Grain refinement of austenitic manganese steels

Abstract

From this work it can be concluded that: With correct additions of cerium and aluminum, grain refinement of austenitic high manganese steel is attainable, as long as the other influencing parameters are within favorable conditions. The effect of the added grain refiner is dependant of good control of the casting temperature. The refiner efficiency reduces with increasing temperature. Types of cerium containing particles formed depend on the amount of cerium added.

Full text

Grain Refinement of Austenitic Manganese Steels
Fredrik Haakonsen
SINTEF
Alfred Getz vei 2
7465 Trondheim, Norway
Phone -+47 95239469
E-mail: fredrik.haakonsen@sintef.no
Jan Ketil Solberg
NTNU
Alfred Getz vei 2
7491 Trondheim, Norway
Phone +47 73592051
E-mail: jan.solberg@material.ntnu.no
Ole Svein Klevan
Elkem AS
Alfred Getz vei 2
7465 Trondheim, Norway
Phone +47 73590713
E-mail: ole-svein.klevan@elkem.no
Casper van der Eijk
SINTEF
Alfred Getz vei 2
7465 Trondheim, Norway
Phone +47 98283989
E-mail: casper.eijk@sintef.no
Key words: Steel, Grain Refinement, Cerium, Inclusions
INTRODUCTION
The objective of the present work has been the improvement of the microstructure and mechanical properties of austenitic manganese
steel through introduction of substrate particles which should nucleate grains during solidification, giving grain refinement. The base
for the experiments was austenitic manganese steels. Cerium was chosen as the grain refinement addition. As a highly reactive
compound, cerium reacts with oxygen and sulfur already in the melt, forming cerium oxides and –sulfides which act as substrate
particles for melted steel to solidify on. Different amounts of additions, sometimes in combination with aluminum, were made. Grain
refinement was obtained in a casting added cerium in combination with aluminum, in which CeO2 was the supposed substrate particle.
Refiner efficiency as a function of temperature is also presented.
THEORY
Table I shows the composition of the austenitic manganese wear steel. The liquidus- and solidus temperatures are approximately
1350°C and 1125°C1. The solidification range is then 225 degrees centigrade. The solidification range will have a large influence on
the casting properties of the steel. The austenite is stable down to 400 °C, after which it is metastable. Thus, the possibility of attaining
grain refinement by heat treatment in the solid state (normalizing), as in low alloyed steels, does not exist.
763
AISTech 2011 Proceedings - Volume II

Table I. Approximate composition of the austenitic manganese steel. Values are in weight percent.
C
Mn
Cr
Si
Al
Ti
O
S
N
P
1.42
19.5
2.4
0.7
0.03
0.15
0.005
0.005
0.01
0.04
Due to the lack of solid-state phase transformations in austenitic manganese steels, grain refining during casting is the only way to
reduce the grain size in a cast component. Grain refining during casting may yield improvements as better resistance to hot tearing
during solidification, higher ductility and toughness, better fatigue resistance, reduced porosity and better feeding of the casting2. To
achieve smaller grain size and a larger fraction of equiaxed grains at the expense of columnar crystals, particles can be introduced for
solid nuclei to form on. By heterogeneous nucleation, the surface-to-volume ratio of the nucleus decreases, and the new surface energy
is reduced. In this way the nucleus becomes stable at a lower undercooling than during homogeneous nucleation. In aluminum, these
substrate particles can be intermetallic compounds like TiB2 that have been added to the melt. In steel, it is more common to add
alloying elements that react with other elements in the melt, often elements that otherwise are considered unwanted. These elements
precipitate as non-metallic inclusions above the liquidus temperature for the alloy in consideration. Examples of non-metallic
inclusions that are known to promote grain refinement in steels are CeS, RE2O3, TiN and Al2O3 3.
Austenitic manganese wear steels are often used in large components that are cast in sand molds. Typical components are crushing
plates, with dimensions 2×1×0.2 m3 and a weight of approximately 3 metric tons, and crushing cones. The casting temperature is
approximately 60-80 degrees centigrade above the liquidus temperature. Solidification starts at the surface against the sand mould, and
coarse columnar crystals grow into the melt towards the centre of the casting. If the total undercooling of the melt, ΔT, is large enough,
equiaxed grains will start to grow ahead of the solidification front, see Figure 1. These will nucleate when ΔT> ΔTn, where ΔTn is the
necessary undercooling for heterogeneous nucleation.
Figure 1. A sketch of the solidification at the transition from columnar growth to growth of equiaxed grains. Columnar grains are
shown on the left and equiaxed grains are shown on the right.
The very large solidification range of this steel results in considerable microsegregation. On the other hand, this will contribute to the
formation of a large constitutionally undercooled zone ahead of the solidification front4. The ability of substrate particles to nucleate a
solid phase is dependent on the constitutional undercooling. As the constitutional undercooling during solidification is given by the
alloy composition and the casting parameters, the substrate particles have to be operational as nucleation sites at this given
undercooling to give grain refinement. For high carbon, high manganese austenitic steels, austenite has to grow epitaxially on the
substrate particles. The lattice parameter and crystal structure of the substrate particles should therefore be compatible with austenite.
764
AISTech 2011 Proceedings - Volume II

