + 6 HF  2 AlF3 + 3 H2O took place on
the surface of alumina grains. Less oxygen and more
fluoride was present there (Figure 3). On the other hand,
alumina content was constant through the whole grain
cross section.
However, the hydrogen bonding reaction was still
taking place in the process. It depended on temperature
in reactors that varied with the surrounding temperature
and with the humidity level in reactors. Results of
chemical analyses of fluoride in fluorinated alumina
from dry scrubbers enabled to determine fluoride
contents in different periods of year. In summer when
the average day temperature of surrounding was between
I. PAULIN ET AL.: MECHANISMS OF HF BONDING IN DRY SCRUBBER IN ALUMINIUM ELECTROLYSIS
Materiali in tehnologije / Materials and technology 43 (2009) 4, 189–193 191
Figure 3: EDS discrete-point line analysis of fluorinated alumina
Slika 3: Linijska analiza fluoriranega zrna glinice z EDS-tehniko
Figure 1: SE image of grain of alumina
Slika 1: Slika sekundarnih elektronov z vrsti~nim elektronskim mi-
kroskopom zrna glinice
Figure 2: SE image of EDS discrete-point line analysis of fluorinated
alumina
Slika 2: Slika sekundarnih elektronov z vrsti~nim elektronskim mi-
kroskopom zrna glinice z ozna~enimi mesti linijske analize EDS
fluoriranega zrna glinice
20 °C and 25 °C (month average; every-day temperature
was measured at 2 p. m.) also temperature in reactors
was higher, above 105 °C. In winter, temperature in
reactors was only about 80 °C since the surrounding
temperature was about 0 °C. When temperature in
reactors was above the temperature of water evaporation
more fluoride was bonded onto alumina. Thus in warmer
months, from June to September, more fluoride was
detected in the fluorinated alumina (Figure 4). In
conditions when humidity was low more fluoride was
bonded as AlF3 compound by the principle of Lewis
acid-base complex. In colder months when the tempe-
rature in reactors did not exceed 85 °C, more H2O was
present and more fluoride was bonded by hydrogen
bonding mechanism. However, relation between the
humidity level and the fluoride bonding was obvious;
higher temperature, less humidity, and more fluoride was
loaded onto alumina. Molecules of water occupied the
surface of alumina and thus less fluoride could be
adsorbed. Mechanisms of possible bonding 13 of
vapour-phase HF on new (primary) alumina are shown
in Figure 5. One possibility was that fluoride was
bonded directly on alumina (aluminium atoms) instead
of on terminal hydroxyls (OH groups) (Figure 5 A). On
the other hand, fluoride bridging to alumina by hydrogen
bonding could exist too. The most efficient bonding of
fluoride was achieved when fluoride was bonded as
bridging fluoride and also as terminal fluoride (Figure 5
B).
Small amounts of fluoride were detected with the
EDS analytical technique also in the interior of grains.
Continuous adsorption of fluoride, in excess of surface
monolayers, took place with diffusion of surface bonded
fluor into the alumina lattice to displace oxygen in
bridging sites. This could also occur by direct continuous
reaction of alumina with vapour-phase fluoride to
displace the bridging oxygen, followed by continuous
diffusion of fluoride into alumina surface. Growth of
aluminium fluoride on surface impeded but did not stop
the continuous diffusion of fluor into the alumina lattice
that could cause alumina decay in reactors due to limited
contact time inside the reactor.
4 CONCLUSIONS
Comparison of data obtained in this study with
observations collected from previously published
examinations of related systems strongly supports
findings on the dry scrubber reactor chemistry and
retention of fluoride. Significant conclusions can be
made from chemical microanalyses of alumina grains
and relations between the fluoride bonding and industrial
conditions depending on external conditions.
Micro-chemical analyses confirmed the theory of
fluoride bonding on the surface of alumina. A very small
amount of fluoride was also found in the interior of
grains, but diffusion gradient was not large since the
layers of bonded fluoride on the surface have hindered
penetration of fluor into the interior of grains.
Important finding of this paper was the correlation
between chemical reaction of vapour-phase HF and
primary alumina in dry scrubber reactors and the
industrial temperature conditions. In the warmer season
when average daily temperature was 20 °C to 25 °C,
temperature in reactors increased to over 100 °C and
chemical analyses confirmed higher contents of fluoride
on reacted alumina. Less fluoride was detected in
periods when the average daily temperature was below 5
°C and temperatures in reactors were around 80 °C. This
temperature difference influenced the fluoride adsorp-
tion, because it was related to chemical reaction that took
place on the surface of alumina grains. Less fluoride was
bonded with hydrogen bridges and more of it was
chemisorbed directly onto the surface as AlF3.
5 REFERENCES
1 M. M. Hyland, B. J. Welch, J. B. Metson, in Book Changing Know-
ledge and Practices Towards Minimising Fluoride and Sulphur
Emissions from Aluminium Reduction Cells, Editor, TMS, Norway,
2000, 119
2 R. Huglen, H. Kvande: Global Considerations of Aluminium Elec-
trolysis on Energy and the Environment, Light Metals 1994, 373–380
3 G. Wedde: Emission Control, 18th Int’l Course on Process Me-
tallurgy of Aluminium, Trondheim, Norway, 1999
I. PAULIN ET AL.: MECHANISMS OF HF BONDING IN DRY SCRUBBER IN ALUMINIUM ELECTROLYSIS
192 Materiali in tehnologije / Materials and technology 43 (2009) 4, 189–193
Figure 4: Chemical analyses of fluoride in fluorinated alumina
Slika 4: Kemijska analiza fluora v fluorirani glinici
Figure 5: Scheme of mechanisms of vapour-phase HF bonding onto
primary alumina13
Slika 5: Shema vezave HF na primarno glinico13
4 M. Karlsen, V. Kielland, H. Kvande, S.B. Vestre: Factors influ-
encing cell hooding and gas collection efficiencies, Light Metals
1998, 303–310
5 M. Karlsen, R. Huglen: Reduction of fluoride emissions, Proc. 6th
Australasian Al Smelting Workshop, 1998, 703–720
6 E. Sum: A study of fugitive emissions in reduction lines, Proc. 6th
Australasian Al Smelting Workshop, 1998, 677–682
7 J. L. Henry: A study of the factors affeting fluoride emissions from
10,000 ampere experimental aluminium reduction cells, Extractive
Metallurgy of Aluminum Vol 2, G. Gerard ed., Interscience, New
York, 1963, 67–81
8 K. Grjotheim, H. Kvande, K. Motzfeldt, B. Welch: The formation
and composition of the fluoride emissions from aluminum cells, Can.
Metall. Quart., 11, (1972) 4, 585–599
9 W. Haupin, H.Kvande: Mathematical model of fluoride evolution
from Hall-Héroult cells, Light Metals 1993, 257–263
10 E. Dernedde: Estimation of fluoride emissions to the atmosphere,
Light Metals 1998, 317–322
11 S. J. Lindsay: Effective techniques to control fluoride emissions,
Light Metals 2007, 199–203
12 O. D. Tkach, A. E. Bazhenov, V. I. Ganchukov, V. F. Kharin, L. V.
Isaeva: Some physicochemical characteristics of dry extraction of
HF from exhaust gases of aluminum cells, Translated from Zhurnal
Prikladnoi Khimii, 49 (1976) 3, 642–644
13 N. R. Dando: Adsorption/entrainment of fluoride in smelting grade
alumina: surface chemical speciation and adsorption mechanism,
Light Metals 2005, 133–139
14 J. F. Coyne, P. J. Wong, M. S. Wainwright, M. P. Brungs: Factors
influencing hydrogen fluoride adsorption on alumina, Light Metals
1989, 113–119
15 W. D. Lamb: A study of the equilibrium adsorption of hydrogen
fluoride on smelter grade aluminas, Light Metals 1978, 425–444
16 J. B. Wagener, J. J. Hancke, P. A. Carstens: Adsorption of HF on
alumina, Journal of Thermal Analysis, 39 (1993), 1069–77
17 M. Baverez, R. De Marco: A study of the adsorption of hydrogen
fluoride on certain smelter grade alumina, Light Metals, 1980,
891–907
18 J. F. Coyne, M. S. Wainwright, M. P. Brungs, A. N. Bagshaw: The
influence of physical and chemical of alumina on hydrogen fluoride
adsorption, AIME Light Metals 1987, 35–38
19 V. S. Burkat, V. S. Dudorova, V. S. Smola, T. S. Chagina: Physico-
chemical properties of alumina used for removing fluorides in the
dry cleaning systems, AIME Light Metals, TMS, USA, 1985,
1443–48
20 A. R. Gillespie, M. M. Hyland, J. B. Metson: Irreversible HF adsorp-
tion in the dry-scrubbing process, JOM-Journal of the Minerals
Metals & Materials society, 51 (1999), 5, 30–32
21 R. G. Haverkamp, J. B. Metson, M. M. Hyland, B. J. Welch: Adsorp-
tion of hydrogen fluoride on alumina, Surface and Interface
Analysis, 19 (1992), 139–144
22 V. Burkat, V. S. Dudorova, V. S. Smola, T. S. Chagina: Physico-
chemical properties of alumina used for removing fluorides in the
dry cleaning systems, AIME Light Metals 1987, 1443–8
I. PAULIN ET AL.: MECHANISMS OF HF BONDING IN DRY SCRUBBER IN ALUMINIUM ELECTROLYSIS
Materiali in tehnologije / Materials and technology 43 (2009) 4, 189–193 193

A. KOCIJAN ET AL.: THE CORROSION BEHAVIOUR OF DUPLEX STAINLESS STEEL ...
THE CORROSION BEHAVIOUR OF DUPLEX
STAINLESS STEEL IN CHLORIDE SOLUTIONS
STUDIED BY XPS
XPS RAZISKAVE KOROZIJSKEGA VEDENJA DUPLEKSNEGA
NERJAVNEGA JEKLA V KLORIDNIH RAZTOPINAH
Aleksandra Kocijan, ^rtomir Donik, Monika Jenko
Institute of metals and technology, Lepi pot 11, 1000 Ljubljana, Slovenia
aleksandra.kocijan imt.si
Prejem rokopisa – received: 2008-12-16; sprejem za objavo – accepted for publication: 2009-03-11
The evolution of the passive films formed on duplex stainless steel 2205 in a chloride solution was studied by X-ray
photoelectron spectroscopy, and their compositions were analysed as a function of depth. The passive films on the duplex
stainless steel contained the oxides of the two main elements, i.e., Fe and Cr. The alloying elements were found to improve the
corrosion resistance of duplex stainless steels; however, their content within the passive layer was negligible. A strong
chromium enrichment was observed in the passive layers with increasing chloride concentration. The potentiodynamic
measurements were used to determine the corrosion behaviour of duplex stainless steel under different concentrations of
chloride solution.
Keywords: duplex stainless steel, XPS, passive films
Z rentgensko fotoelektronsko spektroskopijo (XPS) smo raziskovali nastanek pasivne plasti na povr{ini dupleksnega nerjavnega
jekla 2205 v raztopini natrijevega klorida. Pasivna plast na povr{ini tega jekla je vsebovala oksida dveh glavnih elementov Fe in
Cr. Legirni elementi pove~ajo korozijsko odpornost dupleksnih nerjavnih jekel, njihova vsebnost v pasivni plasti pa je
zanemarljiva. Z ve~anjem koncentracije kloridnih ionov se je mo~no pove~ala vsebnost kroma v pasivni plasti.
Klju~ne besede: dupleksno nerjavno jeklo, XPS, pasivne plasti
1 INTRODUCTION
Duplex stainless steels (DSSs) with a ferrite/austenite
volume ratio of about 1:1 have been recognized as good
corrosion-resistant materials in various aqueous environ-
ments 1. The high Cr content together with the high Mo
and N contents gives rise to a high pitting-corrosion
resistance in chloride solutions. The presence of an
approximately 50 % volume of ferrite results in an
increase in the strength, as compared with the austenitic
stainless steels 2,3. It is generally known that DSS is also
more resistant to stress corrosion cracking in chloride-
containing solutions 4. Molybdenum increases the
stability of the passive film and, therefore, the ability of
the stainless steel to resist the localised corrosion,
including pitting and crevice corrosion, particularly in
environments containing chloride ions 4.
Various spectroscopic techniques have been used to
study the corrosion behaviour of duplex stainless steels.
Souto et al. 5 studied the passivation and the resistance to
pitting corrosion of duplex stainless steel in neutral and
alkaline buffered solutions, with and without chloride
ions. The presence of NaCl enhanced the metal’s electro-
dissolution through the passive layer. The interpretation
of the results was based on the presence of Cr and Ni in
the alloy. Antony et al. 6 showed the aggressiveness of
sulphate-reducing bacteria in a marine environment,
which plays an important role in the corrosion of duplex
stainless steel. The initiation of the attack was evident
from SEM and AFM studies. In addition, ESCA studies
revealed that under anaerobic conditions sulphidation of
the passive film occurs. Abreu et al. 7 pointed out the
stabilising effect of molybdenum on the surface of the
passive film, enhancing the formation of a layer on
duplex stainless steel with a higher Cr/Fe ratio.
In the present work, 2205 duplex stainless steel was
studied in various sodium chloride solutions. The com-
position and the depth profiles of the oxide films formed
on 2205 duplex stainless steel at various potentials were
studied by XPS analysis.
2 EXPERIMENTAL
Duplex 2205 stainless steel was investigated. Its
composition was confirmed by analytical chemical
methods, as shown in Table 1. The experiments were
carried out in 0.9 %, 2 % and 3.5 % solutions of sodium
chloride. All the chemicals were from Merck, Darmstadt,
Germany. The test specimens were cut into discs of 15
mm diameter. The specimens were polished with SiC
emery paper down to 1000 grit prior to the electroche-
mical studies, and to 4000 grit prior to the XPS studies,
and then rinsed with distilled water. The specimens were
then embedded in a Teflon PAR holder and employed as
the working electrode. The reference electrode was a
Materiali in tehnologije / Materials and technology 43 (2009) 4, 195–199 195
UDK 669.14.018.8:620.193:620.18 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 43(4)195(2009)
saturated calomel electrode (SCE, 0.242 V vs. SHE) and
the counter electrode was a high-purity graphite rod.
