Professional Documents
Culture Documents
InterpretationandClassificationofNon MetallicInclusions MPC20160040 DL.000109386 1.Pdf0
InterpretationandClassificationofNon MetallicInclusions MPC20160040 DL.000109386 1.Pdf0
Characterization
A. Kazakov,1 A. Zhitenev,2 and S. Ryaboshuk2
DOI: 10.1520/MPC20160040
Interpretation and
Classification
of Non-Metallic Inclusions
VOL. 5 NO. 5 / 2016
Materials Performance and Characterization
Reference
Kazakov, A., Zhitenev, A., and Ryaboshuk, S., “Interpretation and Classification
of Non-Metallic Inclusions,” Materials Performance and Characterization, Vol. 5, No. 5,
2016, pp. 1–9, doi:10.1520/MPC20160040. ISSN 2165-3992
ABSTRACT
Manuscript received March 31, The non-metallic inclusions in steel are one of the most important parameters
2016; accepted for publication of metallurgical quality. Despite the extensive progress in thermodynamic
May 27, 2016; published online
August 25, 2016.
description of non-metallic inclusions formation in the liquid and solidifying
1
steel, a complete simulation of their evolution during secondary metallurgy,
Peter the Great Saint Petersburg
Polytechnic Univ.,
casting and solidification is impossible. The processes for non-metallic
Polytechnicheskaya str., 29, Saint- inclusion formation, growth and elimination, taking into account macro- and
Petersburg, 195251, Russia
micro-segregation evolution, are too complicated. Therefore, a statistically
(Corresponding author), e-mail:
kazakov@thixomet.ru reliable estimation of the amount, morphology, size distribution and chemical
2
composition of non-metallic inclusions obtained by modern experimental
Peter the Great Saint Petersburg
Polytechnic Univ., approaches could be an important basis for understanding steelmaking
Polytechnicheskaya str., 29, Saint- technology. Automated feature SEM/EDS analysis of the chemical
Petersburg, 195251, Russia
composition, size, and volume fraction of more than 600 non-metallic
inclusions in rail steel was conducted. Analysis of these non-metallic inclusions’
database by developed software found the following clusters of inclusions: (1)
oxides, Al-Ca-Si-Mn-S-O; (2) predominantly sulfides, Mn-S-Al-Ti-O; (3) oxides
and sulfides, Mn-S-O; and (4) complex non-metallic inclusions, Ti-Mn-O(N). All
found compositions of non-metallic inclusions were placed on the ternary
diagram Mn-Al-S and they lined up from the Al corner to the MnS point on the
Mn-S axis. The chemical compositions of the non-metallic inclusions evolved
from Al-Ca-Si-Mn-S-O to Mn-S-Al-Ti-O and, finally, to the Mn-S-O þ Ti-Mn-O(N)
system. Interpretation of the chemical composition of non-metallic inclusions
by thermodynamic modeling revealed the nature of each cluster with
Copyright V
C 2016 by ASTM International, 100 Barr Harbor Drive, P.O. Box C700, West Conshohocken, PA 19428-2959 1
2 KAZAKOV ET AL. ON CLASSIFICATION OF NON-METALLIC INCLUSIONS
Keywords
non-metallic inclusions in steel, automated feature SEM/EDS analysis, cluster analysis of
the non-metallic inclusions composition, automated feature analysis for assessment of
non-metallic inclusions compositions, thermodynamic simulation of non-metallic inclusions
formation in liquid and in solidifying steel
Introduction
Modern technology permits production of very “clean” steel by reduction of oxygen
and non-metallic inclusions. Total oxygen content does not exceed 40–60 ppm in
air-melting processes and 10–40 ppm after vacuum secondary metallurgy [1]. There-
fore, considering that the concentration of oxygen dissolved in molten melt is only
about 1–5 ppm, it is possible to conclude that the bulk oxygen content in the finished
metal is present within the non-metallic inclusions (NIs).