In the literature, there seems to be no report on how the undercooling, ΔT, during solidification is related to the lattice mismatch, δ, for
steels that solidify with an austenitic structure, but for ferritic steels this connection is known. According to an article by Grong5,
substrate particles of the type CeS, CaS, TiN and TiC can nucleate ferrite at an undercooling in the order of 1 centigrade or lower.
These particles have a mismatch to ferrite of less than 5 %.
EXPERIMENTAL AND RESULTS
Cerium in the form of an iron-chromium-silicon-cerium alloy was chosen as a grain refiner for the tests. This alloy is produced by
Elkem AS under the name Elkem Grain Refiner (EGR). EGR is commercially available and have been tested in austenitic stainless
steels with positive results6. Cerium is a very reactive element, on a level with aluminum, and forms oxide- and sulfide compounds.
The EGR alloy was in the form of crushed pieces in a size of 2-20 mm.
The basis for the grain refinement experiments was scrap/returns from production of austenitic manganese steel. The melting was
done in a 2 ton high frequency induction furnace. Steel for the whole test was melted together to reduce differences in chemical
composition between the different castings. The ladle used for casting was preheated for three hours with a propane burner. The ladle
had a capacity of approximately 300 kg steel, and was transported by a crane during the experiment. The samples to be cast were
massive with dimensions 270×150×270 mm3. The sand molds were made of olivine sand. The amount of steel melted was 1300 kg.
This was heated to 1600°C, and for each test, 120 kg steel was pored into the ladle. The grain refiner and deoxidizing agents were
added during tapping from the furnace to the ladle. The temperature was measured prior to casting, and the target casting temperature
was 1420°C. If the temperature of the melt was higher than the wanted casting temperature, the casting was delayed until the
temperature was correct.
The additions to the different castings of the experiment were made with EGR containing 10 % cerium, 20 % silicon, 35 % chromium
and 35 % iron. Aluminum was added as ferroaluminum (FeAl) containing 38 % aluminum, and titanium was added as ferrotitanium
(FeTi) containing 70 % titanium. No elements were added to casting 1. Casting 2 was alloyed with 0.1% aluminum and 0.4 %
titanium. This is the same addition that is used in regular production of this steel. Casting 3, 4 and 5 were alloyed with increasing
amounts of cerium, 0.03 %, 0.06 % and 0.12 % cerium respectively. Casting 6 was alloyed with 0.06 % cerium and 20 ppm boron in
the form 35 % ferroboron. Casting 7 was alloyed with 0.06 % cerium and 0.05 % aluminum. Casting 8 was alloyed with 0.12 %
cerium and 0.02 % aluminum.
After casting, a vertical section of approximately 150 mm width, 270 mm height and 10 mm thickness from the centre of each casting
was cut with a water jet cutting machine. These plates were later cut into smaller samples. The steel was austenitized at 1120 °C for 4
hours, as is industrial practice. A muffle furnace with purging argon gas was used for these heat treatments. After austenitizing, the
samples were water-quenched. For light microscopy, the samples were polished and etched in 2 % Nital for 60 seconds, followed by
washing with ethanol and drying.
Chemical compositions measured with Optical Emission Spectrography (OES) are given in Table II. The castings were made in the
order they are listed. In casting 2, the aluminum and titanium contents are higher than in the others, as expected as this casting was
added these elements. The nitrogen content in this casting is also lower than in the others, because of the reaction with the added
titanium to form TiN.
Table II. Chemical composition measured with Optical Emission Spectrograph. Values are in weight percent.
Casting
C
Si
Mn
S
P
Cr
Al
Ti
N
1
1.29
0.33
18.0
0.005
0.043
2.42
0.002
0.004
0.023
2
1.31
0.33
18.0
0.004
0.042
2.42
0.043
0.098
0.010
3
1.31
0.36
17.9
0.005
0.041
2.48
0.016
0.050
0.035
4
1.31
0.38
17.8
0.004
0.040
2.60
0.005
0.005
0.027
5
1.28
0.49
17.7
0.004
0.038
2.78
0.005
0.004
0.027
6
1.29
0.34
17.8
0.004
0.041
2.60
0.002
0.003
0.029
7
1.30
0.31
17.9
0.004
0.041
2.56
0.015
0.004
0.029
8
1.27
0.40
17.6
0.004
0.042
2.80
0.019
0.003
0.029
765
AISTech 2011 Proceedings - Volume II