Table 1: The composition of the 2205 duplex stainless steel (w/%)
Tabela 1: Sestava dupleksnega nerjavnega jekla 2205 (w/%)
Material Cr Ni Mn Si P S C Mo
duplex 2205 22.74 5.74 1.37 0.38 0.032 0.001 0.03 2.57
The potentiodynamic measurements were recorded
using an EG&G PAR PC-controlled potentiostat/galva-
nostat Model 263 with M252 and Softcorr computer
programs. The specimens were immersed in the solution
1 h prior to the measurement in order to stabilize the
surface at the open-circuit potential. The potentio-
dynamic curves were recorded, starting at 250 mV more
negative than the open-circuit potential. The potential
was then increased, using a scan rate of 1 mV s–1, until
the transpassive region was reached.
The passive layers on the alloy’s surface were
formed under potentiostatic conditions for 1 hour at
potentials of –0.2 V, 0.4 V, 0.8 V and 1.2 V vs. SCE.
These potentials were chosen with reference to the
characteristic features of the polarization curve (see
Figure 1). After the electrochemical preparation, the
specimens were rinsed with distilled water, dried and
transferred to the analyzer chamber within an hour.
The XPS measurements were performed using a VG
Scientific Microlab 310F instrument, non-monochro-
matized Mg Ka radiation (E = 1253.6 eV) and a hemi-
spherical electron analyzer operating at a pass energy of
20 eV. The spectra were collected using Avantage 3.41V
data-analysis software, supplied by the manufacturer.
The thickness of the passive layer was determined by
argon-ion sputtering for different time intervals. The
reference sputtering rate was 0.01 nm/s, and this was
calculated relative to Ta2O5 as the intensity of the
oxygen signal decreased to a half 9. Sputter depth
profiles were measured for the passive films formed at
two selected potentials, –0.2 V and 1.2 V, in all three test
solutions containing a different concentration of NaCl.
After a background subtraction, according to Shirley 10,
and processing with CasaXPS software developed by
Fairley 11, the XPS signals were separated into the
contributions from the different species. The spectra
were evaluated using the parameters of standard peaks.
Figure 2 presents examples of the deconvoluted
spectra. To calculate the composition of the passive layer
the peak areas were corrected with the corresponding
photo-ionization cross-sections: σ (Fe 2p3/2) = 9.68, σ (Cr
2p3/2) = 6.90, σ (Mo 3d5/2) = 5.62, σ (Ni 2p3/2) = 13.04,
and σ (O 1s) = 2.51 12, 13.
3 RESULTS
Figure 1 shows the potentiodynamic curves for 2205
DSS in mass fractions 0.9 %, 2.0 % and 3.5 % of sodium
chloride solutions. After 1 h of stabilization at the
open-circuit potential, the corrosion potential (Ecorr) for
the 2205 DSS in all three solutions was approximately
–0.300 V. Following the Tafel region, the alloy exhibited
passive behaviour. The extent of the passive range
slightly decreased with the increasing chloride concen-
tration. The passive range is limited by the breakdown
potential (Eb), which corresponds to the oxidation of
water and the transpassive oxidation of metal species.
The breakdown potential for the 2205 DSS in a 0.9 %
A. KOCIJAN ET AL.: THE CORROSION BEHAVIOUR OF DUPLEX STAINLESS STEEL ...
196 Materiali in tehnologije / Materials and technology 43 (2009) 4, 195–199
Figure2: Examples of fitted Cr 2p3/2, Fe 2p3/2, Ni 2p3/2 and Mo 3d
XPS-spectra
Slika 2: Primeri obdelanih Cr 2p3/2, Fe 2p3/2, Ni 2p3/2 and Mo 3d
XPS-spektrov
Figure 1: Polarisation curves recorded for the 2205 duplex stainless
steel in chloride solutions with three different concentrations
Slika 1: Polarizacijske krivulje dupleksnega nerjavnega jekla 2205 v
raztopinah natrijevega klorida z razli~nimi koncentracijami
chloride solution was approximately 1.10 V, and this
moved towards more negative values with an increasing
chloride concentration, i.e., to 1.03 V and 1.00 V,
respectively.
The XPS spectra were recorded after exposures at
potentials of special interest, in close correlation to the
polarization curves. The samples were oxidised at the
selected potentials prior to the XPS measurements for 30
min.
Figure 3 presents the cationic fractions of the passive
layer after the oxidation of the 2205 DSS in the presence
of three different chloride solutions. The bulk com-
position of the alloy is indicated by the dotted lines, so
that the relative enrichment and depletion of the
particular elements are clearer. In all three solutions the
fraction of Fe was less than that of the bulk con-
centration, and this effect increased with the chloride
concentration. The passive layer was Cr-enriched
compared to the bulk concentration; the Cr concentration
was somewhat higher at lower potentials and slightly
decreased at higher anodic potentials. The concentration
of the Ni was strongly reduced compared to the bulk
concentration; that of the Mo was slightly increased at
higher anodic potentials compared to the bulk concen-
tration. This result suggests the preferential dissolution
of Fe and Ni from the passive layer in the presence of
chloride ions.
Figures 4 and 5 show the composition depth profiles
for the 2205 DSS after the oxidation at –0.2 V and 1.2 V
in the presence of three different chloride solutions. The
profiles present the relative concentrations of the
A. KOCIJAN ET AL.: THE CORROSION BEHAVIOUR OF DUPLEX STAINLESS STEEL ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 195–199 197
Figure 4: Depth profiles of the 2205 duplex stainless steel as a
function of sputtering time after oxidation at –0.2 V in (a) 0.9 % NaCl,
(b) 2 % NaCl and (c) 3.5 % NaCl
Slika 4: Globinski profili dupleksnega nerjavnega jekla 2205 v
odvisnosti od ~asa jedkanja po oksidaciji pri –0,2 V v (a) 0,9 % NaCl,
(b) 2 % NaCl in (c) 3,5 % NaCl
Figure 3: Cationic fractions of the passive layer on the 2205 duplex
stainless steel in (a) 0.9 % NaCl, (b) 2 % NaCl and (c) 3.5 % NaCl as
a function of the oxidation potential. The dashed lines denote the bulk
values of the atomic fractions for Fe, Cr, Ni and Mo within the alloy.
The oxidation time was 30 min.
Slika 3: Kationske frakcije povr{ine pasivne plasti na dupleksnem
nerjavnem jeklu 2205 v raztopinah natrijevega klorida z razli~nimi
koncentracijami (a) 0,9 % NaCl, (b) 2 % NaCl and (c) 3,5 % NaCl v
odvisnosti od potenciala oksidacije. ^rtkane ~rte prikazujejo vrednosti
atomskih frakcij Fe, Cr, Ni in Mo v osnovnem materialu. ^as oksi-
dacije je 30 min.
metallic and oxidised species of a particular metal. After
each sputtering step the composition of each oxide and
metal species was plotted cumulatively along the
ordinate, so that the total composition amounted to 100
%. The precise stoichiometry of the oxide cannot be
determined due to ion-sputtering-induced effects, such as
the reduction of the oxidation state 16,17. The films were
seen to exhibit a gradual oxide-to-metal transition, with
complete oxidation only at the outermost surface. Never-
theless, the metallic parts of the individual components
appeared at the surface due to a less uniform oxidation or
an effect related to ion-sputtering reduction 16,17.
The main constituent of the passive layer formed at
the surface of the 2205 DSS after oxidation at –0.2 V in
all three solutions was Cr-oxide, and the total amount of
Cr was enriched compared to the bulk concentration
(Figure 4). The second major constituent of the passive
layer was Fe-oxide, although the total amount of Fe was
lower than in the bulk. The amounts of Mo- and
Ni-oxides were very low, especially for the Ni, in com-
parison to the bulk.
After the oxidation at 1.2 V the passive layer
consisted predominantly of Fe-oxides and was slightly
depleted in Cr-oxide (Figure 5), compared to the results
at –0.2 V, due to the effects explained later. The amount
of Ni species was significantly depleted at the outer
layers. With the increasing chloride concentration the
amounts of Fe and Ni species decreased due to the
preferential dissolution in the presence of chloride ions.
Therefore, the amount of Cr-oxide increased.
4 DISCUSSION
The corrosion-passivation processes of a duplex
stainless steel polarised in neutral chloride solutions
were investigated by using electrochemical and surface
analytical techniques. The results obtained in the present
work reveal that the passive film formed on the surface
of the DSS 2205 in the chloride solution at pH 7 contains
oxides of the two main elements, i.e., Fe and Cr. The
oxides of the alloying elements Ni and Mo are negligible
compared to the bulk, except for the increased Mo
concentration in the transpassive region. Previous studies
have confirmed that for stainless steels the amount of
nickel present in the passive film is very small 18,8,19.
Abreu et al. 8 also observed only a small amount of Ni
distributed along a passive film of DSS. Milo{ev et al. 18
reported the strong depletion of Ni in the layer formed
on AISI 316L in a neutral solution compared to the bulk
content. Clayton et al. 19 emphasized that Ni is rarely
observed in the passive films of stainless steels formed
in solutions containing chloride ions. From the equili-
brium diagrams it is expected that at potentials anodic to
0 V, chromium would dissolve preferentially in the
buffer solution 14. The XPS results confirm this predic-
tion: when the DSS is passivated at potentials = 0 V, the
chromium level slightly decreases and that of the iron
increases towards the oxide-solution interface (Figure
5).
Increasing the chloride concentration causes a sub-
stantial reduction in the potential range for passivity and
further growth of the oxide film is hindered as the metal
electro-dissolution becomes the dominant electrochemi-
cal process 3. The passivity of stainless steels arises from
the high corrosion resistance exhibited by the Cr(III)
oxide-hydroxides present in the passivating layers. The
role of the alloyed chromium in enhancing the passivity
of stainless steels is frequently explained in terms of a
percolation model of passivation 3. It is considered that
chromium forms insoluble Cr2O3, and a continuous
network of Cr-O-Cr-O is then produced, which prevents
the dissolution of iron.
The addition of nickel has a beneficial influence on
the corrosion resistance of ferritic steels 3. However, its
influence is less effective than in the case of chromium,
A. KOCIJAN ET AL.: THE CORROSION BEHAVIOUR OF DUPLEX STAINLESS STEEL ...
198 Materiali in tehnologije / Materials and technology 43 (2009) 4, 195–199
Figure 5: Depth profiles of the 2205 duplex stainless steel as a
function of sputtering time after oxidation at 1.2 V in (a) 0.9 % NaCl,
(b) 2 % NaCl and (c) 3.5 % NaCl
Slika 5: Globinski profili dupleksnega nerjavnega jekla 2205 v
odvisnosti od ~asa jedkanja po oksidaciji pri 1,2 V v (a) 0,9 % NaCl,
(b) 2 % NaCl and (c) 3,5 % NaCl
since Ni(II) ions do not anchor the water molecules and
the hydroxyl ions enough to prevent the chloride ions
from migrating through the passivating layer by
replacing the water and hydroxyl groups, and therefore,
cause localised corrosion.
Molybdenum does not significantly change the
composition of the film, but its presence is believed to
contribute to the deprotonation of the hydroxide by
acting as an electron acceptor, creating an oxygen abun-
dance in the inner regions of the passive layer and
helping in the formation of a stable CrO3/Cr2O3 pro-
tecting oxide. The other effect of this reaction is the
production of protons; they will be drawn towards the
solution by the cation-selective properties of the outer
layer. This will increase the proton activity on the
surface and assist in the formation of ammonium ions.
This is, according to Olsson 20, one of the conceivable
mechanisms of synergism between molybdenum and
nitrogen in the DSS. Clayton and Martin emphasised that
nitrogen reactions, particularly at higher potentials, are
extremely sluggish, and that there is a possibility for the
formation of soluble oxidised species, which may be
dispersed in the solution and not observed in the surface
analyses 21. Our results agree with this assertion, since
the concentration of Mo in the passive layer was strongly
diminished compared to the bulk and N was not detected
at all. However, this induced modification in the film can
explain the high Cr/Fe ratio measured in the XPS depth
profiles, which indicates the better corrosion characte-
ristics of molybdenum-containing alloys.
5 CONCLUSIONS
The chemical composition of the film formed on
duplex stainless steel showed the presence of two main
oxides, i.e., Fe and Cr. The oxides of the alloying
elements Ni and Mo were negligible compared to the
bulk. A slight decrease of the chromium content close to
the oxide/solution interface at higher anodic potentials
was also observed. The compositional changes in the
passive film can be explained on the basis of the
selective solubility of the oxides at the applied
potentials. Molybdenum mainly enhances the effect of
other passivating species, i.e., Cr, more than acting
directly in the passivating process in chloride media
since its content in the oxide layer is minor.
