AISTech 2019 — Proceedings of the Iron & Steel Technology Conference

6–9 May 2019, Pittsburgh, Pa., USA

DOI 10.1000.377.220

Development of a Powder Metallurgy Technique for Introducing

Synthetic Oxide Inclusions Into Liquid Steel

A. Smith1, T. Tran1, F. Ji1, W. Tiekink2, G. Abbel2, S. Verdier2

1
Materials Processing Institute
Eston Road, Middlesbrough, Cleveland, UK, TS6 6US
Phone: +44 (0) 1642 382000
Email: andy.smith@mpiuk.com
2
Tata Steel Europe
NL1970CA, IJmuiden, The Netherlands

Keywords: Steel, Inclusions, Cleanness, Powder Metallurgy, Solidification

INTRODUCTION
Non-metallic inclusions are detrimental to the castability of clean steels and to the mechanical properties of the rolled
and finished product. For example, bearing steels require very low inclusion levels, which would otherwise act as stress
raisers promoting the initiation of micro-cracks and spalling leading to premature bearing failure. Tyre wire steel is drawn
down to very small diameters (~150 µm) and incorporated into tyre tread. Excessive levels of non-deformable
inclusions, such as alumina, significantly increases the risk of wire breakage at the drawing stage, which disrupts the
process and impacts on plant productivity. For these reasons it is important to control and make advancements in
steelmaking operating practices to continually improve the overall cleanness of the manufactured steel.
Introducing synthetic inclusions into liquid steel is an important step in the study of inclusion removal and modification at a
laboratory level. However, previous studies have shown that it is extremely problematic to add synthetic
inclusions successfully on a small-scale experimental level and achieve the inclusion densities required to realise meaningful
results. In 2012 Dogan et al. described sealing inclusions in-between two pieces of steel plate.1 The plates were
compressed together using a mechanical press utilising 50 tonnes of pressure to create a “steel sandwich”. The
compression sealed the oxide inclusions into the metal and removed any entrapped gas. The steel sandwich was then added
to liquid steel, which resulted in an inclusion recover of approximately 9%. A Japanese patent (JP2005091017) describes a
similar pressing method, whereby oxide particles are spread on to the surface between a metallic foil and a steel plate
before being pressed together at high pressure.2 The use of powder metallurgy has also been proven to be a viable technique
where oxide particles and powdered metal are compacted together using either hot isostatic pressure or hydraulic pressing.3-6
The main advantage of all of these techniques is that oxide inclusions of known chemical composition, size and
density can be chosen for the intended experiment. This cannot always be achieved with any consistency when
utilising chemical reactions within a steel melt to produce the required oxide inclusions.
This paper describes the experimental materials, associated equipment and methods that have been undertaken by
the Materials Processing Institute (the Institute) and Tata Steel Europe, IJmuiden (Tata) to introduce oxide inclusions into
liquid steel at a batch size of 400 g and achieve acceptable inclusion recovery and distribution.

EXPERIMENTAL MATERIALS AND EQUIPMENT

Selection of Synthetic Oxide Inclusions

After careful consideration of the different types of compounds that could be used as a synthetic inclusion source it
was decided that cerium oxide (CeO2) was the most suitable for the experiments being undertaken. CeO2 is a readily
available and affordable compound for experimental purposes of this nature and scale. The compound has a high melting
temperature of 2400 °C, so will not dissolve into the steel when melted out in a crucible using an induction furnace. CeO2
also has a density

© 2019 by the Association for Iron & Steel Technology. 2137

closer to liquid steel of 7.22 g/cm3, so is less likely to float to the surface of the liquid steel during the melting process, as
might be expected to happen with alumina, which has a much lower density of 3.95 g/cm3. In these experiments, CeO2
inclusions in the range of 1 m to 14 m were incorporated into the metallic charge for induction furnace melting. Figure 1
details the size and chemical composition of the CeO2 inclusions used in the experiments. Two batches of CeO2 inclusions
were used, one at 5 m and one at 14 m. The inclusions in each batch were not uniform in size, with much smaller
inclusions present due to breakup of the inclusions.

100 m

100 m

Figure 1. CeO2 Inclusions for batches sizes of 5 m and 14 m

Induction Furnace Design

An induction furnace based at the Institute was used for melting the steel samples. The furnace is shown in Figure 2 and has
the following specifications:

Manufactured by Ekoheat: 45-kW / 50-150 kHz Annealed copper electromagnetic coils

Variable temperature control proven to 1800 °C Temperature controlled using x2 thermocouples
Controlled inert gas atmosphere (Ar, He) Water cooled top and bottom sealing
Crucible diameter up to 100 mm 3 kg melting capacity

2138 © 2019 by the Association for Iron & Steel Technology.