The formation and evolution of non-metallic inclusions is a complex multistage
process [2] that continuously proceeds along all metallurgical treatments. Non-
metallic inclusions can be classified by thermo-time nature as follows: (1) primary
NIs, which are formed under isothermal conditions of secondary metallurgy after
addition of deoxidants, when concentrations of these elements exceed the equilib-
rium values; (2) secondary NIs, which are formed because of the decrease of an ele-
ment’s solubility in the molten metal after its cooling from steelmaking temperature
to the liquidus temperature; (3) tertiary NIs, which are formed during solidification
not only because of the temperature decrease, but mainly because of segregation
processes; and, (4) quarternary NIs, which are formed in solid metal.
A comprehensive description of non-metallic inclusions formation, their growth
and the removal processes in connection with hydrodynamics of the liquid and sol-
idifying steel, micro- and macro-segregation, and the formation of various billet
structural zones is impossible [3–7]. Therefore, a statistically reliable estimation of
the amount, size, and composition of NIs in steel obtained by a modern experimen-
tal approach could be an important basis for detailed interpretation of steelmaking
technology. Conventional methods of metallographic analysis of NIs using light op-
tical microscopes (LOM) make it possible to determine their volume fraction, size,
and distribution pattern (ASTM E1245) [22]. An attempt at NIs composition esti-
mation was made in ASTM E45 [23]. However, the classification of non-metallic
inclusions in ASTM E45 does not reflect the whole variety of inclusions formed
under the conditions of modern steelmaking technology. Therefore, an inclusion rat-
ing using ASTM E45 is carried out rather by their external features than by their
chemical composition. For example, deoxidation products in steel formed during
TABLE 1
Chemical composition of steel.
Element C Mn Si P S Cr Ni Cu Ti Al Ca
Wt. % 0.82 0.84 0.38 0.015 0.012 0.028 0.04 0.01 0.008 0.005 0.003
reliable information, not only for the volume fraction and size but also of the ele-
mental composition of each of the detected inclusion. However, these databases are
complicated and require inclusions to be separated into groups according to their
composition. The procedure for the interpretation of the nature of non-metallic
inclusions by SEM-EDS microanalysis method and thermodynamic modeling was
previously proposed [17]. Then, this procedure was implemented using automated
feature analysis [11]. In this paper, we used the database cluster analysis [18] by
non-metallic inclusions chemical compositions, which have been created as a soft-
ware package. The clusters were identified from the raw data by k-means clustering
algorithm [18].
The cluster analysis is a breakdown of object sets into disjoint subsets (clusters)
in the way that each cluster contains similar objects and objects of different clusters
are distinguished [18,19]. In our case, the cluster analysis method means that inclu-
sions are separated according to the similarity of their chemical composition. An
important issue of the cluster analysis is a selection of a finite number of clusters on
which non-metallic inclusions are divided. The minimum is one cluster that includes
all observed inclusions; the maximum is the number of all detected non-metallic
inclusions. In the present work, the number of clusters was chosen in accordance
with a change in distortion. Distortion is the sum of the root-mean-square Euclidean
distances between points of NIs chemical compositions and center of cluster to
which these points belong [20]. Distortion decreases with an increasing amount of
clusters. Number of the clusters is assigned by the analysis of relationship of distor-
tion and the number of clusters. Separation of inclusions into clusters is accom-
plished when it does not lead to the appearance of clusters with significantly
different compositions.
TABLE 2
Cluster analysis results from the non-metallic inclusions database.
TABLE 3
Non-metallic inclusions composition in clusters.
Clusters Al2O3 CaO SiO2 MnO MgO CaS MnS FeS TiCxOyNz
Primary and secondary
Al-Ca-Si-Mn-S-O 50 15 15 7 4 5 4 – –
Tertiary, the first half of solidification
Mn-S-Al-Ti-O 19 – – 4 – – 50 8 19
Tertiary, the second half of solidification
Mn-S -O – – – 3 – – 80 17 –
Ti-Mn-O(N) – – – 5 – – 7 – 88
inclusions were combined into clusters by their chemical composition (Table 2).