6 REFERENCES
1 S. T. Tsai, K. P. Yen, H. C. Shih, Corros. Sci., 40 (1998), 281–295
2 F. Tehovnik, F. Vodopivec, L. Kosec, M. Godec, Mat. Tech., 40
(2006), 129–138
3 A. Kocijan, ^. Donik, M. Jenko, Corros. sci., 49 (2007), 2083–2098
4 J. A. Platt, A. Guzman, A. Zuccari, D. W. Thornburg, B. F. Rhodes,
Y. Ossida, D. W. Thornburg, B. F. Rhodes, Y. Ossida, B. K. Moore,
American Journal of Ortodontics and Dentofacial Orthopedics, 112
(1997), 69–79
5 R. M. Souto, I. C. Mirza Rosca, S. Gonzales, Corrosion, 57 (2001),
300–306
6 P. J. Antony, S. Chongdar, P. Kumar, R. Raman, Electrochim. Acta
52 (2007), 3985–3994
7 C. M. Abreu, M. J. Cristóbal, R. Losada, X. R. Nóvoa, G. Pena, M.
C. Pérez, Electrochim. Acta 49 (2004), 3049–3056
9 K. Hashimoto, K. Asami, Corros. Sci. 19 (1979) 3, 251–260
10 D. A. Shirley, High-resolution X-ray photoemission spectrum of
valence bands of gold, Phys. Rev. B5 (1972), 4709
11 N. Fairley, CasaXPS VAMAS Processing Software, http://www.
casaxps.com/
12 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom. 8 (1976),
129–137
13 R. F. Reilman, A. Msezane, S. T. Manson, J. Electron Spectrosc.
Relat. Phenom. 8 (1976), 389–394
14 N. Ramasubramanian, N. Preocanin, R. D. Davidson, J. Electrochem.
Soc. 132 (1985), 793–798
15 P. Delahay, New instrumental methods in electrochemistry Inter-
science, New York, London, 1954, 115–145
16 W. H. Hocking, F. W. Stanchell, E. McAlpine and D. H. Lister,
Corros. Sci. 25 (1985), 531–557
17 N. S. McIntyre, D. G. Zetaruk, E. V. Murphy, Surf. Interface Anal. 1
(1979) 105–110
18 I. Milo{ev, H.-H. Strehblow, J. Biomed. Res. 52 (2000) 2, 404–412
19 C. R. Clayton, Y. C. Lu, J. Electrochem. Soc. 133 (1986) 2465–2472
20 C. A. Olsson, Corros. Sci. 37 (1995) 467–479
21 C. R. Clayton, K. G. Martin, in Proc. High Nitrogen Steels-HNS 88,
The Institute of Metals, London (1988) p. 256
A. KOCIJAN ET AL.: THE CORROSION BEHAVIOUR OF DUPLEX STAINLESS STEEL ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 195–199 199

S. RE[KOVI], Z. GLAVA[: THE APPLICATION OF AN ARTIFICIAL NEURAL NETWORK ...
THE APPLICATION OF AN ARTIFICIAL NEURAL
NETWORK FOR DETERMINING THE INFLUENCE OF
THE PARAMETERS FOR THE DEPOSITION OF A ZINC
COATING ON STEEL TUBES
UPORABA UMETNIH NEVRONSKIH MRE@ ZA DOLO^ITEV
DEBELINE CINKOVE PLASTI NA JEKLENIH CEVEH
Stoja Re{kovi}, Zoran Glava{
Faculty of Metallurgy, University of Zagreb, Aleja narodnih heroja 3, 44000 Sisak, Croatia
reskovic simet.hr
Prejem rokopisa – received: 2009-02-09; sprejem za objavo – accepted for publication: 2009-03-04
The influence of deposition temperature and time on the thickness of a zinc coating on tubes with different dimensions was
investigated. Backpropagation neural networks (BPNNs) were established to predict the zinc-coating thickness on the tubes
using the temperature of the zinc melt, the deposition time and the tube diameter as the inputs. A BPNN was used to determine
the optimal temperature range for zinc deposition on tubes with a diameter of 48.3 mm and determine the influence of the
deposition time on the zinc-coating thickness.
Key words: zinc-coating deposition, temperature, coating thickness, artificial neural networks
Povratna nevronska mre`a (BPNM) je bila uporabljena za oceno debeline plasti cinka na podlagi temperature cinkove kopeli,
trajanja cinkanja in premera cevi kot vstopnih podatkov. Z mre`o BPNM je bilo dolo~eno optimalno temperaturno podro~je za
cinkanje cevi s premerom 48,3 mm in vpliv trajanja cinkanja na debelino cinkove plasti.
Klju~ne besede: cinkanje cevi, temperatura, debelina cinkove plasti, umetne nevronske mre`e
1 INTRODUCTION
Zinc deposition is the most efficient and ecologically
most acceptable method of protecting steel from
corrosion. Because of the high affinity of zinc for
oxygen, in the presence of moisture and carbon oxide in
the atmosphere, the outer layer of the zinc is changed to
alkaline zinc carbonate, a layer (known as "white rust")
that prevents any advance of the corrosion process. In
comparison to other corrosion-protection methods, zinc
deposition provides long-term protection (20–50 years),
the process is automated and fast, the protective layer is
resistant to mechanical damage, the protection of
complex shapes is possible and a uniform thickness of
the protective layer is achieved.
In the last few years the demand for zinc-coated
tubes has increased significantly. Today, more than 75.0
% of the tubes produced use some sort of corrosion
protection. For certain types and applications of tubes,
the minimum thickness of the zinc coating is defined in
standards. However, the price of zinc is high and
increasing the thickness of the zinc coating significantly
increases the production costs. The zinc metal costs
amount to approximately 50.0 % of the costs of the
deposition process.
Zinc deposition is a very complex process 1,2. Molten
zinc in contact with steel forms alloys, which have
different compositions (different phases). Some of them
are brittle and are detrimental to the quality of the zinc
coating. The zinc-coating thickness depends on several
factors: the temperature of the molten-zinc bath, the
deposition (immersion) time and the chemical
composition of the steel 3,4. The process of forming the
zinc coating by reacting the zinc with iron is influenced
by carbon, phosphorus, silicon and aluminum. Carbon,
phosphorus and silicon have a significant, negative
influence on the quality of the zinc coating; they
promote the formation of brittle phases that cause
peeling of the zinc coating from the pipe 3.
The melting point of zinc is 419.4 °C and the
deposition process is carried out in the temperature range
from 430 °C to 460 °C. After immersion of the tubes the
zinc-melt temperature decreases, and it must not fall
below 430 °C. At higher temperatures the thickness of
the zinc coating increases and the share of the brittle
phases in the coating rapidly increases, also 1,3.
The deposition time depends on the mass of the tube
and must be equal to the time needed to equalize the
temperature of the tube and of the zinc bath. If the
immersion time is shorter, the zinc coating will be thick
and the surface of the tube will be rough because of the
crystallization of the zinc on the colder tube surface. If
the immersion time is too long, the zinc coating will be
brittle, see also 1–4. The angle of the tube extraction from
the zinc bath affects both the zinc-coating thickness and
its appearance. By increasing the angle of extraction, the
draining of the zinc from the surface of the tubes
Materiali in tehnologije / Materials and technology 43 (2009) 4, 201–205 201
UDK 669.586:681.32 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 43(4)201(2009)
increases and the thickness of the zinc coating is reduced
3. The temperature of the zinc bath is a very significant
parameter with regard to the zinc-coating thickness.
Thus, it is clear that determining the optimal process
parameters to obtain the desired zinc-coating thickness is
of great importance.
Based on its characteristics, the deposition of a zinc
coating from a zinc bath, as with most technical
processes, belongs to the group of complex and
nonlinear systems. This means that the temperature of
the zinc melt (i.e., the bath), the dependence of the
immersion time and the thickness of the zinc coating on
the tubes is not linear. Models that describe this system
based on a multiple linear-regression technique could be
applied more or less successfully, but only for specific
process conditions under which they have been
developed.
2 APPLICATION OF THE NEURAL NETWORK
In recent years, rapid progress in artificial intelli-
gence has enabled us to use a new method for processing
information – artificial neural networks (ANNs) (5).
ANNs are complex systems consisting of simple
elements (artificial neurons) operating in parallel, which
can successfully solve the system’s nonlinearity. These
elements (neurons), inspired by biological nervous
systems, are in a specific interaction, mutually and with
the environment of the system (the weights of the
artificial neural networks), in this way building a
functional unit. Two or more neurons may be combined
in a layer, and a particular network might contain one or
more layers (input layers, output layers and hidden
layers, also). The number of inputs to the network is
determined by the problem and the number of neurons in
the output layer is defined by the number of outputs
required by the problem. However, the designer has to
define the number of layers between the network input
and the output layer and the size of the layers (the
number of neurons).
The network function is determined by the
connections between the elements. We can train (i.e.,
teach) the ANN to perform a particular function by
adjusting the values of the connections (i.e., the weights)
between the elements. Each input to the neuron is
weighted with an appropriate weight. The sum of the
weighted inputs and the bias forms the input to the
neuron transfer function, which maps a neuron’s (or
layer’s) net output to its actual output. The most popular
transfer functions are linear, log sigmoid, hyperbolic,
tangent, sigmoid, etc. The bias is much like the weight,
except that it has a constant input of 1.
A properly trained ANN can map input to output
patterns with minimal error between the modeled and the
measured output values. The testing of an ANN follows
its training and it is performed with a new input data set,
which is not included in the input data set for training the
ANN.
Currently, backpropagation is the most important and
the most widely used algorithm for neural network
training. This algorithm uses the mean squared error and
the gradient descent for training.
The goal of these examinations was to determine the
influence of the immersion parameters, in the first place
the influence of the zinc-bath temperature and the
immersion time on the zinc-coating thickness and
establish neural network models for predicting of the
zinc-coating thickness using the deposition parameters as
the inputs.
3 EXPERIMENTAL
The information that the zinc-coating thickness on
the tubes was much greater than that prescribed in the
EN10240 standard was confirmed in a preliminary study
6. In the frame of this study the thickness of the zinc
coating was assessed for six different tube dimensions
and a larger number of samples.
The technological process of the zinc deposition on
the tubes consists of a chemical treatment, drying, zinc
deposition, removal of the excess zinc and cooling. The
tubes are immersed in a standard detergent suspension
for 10 minutes, washed out in water with two to three
immersions in an inclined position for draining the
detergent and pickled in (50 g/L) sulphuric acid at 65 °C
for 20 min. Then the tubes are rinsed three times and
treated with suspensions of ZnCl2 (155–185 g/L) and
NH4Cl (180–200 g/L) and finally dried at 120 °C and
coated.
The chemical composition of the zinc bath was
controlled and maintained in the prescribed range. The
temperature of the bath was only changed during the
process if required by the processing, for example, a
change of the tube dimensions or the coating thickness, a
standstill of the coating line, etc. The immersion time
was selected with regard to the tube dimensions and then
varied to determine its influence on the zinc-coating
thickness. After the zinc deposition, the excess zinc from
the tubes was removed with compressed air at a pressure
of 2 bar and cooled in water. Also, on rejected tubes the
coating thickness was not assessed for the collection of
the required data. The tubes’ assessment data are shown
in Table 1.
Table 1: Overview of the performed measurements
Tabela 1: Pregled opravljenih meritev
Tube
diameter,
mm
Immersion
time, s
Number of
measure-
ments
Temperature, °C Zinc-coatingthickness, g/m2
min. max. min. max.
26.9 10.63 132 437 452 415 599
33.4 10.35 85 435 462 463 649
42.4 8.80 111 445 454 429 649
48.3 7.72 150 449 568 414 680
60.3 5.63 66 445 460 454 689
71.6 4.25 57 450 661 420 569
S. RE[KOVI], Z. GLAVA[: THE APPLICATION OF AN ARTIFICIAL NEURAL NETWORK ...
202 Materiali in tehnologije / Materials and technology 43 (2009) 4, 201–205
All the assessments were performed in real industrial
conditions on a quantity of 3830 t of coated tubes of six
different diameters. A total of 601 measurements of the
zinc-deposition temperature and the zinc-coating thick-
ness were performed.
4 RESULTS AND DISCUSSION
In the first phase of the examination all the measured
data were used to create the ANN (all the tubes’
diameters together) and the performance of the ANN in
predicting the thickness of the zinc coating on tubes was
checked with the coefficient correlation (R). The input
parameters for the ANN were the tube diameter, the
zinc-bath temperature and the deposition time, and the
output parameter was the thickness of the zinc coating.
When creating a neural network, it is important to
prevent any overfitting of the data. In this work, the
early-stopping method was used to improve the network
generalization and prevent the overfitting. According to
this method, the experimental data set is divided into
three subsets: the training data set, the validation data set
and the test data set. The training data set was used for
computing the gradient and updating the network’s
weights and biases. The validation data set was not
included in the training data set and was used to decide
when to stop the training. The test data set error was not
used during the training; it was, however, used for the
comparison of the different models, i.e., for the
evaluation of the performance of the networks. To
achieve the most efficient training, the input and the
output data are normalized before the training.
In this paper, different network architectures are
examined, with the aim to determine the networks with
the minimal generalization error. The best results were
achieved with the multilayer backpropagation neural
networks (BPNNs) trained using the Levenberg-Mar-
quardt algorithm.
The performances of the trained BPNNs were deter-
mined with regression analysis of the networks’ outputs
(predicted values of the zinc coating) and the corres-
ponding target values of the zinc coating were obtained
experimentally (Table 2). Figure 1 shows the perfor-
mance of the BPNN on the test data set.
Table 2: Coefficients’ correlation (R) for training, validation, tests
and the entire data set (all diameters of the tubes together)
Tabela 2: Koeficienti korelacije (R) za trening, preverjanje in preiz-
kuse ter pregled vseh podatkov (za vse cevi)
Data set Trainingdata set
Validation
data set
Test data
set
Entire data
set
R 0.823 0.828 0.809 0.821
Figure 1 shows a good network generalization,
which is confirmed by a high value of the coefficients’
correlation between the networks’ outputs and the
corresponding target measured values, confirming the
correctness of the network architecture and the proper
selection of the input network parameters.
With the goal of achieving a higher accuracy of
prediction of the zinc-coating thickness, a separate
neural network was established for each tube diameter.
The input parameters for the ANN were the temperature
of the zinc melt and the immersion time, and the output
parameter was the zinc-coating thickness. The
early-stopping method was used to improve the network
generalization and prevent the overfitting. To achieve the
most efficient training, the input and the output data
were normalized before the training.
Different networks’ architectures were examined to
determine the networks with a minimal generalization
error. The best results were achieved with the multilayer
backpropagation neural networks (BPNNs) trained using
the Levenberg-Marquardt algorithm.
The performances of the trained BPNNs were
checked with regression analyses between the networks’
outputs (predicted values of zinc coating) and the
corresponding target values of the zinc coating obtained
with the measurements (Table 3). Figure 2 shows the
performance of the BPNN on the test data set for each
tube diameter.
Figure 2 shows the good networks’ generalization,
confirmed by high values of the coefficients’ correlation
between the networks’ outputs and the corresponding
target values obtained by measuring (from 0.759 for
smallest diameter tubes (26.9 mm) to 0.947 for tubes
with a diameter of 60.3 mm). The correlation confirms
the correctness of the networks’ architectures and the
proper selection of input networks’ parameters.