Table III summarizes the different casting temperatures, the measured grain sizes, the cerium and boron contents, as measured with
Inductively Coupled Plasma Mass Spectroscopy (ICP-MS), the yield of cerium, the measured number densities of cerium-containing
particles, and the measured lengths of the columnar zone. The casting temperature for casting 1 was not measured, but is expected to
be in the lower range as this was in the start of the experiment and the ladle was not fully heated. Castings 3, 6 and 8 had lower casting
temperatures than the other castings. The grain sizes of casting 1 and 8 were quite small, the grain size in casting 2 was intermediate,
and the grain sizes of casting 3, 4, 5, 6, and 7 were large. Increasing amounts of cerium from casting 3 to casting 5 are as expected
from the amounts being added. Castings 4, 6 and 7 were added the same amount of cerium, with small differences in the remaining
content. Castings 5 and 8 were also added the same amount of cerium, with a little higher content remaining in casting 8, probably due
to the extra aluminum added to casting 8, being available to react with oxygen. The yield of cerium is for all castings between 30 %
and 52 %. The boron content was unexpectedly high for all castings, with the highest content in casting 6. Casting 6 was added 20
ppm boron extra. The number densities of particles in the castings that were added cerium are given in the right column. As only
cerium based particles were counted, no results are reported for castings 1 and 2. The highest number of particles was found in casting
8. The columnar zone was shortest for casting 1 and 8.
Table III. Casting temperature, measured grain size, weight percent cerium and boron with cerium yield listed in brackets, number
density of cerium-based particles, and length of columnar zone for the different castings.
Casting
Casting temp
Grain size
Ce (yield)
B
Part./mm2
Col. zone
1
-
320 μm
-
0.015
-
0.66 mm
2
1425°C
570 μm
-
0.014
-
1.3 mm
3
1411°C
930 μm
0.009 (30%)
0.014
49
1.0 mm
4
1425°C
820 μm
0.025 (42%)
0.018
81
1.2 mm
5
1427°C
970 μm
0.037 (31%)
0.020
157
0.83 mm
6
1407°C
810 μm
0.031 (52%)
0.022
50
1.2 mm
7
1424°C
1060 μm
0.021 (35%)
0.017
97
1.3 mm
8
1402°C
270 μm
0.050 (42%)
0.020
188
0.69 mm
Table IV shows the result of additional Wavelength Dispersive X-ray Spectroscopy (WDS) performed in an Electron Probe Micro
Analyzer (EPMA) of an arbitrary selection of particles found in the six castings added cerium. The particles were recognized from
their chemical composition. All the particles found in casting 3 were of the CeAlO3 type. In casting 4 and 6 there were a
predominance of CeO2 particles, but a fraction of CeAlO3 and Ce2O2S was also found. Most particles in casting 5 and all particles in
casting 8 were of the CeO2 type. Casting 7 had a slight predominance of CeAlO3, in addition to Ce2O2S and CeAlO3. The mean
particle size was between 1.0 and 1.8 μm. There seemed not to be any difference in size between the different types of particles.
Table IV. The percentage and type of detected oxides, number density of particles analyzed in the EPMA and mean size of these
particles.
Casting
Phase
Number of particles
analyzed
Mean particle
size
CeO2
CeAlO3
Ce2O2S
3
-
100 %
-
10
1.0 μm
4
50 %
20 %
30 %
10
1.3 μm
5
90 %
-
10 %
20
1.6 μm
6
60 %
20 %
20 %
10
1.8 μm
7
20 %
45 %
35 %
20
1.2 μm
8
100 %
-
-
20
1.6 μm
766
AISTech 2011 Proceedings - Volume II