Elements were arranged in order of their decreasing concentration in clusters: (1)
predominantly oxides, Al-Ca-Si-Mn-S-O; (2) predominantly sulfides, Mn-S-Al-Ti-O;
(3) oxysulfides, Mn-S-O; and (4) non-metallic inclusions of Ti-Mn-O(N) system. For
each of the 634 non-metallic inclusions studied, the chemical composition was
recalculated into the phase composition in accordance with the method in Ref 8.
Then, the average phase composition of each cluster was determined (Table 3).
The results of NIs assessment by AFA analysis are as follows: total volume frac-
tion of all NIs was 0.046 %; maximum average area of non-metallic inclusions based
on clusters: 16.5 lm2 for Al-Ca-Si-Mn-S-O, 12 lm2 for Mn-S-O, and 6.2 lm2 for
Ti-Mn-O(N). The volume fraction and size of non-metallic inclusions estimated by
LOM and AFA agreed well with each other.
Cluster analysis estimates the diversity of all non-metallic inclusions formed
during ladle treatment, casting, and solidification. But, to improve steelmaking tech-
nology, the temperature at the beginning of NIs formation and the concentration of
the elements in the molten metal at this point are necessary. To solve this task, ther-
modynamic modeling was carried out. The calculated oxygen equilibrium concen-
tration for the investigated steel at 1550 C was 8 ppm. However, the experimental
oxygen content in steel estimated using the gas analyzer Eltra ON900 was 35 ppm.
This total oxygen (dissolved in steel and fixed in non-metallic inclusions) was used
for the thermodynamic simulation.
An example of the simulation of the non-metallic inclusions formation process
is shown in Fig. 1. When the initial concentration of oxygen [O] selected was 35 ppm
at 1550 C, the primary NIs are formed. During subsequent cooling of the molten
metal, secondary NIs of the same composition continued to grow: alumina and liq-
uid oxysulfide inclusions (Fig. 1a). Below 1550 C, during the solidification of these
liquid inclusions, the solid oxides and sulfides were formed (Fig. 1b). Exactly the
same complex inclusions combined into clusters were found in the investigated
specimens. So, the thermodynamic simulation for chemical composition interpreta-
tion and revealing the thermo-time nature of NIs clusters was used.
CLUSTER OF AL-CA-SI-MN-S-O
At 1550 C, primary NIs represented by alumina and liquid oxysulfide solutions
(Fig. 1) are formed. It should be noted that there were practically no inclusions of
FIG. 1 Thermodynamic simulation of non-metallic inclusions formation (a), and crystallization of liquid inclusions (b) in steel:
[Al] ¼ 0.004 %, [Ca] ¼ 0.00035, [Ti] ¼ 0.006, [S] ¼ 0.01, [O] ¼ 35 ppm, and [N] ¼ 40 ppm.
pure alumina in the investigated specimens, because of their active removal and dis-
solution during coalescence with liquid inclusions during ladle treatment. With fur-
ther cooling of the liquid steel, secondary inclusions with compositions similar to
the primary inclusions are formed. Additionally, solid compounds of Al2O3,
CaOAl2O32SiO2, MnOAl2O3, MnS, CaS, and 6Al2O3CaO crystallize from
liquid slag inclusions (Fig. 1b). All of these compounds were found in the Al-Ca-Si-
Mn-S-O cluster (Table 3): 50 % Al2O3, 15 % CaO, 15 % SiO2, 7 % MnO, 5 % MnS,
and 4 % CaS. A small amount of magnesia spinel was found in non-metallic inclu-
sions of this cluster type. The origin of this compound is a reduction of magnesium
from MgO-containing refractories by the aluminum in the molten steel [21].
CLUSTER OF MN-S-AL-TI-O
During steel solidification, elements such as oxygen, titanium, sulfur, and others
enrich the liquid fraction of a dendritic cell and accelerate formation of tertiary tita-
nium oxides and manganese sulfides (Fig. 1a). These non-metallic inclusions, along
with the remaining aluminum oxides, form cluster of Mn-S-Al-Ti-O: 19 % Al2O3,
19 % TiCxNyOz, and (50 % MnS þ 8 % FeS) (Table 3).