Tubes with diameter of 48.3 mm represent the
majority of the zinc-coated product. For this reason, the
largest number of measurements (150 cases) was carried
S. RE[KOVI], Z. GLAVA[: THE APPLICATION OF AN ARTIFICIAL NEURAL NETWORK ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 201–205 203
Figure 1: Performance of the BPNN for the test data set. R – coeffi-
cient correlation, A – predicted values of zinc-coating thickness
(g/m2), T – target values of zinc-coating thickness (g/m2). All tested
tubes together
Slika 1: Performance BPNN za sklop podatkov: R – koeficient korela-
cije, A – napovedane debeline cinkove plasti (g/m2), T – ciljne debe-
line cinkove plasti (g/m2), vse cevi skupaj
out on tubes of this diameter (Table 1). In Table 4 some
measurements are shown.
Table 3: Coefficients’ correlation (R) for training, validation and test,
and the entire data set
Tabela 3: Koeficienti korelacije (R) za trening, preverjanje in preiz-
kuse
Tube
diameter,
mm
BPNN
label
Coefficient correlation, R
Training
data set
Validatio
n data set
Test data
set
Entire
data set
26.9 BPNN1 0.727 0.741 0.844 0.759
33.4 BPNN2 0.857 0.896 0.904 0.878
42.4 BPNN3 0.803 0.802 0.767 0.802
48.3 BPNN4 0.872 0.858 0.871 0.868
60.3 BPNN5 0.967 0.902 0.952 0.947
71.6 BPNN6 0.806 0.948 0.905 0.864
The high coefficients’ correlation between the net-
work outputs and the corresponding target values
obtained with the measurements (R = 0.868) confirms
204 Materiali in tehnologije / Materials and technology 43 (2009) 4, 201–205
S. RE[KOVI], Z. GLAVA[: THE APPLICATION OF AN ARTIFICIAL NEURAL NETWORK ...
Figure 2: Performance of the BPNN on the test data set for tubes of
different diameters: a) 26.9 mm (BPNN1), b) 33.4 mm (BPNN2), c)
42.4 mm (BPNN3), d) 48.3 mm (BPNN4), e) 60.3 mm (BPNN5), f)
71.6 mm (BPNN6). R – coefficient of correlation, A – predicted zinc-
coating thickness (g/m2), T – target zinc-coating thickness (g/m2)
Slika 2: Performance BPNN za sklop podatkov za cevi z razli~nim
premerom: a) 26.9 mm, b) 33,4 mm, d) 48,3 mm, e) 60,3 mm, f) 71,6
mm. R – koeficienrt korelacije, A – napovedana debelina plasti cinka
(g/m2), T – ciljna debelina plasti cinka (g/m2)
Figure 3: Dependence of zinc-coating thickness on zinc-melt tem-
perature. Immersion time: 84.0 s. Tube diameter: 48.3 mm. Results
obtained with BPNN4.
Slika 3: Odvisnost med debelino cinkove plasti d in temperaturo
taline; ~as v kopeli 84 s, premer cevi 48,3 mm. Rezultati BPNN4.
Figure 4: Dependence of zinc-coating thickness on zinc-melt tem-
perature. Immersion time: 92.0 s. Tube diameter: 48.3 mm. Results
obtained by BPNN4.
Slika 4: Odvisnost med debelino plasti cinka d in temperaturo kopeli;
~as v kopeli 92 s, premer cevi 48,3 mm. Rezultati BPNN4.
Table 4: Measured values of the zinc-deposition parameters and the
zinc-coating thickness (tubes of diameter 48.3 mm)
Tabela 4: Izmerjeni parametri za nanos cinka in debelina plasti cinka
(cevi s premerom 48,3 mm)
Ordinal
number
Immersion
time, s
Temperature of
zinc melt, °C
Zinc-coating
thickness, g/m2
1
84.0
446 398
2 455 625
3 450 418
4 460 652
5 459 482
… … … …
145
92.0
448 525
146 455 536
148 450 475
149 460 524
150 453 569
the suitability of the networks’ architecture. The optimal
deposition temperature was obtained for this tube
diameter (48.3 mm) with an ANN for both immersion
times of 84.0 s and 92.0 s (figures 3 and 4). The
zinc-melt temperature was varied in the interval from
430 °C to 456 °C for both immersion times.
According to the norm EN10240, the minimum
thickness of the zinc coating on the tube (the mass of
zinc coating per unit of surface) is 400 g/m2. In Figures
3 and 4 it is shown that the optimal temperature interval
for coating tubes with a diameter of 48.3 mm is 430–440
°C. In this temperature interval the immersion time did
not have an important influence on the thickness of the
zinc coating. The obtained results show that it is also
possible to decrease the immersion time.
5 CONCLUSIONS
– the investigation shows that it is possible to
determine the influence of the deposition parameters
on the thickness of the zinc coating and optimize the
coating by applying the artificial neural networks;
– different network architectures were examined to
determine the networks with a minimum genera-
lization error. The best results were achieved by
applying the multilayer backpropagation neural net-
works (BPNNs) trained using the Levenberg-Mar-
quardt algorithm;
– high values of the coefficients’ correlation between
the networks’ outputs and the corresponding target
values obtained by measuring show a good net-
works’ generalization for all the tube dimensions, all
together and for each tube dimension separately;
– for most of the tube stock produced, a tube of
diameter 48.3 mm, the dependence of the deposition
temperature and the zinc-coating thickness were
obtained. The optimal temperature range for the zinc
deposition on these tubes was 430–440 °C;
– additionally, it was found that the used immersion
time at the lower galvanization temperatures did not
have a significant influence on the zinc-coating
thickness, while at the temperatures above 442 °C,
with an increase in the immersion time the zinc-
coating thickness was increased too.
6 REFERENCES
1 D. Ash, Elektrogalvanising for the tube and pipe industry, Tube
specialty, 2008, 1–7
2 D. Gotal: Prakti~ni priru~nik, @eljezara Sisak, Sisak 1982, 184–205
3 www.ohgalvanizedtube.com
4 www.thermix.com
5 M. T. Hagan, H. B. Demuth, M. H. Beale, Neural Network Design.
Boston, PWS Publishing Company 1996
6 S. Re{kovi}, Analiza potro{nje cinka, PC [avne cijevi, @eljezara
Sisak, Sisak 2000
S. RE[KOVI], Z. GLAVA[: THE APPLICATION OF AN ARTIFICIAL NEURAL NETWORK ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 201–205 205

A. V. SHRAMKO ET AL.: THE APPLICATION OF THE PROGRAM QFORM 2D IN THE STAMPING ...
THE APPLICATION OF THE PROGRAM QFORM 2D IN
THE STAMPING OF WHEELS FOR RAILWAY
VEHICLES
UPORABA PROGRAMA QFORM 2D PRI KOVANJU KOLES ZA
@ELEZNI[KA VOZILA
Aleksandr Shramko1, Ilija Mamuzi}2, Valentin Danchenko1
1National Metallurgical Academie of Ukraine, Prospekt Gagarina 4, Dnepropetrovsk, Ukraine
2Faculty of Metallurgy, University of Zagreb, Aleja narodnih heroja 3, 44000 Sisak, Croatia
mamuzic simet.hr
Prejem rokopisa – received: 2008-11-20; sprejem za objavo – accepted for publication: 2009-04-15
Computers allow us to obtain and process data to improve and accelerate the analysis of the information required to optimize
the processes of metal forming and to minimize production costs. The computer program Qform 2D, adapted to wheel forming,
has been used for several years in the company "Interpipe – Nikopolski tube rolling plants" for solving problems related to the
selection of the optimal schemes for the hot plastic deformation of wheel blanks, the analysis of existing shapes and the
development of new shapes for the die-forging tool.
Key words: stamping, railway wheels, computer simulation, distribution and rate of deformation
Ra~unalniki omogo~ajo, da pridobimo in procesiramo podatke za izbolj{anje in pospe{enje analize podatkov za optimizacijo
procesa oblikovanja kovin in zmanj{anje stro{kov izdelave. Ra~unalni{ki program Qform, prilagojen za kovanje koles, se
uporablja `e ve~ let v podjetju "Interpipe- Nikopolski tube rolling plant" za re{evanje problemov, povezanih z izbiro optimalne
sheme za vro~e kovanje surovcev za kolesa, analizo uporabljenih oblik in za razvoj novih oblik orodij za utopno kovanje.
Klju~ne besede: kovanje, `elezni{ka kolesa, ra~unalni{ka simulacija, porazdelitev in hitrost deformacije
1 INTRODUCTION
Railway transport and the manufacturers of wheels
for railway transport are constantly facing more and
more complex problems related to the reliability and
durability of the rolling stock under conditions that are
becoming increasingly strict.
From an analysis of the development of the
technology for manufacturing the wheels for railway
vehicles of major producers it is clear that it is necessary
to increase the exploitation properties of wheels and to
optimize the hot forming of wheel blanks. The use of
modern computer facilities and methods for obtaining
and processing data allows us to improve and accelerate
the analysis of the data necessary to optimize the
processes of metal forming and, at the same time, to
minimize the costs of production.
The computer program Qform 2D, adapted to the
conditions of wheel manufacturing, has been used for
several years in the company "Interpipe – Nikopolski
tube rolling works" for solving problems connected with
the choice of the optimal schemes for the hot plastic
forming of wheel blanks as well as the analysis of
existing, and the development of new, grooves for the
die-forging tool.
2 EFFECT OF THE PARAMETERS OF HOT
PLASTIC DEFORMATION ON THE WHEEL
RIM’S MECHANICAL PROPERTIES
The effect of deformation on the mechanical
properties of the wheel’s steel was determined with
laboratory specimens and from stamped wheels. It is
related to the steel’s macrostructure and microstructure1.
In addition, the degree of deformation, the temperature
and the rate of deformation significantly affect the
mechanical properties as well as the "hardening-soften-
ing" processes occurring during the deformation,
including the time of the inter-deformation pauses.
During the hot plastic deformation at a temperature
above the temperature of recrystallization, the processes
of metal hardening and softening occur in parallel and
compete with each other. As a rule, during the hot
deformation, softening prevails over hardening.
A distinctive feature of the hot-forming process is the
stage of stable flow, when above a critical level the
deformation resistance becomes independent of the
increasing plastic deformation. The critical deformation
depends on the rate and the temperature of deformation
that affect the mechanism and the kinetics of the
hardening and softening processes.
The strain resistance increases with the increase of
the extent of hot deformation and the rate, which may
Materiali in tehnologije / Materials and technology 43 (2009) 4, 207–211 207
UDK 621.73.042:629.4 ISSN 1580-2949
Original scientific article/Izvirni znanstveni ~lanek MTAEC9, 43(4)207(2009)
even hinder the softening processes and increase the
deformation hardening, while the rate of deformation
also increases the temperature that then increases the rate
of metal softening.
Therefore, it is possible to regulate the steel’s micro-
structure, its homogeneity and stability with changes of
the deformation schemes’ schedules and temperature and
to affect, in this way, the mechanical properties of the
metal. In the company "Interpipe – Nikopolski tube
rolling works" the optimizing of the schedules of the
wheel blanks’ hot deformation was solved on the basis
of the results of a series of laboratory and industrial
experiments in cooperation with scientists from the Iron
and Steel Institute of the National Academy of Sciences
of Ukraine and the National Metallurgical Academy of
Ukraine using the computer program Qform.
In the investigations the effects of the degree, rate
and temperature of deformation on steel hardening were
determined with a Type 805 A/D2-dilatometer. The
investigations were performed on samples of steel cut
out from the ingot zone intended for the wheel rim. In
the process of plant forming the maximum steel
hardening is achieved during the initial stages of the
wheel blank’s deformation, while the deformation in the
last stages of forming must be sufficient to compensate
for the steel softening during inter-deformation pauses
with a decreasing time-length or with increasing the rate
of deformation in the following stages. The influence of
deformation pauses on the steel softening was studied
with dilatometry and the deformation of specimens at
different temperatures with various degrees of defor-
mation and different inter-deformation pauses2.
Various versions of the wheel blanks’ hot-defor-
mation schemes were suggested and investigated with
simulation by applying the program Qform 2D. The
computer model used in the present work is based on the
plastic-flow theory. In this theory, the deformed metal is
assumed to be an incompressible and rigid plastic body.
The model describing the plastic deforming includes:
– the differential equation of motion:
σ ij j, = 0 (1)
– the kinematic correlation:
ω ν νij i j j i= +
1
2
( ), , (2)
– the state equation:
σ σ δ
σ
ω
ωij ij ij= +0
2
3
) (3)
– the expression for incompressibility:
ν i i, = 0 (4)
– the heat-balance equation:
c T T
T
k T T( ) ( ) ( ( ) ( ))ρ
τ
βσω
d
d
div grad= + (5)
– the model of material flow stress in dependence on
the deformation parameters:
σ ε ω= f T( , , ) (6)
With σij, εij, νi being the components of the tensors of
the stress and the rate of deformation and the vector of
the flow rate; σ ε ω, , being the intensities of the stresses,
deformations and rates of deformation; σ0 the average
normal stress; δij the Kronecker symbol; T the tempera-
ture; β the coefficient of transition of mechanical energy
into heat (β = 0.9–0.95); ρ the density; c the thermal
capacity and k the thermal conductivity.
In equations (1)–(4) the rule of summation with
repeating indexes is used. The indexes i, j change from 1
to 2 for two-dimensional calculations and from 1 to 3 for
three-dimensional calculations.
The cooling of the blank during the transport from
the furnace to the press, as well as between the presses
and the rolling mill, is described with the equation of
thermal conductivity (5).
The heat boundary conditions on the free surface
include the heat exchange by convection and radiation.
The heat exchange on the contact surface is considered
with a heat-transfer coefficient, while the friction on the
contact surface is determined by applying Levanov’s
relation:
τ σ σ= −K s e n s( )
, /1 1 25 (7)
with Ks being surface constant; σn the pressure constant;
σs the yield stress.
The equations (1)–(4) are transformed into a system
of algebraic equations on the basis of virtual speed
principles and a FEM, where the components of the rate
vector and the average stress are nodes of unknown
values. The rate and the average stress are approximated
with square and linear functions of the form on
triangular elements. The generation and rebuilding of the
FE of the network are automated.