DISCUSSION
Table III shows the measured contents of cerium in the experimental castings. The cerium yield varied between 30 and 52%. The
cerium yield for a given casting is to some extent influenced by the addition done in the ladle during the previous casting, as there was
an increase in yield when the addition to the previous casting was high, as for casting 6 that has the highest yield. This is probably
caused by re-entrainment of cerium particles transferred to the ladle wall during the previous trial.
The casting temperatures, grain sizes and lengths of the columnar zone are listed in Table III. Casting 1 is omitted in the following
evaluation as its casting temperature was not measured. The casting temperatures have unfortunately a larger scatter than intended.
This scatter will be the most significant uncertainty in the assessment of the grain refining results. Except for casting 8 that had the
smallest grain size, the shortest columnar zone, and also the lowest registered casting temperature, there is little coherence between the
casting temperature and the grain size and columnar zone. The small grain size and shorter columnar zone in casting 8 is therefore
probably an effect of another mechanism than just the low casting temperature.
By comparing cerium content and grain size as is done in Figure 2, there seems to be a sudden drop in the grain size as the cerium
content reaches a critical value. This indicates that cerium is an effective grain refiner in this steel, but that certain critical cerium
content in the steel is required. As is seen from Figure 3, the width of the columnar zone seems to be little influenced by the cerium
content at low cerium levels, but is reduced when the cerium content increases above approximately 0.035 %.
Figure 2. Cerium content versus grain size.
Figure 3. Cerium content versus length of columnar zone.
767
AISTech 2011 Proceedings - Volume II

