of inclusions and their development during secondary steelmaking217063/... · 2009-05-13 ·...
TRANSCRIPT
Formation of Inclusions and their Development during Secondary Steelmaking
Mikael Thunman
Doctoral Thesis
School of Industrial Engineering and Management Department of Materials Science and Engineering
KTH Royal Institute of Technology SE‐100 44 Stockholm
Sweden
Akademisk avhandling som med tillstånd av Kungliga Tekniska högskolan framlägges till offentlig granskning för avläggande av teknologie doktorsexamen fredagen den 24 april 2009, kl. 10.00 i sal F3, Lindstedtsvägen 26, Kungliga Tekniska högskolan, Stockholm
ISRN KTH/MSE‐ ‐09/13‐ ‐SE+MICROMODMETU/AVH
ISBN 978‐91‐7415‐276‐0
i
Mikael Thunman Formation of inclusions and their development during secondary steelmaking
KTH School of Industrial Engineering and Management Division of Micro‐Modelling Department of Materials Science and Engineering School of Industrial Engineering and Management KTH Royal Institute of Technology SE‐100 44 Stockholm Sweden
ISRN KTH/MSE‐ ‐09/13‐ ‐SE+MICROMODMETU/AVH ISBN 978‐91‐7415‐276‐0
© The Author
ii
iii
And we should consider every day lost on which we have not danced at least once. And we should call every truth false which was not accompanied by at least one laugh.
‐ Friedrich Nietzsche
ABSTRACT Al–O relations in iron were investigated. Pure iron with varying Al content was equilibrated at 1873 K. The oxygen content of iron with higher Al content than 1.0 mass % was found to be much lower than previous works, while the oxygen content for Al content less than 1 mass% was found to be slightly higher. Further, a classification procedure of inclusions was developed using the commercial software INCA Feature. Three classes were made, spinel, TP‐(CaO‐Al2O3), and (CaO‐Al2O3) class, corresponding to the inclusions found during degassing at Uddeholm Tooling. The results showed that the spinel phase disappeared after degassing along with a reduction in numbers for the two phase inclusion (TP‐(CaO‐Al2O3)). Pure calcium aluminates however showed an increasing trend in a majority of the heats. The chemical development of inclusions at OVAKO Steel in Hofors, Sweden was also established. According to the morphologies and compositions, the inclusions were classified into 5 different types, namely, (1) alumina inclusions, (2) calcium aluminate, (3) spinel+calcium aluminate, (4) calcium aluminate surrounded by a CaS shell, and (5) spinel+calcium aluminate surrounded by a CaS shell. Thereafter refractory lining samples with attached slag layer were taken from used ladles at the two steel plants. The morphologies of the slag layers and the phases present were examined. The precipitated phases found in the refractory were 3CaO.Al2O3, MgO.Al2O3 and CaO in the case of Ovako Steel and 3CaO.Al2O3 and 2CaO.SiO2 in the case of Uddeholm Tooling. To help the understanding, model calculations using THERMOCALC were carried out. The model predictions differed somewhat from the experimental observation, the predicted major phases were in line with the EDS analysis on the refractory samples. Finally experiments were carried out to study the slag entrainment related to the open‐eye during ladle treatment. Ga‐In‐Sn alloy was used to simulate the liquid steel, while MgCl2‐Glycerol(87%) solution was used to simulate the ladle slag. No noticeable amount of top liquid was observed in any of the samples taken from the metal bulk during gas stirring. To confirm this aspect, slag‐metal interface samples were taken from an industrial gas stirred steel ladle. No entrapment was found in the steel. The accordance of the laboratory and industrial results suggests that the entrainment of slag into the steel bulk around the open‐eye cannot be considered as the major contribution to inclusion formation.
v
ACKNOWLEDGEMENTS First I would like to thank my main supervisor Prof. Du Sichen for your outstanding guidance and support during these years of studies. You have truly taught me a lot about how to conduct research.
Second I would like to thank my co supervisors Prof. Kazuki Morita and Dr. Johan Björkvall for your encouragement and discussions about research and other topics.
Dr. YoungJo Kang with all the others in Morita Laboratory at The University of Tokyo are truly acknowledged for making my stay in Japan memorable and fun even though it was 320 Kelvin in the lab.
I also like to thank Prof. Pär Jönsson, Prof. Seshadri Seetharaman, Dr. Ragnhild Aune and Dr. Margareta Andersson for fruitful suggestions.
Dr. Eckert with co‐workers at Forschungszentrum Dresden Rossendorf also deserves thanks for their hospitality and valuable research input.