The heat-balance equation (5) is transformed by
means of Galerkin’s method into a system of differential
equations that are numerically integrated with respect to
time.
The thermomechanical problem is solved by
applying the method of successive approximations of the
mechanical and heat problem by entering the boundary
values.
In the strength calculation the tool is assumed to be
an elastic-plastic body with linear strain hardening and
submitted to the blank contact forces that depend on the
boundary conditions.
The results of the calculation of the deformation
parameters in the central zone of the wheel rim with a
mass of 400 kg according to the different schedules are
shown in Figure 1.
With the options of the program the deformation
parameters influencing the mechanical properties in
different zones of the wheel are investigated, e.g., the
A. V. SHRAMKO ET AL.: THE APPLICATION OF THE PROGRAM QFORM 2D IN THE STAMPING ...
208 Materiali in tehnologije / Materials and technology 43 (2009) 4, 207–211
rate and degree of deformation and the rate of
accumulation of the deformation can be determined. For
the presented case, the rate and degree of deformation
calculated by applying the program Qform are shown in
Figure 2.
The comparison of the experimental data on the
rheological properties of the wheel steel with the results
of the computer simulation allowed us to determine, and
afterwards to realize in the stamping process also, the
most suitable scheme of the distribution of rate and the
degree of deformation by passes for the multistage
stamping of the wheel blank and to obtain improved
mechanical properties in the central wheel-rim zone3.
3 INFLUENCE OF DEFORMATION SCHEME
ON THE HARDNESS OF THE WHEEL-HUB
METAL
The practice of manufacturing wheel pairs with
increased wheel hardness (HB = 320) in the train car
buildings plant revealed a substantial spread in the metal
hardness values (above HB = 20) around the opening
(bore) for the wheel hub on the internal and external
sides of the wheel. An irregular hardness distribution
over the length of the wheel hub bore generatrix affects
the accuracy of the geometrical parameters of the bore
turning before the wheel is pressed on the axis. The
irregular hardness distribution is often connected with
the coning of the bore of more than 50 µm, which is a
reason for rejecting the wheel.
The hardness (HB) of carbon steels is related to the
flow stress characterizing the plastic strain resistance of
steel ( f) by a linear dependence through the factor of
proportionality (C):
 f = C × HB
The steel’s plastic strain resistance depends on its
microstructure. In the present case, besides the micro-
structure, it includes the changes to the intra-grain and
dislocation structure, too.
The various mechanisms hindering the dislocations
slip that are widely used for increasing the plastic strain
resistance and hardness are related to the interaction of
dislocations with the presence of effective and uniformly
distributed obstacles, as well as grain boundaries.
The parameters of the hot plastic steel deformation,
such as the rate and degree of deformation, the degree of
total deformation, the temperature of deformation
including the length of inter-deformation pauses,
significantly affect the grain size of the wheel steel 2.
The influence of the these parameters on the
distribution of hardness along the generatrix of the
wheel-hub bore was investigated in the plant in the
frame of the production of wheels with D = 957 mm
A. V. SHRAMKO ET AL.: THE APPLICATION OF THE PROGRAM QFORM 2D IN THE STAMPING ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 207–211 209
Figure 1: Results of the calculation of the wheel-blank shape change
at different stages of the existing (a) and the experimental (b)
technology: 1 – upsetting in the 20-MN press; 2 – upsetting to a ring
in the 50-MN press; 3 – ring expansion (dilatation) with a punch on
the 50-MN press; 4 – stamping in the 100-MN press; 5 – stamping in
the 50-MN press
Slika 1: Rezultati izra~unov spremembe oblike surovca pri razli~nih
stopnjah sedanje (a) in poskusne (b) tehnologije. 1 – kr~enje v 20
MN-pre{i, 2 – kr~enje v obro~ v 50 MN-pre{i, 3 – {irjenje obro~a, s
trnom v 50 MN-pre{i, oblikovanje v 100 MN pre{i, oblikovanje v 50
Mn pre{i
Figure 2: Distribution of deformation (a) and of deformation rate (b)
in the central wheel-rim zone for the stages of the blanks’ forming
using different schemes of stamping
Slika 2: Porazdelitev deformacije (a) in hitrosti deformacije (b) v
sredi{~ni zoni oboda kolesa pri fazah oblikovanja surovca in pri
uporabi razli~nih shem oblikovanja
according to the standard GOST 10791-2004 as well as
wheels with increased hardness according to the techni-
cal conditions TU V 35.2-23365425-600; 2006. The
parameters of the deformation of the internal, central and
external zones in the wheel hub were determined with a
computer simulation of the wheel-stamping process
applying the program Qform 2 (Figure 3).
As shown in Figure 3, the true (logarithmic) defor-
mation of steel (∆e) on the hub internal side during
stamping the blank in the 100-MN press achieves a value
of 2.4; it decreases in the central zone and on the
external hub side to values of 1.2–1.4. At the same time,
the deformation rate () increases from the internal to
external side of hub from 0.1 s–1 to 8 s–1.
Earlier investigations of the rheological particula-
rities of the wheel’s steel have shown a softening of the
metal, while deforming it with a rate of deformation of
0.1 s–1 and values of the logarithmic deformation of
above 0.15 at a temperature of 1100 °C (the temperature
of the wheel blank stamping in the 100-MN press). At
the same temperature, steel hardening still takes place,
while deforming with a rate 10s–1 and a logarithmic
deformation of 1.22. Thus, applying the existing scheme
of deformation for the wheel blank in the 100-MN press,
steel softening will take place on the internal side of the
wheel hub, while hardening will take place on the hub’s
external side. Metallographic studies of the changes in
the microstructure of steel for different wheel-hub parts
confirmed that under these conditions a different steel
hardness on the opposite ends of the wheel hub is
obtained4.
4 DETERMINATION OF THE POSITION OF
THE NEUTRAL PLANE BY STAMPING WHEEL
BLANKS
The stamping process must ensure a precise mass
distribution of the initial blank for the part of the disc
and the wheel hub as well as other parts of the wheel. An
erroneous balance of the masses of different parts of the
blanks leads to the formation of defects in the wheel’s
geometry, e.g., rim shrinkage, non-filling of the hub, and
a "thick disc". For this reason, when calculating
groovings it is of great importance to know the position
of the plane relative to which the metal flows to both
sides during the wheel forming in the press. The position
of this plane changes in the process of blank forming;
however, it is not clear how the change occurs. When
calculating the groovings for wheels with a flat disc it is
assumed that the neutral plane corresponds to the middle
of the wheel disc length. No recommendations have been
given about how to determine the neutral plane position
in the calculation of the groovings for the wheel with a
disc of curvilinear form.
With the program Qform it is possible to determine
the value and the direction of the rate of deformation at
any point of the wheel during the whole cycle of the
blank’s deformation5. From the vectors of the directions
of deformation rate it is possible to establish the position
of the neutral section of the wheel as the plane relative to
which the metal flows to both sides (Figure 4.1–4.2).
The program Qform was used to study the position of the
neutral radius (Rn) and the metal flow rate at any point
during the cycle of forming the blank to various shapes
of the disc.
A. V. SHRAMKO ET AL.: THE APPLICATION OF THE PROGRAM QFORM 2D IN THE STAMPING ...
210 Materiali in tehnologije / Materials and technology 43 (2009) 4, 207–211
Figure 3: Distribution of the degree (A) and the rate (B) of steel
deformation along the generatrix of the wheel-hub bore for stamping
the blank in the 100-MN press
Slika 3: Porazdelitev stopnje (a) in hitrosti (b) deformacije po obodu
izvrtine pest kolesa za oblikovanje surovca v 100 MN pre{i
Figure 4.1: Variation of the position of the neutral plane (Rn) relative
to the middle of the disc (Rav) in the process of forming the wheel
blank with D = 957 mm in the 100-MN press
Slika 4.1: Sprememba polo`aja nevtralne ravnine (Rn) glede na
sredino diska (Rav) pri oblikovanju surovca za kolo z D = 957 mm v
100 MN-pre{i
Two types of wheels were chosen as the object of
investigation: a wheel with a flat-conical disc and a
wheel with a curvilinear disc. The process of stage-by-
stage deformation of the blanks in the 100-MN press was
simulated considering the groovings of the deforming
tool for stamping both wheels. The value of the "neutral"
radius (Rn) was determined for each stage of the blank
deformation and the results of these investigations
served as a basis for building the curves of the change of
the position of the neutral plane (Rn) relative to the
middle of the disc (Rav).
As shown in Figure 4.1, during deforming the blank
for the wheel with a flat-conical disc, the neutral plane,
defined with the parameter Rn, constantly changes its
position (place) relative to the middle of the wheel disc
defined with the parameter Rav. The average of all the
values of Rn (Rn.av) is displaced toward the side of the
wheel axis by 8 % for wheel blanks with a mass of 500
kg and by 11 % for wheel blanks with a mass of 700 kg.
A different picture is obtained for the stamping of
wheels with a curvilinear disc shape (Figure 4.2). In this
case the character of metal flow in the deformation
process is determined by the surface of the upper die
forming the wheel disc with the greatest angle to the
horizontal. The neutral plane then passes through the
point close to its middle and not to the middle of the
disc, as in the previous case. In this connection, Rn is
displaced along the horizontal relative to Rav for a much
greater value, equal to 16 %.
Investigations of the influence of different factors on
the position of the neutral plane showed that while
calculating the groovings of the press tool for
manufacturing wheels for various disc shapes, one
should be initially guided by the disc geometry and
consider the findings in6.
5 CONCLUSIONS
– The program QForm 2D can be successfully used for
the computer simulation of the deformation process
for stamping railway wheels with different shapes
and sizes.
– The program allows us to determine the intensity of
deformation and the rate of metal flow at any point
of the wheel and to determine the position of the
neutral plane of metal flow.
– The program was tested with calculations of the
shape of forming tools and verified with the indu-
strial production of wheels of different shapes and
sizes.
– The authors are indebted to Mr. Franc Vodopivec for
his revision of the manuscript.
6 REFERENCES
1 I. G. Uzlov, A. I. Babachenko, A. V. Shramko et al. Issledovanie
vliyaniya deformatsinnoy obrabotki kolesnoy zagotovki na mekhani-
cheskie svoystva zheleznodorozhnykh koles. Metal i litiye Ukrainy,
(2005) 9–10, 54–57
2 A. A. Milenin, A. B. Shramko, A. G. Stupka. Fizicheskoe modeli-
rovanie mnogostupenchatoy deformatsii stali v protsesse prokatki
zagotovok zheleznodorozhnykh koles. Metalurgicheskaya i gorno-
rudnaya promyshlennost, (2005) 2, 37–40
3 A. Shramco, A. Kozlovsk’yy, V. Danchenco. Effect of hot defor-
mation characteristics on mechanical properties of metal of the
railway wheel rim. International Heavy Haul Conference, Specialist
Technical Session – Kiruna, Sweden, (2007), 713–718
4 V. P. Esaulov, A. V. Shramko, V. N. Danchenko et al. Neravno-
mernost raspredeleniya tverdosti po dline stupitsy zheleznodo-
rozhnogo kolesa. Metalurgicheskaya i gornorudnaya promyshle-
nnost, (2007) 5, 90–93
5 Danchenko V. N., Milenin A. A., Kuzmenko V. I., Grinkevich V. A.
Kompiyuternoe modelirovanie processov obrabotki metalov davle-
niem. Dnepropetrovsk: Sistemnye tekhnologii, (2005), 448
6 A. V. Shramko, V. A. Afanasjev, A. G. Stupka, D. V. Sorokin.
Issledovanie techeniya metala pri formovke zheleznodorozhnykh
koles. Sovershenstvovanie protsessov i oborudovaniya obrabotki
davleniem v metalurgii i mashinostroenii: Tematicheskiy sbornik
nauchnykh trudov. – Kramatorsk: DGMA, (2006), 307–311
A. V. SHRAMKO ET AL.: THE APPLICATION OF THE PROGRAM QFORM 2D IN THE STAMPING ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 207–211 211
Figure 4.2: Change of the neutral plane position (Rn) relative to the
middle of the disc (Rav) in the process of forming the wheel blank with
D = 932 mm in the 100-MN press
Slika 4.2: Sprememba polo`aja nevtralne ravnine (Rn) glede na
sredino diska (Rav) pri oblikovanju surovca za kolo z D = 932 mm v
100 MN-pre{i

M. TORKAR: INFLUENCE OF WORKING TECHNOLOGY ON Al-ALLOYS ...
INFLUENCE OF THE WORKING TECHNOLOGY ON
Al-ALLOYS IN SEMI-SOLID STATE
VPLIV TEHNOLOGIJE PREOBLIKOVANJA Al-ZLITIN V
TESTASTEM STANJU
Matja` Torkar, Mirko Dober{ek, Iztok Nagli~
Institute of Metals and Technology, Lepi pot 11, SI-1000 Ljubljana, Slovenia
matjaz.torkar imt.si
Prejem rokopisa – received: 2009-01-05; sprejem za objavo – accepted for publication: 2009-02-27
We present two industrial technologies for working Al-alloys in the semi-solid state: thixocasting and rheocasting. The
development of these processes has enabled significant progress and improved market competitiveness for the
aluminium-casting industry. During both processes the share and the form of the globular solid phase are very important and
they influence the flow of material in the tool. The comparison of thixocasting, rheocasting and the conventional casting process
characteristics must be considered during the choice of one of these technologies. The performed investigations of rheocast
components of A357 alloy revealed the important reasons for failures were the non-optimised parameters of the semi-solid
working technology, the rapid opening of the tools and the non-optimised shape of the tool cavities.