According to the theory of grain refinement, cerium based particles should be able to nucleate equiaxed grains, giving a finer grain
structure and a more narrow columnar zone. In Figures 4 and 5 the grain size and the width of the columnar zone are plotted against
the number density of cerium-containing particles. From a comparison with Figures 2 and 3, it is evident that the grain size and
columnar width depend on the number density of cerium-based particles in the same way as they depend on the cerium content. Thus
to obtain a fine grain structure and a small width of the columnar zone, the number density of cerium-containing particles in the steel
has to be above a critical value.
Figure 4. Number density of cerium-containing particles versus grain size.
Figure 5. Number density of cerium-containing particles versus length of columnar zone.
Different types of cerium-containing particles were found. These particles are deemed to be CeO2, CeAlO3 and Ce2O2S, as the atomic
ratios measured match these known cerium compounds fairly close. By looking at the added cerium and aluminum amounts, the
residual cerium content (Table III), and the particles found in the different castings (Table IV), there seem to be an effect that small
additions of cerium give CeAlO3 inclusions. Even though no aluminum was added, the residue in the melt was enough to be
incorporated in the cerium oxide phase. In the cases where moderate amounts of cerium were added (castings 4, 6 and 7), there are a
mix of CeO2, CeAlO3 and Ce2O2S. In casting 7 a relatively large amount of aluminum was also added, but this only gave a slight
predominance of CeAlO3. The highest additions of cerium were applied to castings 5 and 8, and in these castings almost all particles
were CeO2. Surprisingly, in casting 8 where aluminum also was added, no CeAlO3 was detected among the 20 analyzed particles. The
reason for this correlation between amount of cerium added and type of inclusion precipitated is not understood, but this is an
important aspect which needs to be clarified if specific cerium compounds are required.
768
AISTech 2011 Proceedings - Volume II

If the theoretical grain size is calculated by assuming that all particles act as a nucleation point and are evenly distributed in the steel,
the average grain size g for a two dimensional case is given by:
1
()gN
(1)
where N is measured number of particles per square millimeter. The estimated grain size for the castings that were added cerium is,
together with the obtained grain size, given in Table V. It is seen that, for all alloys, there is a large difference between the theoretical
and the real grain size. Only a few particles have therefore been active in the grain nucleation process. By replacing g in Equation 1 by
the measured grain size, the number of active particles per mm2, Na=1/g2, can be calculated, assuming that all grains have been
nucleated by a cerium-based particle. The calculated Na- values are given in Table V. The table also gives the percentage of particles
that have nucleated grains, i.e. the refiner efficiency.
Table V. Summary of the experimental results obtained in the grain refining experiments.
Casting
Measured
grain size
Estimated
grain size
N,
mm-1
Na,
mm-1
Refiner
efficiency
3
930 μm
143 μm
49
1.2
2.4 %
4
820 μm
111 μm
81
1.5
1.9 %
5
970 μm
80 μm
157
1.1
0.7 %
6
810 μm
141 μm
50
1.5
3.0 %
7
1060 μm
102 μm
97
0.9
0.9 %
8
270 μm
74 μm
188
13.7
7.3 %
It is seen that the refiner efficiency varied a lot between the different castings, and is generally quite low. For casting 8, which has the
smallest grain size, the refiner efficiency is by far the highest. Compared to other alloy systems, the obtained yield values are not
small, however. For instance, in a paper by Greer7 on grain refinement of aluminium by means of TiB2 particles, refiner efficiencies in
the range 0.09% - 0.6 % are reported. The reason for low efficiency values is that most of the grain refining particles are too small to
nucleate grains from the melt. Greer reports a minimum particle size in the order of 0.8-3 μm, depending on the type of grain refiner
7,8. Table IV shows that most of the cerium oxide/sulfide particles in castings 3-8 are in the range 1-2 µm. This could indicate that the
critical size of cerium based grain refiners in austenitic manganese steels is somewhat larger that anticipated.
From Table III and IV is difficult to draw any decisive conclusion regarding the efficiency of the different oxides as a grain refiner.
Based on the mismatch to the austenite lattice at 1350°C 9, where the solidification of this steel starts to occur, the approximate misfit
is 3.2 % for CeAlO3, and 4.8 % for CeO2. CeAlO3 is therefore expected to be a better substrate for nucleating austenite than CeO2 is.
However, from the types of particles found (Table IV) no such correlation between obtained grain size (Table IV) and type of grain
refining particle can be deduced. For instance, a majority of CeO2 particles were detected in both casting 5 and casting 8 which have
the lowest and the highest refiner efficiency of all alloys respectively. There is not any prominent difference in the mean particle size
that can explain the difference in refiner efficiency between the two alloys.
The only parameter that could have influenced the refiner efficiency and the obtained grain sizes in the observed way is the casting
temperature which varied slightly between the alloys. If refiner efficiency is plotted against casting temperature, as is done in Figure 6,
one see that the refiner efficiency increases with decreasing casting temperature. Thus, the oxides are not active as grain refiners if the
casting temperature is too high. A lower casting temperature will give a lower temperature gradient during solidification, this will in
turn give a wider undercooled zone ahead of the solidification front and increase the effect of the substrate particles as nucleation sites.
769
AISTech 2011 Proceedings - Volume II