Peter Kling needs a special thank for his technical expertise and help in experimental set‐ups. Without them there would have been more hang‐ups.
I specially would like to thank S‐O Eriksson, Alf Sandberg, Anders Gustavsson, Anders Lind, Dr. Robert Eriksson, Dr. Mselly Nzotta and all other people from the industry who have helped me and taught me valuable things about the steel industry. Jernkontoret are further acknowledged for their financial support of the research.
I thank all my friends and colleagues at the Department of Materials Science and Engineering for their friendship, specially my roommates Jimmy Gran and MinHo Song.
Last but definitely not least, I would like to thank my girlfriend Nina, my parents Lars and Gunilla, my baby sister Anna‐Maria, and of course the rest of the family members for your continuous love and support.
Stockholm, 22 March 2009.
Mikael Thunman
vii
SUPPLEMENTS This thesis is based on the following supplements:
Supplement 1: “Aluminum deoxidation equilibrium of molten Fe‐Al alloy with high aluminum content at 1873K” Y. Kang, M. Thunman, Du Sichen, T. Morohoshi, K. Mizukami and K. Morita Submitted to ISIJ International
Supplement 2: “Identification of inclusions using the results of INCA feature analysis and its application on the inclusions found in ladle treatment” M. Thunman Submitted to Steel Grips
Supplement 3: ”Origins of non‐metallic inclusions and their chemical development during ladle treatment” M. Thunman and Du Sichen Published in Steel research int. 79 (2008) No. 2, p. 52‐60
Supplement 4: “Slag‐refractory reaction during ladle refining and teeming” M. Thunman, J. Gran and Du Sichen Submitted to Steel Grips
Supplement 5: “Study on the formation of open‐eye and slag entrainment in gas stirred ladle” M. Thunman, S. Eckert, O. Hennig, J. Björkvall, and Du Sichen Published in Steel research int. 78 (2007) No. 12, p. 847‐854
ix
CONTENTS INTRODUCTION 1
LAB AND INDUSTRY STUDY 5
Thermodynamic study on the Al‐Fe‐O system 5
Inclusion classification procedure 5
Physical modeling of slag entrainment 8
Industrial experiments 9
RESULTS 13
Al‐O relation in liquid iron at 1873 K 13
Application of inclusion classification procedure on a tool‐steel 14
Chemical development of inclusions in a bearing steel 16
Inclusion formation from attached slag layers 18
Top phase entrainment in Ga‐In‐Sn model 21
Top slag entrainment in metal phase in an industrial ladle 21
DISCUSSION 23
Evaluation of the Al‐O parameters in liquid iron at 1873 K 23
Classification procedure with the help from INCA Feature 25
Origins and chemical development of inclusions 27
Inclusion formation from attached slag layers 28
Inclusion formation from slag entrainment at the open‐eye 31
CONCLUSIONS 33
FUTURE WORK 35
REFERENCES 37
xi
INTRODUCTION In order to minimize the number of detrimental inclusions in the final steel products it is important to know the mechanisms for the formation, chemical development and separation of inclusions during the steel making process, starting from tapping the electric arc furnace (EAF) to the solidification of the steel after casting. Lots of efforts have been made in the past to reveal the origins of the inclusions, which may come from many sources, such as deoxidation, reoxidation, slag entrapment and attached slag layers.[1‐14]
Deoxidation is a very important process during steelmaking and is in many cases done by adding aluminum. It lowers the dissolved oxygen content in steel to a great extent which is important when alloying with expensive alloying material such as chromium, vanadium, titanium etc. so it does not get oxidized. The drawback is that alumina inclusions are created. Normally they are in big clusters and are separated to the top slag quite quickly. However, some small ones might stay in the melt as inclusions. Good understanding of the deoxidation behavior is thus essential. The Al deoxidation equilibrium has been experimentally and thermodynamically studied since the middle of the 1950s, as summarized in Figure 1.[15‐29] However, some discrepancies are still evident among the results and a trend can hardly be found for the scattered values.
1
10-2 10-1 100 10110-4
10-3
10-2
10-1
H i l t y e t a l .
Fruehan Schenck et al. Rohde et al. Janke et al. McLean et al. (1823K) d'Entremont et al. (2013K) Swisher (1853K) JSPS recommended Itoh et al.
mas
s%O
mass%Al
Figure 1 Al‐O equilibrium in liquid iron at 1873 K (a few at other temperature) in the previous studies.