Key words: thixocasting, rheocasting, semi-solid state, microstructure, Al-alloys
Predstavljeni sta dve vrsti tehnologije ulivanja oz. preoblikovanja v testastem stanju, thixocasting in rheocasting. Razvoj teh
dveh postopkov je povzro~il preobrat pri postopkih ulivanja aluminija in omogo~il bolj{o konkuren~nost na trgu. Klju~nega
pomena sta dele` trdne faze in oblika te faze, ki vplivata na tok materiala v orodju. Primerjava thixocastinga, rheocastinga in
konvencionalnih postopkov pri preoblikovanju Al-zlitine poka`e njihove zna~ilnosti, ki jih je treba upo{tevati pri izbiri za
uvedbo ene od teh tehnologij. Opravljene raziskave rheocast-komponent iz A357-zlitine so pokazale, da so med glavnimi vzroki
za ugotovljene napake pri preoblikovanju Al-zlitine premalo optimirani parametri tehnologije preoblikovanja v testastem stanju,
tudi prehitro odpiranje orodja in neprilagojena oblika orodij.
Klju~ne beside: thixocasting, rheocasting, testasto stanje, mikrostruktura, Al-zlitine
1 INTRODUCTION
Several methods for casting Al-alloys were deve-
loped in the past, for example, gravitational casting,
low-pressure die casting, high-pressure die casting and
squeeze casting in porous preforms. All these methods
have some advantages and some disadvantages that
influence their application. The automotive industry
demands better parts with lower weight (which means
thinner walls) and lower price. To improve competi-
tiveness in the market, new and improved processes are
being developed.
Basic and applied research work in the past also
enabled the development of new technologies for
working Al-alloys in the semi-solid state. The basis of
these technologies was Fleming’s1 discovery that a
material with a globulitic microstructure in a two-phase
region (L+α) behaves in a thixotropic way. In the
conventional casting process the local supercooling is
responsible for the evolution of a dendritic microstruc-
ture and for the evolution of a globular microstructure a
lower supercooling of the melt is favourable. This can be
achieved either with forced convection of the melt
during the solidification or with slow cooling. Lower
cooling rates support the spherical growth of the solid
phase. A more detailed background to these processes is
described in 1–6. Semi-solid metal processing offers
several advantages over conventional technologies such
as casting, forging and powder metallurgy. Semi-solid
metal processing enables the manufacturing of
components with complex shapes, with thin walls, with
good mechanical properties and with a high dimensional
tolerance and accuracy.
The thixocasting process uses stirring of the melt
during the solidification of a continuous cast bar to
obtain the globulitic microstructure. These bars are then
cut to the required pieces and reheated to the hot-work-
ing temperature. During the rheocasting process the
globular primary αAl phase is obtained by rapid cooling,
followed by controlled cooling to the temperature range
of the hot working. With the conventional casting
process, the development of the microstructure depends
on the cooling rate.
The comparison of the rheocasting, thixocasting and
conventional casting processes is schematically pre-
sented in Figure 1. From this figure it is evident that
during rheocasting the material is obtained with a
controlled cooling of the melt, while thixocasting needs
stirring of the melt during solidification and additional
heating for the hot working. The analyses of both
processes on real components were carried out5 and a
comparison of the production parameters is presented in
Table 1. The production parameters in Table 1 allow us
to compare both casting processes on the basis of data
Materiali in tehnologije / Materials and technology 43 (2009) 4, 213–217 213
UDK 669.715:620.18 ISSN 1580-2949
Professional article/Strokovni ~lanek MTAEC9, 43(4)213(2009)
obtained from industrial components. The comparison of
the costs for the different hot-working technologies of
Al-alloys is presented in Figure 2, where it is assumed
that the cost per kg of the rheocast component is 100 %.
Figure 2 shows that the price per kg of a thixocasting
component, compared with rheocasting, is lower only for
high-pressure die casting (HPDC), which ensures the
properties of the components of lower quality. The
thixocasting is 22 % more expensive due to the more
expensive technology, squeeze casting is 13 % more
expensive because of the lower productivity and gravity
casting is 4 % more expensive because of the cost of
machining the components. From this comparison it is
evident that rheocasting is a more competitive techno-
logy than thixocasting. Therefore, rheocasting is advan-
tageous from the energy- and costs-saving points of
view, when compared to thixocasting.
M. TORKAR: INFLUENCE OF WORKING TECHNOLOGY ON Al-ALLOYS ...
214 Materiali in tehnologije / Materials and technology 43 (2009) 4, 213–217
Table 1: Comparison of production parameters for thixo and rheo 8V bracket 5
Tabela 1: Primerjava parametrov izdelave nosilca 8V-motorja po thixo- in rheo- postopku 5
CHARACTERISTIC THIXOCASTING NEW RHEOCASTING
Material
A357 alloy in billets, obtained by
electromagnetic stirring during the casting
(Supplier: Pechiney)
A357 alloy ingots, without any special
preparation (Supplier: various)
Semi-solid slurry
production
– billet sizing
– induction reheating in vertical
medium-frequency furnaces up to the
semi-solid state
– melting in a gas furnace
– metal preparation in the holding furnace
– pouring into specific steel cups and cooling
up to the semi-solid state
Slurry Temperature 577 °C ± 2 °C* 579 °C ± 2 °C *
Metal need 4000 g 4700 g
Metal losses 10 % 1 %
Tool 2-cavity die with 4 hydraulic cylinders 2-cavity die with 2 hydraulic cylinders
Injection Horizontal, with the slurry laying on the shotsleeve
Vertical, with the slurry inverted in the shot
sleeve
Cycle time 59 s 52 s
Scraps recycling scraps recycled by the supplier scraps recycled in-house
Scraps rate 3 % visual + 2 % X-Ray 1 % visual + 0.5 % X-Ray
* temperature measured in the centre of the slug
Table 2: Chemical composition of A 357 alloy in mass fractions (w/%)
Tabela 2: Sestava A 357-zlitine v masnih dele`ih (w/%)
Alloy Cu Mg Si Fe Mn Ti Zn Sr Al
A 357 0.2 0.4 – 0.7 6.5 – 7.5 Max. 0.2 Max. 0.2 0.05 –0.2 Max. 0.2 0.03 Rest
Figure 2: Costs comparison of technologies for Al-alloy component
production. The cost per kg of the rheocast component is taken as
100% 5
Slika 2: Primerjava stro{kov tehnologij izdelave komponente iz
Al-zlitine. Stro{ki na kilogram izdelave rheocast-komponente so
prikazani kot 100 % 5
Figure 1: Schematic presentation of the rheocasting, thixocasting and
conventional casting processes and the obtained microstructure.
Slika 1: Shematski prikaz rheocastinga, thixocastinga in klasi~nega
ulivanja ter dobljena mikrostruktura
Besides the presentations of the basic differences and
characteristics of the thixocasting and rheocasting
technologies, the aim of the experimental work was to
evaluate and to detect typical failures in the rheocast
components.
2 EXPERIMENTAL WORK
The rheocast components produced on an 800 t UBE
rheocasting device were evaluated, and the micro-
structure of the slurry was investigated. Before the hot
working the slurry was knife cut and quenched into
water. After the manufacturing the components were
quenched into water and investigated. The components
were checked with an industrial x-ray device, YXLON
SMART 225 kV (Andrex), for internal soundness. The
surface and internal defects were also checked with
metallography. The samples for metallography were cut
from the components and prepared using a standard
metallographic procedure. The microstructure was
investigated with a light microscope (Nikon Microphot
FXA) equipped with a 3CCD video camera (Hitachi
HV-C20A) and software (analysis) for a quantifiable
assessment of the microstructural characteristics.
3 RESULTS AND DISCUSSION
The slurry of A357 Al-alloy (Table 2) at the tempe-
rature of hot working and with an approximately 50 %
content of liquid phase behaves like a butter. It can thus
be cut with a knife.
The microstructure of the slurry consists of the
globular αAl and eutectic phases (Figure 3a). With
inadequate thermal conditions the slurry has a dendritic
or mixed dendritic-globulitic microstructure (Figure 3b).
One of the investigated rheocast components is pre-
sented in Figure 4a, and the microstructure of the com-
ponent quenched in water, in Figure 4b. The comparison
M. TORKAR: INFLUENCE OF WORKING TECHNOLOGY ON Al-ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 213–217 215
Figure 4: Component (a) and microstructure (b) of the component,
quenched in water
Slika 4: Komponenta (a) in mikrostruktura (b) komponente, ohlajene
v vodi
Figure 3: Suitable (a) and less-suitable (b) slurry microstructure.
Quenched in water
Slika 3: Primerna (a) in manj primerna (b) mikrostruktura surovca.
Ohlajeno v vodi
Figure 5: X-ray picture of shrinkage porosity in one of the compo-
nents
Slika 5: Rentgenski posnetek poroznosti v eni od komponent
of the slurry (Figure 3a) and the component micro-
structure (Figure 4b) did not show any difference. As
expected, in both cases the microstructure consisted of
the globulitic αAl primary phase, surrounded by the
eutectic phase.
To detect the internal defects the components were
examined with x-rays. A typical x-ray picture of the
shrinkage porosity in the component is presented in
Figure 5. The metallography confirmed the presence of
internal porosity (Figure 6). The possible causes of
porosity are a lack of melt, a too high content of trapped
gases or a too low pressure in the die. Besides porosity in
the critical regions of the component, cold joints were
also observed. In these joints two fronts of the melt
(Figure 7) in the die cavity contact with an intermediate
oxide layer that greatly lower the local tensile strength of
the alloy.
Metallographic examinations also revealed some
other types of defects in industrially produced rheocast
components. Besides microporosity, an increased share
of eutectic was often observed near the surface (Figure
8). In some cases an overflow of the melt and pull cracks
due to rapid opening of the die (Figure 9) were found.
The increased share of eutectic near the surface is the
consequence of an unequal material flow in the die
cavity, where the die pressure squeezed the liquid
eutectic to the surface.
4 CONCLUSIONS
A comparison of the thixocasting and rheocasting
processes for the hot working of the Al-alloy A357
revealed, for both methods, some basic differences that
should be considered during the choice of technology.
The main reasons for the failures during the hot
working of the A357 Al-alloy in the semi-solid state by
rheocasting were the non-optimised parameters of the
semi-solid technology, the too fast opening of the tool
and the non-optimised shape of the die cavity.
Acknowledgement
This work was supported by EC, under contract no.
G1RD-CT-2002-03012.
M. TORKAR: INFLUENCE OF WORKING TECHNOLOGY ON Al-ALLOYS ...
216 Materiali in tehnologije / Materials and technology 43 (2009) 4, 213–217
Figure 9: Overflow and pull crack due to rapid opening of the die
Slika 9: Prelitje in razpoka zaradi prehitrega odpiranja orodja
Figure 8: Microporosity and increased share of eutectic near the
surface
Slika 8: Mikroporoznost in pove~an dele` evtektika ob povr{ini
Figure 7: Cold joint of two melt fronts
Slika 7: Hladni spoj ob stiku dveh tokov taline
Figure 6: Internal porosity – lack of melt, too high content of gases or
to low pressure in the die
Slika 6: Notranja poroznost – manko taline, prevelika vsebnost plinov
ali prenizka sila stiskanja orodja
5 REFERENCES
1 M. C. Flemings, R. G. Riek, K. P. Young, Rheocasting, Materials
Science and Engineering, 25 (1976), Sep.-Oct., 103–117
2 M. F. Zhu, J. M. Kim, C. P. Hong, Modeling of globular and den-
dritic structure evolution in solidification of an Al-7 mass % Si alloy,
ISIJ Intern., 41 (2001) 9, 992–998
3 P. J. Uggowitzer, H. Kaufmann, Evolution of globular microstructure
in new Rheocasting and super Rheocasting semi-solid slurries, Steel
Research Int., 75 (2004) 8/9, 525–530
4 M. Torkar, B. Breskvar, M. Godec, P. Giordano, G. L. Chiarmetta,
Microstructure evaluation of the NRC-processed automotive
component, Mater. Tehnol. 39 (2002) 6, 73–78
5 P. Giordano, G. L. Chiarmetta, Thixo and Rheo casting: comparison
on a high production volume component, Proceedings of the 7th S2P
Advanced Semi-solid Processing of Alloys and Composites, ed.
Tsutsui, Kiuchi, Ichikawa, Tsukuba, Japan, 2002, 665–670
6 F. Taghavi, H. Saghafian, Y. H. K. Kharrazi, Study on the ability of
mechanical vibration for the production of thixotropic microstructure
in A356 aluminum alloy, Materials & Design , 30 (2009) 1, 115–121
M. TORKAR: INFLUENCE OF WORKING TECHNOLOGY ON Al-ALLOYS ...
Materiali in tehnologije / Materials and technology 43 (2009) 4, 213–217 217

Z. ADOLF ET AL.: RAFINACIJA JEKLA V VAKUUMSKI NAPRAVI Z VPIHOVANJEM LEGIRNIH DODATKOV
STEEL REFINING IN A VACUUM UNIT WITH
CHEMICAL BOOSTING
RAFINACIJA JEKLA V VAKUUMSKI NAPRAVI Z VPIHOVANJEM
LEGIRNIH DODATKOV
1Zden k Adolf, 1Miroslav Dostál, 2Zden k [ána
1 V[B-Technical University of Ostrava, Faculty of Metallurgy and Materials Engineering, 17. listopadu 15/2172, 708 33 Ostrava-Poruba,
Czech Republic
2 EVRAZ Vítkovice Steel, a. s., 708 33 Ostrava, Czech Republic
zdenek.adolf vsb.cz
Prejem rokopisa – received: 2008-10-13; sprejem za objavo – accepted for publication: 2009-04-15
This paper describes an integrated system of secondary metallurgy (ISSM), which is exploited at the company EVRAZ
Vítkovice Steel a.s. Ostrava Czech Republic. The system consists of a caisson-type vacuum unit, enabling chemical boosting,
decarburisation, desulphurisation, chemical homogenisation and the modification of a liquid steel’s chemical composition
simultaneously in two ladles. The experimental part was focused on observation and evaluation with a statistical analysis of the
parameters, which influence substantially the final hydrogen content after a vacuum treatment of steel. The most important
of the monitored parameters appeared to be the vacuum treatment time, the sulphur content in the refined steel and the
difference between initial and final temperatures of the steel in the ISSM.