 Figure 2. Induction furnace based at the Institute

Crucible Design
The crucibles used in the experiments were flat bottomed, isostatically pressed alumina. The crucibles measured an
outer/inner diameter of 105/85 mm and an outer/inner length of 130/120 mm. The crucibles are designed for high hot
strength, having excellent thermal shock resistance and can operate to 1800 °C before failure. The maximum working
temperature is approximately 1620 °C.
However, during the first experiments using these crucibles it was found that their integrity was being compromised resulting
in severe cracking, only evident once the crucible was removed from the induction furnace. In one extreme case the cracking
resulted in a crucible breakout which could have severely damaged the induction furnace, as shown in Figure 3. After
thorough investigation, modifications were made to the position of the induction coils to ensure the whole area of the crucible
received a more uniform temperature gradient during heating and melting. In addition, the heating and melting gradient was
significantly reduced to prevent the alumina material heating up too quickly. Having made these changes, no further crucible
cracking was experienced.

Figure 3. Crucible failure during the early melting experiments

© 2019 by the Association for Iron & Steel Technology. 2139

Sintered Cerium Oxide Ingots
To produce material with the required quantities of synthetic CeO2 inclusions that could then be melted out in a crucible
within an induction furnace, it was decided to use a powder metallurgy to produce sintered ingots. These ingots were
manufactured by Tata’s Ceramics Research Centre (CRC), IJmuiden. In this study, powder metallurgy was used to distribute
CeO2 inclusions homogeneously in an iron matrix. Carbonyl iron powder with a grain size of approximately 10 µm was
mixed together with 0.1 wt.% CeO2 inclusions with a grain size either 5 µm or 14 µm and 1.0 wt.% of the binder
bisstearoylethylene diamide in a container called a Turbula Mixing System. The amount of iron powder, CeO2 inclusions and
binder were mixed for approximately 1 hour.
Approximately 200 g of the powder mixture was then pressed in a cylindrical mould (Ø 50 mm) with 60 MPa of pressure
using a hydraulic press. The pressed sample was then heated to 600 °C at a heating rate of 3 °C/min in atmospheric
conditions to remove the binder. The sample stayed at this temperature for 1 hour before it was allowed to cool down
following a natural cooling curve.
After debinding, the sample was put in a hydrogen furnace (see Figure 4a), to sinter the sample. A temperature program was
applied with a heating rate of 5 ºC/min, and the sample held at 600 °C for one hour, before further heating until a top
temperature of 1300 °C was reached. The holding time at the top temperature was 30 minutes, before the sample was cooled
down following a natural cooling curve. The density of the ingot was approximately 7.2 g/cm3 after sintering. The
manufactured ingots can be seen in Figure 4b.

Figure 4. (a) Hydrogen furnace, (b) Manufactured CeO2 sintered ingots

Samples of the sintered ingots have been cut, ground and polished to a 1µm finish (refer to small sample in Figure 4b),
suitable for both light optical metallography and field emission gun scanning electron microscopy (FEG-SEM) investigation.
After light optical investigation the samples were then examined using the Institute’s FEG-SEM, equipped with an Oxford
Instruments energy-dispersive x-ray spectroscopy (EDX) analyser and INCA Feature® software. Figure 5 shows examples of
the SEM microphotographs taken of the 5 m and 14 m sintered CeO2 ingots respectively. It was found that the CeO2
inclusions appeared well dispersed throughout the ingots, in amongst iron oxide (FeO) particles.

2140 © 2019 by the Association for Iron & Steel Technology.

 3

6 5
1
7

50 m 50 m
2

Label O Fe Ce Comment Label O Fe Ce Comment

1 17.03 5.92 77.60 CeO2 Inc. 4 16.62 2.33 81.05 CeO2 Inc.
2 16.22 13.76 70.01 CeO2 Inc. 5 16.23 4.46 79.31 CeO2 Inc.
3 15.57 14.69 69.74 CeO2 Inc. 6 12.04 16.59 71.38 CeO2 Inc.
7 12.52 38.29 49.19 CeO2 Inc.
Figure 5. FEG-SEM microphotographs and analysis, 14 m CeO2 sintered ingot
The sintered samples were then examined using the Institute’s FEG-SEM, equipped with an Oxford Instruments energy-
dispersive x-ray spectroscopy (EDX) analyser and INCA Feature® software. The software controls an automated procedure
that analyses particle features detected within a designated sample area. The data can then be processed to give several
parameters including (a) inclusion counts/mm2, (b) inclusion area %, (c) inclusion size and (d) particle shape for each
detected phase analysed. The analyser was able to measure very small sized particles < 1µm upwards within the designated
analysis areas.
Samples of the sintered ingot have been mapped to detect every CeO2 inclusion, so that a distribution of CeO2 inclusions
within the ingot could be determined. Figure 6 shows the location of CeO2 inclusions from a sample of 14 m sintered ingot.
The area for inclusion detection covered approximately 17 mm2. It was seen that the CeO2 inclusions were dispersed
homogenously in the sintered ingot, from a small inclusion size of 1 µm through to 50 µm. The increase in inclusion size to
50 m was caused by agglomeration during the manufacture of the sintered ingots. The INCA Feature® analysing software
was able to detect the size of each CeO2 inclusion, so that a size distribution could be made for the ingot, as shown in Figure
7. The major inclusion sizes were from 1 µm to 7 µm. No contamination of the inclusions in the sintered ingot was found.