To monitor the inclusions, steelmakers have during the years worked out standards for detecting and classifying the inclusion types and quantities in their final products. This has in most cases been made manually with the light optical microscope (LOM) and in some cases scanning electron microscope (SEM) equipped with energy dispersive x‐ray spectroscopy (EDS) for chemical micro analysis. During manual investigation it is somewhat easy for an experienced microscopist to retrieve this information, although it is time consuming.
Recently, a new approach has been put forward enabling reasonable fast automatic computerized detection and classification of inclusions by using SEM‐EDS x‐ray together with the commercial software INCA Feature. Even though some considerations are made regarding the chemical composition of the inclusions, currently, focus is mainly on their size and shape in the final product and not on the chemical phases present during production of the liquid steel. Knowing the chemical composition and the present phases inside the inclusion is important when steelmakers want to find the formation mechanisms for different inclusion types. With this information, thermodynamic assessments can be made and the process parameters can be optimized. If the inclusions could be classified in a reproducible
2
way and in great numbers it would be somewhat easier to draw conclusions about if the non‐metallic inclusions derive from the major inclusion formation sources previously suggested by other researchers. At present time, ladle slag attached to the refractory material from the earlier heat has been reported as one of the major sources of inclusions[1‐10]. Researchers also believe that inclusions might come from top slag entrapment in the open‐eye region in the ladle during stirring.[12‐14]
In this thesis, the aim is to study some formation, chemical development and separation mechanisms of non‐metallic inclusions in liquid steel during secondary steelmaking by re‐evaluate the deoxidation curve for Al‐O in liquid steel at 1873 K. Thereafter, find a new inclusion classification approach to evaluate inclusions in steel samples taken from the metal bath. Also the chemical development of inclusions during production of bearing steel is investigated. Finally some major inclusion formation sources, such as ladle glaze and entrained slag caused by gas stirring are looked into.
3
4
LAB AND INDUSTRY STUDY Thermodynamic study on the AlFeO system In order to re‐evaluate the Al‐O relation in liquid iron at 1873 K some equilibrium experiments were made. Pre‐melted alloy samples with various Fe‐Al contents were prepared in the composition range of Al 0.01~10.0 mass % using pure iron and aluminum. The iron was placed in a pure alumina crucible and melted in an induction furnace under Ar‐3 vol % H2 gas mixture, to ensure reducing atmosphere. After the iron was completely melted, controlled amounts of aluminum were added. The furnace was kept at 1873 K for about 5 minutes after the aluminum addition to ensure that a homogeneous composition had been established. The pre‐melt samples were cut, polished and put into a Tanmann tube for the following equilibrium experiments in the MoSi2 (super kanthal) electric resistance furnace. The samples were equilibrated at 1873 K for 1 hour, which was regarded enough, based on a preliminary investigation. After equilibration the samples were quenched in water, cut and polished for the following chemical analysis.
The aluminum in the samples was analyzed using an Inductively Coupled Plasma‐Atomic Emission Spectrometry (ICP‐AES), Seiko Instrument Co. For the oxygen, Inert Gas Fusion‐Infrared Absorption Method, TC‐436, LECO Co., St Joseph, MI, USA, was employed.
Inclusion classification procedure To be able to process vast amounts of inclusion data a new inclusion classification approach is needed. In the classification only inclusions found in liquid steel would be considered. In fact, only the inclusion types found before and after vacuum treatment is initially paid attention to. Extensive work has been done in the past at Uddeholm Tooling to find the inclusion amounts, types and their chemical compositions during steelmaking.[ 3,5,6,9,30‐34] This amount of work has generated a great knowledge about the inclusion situation during the production of their tool steel. Tripathi et al. has categorized the inclusions found at different steps during the production in a table.[6] This table shows that the present inclusions in liquid
5
steel, before and after vacuum degassing are spinel (MgO.Al2O3), calcium aluminate with spinel core, calcium aluminate with MgO particles inside, and the pure calcium aluminate. It should be mentioned that the calcium aluminate phases also consists of lower amounts of MgO and SiO2.
Three classes were assumed to be enough to represent the most typical inclusions found, namely, the spinel (MgO.Al2O3) inclusion, a two phase calcium aluminate inclusion consisting of a spinel core or MgO particle(s) (TP‐(CaO‐Al2O3)), and the single phase calcium aluminate inclusion ((CaO‐Al2O3)). In this study the multi phase calcium aluminate inclusion with spinel core or MgO particle has been put together in the same class because of the difficulties it would involve to distinguish them from one another when comparing average chemical compositions. These inclusions corresponds to inclusion types 3, 6, 8 and 7 in the work of Tripathi et al.[6]. For this reason the multiphase inclusion class is called TP‐(CaO‐Al2O3), where TP stands for “two phase”.