Keywords: vacuum refining of steel, ladle with chemical boosting, reduction of hydrogen content
V ~lanku je opisan integrirani sistem za sekundarno metalurgijo (ISSM), ki je v uporabi v podjetju EVRAZ Vitkovice Steel,
Ostrava, ^e{ka Republika. Sistem sestavljajo vakuumska posoda, v kateri je mogo~e legiranje dodatkov, razooglji~enje,
raz`veplanje, kemi~na homogenizacija in sprememba kemi~ne sestave taline v dveh ponovcah isto~asno. Eksperimentalno delo
je bilo namenjeno spremljanju in statisti~ni analizi parametrov, ki pomebno vplivajo na vsebnost vodika po vakuumski obdelavi
jeklene taline. Najve~ji vpliv je bil ugotovljen pri trajanju vakuumske obdelave, kon~ni vsebnost `vepla in razliki med za~etno
in kon~no temperaturo procesa v ISSM.
Klju~ne besede: vakuumska rafinacija jekla, ponovca z legiranjem, zmanj{anje vsebnosti vodika
1 INTRODUCTION
In 2007 the company EVRAZ Vítkovice Steel put
into operation the integrated system of secondary
metallurgy (ISSM). The equipment enables chemical
boosting, vacuum treatment, degassing, decarburisation,
desulphurisation, temperature and chemical homogeni-
sation and modification of the chemical composition of
molten steel simultaneously in two ladles.
2 EQUIPMENT ISSM
The basic part of the ISSM technological equipment
consists of a caisson, which contains a refining ladle
with molten steel stirred by an inert gas. The design of
the equipment is shown in Figure 1.
After the insertion of the ladle, the caisson is
hermetically closed and the treatment of hot metal
begins. The evaluation of the technological parameters
of the ISSM is shown in Figure 2. Figure 3 shows a
timeline of the metallurgical parameters of the molten
steel during its refining.
3 EXPERIMENTAL PART
The steel manufactured in a bottom blown converter
OBM is characterised by an increased content of hydro-
gen. Therefore, the objective of the experimental work
was to evaluate the connection of the measured
parameters in terms of the final content of hydrogen
during the steel’s treatment in the vacuum unit. The
investigated X70 steel is intended for large-diameter
welded tubes used for long-distance ducts. The required
chemical composition is given in Table 1.
Materiali in tehnologije / Materials and technology 43 (2009) 4, 219–222 219
UDK 669.187:669.788 ISSN 1580-2949
Professional article/Strokovni ~lanek MTAEC9, 43(4)219(2009)
Figure 1: Design of the ISSM vacuum unit
Slika 1: Shema vakuumske naprave ISSM
The evaluation of partial influences of the individual
parameters on the hydrogen content of steel was made
by the method of pairing linear regression. The evaluated
data were obtained from heat sheets. The variables, their
maximum, minimum and average values, are given in
Table 2.
The extent of the impact of the independent variable
(regressant) on the hydrogen content in the steel
(regressor) was evaluated with the help of:
– the correlation coefficient (R), characterising close-
ness of dependence,
– the slope of a straight line, which was recalculated to
an angle, which is formed by the straight line with
the axis x ( ), characterising the closeness of the
dependence,
– the testing parameter (P), the values of which should
be lower than 0.05.
In order to obtain a comparison of the slope of the
dependence characterising the intensity of the effect of
the given parameter on the hydrogen content, the values
of the regressant were re-calculated to a dimensionless
form in the interval from 0 to 1 using the equation
x
x x
x xi
i=
−
−
min
max min
where xi is the concrete value of the parameter, x i is the
re-calculated value of the parameter, 1, xmax, xmin are the
maximum and minimum values of the parameter.
The results of the regression statistics are given in
Table 3.
3.1 Evaluation of the results of the regression analysis
From Table 3 it follows that the final content of
hydrogen in the steel is, as expected, influenced mostly
Z. ADOLF ET AL.: RAFINACIJA JEKLA V VAKUUMSKI NAPRAVI Z VPIHOVANJEM LEGIRNIH DODATKOV
220 Materiali in tehnologije / Materials and technology 43 (2009) 4, 219–222
Figure 3: Metallurgical parameters of the ISSM processing
Slika 3: Metalur{ki parametri procesiranja ISSM
Figure 2: Technological parameters of the ISSM processing
Slika 2: Tehnolo{ki parametri procesiranja v ISSM
Table 1: Chemical composition of the steel X70 in mass fractions (w/%)
Tabela 1: Kemi~na sestava jekla X 70 (w/%)
X 70 C Mn Si P S Cu Ni Cr V Ti Al N Nb
min 0.020 0.03
max 0.11 1.80 0.40 0.018 0.006 0.30 0.30 0.30 0.08 0.035 0.050 0.010 0.06
Table 2: Table of variable values
Tabela 2: Tabela z vrednostmi spremeljivk
Minimum Maximum Average Unit
Regressor Hydrogen content 1.2 2 1.58 µg/g
Regressant
Duration of vacuum treatment 525 1595 937.1 s
Temperature of steel at the beginning of the treatment
(ISSM) 1589 1624 1605.5 °C
Temperature of steel at the end of the treatment (ISSM) 1557 1576 1564.5 °C
Difference between the initial and final temperatures of the
steel (ISSM) 28 55 40.9 °C
Sulphur content in steel during tapping from the converter
in mass fractions (w) 0.0047 0.012 0.008 %
Sulphur content in steel at the end of the treatment (ISSM)
in mass fractions (w) 0.0011 0.0038 0.002 %
Mass of charged ferro-alloys 20 214 100 kg
Difference between overall time of treatment and duration
of vacuum treatment in ISSM 16.5 38 23.9 min
Regressor Sulphur content in ISSM in mass fractions (w) 0,0011 0,0038 0,002 %
Regressant Duration of vacuum treatment 525 1595 937,1 s
by the vacuum treatment time (Figure 4). A higher
initial temperature of the steel, which accelerates the
diffusion of hydrogen in the melt, also has a positive
influence. While the temperature of steel at the end of
treatment in the ISSM has no influence (it must
correspond exactly to the specified temperature of the
casting on the CCM), the difference between the initial
and final temperatures of the steel in the ISSM shows a
substantial influence on the reduction of the hydrogen
content (Figure 5).
It is evident from Figures 6 and 7 that the final
content of hydrogen is closely related to the sulphur
content in the steel. This is explained by the surface
activity of the sulphur, which prevents the transfer of
hydrogen into the gaseous phase. The higher slope and
the closeness of the dependence in Figure 7 shows that
the negative influence of sulphur on the degassing is
already evident at low contents of sulphur at the end of
the vacuum treatment. At the beginning of the steel
treatment in the ISSM, the steel contains, apart from
sulphur, also surface-active oxygen, which also restricts
the de-sulphurisation. However, its content was not
monitored.
Ferro-alloys containing humidity are always a source
of hydrogen. The mass of ferro-alloys shows the com-
paratively insignificant influence on the final content of
hydrogen due to the intensive vacuum treatment.
The intensive contact of steel with slag during argon
blowing into the melt also causes deep desulphurisation,
which is evident from Figure 8.
Z. ADOLF ET AL.: RAFINACIJA JEKLA V VAKUUMSKI NAPRAVI Z VPIHOVANJEM LEGIRNIH DODATKOV
Materiali in tehnologije / Materials and technology 43 (2009) 4, 219–222 221
Table 3: Parameters of evaluated dependencies for hydrogen and sulphur on different variables
Tabela 3: Parametri ocenjenih odvisnosti za vodik in `veplo od razli~nih spremenljivk
Hydrogen content R P α
Duration of the vacuum treatment 0.381 0.0005 −20°
Temperature of steel at the beginning of the treatment (ISSM) 0.204 0.071 −9.8°
Temperature of steel at the end of the treatment (ISSM) 0.031 0.791 −1.5°
Difference between the initial and final temperatures of the steel (ISSM) 0.2725 0.0172 −13.8°
Sulphur content in steel during tapping from the converter 0.220 0.050 10.6°
Sulphur content in steel at the end of the treatment (ISSM) 0.3304 0.0031 15.6°
Mass of charged ferro-alloys 0.216 0.120 9.2°
Difference between the overall time of the treatment and the duration of the vacuum
treatment in ISSM 0.1466 0.2094 7.4°
Sulphur content in ISSM R P α
Duration of vacuum treatment 0.4477 0.00001 −26.8°
Figure 6: Dependence of hydrogen content on the sulphur content in
steel during tapping from the converter
Slika 6: Odvisnost vsebnosti vodika od vsebnosti `vepla ob izlivu
jekla iz konvertorja
Figure 5: Dependence of the hydrogen content on the difference
between the initial and final temperatures of the steel in the ISSM
Slika 5: Odvisnost vsebnosti vodika od razlike med za~etno in kon~no
temperaturo jekla v ISSM
Figure 4: Dependence of hydrogen content on the vacuum treatment
time
Slika 4: Odvisnost med vsebnostjo vodika in trajanjem vakuumiranja
The positive value of the slope in Figure 9 (the
dependence of hydrogen content on the time of vacuum
treatment and argon blowing of the steel only) shows
that during the time of argon blowing and alloying the
hydrogen content increases.
4 CONCLUSION
The aim of the work was to compare the influence of
selected parameters on the final content of hydrogen
during steel refining in the ISSM vacuum unit. The
investigation was made on this equipment installed in the
company EVRAZ Vitkovice Steel. The experimental
findings of the tests are as follows:
• the greatest effect on the final hydrogen content in
steel after treatment in the ISSM is found for the
vacuum treatment time. This dependence shows both
significant closeness and slope;
• the increased initial temperature of the steel in the
ISSM has a positive influence on the hydrogen
diffusivity and, therefore, also a positive influence on
steel degassing;
• the final temperature of the steel after treatment in
the ISSM has no influence on the degassing
efficiency, since, it is specified in a very narrow
interval depending on the casting temperature of the
given grade on the CCM;
• the mass of ferro-alloys showed a certain influence
on the final hydrogen content in the steel, but it was
less significant than the effect of the vacuum
treatment time;
• the content of sulphur influences significantly the
final hydrogen content in the steel because of it high
surface activity;
• from the comparison of the dependence of the
hydrogen content on the duration of the vacuum
treatment and argon treatment it was found that
during the time when the steel was argon treated
only, the degassing is almost arrested;
• the influence of the vacuum treatment time on the
reduction of sulphur content was found to be
important. Although the vacuum does not affect the
desulphurisation, the increase of the intensity of the
stirring of the steel and the slag as a result of argon
blowing into the vacuum-treated steel has signifi-
cantly reduced the content of sulphur.
The work was undertaken within the frame of the
project EUREKA 3580! and the project FI-IM4/110 with
the financial support of the Ministry of Education, Youth
and Sports and Ministry of Industry and Trade of the
Czech Republic.
5 REFERENCES
1 Dostál, M. Pánvová pec ISSM – hodnocení vybraných parametr
taveb Ladle furnace ISSM – evaluation of selected parameters of
heats. Bachelor degree thesis, V[B-TU Ostrava, 2008, 40 pp.
Z. ADOLF ET AL.: RAFINACIJA JEKLA V VAKUUMSKI NAPRAVI Z VPIHOVANJEM LEGIRNIH DODATKOV
222 Materiali in tehnologije / Materials and technology 43 (2009) 4, 219–222
Figure 9: Dependence of the hydrogen content on the time when the
steel is not vacuum treated in the ISSM
Slika 9: Odvisnost vsebnosti vodika od ~asa prepihovanja z argonom
brez vakuumiranja
Figure 8: Dependence of the sulphur content on the vacuum treatment
time
Slika 8: Odvisnost vsebnosti `vepla od trajanja vakuumiranja
Figure 7: Dependence of the hydrogen content on the sulphur content
in the ISSM
Slika 7: Odvisnost vsebnosti vodika od vsebnosti `vepla v ISSM
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
HISTORICAL SURVEY OF IRON AND STEEL
PRODUCTION IN BOSNIA AND HERZEGOVINA
ZGODOVINSKI PREGLED PROIZVODNJE @ELEZA IN JEKLA V
BOSNI IN HERCEGOVINI
Sulejman Muhamedagi}1, Mirsada Oru~2
1University of Zenica, Faculty of metallurgy and materials, Travni~ka c. 1, 72000 Zenica, Bosna i Hercegovina
2University of Zenica, Institute of Metallurgy "Kemal Kapetanovi}", Travni~ka c. 1, 72000 Zenica, Bosna i Hercegovina
sulejman.muhamedagic famm.unze.ba
Prejem rokopisa – received: 2009-01-08; sprejem za objavo – accepted for publication: 2009-01-16
Cast-iron and steel production facilities were, and still are, frequently located on sites with deposits of iron ore and coal. The
center of steel metallurgy in Bosnia and Herzegovina, and of the former Yugoslavia, is located in the Iron and Steel Plant
Zenica, today known as Arcelor Mittal Zenica. In this paper the beginning, the development and the planned growth of the iron
and steel plant in Zenica is presented with periods of success and periods of crisis.
Key words: Iron and Steel Plant Zenica, developmentr, pig iron, steel.
Proizvodne naprave za grodelj in jeklo so pogosto zgrajene na le`i{~ih `elezove rude in premoga. Sredi{~e proizvodnje jekla v
Bosni in Hercegovini ter v nekdanji Jugoslaviji je bilo v @elezarni Zenica, danes Arcelor Mittal Zenica. V tem sestavku so
predstavljeni za~etek, razvoj in na~rtovana rast @elezarne Zenica z obdobji krize in uspeha.
Klju~ne besede: @elezarna Zenica, razvoj, grodelj, jeklo
1 INTRODUCTION
Metal materials based on iron have been used for
millennia; first as natural iron metal and then extracted
from iron ores 1. Pig iron is produced in blast furnaces
and is the basis for steel production, with a share of more
than 60 %. Pig iron is produced from ores with 40 % to
65 % Fe. Besides the content of iron, the possibility of
using it in a blast furnace without previous ore pro-
cessing is of essential importance 2.
Steel is an alloy that can be plastically worked; it is
of strategic importance for every country and has many
applications 3. Its wide range of technological and
mechanical properties make steel the most important
metallic material with a steady growth in annual
production. Today’s technical society would not be
possible without steel, which is produced in an annual
quantity that is five times greater than the total
production of all the other metallic materials 4.