Figure 6. SEM map of CeO2 inclusions using INCA Figure 7. Size distribution of CeO2 inclusions in the sintered
Feature® ingot

© 2019 by the Association for Iron & Steel Technology. 2141

The INCA Feature® results showed that the number of CeO2 inclusions was approximately 45 number counts/mm2. Based on
these results it was decided to use only the sintered ingots, which were manufactured from the 14 m CeO2 inclusions. These
ingots contained enough quantities of 1 m up to 50 m CeO2 inclusions that there was no requirement to use the material
that had been manufactured from the 5 m sintered ingots.

EXPERIMENTAL RESULTS AND DISCUSSION

Experimental methodology at 400 g scale

The experiments were undertaken at a relatively small scale of 400 g with an aim to develop a suitable technique to introduce
known quantities of sintered CeO2 ingot into the melt with the oxide inclusions homogenously dispersed throughout the ingot
after solidification. Due to the cost and time to manufacture the sintered ingots it was decided to incorporate the material into
the crucible with a scrap feedstock. To limit chemical reactions that would have happened if carbon-steel scrap was used as
the feedstock, a decision was made to use laboratory grade electrolytic iron. The chemical composition of this material is
detailed in Table I.

Table I. Chemical composition of electrolytic iron (ppm)

C P S Si Mn Cu N O H As Sb
5 7 5 <5 1 1 <5 90 3 <1 1

Sn B Ni Co Zn Al Pb Cr Ca Mg Fe
<1 1 1 3 3 1 1 3 1 1 Balance

For the 400 g scale experiments, 120 g of 14 m sintered CeO2 ingot was added together with 280 g of electrolytic iron into
an alumina crucible. The 45-kW induction furnace was then used for melting the samples. The induction furnace was purged
with argon to remove any traces of oxygen. The alumina crucibles used in these experiments were surrounded by a graphite
sleeve manufactured in the workshop of the Institute to improve melting performance. Several techniques of placing the
sintered cerium oxide ingot within the electrolytic iron have been tested. The electrolytic iron and sintered cerium oxide ingot
were melted out in the induction furnace and the temperature raised to 1610 °C and held for 5 minutes. The melt temperature
was measured using a type-S thermocouple submerged in the liquid steel (Figure 2). The induction furnace was then shut
down and the melt temperature quickly cooled down to room temperature.
Once the experiments were completed the solidified ingot was left to cool naturally within the induction furnace. Once
cooled to ambient temperature the crucible was taken out the furnace and the ingot removed from the alumina crucible,
cleaned and sectioned for metallographic assessment and FEG-SEM EDX analysis. Samples of the induction ingots were
mapped using the INCA Feature® software to detect every CeO2 inclusion, so that a distribution of CeO2 inclusions (top,
middle and bottom) could be determined for each of the experiments undertaken.

Results of the induction furnace experiments

To determine the best yield of CeO2 inclusions in the solidified ingot, three experiments using 120 g of the sintered ingot and
280 g of electrolytic iron were undertaken. Each experiment used a different technique of placing the samples of sintered
ingot within the alumina crucible, as shown schematically in Figure 8 and were as follows:
1. Sintered ingot at the bottom, electrolytic iron at the top.
2. Sintered ingot in the middle, electrolytic iron at the top and bottom
3. Sintered ingot close to the bottom surrounded by electrolytic iron.

2142 © 2019 by the Association for Iron & Steel Technology.

 E - iron E - iron
E - iron
CeO2 Ingot
CeO2
CeO2 Ingot E - iron Ingot

1st Experiment 2nd Experiment 3rd Experiment

Figure 8. Position of material for the three induction furnace experiments

Figure 9 shows the condition of the induction furnace ingots after they had been removed from the alumina crucibles.

1st Experiment 2nd Experiment 3rd Experiment

Figure 9. Induction furnace ingots from the three experiments

After sectioning the induction furnace ingots and examining using the FEG-SEM and EDX analyser, it was found that the
position of the sintered material within the crucible before induction furnace melting affected the agglomeration, floatation
and distribution of the CeO2 inclusions within the solidified ingot. Table II shows the inclusion number counts/mm2 for the
three induction furnace ingots taken at top, bottom and centre positions. It can be seen in the table that the third experiment
showed the maximum recovery of CeO2 inclusions.