Classification The boundary conditions put into the classification are based on empirical knowledge at the specific steel mill. The data retrieved from the INCA Feature software can be divided into two main categories; chemical‐ and physical attributes. The chemical attributes are given as detected amounts of elements present in the feature. The physical attributes are given as measurements of length, shape, equivalent circle diameter (ECD) etc. The classes defined are presented in more detail in Supplement 2. In short the class names are Spinel‐, TP‐(CaO‐Al2O3)‐ and (CaO‐Al2O3) class respectively. Figure 2 below gives a visual impression of the defined classes.
6
Figure 2 Schematic view of the boundary settings for the three defined inclusion classes.
Actually there is also a fourth class which is a class where all inclusions end up that does not meet the requirements for the previous mentioned classes. This class is regarded as a scrap class.
Application of the classification algorithm on steel samples taken before and after vacuum degassing To evaluate the classification algorithm on the inclusion number, size and type, four heats of the tool steel ORVAR 2M at Uddeholm Tooling in Hagfors, Sweden was chosen. Lollipop samples were taken before and after the vacuum degassing station. The steel samples were ground and polished for the following SEM‐EDS x‐ray investigation using INCA Feature for detection and classification of the present inclusions by the new classification classes. The file was retrieved from the microscope and processed on a PC with the installed software.
7
Physical modeling of slag entrainment
Cold model study on possible slag entrainment at openeye region In the past, researchers have claimed that the open‐eye created by gas stirring in the ladle would be a major source of inclusions and top slag entrainment.[12‐14] It is thus important to establish what is happening in the open‐eye region. The experimental setup employed is schematically shown in Figure 3. The vessel, having dimensions of 150 mm x 350 mm x 250 mm (L, H, W), was filled with Ga‐In‐Sn alloy to a predetermined height of 180 mm.
Figure 3 Experimental set‐up for the study on slag entrainment at the open‐eye region.
To simulate the top slag in the ladle a mixture between manganese (II) chloride‐tetra‐hydrate salt solution and glycerol (87%) were chosen. The viscosities and densities of the two liquids are fund in Supplement 5.
Sampling was made in the metal bulk at 9 different positions in the case of 3 different gas flow rates to determine whether or not top phase entrainment could be observed in the bulk phase. The samples were at rest for a certain time before the free surface was examined for any emerging fraction of the top liquid.
8
During the experiments, the top view of the vessel was recorded with a video camera.
The video recordings of the open‐eye were processed and pictures were extracted from the recording at different time steps and the area of the open‐eye was evaluated. This is further presented in Supplement 5.
Industrial experiments
Chemical development of inclusions in a bearing steel To determine the chemical development of inclusions over the process steps during secondary steelmaking, some experiments were carried out at Ovako Steel in Hofors, Sweden. The steel mill of Ovako Steel operates on recycled steel scrap. The process is schematically presented in Figure 4.
Figure 4 Schematic sketch of the process flow at Ovako Steel.
Scrap metal is melted in the electric arc furnace (EAF) which has a capacity of 100 tons. The molten steel is tapped through the bottom of the EAF into a ladle. During tapping, deoxidation takes place by adding Si and Al. The slag is raked off before the addition of synthetic slag and alloying materials, which is taking place at arrival at the ladle station.
9
The ladle is then transferred to the degassing station (ASEA‐SKF). Argon gas is introduced through the two porous plugs at the bottom of the ladle to ensure the removal of sulphur, nitrogen and hydrogen. At OVAKO Steel, ingot casting is used. Ingots of 4.2 tons are filled using uphill teeming.
Steel and slag samples were taken at different stages of the process. These stages are presented Table 1. Most of the steel samples were taken using the automatic sampling equipment.
Table 1 Stages of sampling along the process.
Process step
After Tapping EAF
Arrival at Ladle
Furnace
5 min after slag/alloy
Before Degassing
After Degassing
BeforeCasting
Samples taken
Steel Steel Steel Steel Steel Steel
Slag Celox® Slag Slag Celox®
Celox® Celox®
Label B C D E F G
All the samples were of the lollipop type. Slag samples were collected using a scoop. The steel temperatures and the dissolved oxygen activities were also measured using the Celox® equipment[35] at different stages of ladle treatment in accordance with Table 1 and are presented in Supplement 3.
The lollipop samples were prepared and examined in a light optical microscope. The inclusions were marked for easy identification in the later scanning electron microscope investigation to determine the different phases and their chemical compositions present in the inclusions.