2 HISTORICAL SURVEY
The Iron and Steel Works Zenica is the basis of the
Bosnian and Herzegovinian (BH) economic mosaic, with
a development based on the advantages determined by
its location and which are the basis for the future
technological development: the millennium tradition of
iron production, the location and the natural advantages
of its central location in BH, the deposit of quality coal,
the proximity of the Vare{ iron-ore deposit and the
advantage of road and railway communications along the
Bosna valley.
The concession for the erection of the Ironworks
Zenica was given in 1892 to the Austrian industrialists
Leon Bondy (Prag), Moritz and Adolf von Schmit
(Wilhemsburg) and Hans von Peng (Thorl).The first
production facilities were erected in a single year, and in
1893 the production of small profiles and wire rod was
started. The official name of the company was Eisen und
Stahlgewerkschaft Zenica.
In the rest of the paper a review of the iron works is
given for the period 1892–2008, with the emphasis on
different stages. In 1898 the owners formed the share
company Eisenindustrie – Aktiengesellschaft Zenica,
with the aim to strengthen the company, and the
increased capital and profit enabled the further con-
struction and modernization of the production facilities
in the period 1898 to 1913. In 1908 the iron works
became a member of the Central European cartel of the
metallurgical industry. A maximum production of 38
583 t of steel was achieved in 1912, and this was not
exceeded until 1936, when a substantial enlargement of
the production facilities was introduced. In 1936 the
Kingdom of Yugoslavia backed the restructuring and the
enlargement of the production facilities for steel. In this
year, the characteristic growth of production facilities by
stages was started with the erection of the rolling mill for
heavy profiles. The production increased to 80 000 t of
steel, and more than 72 000 t of final rolled products.
In the period of capital erection from 1947 to 1958,
in the time of Socialist Yugoslavia, the Ironworks Zenica
Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229 223
UDK 669.1(497.15)(091) ISSN 1580-2949
Professional article/Strokovni ~lanek MTAEC9, 43(4)223(2009)
was enlarged with the erection of the integrated process,
and by the 60th year the project to build facilities with an
annual production of 265 000 t of steel was prepared.
This program was planned to be realized in three stages.
The production facilities enlarged in the period from
1892 to 2008 occurred in several stages. As an integrated
steel producer, it includes all the processing stages: the
production of coke, agglomerate, pig iron, steel, power
stations and steel transformation to final products with
hot rolling and forging.
3 STAGES OF DEVELOPMENT OF THE
IRONWORKS ZENICA
The stages of development are shown by year
together with the annual production for all works and for
110 years of activity. In this period of time the name and
the owners changed, and presently the name is Arcelor
Mittal Zenica.
1892–1893
Erection and revamping of production facilities
35 000 t per year
Erection of 2 puddling furnaces, 2 rolling mills,
drives and the boiler
1895–1898
55 000 t per year
Completion and reconstruction of production faci-
lities. Erection of 3 Siemens-Martin (SM) furnaces with
a capacity of 15 t, electrical power station and a new
light profiles rolling mill (Figures 4, 5, 6 and 7).
1936−1940
Erection of the heavy rolling mill, 2 SM furnaces of
capacity 40 t and 50 t, electric arc furnace of capa-
city 3 t, drive generators and the mechanical work-
shop 100 000 t per year
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
224 Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229
Figure 1: Technological scheme of the integrated production of steel in the Ironworks Zenica (2007)
Slika 1: Tehnolo{ka shema integrirane proizvodnje jekla v @elezarni Zenica (leto 2007)
Figure 3: Light rolling mill at the beginning of the 20th century
Slika 3: Lahka valjarna v za~etku XX. stoletja
Figure 2: The Ironworks Zenica in 1895
Slika 2: @elezarna Zenica leta 1895
1947–1958
Pig iron: 600 000 t per year
Steel 75 000 t per year
Finished rolled and forged products 540 000 t
Erection of:
• 4 coke batteries with 39 furnaces per battery, with an
annual capacity of 650 000 t and with auxiliary faci-
lities,
• the complex of blast furnaces consisting of the ore
treatment with a capacity of 400 000 t of burden
materials, agglomeration with 8 Greenawald pans
with surfaces of 22 m2 and a capacity of 800 000 t
per year of agglomerate,
• 3 blast furnaces with volumes of (850, 750 and 800)
m3 and a capacity of 600 000 t of pig iron,
• the new steelworks with 4 fixed SM-furnaces with a
capacity of 70 t, 4 tilting 180 t SM furnaces and the
10 t electric arc furnace,
• complex of rolling mills with a Blooming mill, a
continuous mill for half products and 3 finishing
rolling mills: medium, light and wire rod rolling
mills,
• forging shop with presses of 6 MN, 18.5 MN and 51
MN, hammers and the rolling mill for rings and
wheels,
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229 225
Figure 5: Taking specimens from the SM furnace before the First
World War
Slika 5: Odvzem vzorcev iz SM-pe~i pred prvo svetovno vojno
Figure 4: Heating furnaces for the light rolling mill before the First
World War
Slika 4: Ogrevne pe~i za lahko valjarno pred prvo svetovno vojno
Figure 8: Manual charging of additions in the SM furnace
Slika 8: Ro~no zakladanje dodatkov v SM-pe~
Figure 7: Medium rolling mill
Slika 7: Srednja valjarna
Figure 6: Ore bedding with a view of the blast furnaces
Slika 6: Rudni dvor s pogledom na plav`
• facilities for thermal treatment and machining, and
the power station
1960–1961
A draft project was prepared for new investments for
a realization in three stages 2 650 000 t per year
1965–1968 (stage I)
1 000 000 t per year
Modernization and a production-capacity increase.
With the completion and reconstruction of the ore
treatment and the blast furnaces for the production of
700 000 t per year of pig iron, the erection of a new
system for steel-scrap conveying and the introduction of
oil heating for the SM-furnaces, the erection of slag
bedding, the projected steel plant annual production was
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
226 Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229
Figure 11: Hammer forging
Slika 11: Kovanje s kladivom
Figure 10: Rolling on the Blooming rolling mill
Slika 10: Valjanje na valjarni Bluming
Figure 9: Charging of ingots in the pit furnace
Slika 9: Zakladanje ingotov v talno pe~
Figure 12: Heavy forging piece on the 18.5 MN press
Slika 12: Te`ak odkovek na pre{i 18,5 MN
Figure 13: Casting in moulds
Slika 13: Ulivanje kokil
achieved. For the plants own needs, a steel casting shop
was built.
1965
The seconds stage of the project was finished
1 250 000 t per year
1970–1989 (Stage II)
A new integrated ironwork was erected with the
following facilities:
• coke battery no. 5 with 65 ovens with a capacity of
720 000 t per year,
• unloading stations and the transport system for raw
materials,
• agglomeration shop with 6 units, each with a surface
of 75 m2,
• blast furnace 4 with a volume of 1 756 m3 and a
capacity of 1 250 000 t per year of pig iron,
• facilities for the treatment of ore materials
• a steel plant with 100 and 130 LD convertors and
mixers 21 300 t,
• continuous casting for blooms of section (265 × 340)
mm,
• billet-rolling mills with an annual capacity of
1 500 000 t, a light rolling mill with a capacity of 650
000 t per year and a wire rod rolling mill with a
capacity of 430 000 t per year,
• power complex with 2 220 t/h steam boilers, a heat
shop and a pumping station for water, turbogene-
rators of 7 MW and 25 MW, turbotuyeres TD-4 and
TD-5 with 18 MW for technological air for the blast
furnace 4, transformers and the net for the distri-
bution of energy.
1988
The productions was 1 118 780 t of convertor,
762 886 t of SM and 24 334 t of electro steel
1 906 000 t per year
Total (stages I and II) 2 250 000 t per year
1990–1991
The British company British Steel Consultants,
London, and the World Bank suggested several
development options for the reconstruction of the
Ironworks Zenica. They established that the ironworks
may became profitable after the realization of the
proposed program and become competitive in the
market. A study was prepared on the basis of the total
steel production in Yugoslavia.
1992
The production in Ironworks Zenica, with a tradition
of over 100 years of production and working with steel,
was halted in the second half of 1992 because of the war
and all the facilities were preserved. In September and
October 1992 the ironworks was heavily damaged by air
bombing.
The realization of the reconstruction program
proposed by the company British Steel Consultants,
London was stopped.
1993–1995
The protection of the facilities was ensured and the
air-bombing damage was repaired.
The proposal for the revitalization and the start up of
production was prepared in August 1994 with the
cooperation of local specialists and those from
development organizations in Zenica and with the
optimistic anticipation of the employees and the
managers for the future. They were convinced that the
production of steel was justified by economic, market
and development factors.
In the study, all the base, technical, market, economic
and financial elements were considered for an optimized
production program. In particular, the following were
analyzed:
• the permanent halting of old and loss-making facili-
ties,
• the economic and technical justification for the
reactivation of facilities erected after 1976,
• the necessity of production for supplies to the manu-
facturing industry and because of market trends,
• the number of employees,
• the required financial funds for the restructuring of
production facilities and for the working capital,
• the economic evaluation of the technical, techno-
logical, production and market parameters, the role of
the Ironworks Zenica in the economy of the Republic
of Bosnia and Herzegovina,
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229 227
Figure 14: Discharging of red-hot coke
Slika 14: Praznjenje `are~ega koksa iz pe~i
• the development potential and the change of owner-
ship.
The following were proposed: to permanently stop
the old and loss-bringing facilities from the 1950s: the
coke battery 1 to 4, the ore-treatment facilities, the old
agglomeration with 8 Greenawald pans, the blast
furnaces 1 to 3, the SM steel plant, the electric arc
furnace 3 t and the cast-steel factory, the heavy and light
mills and the wire rod mills II.
• Revitalization and the start of facilities erected after
1976 aimed at an integrated cycle production with the
production of coke, agglomerate, pig iron, steel and
rolled and forged products.
• The increase of the capacity of the old electric arc
furnace from (10 to 15) t per year and the reconstruc-
tion of the vacuum facility for molten steel.
1998–2000
The moral codex and the traditional perseverance of
the Zenica metallurgists maintained the cadres, the
facilities and the start of the production of steel with the
remelting of steel scrap from dismantled and
permanently halted facilities in the SM and electric arc
furnaces. This forced solution was accompanied by
significant reconstructions to cover the marketing
requirements. The production was started in the rolling
mill and the forging shop and with the manufacturing of
reinforcing nets, building armature and lattice beams. In
parallel, in 1998 the certification audit confirmed the
management quality conformed to the requirements of
ISO 9001 for the rolling and forging products
2000
SM furnace 53 118 tons,
electric arc furnace 23 533 t
76 651 t
2003–2004
Planned production 1 000 000 t
In 2003 and 2004 the EAF 100 t, the continuous
casting, a new formill and a new heating furnace for the
light rolling mill were built.
The production of approximately 1 000 000 tons of
steel was achieved in the electric arc 15 t and in the
SM-furnaces. The operation of the forging shop of
approximately 15 000 t was based on the plants own
steel, while, for the rolling mill beside the SM furnaces
production, a limited quantity of billets was also
imported.
In August 2004 a contract was signed with the
company LNM Holdings N.V, which became the owner
of 51% of the Zenica company. By the end of 2004 the
operation of the SM-furnaces was closed permanently.
2004−2006
The production with the electric arc furnace 15 t was
used for the forging shop and, with the electric arc
furnace 100 t, for the rolling mills. In parallel, a
revision of projects was carried out in the frame of
the starting up of the integrated steel production.
2006
Annual production of electric steel 480 035 t
2008
Production was started with the reconstructed blast
furnace, agglomeration and the coking plant.
2007−2008
The "Feniks" project aimed at starting up the inte-
grated production was approved and its realization
was started. In parallel, activities were started up
for the preparation of measures and time limits for
the decrease of emissions and pollution, applying the
best available methods, with the aim to obtain an
environmental license.
The work with reconstruction and overhauling of
the facilities for primary processing continued:
agglomeration, blast furnace, steel plant, energy
supply and quality control, for infrastructure and
auxiliaries. By the second quarter of 2008 the test
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
228 Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229
Figure 16: The new continuous-billets casting facility
Slika 16: Nova naprava za kontinuirno ulivanje gredic
Figure 15: The new electric arc furnace 100 t
Slika 15: Nova elektroblo~na pe~
production of the coking plant and by July also the
test production with agglomeration, blast furnace,
convertor steel plant and the auxiliaries were
started.
2007
Annual electric steel production 533 289 t
2008
Annual electric steel production planned 780 000 t
2009
Annual steel production planned 1 140 000 t
2012
New investments and the construction of continuous
casting for slabs planned for the annual production
of 2 000 000 t
4 CONCLUSION
Steel has for a long time been connected to Central
Bosnia and Herzegovina, and especially to the area of
Zenica, a distance of 70 km from Sarajevo. With the start
of the integrated production process in the Ironworks
Zenica, this area again acquired a strong base industry
for the production and transformation of iron ore, steel
and market products.
5 REFERENCES
1 Z. Pa{ali}: Metalurgija ~elika (Steel metallurgy); Fakultet za meta-
lurgiju i materijale, Univerzitet u Sarajevu, 2002
2 S. Muhamedagi}: Metalurgija gvo`|a-Visoka pe} (Pig iron metallur-
gy-blast furnace), Fakultet za metalurgiju i materijale, Univerzitet u
Zenici, 2005
3 S. Muhamedagi}: Metalurgija gvo`|a-Priprema zasipa (Pig iron
metallurgy, burden preparation), Fakultet za metalurgiju i materijale,
Univerzitet u Zenici, 2005
4 M. Oru~: Savremeni metalni materijali (Modern metallic materials),
Fakultet za metalurgiju i materijale, Univerzitet u Zenici, 2005
5 Fotomonografija – @eljezara Zenica, "Zadruga" Sarajevo, 1968
6 Mittal Steel Zenica – Poduhvat za budu}nost BH ~elika (Venture for
the future of the BH steel), Zenica, 2005
S. MUHAMEDAGI], M. ORU]: HISTORICAL SURVEY OF IRON AND STEEL PRODUCTION IN BiH
Materiali in tehnologije / Materials and technology 43 (2009) 4, 223–229 229