Table II. Inclusion counts for different sintered ingot positions

Experiment Ingot Bottom Ingot Centre Ingot Top Ingot Average
(#) (number counts/mm2) (number counts/mm2) (number counts/mm2) (number counts/mm2)
1 9.0 12.4 10.4 10.6
2 12.1 15.5 15.9 14.5
3 15.2 16.7 20.3 17.4

A sample area of 17 mm2 was scanned using INCA Feature® for the three experiments. Figure 10 shows the distribution of
the CeO2 inclusions at the centre of each sample. The SEM maps showed similar results to the sintered CeO2 ingot (Figure 6),
in that the CeO2 inclusions were homogenously distributed within the samples of the three experiments.

© 2019 by the Association for Iron & Steel Technology. 2143

 1st Experiment 2nd Experiment 3rd Experiment
Figure 10. INCA Feature® map of CeO2 inclusions at the centre position of the induction furnace samples
The second and third experiments showed that the number of CeO2 inclusions were approximately 16 number counts/mm2,
which had dropped from 45 number counts/mm2 for the sintered CeO2 ingot, which was initially added to the melt. This was
consistent with 30% of the sintered CeO2 ingot having been added to the induction furnace with the electrolytic iron. It was
therefore concluded that approximately 95% of the CeO2 inclusions had been recovered from the sintered CeO2 ingot into the
induction furnace ingots.

Figure 11. Size distribution of CeO2 inclusions at the centre of the induction furnace samples
Figure 11 shows the size distribution of CeO2 inclusions for the three experiments at the centre location of the melted-out
ingot. Analysis showed that the major CeO2 inclusion sizes were from 1 µm to 7 µm. No larger inclusion sizes from 14 µm to
over 30 µm were recorded from the melted-out ingot. This can be explained if the larger size inclusions detected from the
original sintered ingot were due to the agglomeration of particles during the sample preparation process. Once these large
oxide inclusions got into the liquid melt, the agglomerated inclusions were separated into their original particle size.
More in-depth calculations of the stability of the CeO2 inclusions in the liquid melt was conducted using the thermodynamic
software FactSage 7.2, utilising databases FactPS, FToxid and FTmisc. The oxygen activity was calculated with an input O2
content of 0.01 ppm, coming from the argon used in the induction furnace. The temperature range used in the calculations
was 1200 °C to 1650 °C. Figure 12 shows the results of the FactSage calculations, indicating that Ce2O3 is stable in the liquid
iron and would be duplexed with FeO.

2144 © 2019 by the Association for Iron & Steel Technology.

 Figure 12. FactSage calculations of cerium oxide in electrolytic iron
SEM elemental mapping confirmed that dual phase FeO-Ce6O11 inclusions were the predominant inclusion composition in
the induction furnace samples. An example of the SEM elemental mapping is shown in Figure 13.

Figure 13. SEM elemental mapping confirming inclusion composition

CONCLUSIONS
A powder metallurgy technique has been developed to enable the addition of synthetic inclusions of known size, composition
and density into liquid steel. The sintered cerium oxide ingot was added with electrolytic iron and melted-out in an induction
furnace. Metallographic assessment of the samples showed that the FeO-Ce6O11 inclusions were homogenously dispersed
throughout the sample from top, centre and bottom. Analysis suggested that more than 95% of the CeO2 inclusions from the
sintered ingot had been distributed throughout the solidified ingots produced during the experiments. This technique is
therefore useful for future steel cleanness studies for high value alloyed steels.

© 2019 by the Association for Iron & Steel Technology. 2145

 ACKNOWLEDGEMENTS
The authors would like to thank Tata Steel Europe for funding this research work and for kind permission to present this
paper. The authors also would like to thank Tata Steel Europe, Ceramics Research Centre (CRC), IJmuiden for their support
in manufacturing the sintered Ce2O3 ingots.

REFERENCES
1. N. Dogan et al., Scanmet IV, 2012, pp. 1-10
2. K. Pruvcheva and T. Tsolov, Metallurgiya (Sofia), 33 (9), 1978, pp. 21-23
3. K. Flórián et al., Fresenius J. Anal. Chem, 349, 1984, pp. 174
4. K Flórián et al., J. Anal. At. Spectrom, 9, 1994, pp. 257
5. H. Mittelsädt, 2002, Patent EP1227312
6. F. Boué-Bigne F et al., European Commission Research Fund for Coal and Steel, ISBN 978-92-79-14228-4, 2010

2146 © 2019 by the Association for Iron & Steel Technology.