FIG. 2
Ternary diagram of chemical
compositions of non-metallic
inclusions in steel:
[Al] ¼ 0.004 %, [Ca] ¼ 0.00035,
[Ti] ¼ 0.006, [S] ¼ 0.01,
[O] ¼ 35 ppm, and
[N] ¼ 40 ppm.
Results of the simultaneous analysis of the composition and size of the different
thermo-time nature NIs are presented in Fig. 3. The size of primary and secondary
NIs decreases correspondingly with the decrease of the temperature at the beginning
of their formation. At the same time, the content of sulfur in the NIs grows. Indeed,
more high-temperature formation of these inclusions results in their larger size
because of their growth and coagulation. The non-metallic inclusions formed near
the liquidus temperature have the minimal size, because they were frozen in solid
steel as soon as they are formed. There are beneficial conditions for growth of NIs at
further slow solidification in the liquid fraction of a dendritic cell: the impurities
liquate and there is enough time for the growth mainly of sulfides as it is manganese
and sulfur that are accumulated here.
FIG. 3
Composition and size of the
formed NIs with different
thermo-time natures.
Conclusions
1. A database for non-metallic inclusions formed in continuous cast rail steel
billets using automated feature analysis has been created. The technique for
processing of this database for identification of non-metallic inclusions clus-
ters using the principle of chemical composition similarity has also been
developed. Thermodynamic simulation of non-metallic inclusions formation
in liquid and in solidifying steel has been used for interpretation of the
thermo-time nature of the identified clusters.
2. All compositions plotted on the ternary diagram Mn-Al-S lined up from the
Al corner to the MnS point on the Mn-S axis. The evolution of non-metallic
inclusions compositions during the decrease of the start temperature of their
formation proceeds from the cluster Al-Ca-Si-Mn-S-O at the start tempera-
ture of their formation in liquid steel to the Mn-S-Al-Ti-O cluster at the
beginning of the steel solidification, and finally to the inclusions of the system
Mn-S-O þ Ti-Mn-O(N) at the end of solidification.
3. The methods developed can be used for the improvement of steel deoxidation
and inclusions modification technologies. Further accumulation of data on
non-metallic inclusions clustering in various steels and various technologies of
deoxidation can form a basis for the development of universal classification
system for non-metallic inclusions.
References
[1] Zhang, L. and Thomas, B. G., “State of the Art in Evaluation and Control of
Steel Cleanliness,” ISIJ int., Vol. 43, No. 3, 2003, pp. 271–291.
[2] Zhang, L., “Nucleation, Growth, Transport, and Entrapment of Inclusions Dur-
ing Steel Casting,” JOM, Vol. 65, No. 9, 2013, pp. 1138–1142.
[3] Pfeiler, B. G., Thomas, M., Wu, A., Ludwig, A., and Kharicha, C., “Solidification
and Particle Entrapment During Continuous Casting of Steel,” Steel Res. Int.,
Vol. 77, No. 7, 2006, pp. 1–10.
[4] Matsumiya, T., “Mathematical Analysis of Segregation and Compositional
Changes of Nonmetallic Inclusions in Steel During Solidification,” Mater.
Trans, Vol. 33, No. 9, 1992, pp. 783–794.
[5] Banaszek, J., Mcfadden, S., Browne, D. J., Sturz, L., and Zimmermann, G.,
“Natural Convection and Columnar-to-Equiaxed Transition Prediction in a
Front-Tracking Model of Alloy Solidification,” Metal. Mater. Trans. A, Vol.
38A, No. 1476, 2007, pp. 1476–1484.
[6] Wang, C. and Beckermann, C., “Prediction of Columnar to Equiaxed Transi-
tion During Diffusion-Controlled Dendritic Alloy Solidification,” Metal. Mater.
Trans. A, Vol. 25A, No. 5, 1994, pp. 1081–1093.
[7] Volkova, O., Heller, H. P., and Janke, D., “Microstructure and
Cleanliness of Rapidly Solidified Steels,” ISIJ Int., Vol. 43, No. 11, 2003,
pp. 1724–1732.
[8] Kazakov, A., Kovalev, P., Ryaboshuk, S., Zhironkin, M., and Krasnov, A.,
“Control of Nonmetallic Inclusions Formation During Converter Steel
Production,” Chernye Metally (Ferrous Metals), Vol. 2, No. 4, 2004, pp. 37–41.