Figure 6. Refiner efficiency plotted against casting temperature. Each point is indicated with the appropriate casting number. Refiner
efficiency increases with decreasing casting temperature.
The information in Figure 6 does not give any further indication of a possible correlation between refiner efficiency and oxide type.
This might indicate that the refiner efficiency of the two types of cerium oxides is relatively equal. A possible reason for this can be
the relatively low mismatch of both oxides towards the austenite (3-5 %). Another reason could be that the surface energies of the
oxides towards the melt and towards the austenite embryo are of greater importance for their nucleation ability than the lattice
mismatch is. More work is needed to confirm any of these suggestions.
SUMMARY
From this work it can be concluded that:
With correct additions of cerium and aluminum, grain refinement of austenitic high manganese steel is attainable, as long as the other
influencing parameters are within favorable conditions.
The effect of the added grain refiner is dependant of good control of the casting temperature. The refiner efficiency reduces with
increasing temperature.
Types of cerium containing particles formed depend on the amount of cerium added.
ACNOWLEGEMENTS
The author wishes to thank Scana Steel Stavanger AS and Elkem AS for contributing to this work. I would also like to thank Professor
Øystein Grong and Morten Raanes at Norwegian University of Science and Technology.
REFERENCES
[1] Handbook of ternary alloy phase diagrams, ASM International, 1995.
[2] John Campbell, “Structure, defects and properties of the finished casting” Castings, Elsevier, 2004, pp. 267-305.
[3] D.A. Porter, K.E. Easterling, “Solidification”, Phase transformations in metals and alloys, Nelson Thornes Ltd, Cheltenham,
UK, 2004, pp. 185-262.
[4] W. Kurz, D.J. Fisher, “Fundamentals of solidification”, Trans tech publications ltd., 1998.
[5] Ø. Grong, P. Jonsson, O.S. Klevan, “The future role of ferroalloys in iron and steel”, The Ninth International Ferroalloys
Congress and the Manganese 2001 Health Issues Symposium, Quebec City, Canada, 2001, p 562-572.
770
AISTech 2011 Proceedings - Volume II

[6] C. Eijk, J. Walmsley, Ø. Grong, O.S. Klevan, ”Grain refinement of fully austenitic stainless steels using a Fe-Cr-Si-Ce
master alloy”, 59th Electric Furnace and 19th Process Technology Conferences, Phoenix AZ, 2001.
[7] A.L. Greer, A.M. Bunn, A. Tronche, P.V. Evans, and D.J. Bristow, “Modelling of inoculation of metallic melts: Application
to grain refinement of aluminium by Al-Ti-B”, Acta Mater. 48, 2000, p. 2823-2835.
[8] A. Tronche and A.L. Greer, “Electron-backscatter diffraction study of inoculation of Al”, Philos. Mag. Lett. 81, 2001, p. 321-
328.
[9] C. Eijk, Ø. Grong, F. Haakonsen, L. Kolbeinsen, G. Tranell, "Progress in the development and use of grain refiner based on
cerium sulfide or titanium compound for carbon steel", ISIJ International, Vol. 49, No.7, 2009, pp. 1046-1050.
771
AISTech 2011 Proceedings - Volume II