[9] Coletti, B., Blanpain, B., Vantilt, S., and Sridhar, S., “Observation of Calcium
Aluminate Inclusions at Interfaces Between Ca-Treated, Al-Killed Steels and
Slags,” Metal. Mater. Trans. B, Vol. 34, No. 5, 2003, pp. 533–538.
[10] Story, S., Smith, S., Fruehan, R., Casuccio, G., Potter, M., and Lersch, T.,
“Application of Rapid Inclusion Identification and Analysis,” Iron and Steel
Technology Conference, AISI Tech, Sept. 15–17, 2004, Nashville, TN, pp. 41–49.
[11] Kazakov, A., Lyubochko, D., Ryaboshuk, S., and Chigintsev, L., “Investigation
of the Nature of Nonmetallic Inclusions in HSLA Steels Using an Automatic
Particle Analyzer,” Chernye Metally (Ferrous Metals), Vol. 2, No. 4, 2014,
pp. 37–41.
[12] Ren, Y., Wang, Y., Li, S., Zhang, L., Zuo, X., Lekahn, S., and Peaslee, K.,
“Detection of Non-Metallic Inclusions in Steel Continuous Casting Billets,”
Metal. Mater. Trans. B, Vol. 45, No. 1292, 2014, pp. 1291–1303.
[13] Kazakov, A., Zhitenev, A., and Kovalev, P., “Distribution Pattern of Nonmetal-
lic Inclusions on a Cross Section of Continuous Cast Steel Billets for Rails,”
Microsc. Microanal., Vol. 21 (Suppl. 3), 2015, pp. 1751–1752.
[14] DIN EN 13674-1, Railway Applications - Track - Rail - Part 1: Vignole Railway
Rails 46 kg/m and Above, Austrian Standards Institute, Vienna, Austria.
[15] Garber, A., Arsenkin, A., Grigorovich, K., Shibaev, S., Kushnarev, A., and
Petrenko, Y., “Analysis of Various Versions of the Deoxidation of Rail Steel at
OAO NTMK,” Russian Metal. (Metally), Vol. 2009, No. 7, 2009, pp. 581–586.
[16] Dhua, S., Amitava, R., Sen, S., Prasad, M., Mishra, K., and Jha, S., “Influence of
Nonmetallic Inclusion Characteristics on the Mechanical Properties of Rail
Steel,” J. Mater. Eng. Perform., Vol. 9, No. 700, 2000, pp. 700–709.
[17] Kazakov, A., Kovalev, P., Ryaboshchuk, S., Mileikovsky, A., and Malakhov, N.,
“Study of Thermo Time Nature of Non-Metallic Inclusions in Order to
Improve Metallurgical Quality of High-Strength Tube Steels,” Chernye Metal.
(Ferrous Metals), No. 12, 2009, pp. 5–11.
[18] Jain, A. K., Murty, M. N., and Flynn, P. J., “Data Clustering: A Review,” ACM
Comput. Surveys, Vol. 31, No. 3, 1999, pp. 1–69.
[19] Jain, A. K. and Dubes, R. C., Algorithms for Clustering Data, Prentice Hall, Eng-
lewood Cliffs, NJ, 1988, 334 pp.
[20] Gersho, A. and Gray, R. M., Vector Quantization and Signal Compression,
Kluwer, Boston, 1992, 760 pp.
[21] Yang, S., Li, J., Zhang, L., Peaslee, K., and Wang, Z., “Evolution of MgOAl2O3:
Based Inclusions in Alloy Steel During the Refining Process,” Metal. Mining
Ind., Vol. 2, No. 2, 2010, pp. 87–92.
[22] ASTM E1245-03, Standard Practice for Determining the Inclusion or Second-
Phase Constituent Content of Metals by Automatic Image Analysis, ASTM
International, West Conshohocken, PA, 2016, www.astm.org
[23] ASTM E45-13, Standard Test Methods for Determining the Inclusion Content of
Steel, ASTM International, West Conshohocken, PA, 2013, www.astm.org