Refractory samples To determine the attached slag layers effect on the formation of inclusions, refractory samples were taken from old ladles at the two steel plants, Uddeholm Tooling and Ovako Steel. Their steelmaking processes have been described both
10
here and in previous work.[36] Both industries uses carbon reinforced magnesia (MgO‐C) as their lining material. The samples were collected from different positions of the walls on ladles taken out of the production. For each steel melting shop, two samples were collected from the slag line (named as S‐sample) and two samples (named as M‐sample) at different heights of the ladle between the bottom and the slag line. The refractory samples were baked in conductive Bakelite, ground and polished to obtain a smooth surface. Also they where gold sputtered before a scanning electron microscope (JEOL JSM‐840) with energy dispersive spectrometer (EDS X‐ray Link ISIS Series 300, Oxford Instruments) were used for the morphological and compositional analysis. The results were evaluated with the help from the literature and the commercial software THERMOCALC.
SteelSlag Samples taken from industrial ladle To compare the findings in the physical modeling of the possible inclusion formation around the open‐eye, samples of the slag‐metal interface taken in an industrial 65‐ton ladle at Uddeholm Tooling was investigated. The sampler and sampling technique are described in a previously published paper.[37]
The samples were collected from the slag‐steel interface, just outside the open‐eye boundary, in a gas stirred ladle (with two porous plugs). Figure 5 shows an approximate sketch of the position of the open‐eye (Note that the two open‐eyes merge together) and the sampling positions.
Figure 5 Schematic sketch of the sampling positions around the open‐eye.
11
If the slag entrainment would be an important mechanism of inclusion formation a considerable amount of tiny slag droplets would be found in the steel part close to the interface. To examine this aspect, the present study was focused on the examination of the presence of tiny slag droplets in the steel part.
After sampling and quenching, each sample was mechanically opened using lathe and milling machine. The metal, slag and slag‐metal interface could easily be identified as shown in Figure 6. The steel part was cut, polished and examined in a light optical microscope.
Figure 6 Opened sample f slag‐steel interface. o
12
RESULTS AlO relation in liquid iron at 1873 K Aluminum and oxygen contents in liquid steel at 1873 K determined in the present study, in the range of interest, are presented in Figure 7. For better comparison, the present results are shown including the results of Itoh et al.[24]
10-2 10-1 100 10110-4
10-3
10-2
10-1
initial oxygen content
present study Itoh et al. JSPS recommended
mas
s% O
mass% Al
Figure 7 Equilibrium contents of aluminum and oxygen in molten steel at 1873 K found in this study.
When the Al content was 1.0–10.0 mass %, the equilibrium O content was less than 10 ppm, as shown in the figure. Compared with the JSPS recommended values and other data presented in Figure 1, the O content at high Al content in this study was considerably low. On the other hand, for 0.01–1.0 mass % Al, it is seen that the O contents in the present results are higher than the JSPS recommended values.
13
Application of inclusion classification procedure on a toolsteel When the inclusions become more complicated in chemical composition and morphology a need for a good and consistent classification is needed. In Figure 8, the results from the chemical classification are shown. Plotted in the figure are 2642 inclusions detected and classified from samples taken before and after vacuum treatment at Uddeholm Tooling. The class with pure spinel phase (green triangles) are only found before vacuum treatment (BV) while the TP‐(CaO‐Al2O3) and (CaO‐Al2O3) class inclusion are found both before and after vacuum degassing. However, it is seen from the figure that the multiphase calcium aluminate (TP‐(CaO‐Al2O3)) is decreasing in numbers after vacuum.
Figure 8 classified inclusion types plotted.
Figure 8 also presents the size distribution for different size groups. The figure (to the right) shows that the smaller inclusions (blue squares) are present everywhere although slightly higher numbers are concentrated in the spinel and multiphase regions. The bigger inclusions (red and black) are more found in the liquid calcium aluminate region closer to the lower corner of the liquidus region. Nevertheless, there are also some bigger inclusions located in the two phase region.
The pure calcium aluminate inclusions are the only inclusion class that increase in all cases except in one, the DV56781 as can be seen in Figure 9. This heat is basically showing no change before‐ and after degassing, while the other heats are showing
14
major changes. The causes for this will be further elaborated in the discussion part. If looking before degassing in Figure 9, it can be seen that all heats except heat DV56030 are having quite similar inclusion number per mm‐2. On the other hand that heat has a very low number of spinel inclusions in comparison to heats DV56663 and DV56805. All three of those heats have nevertheless approximately the same level of two phase calcium aluminate inclusion after degassing. In the case of DV56030 the increase in pure calcium aluminates (CaO‐Al2O3) is almost 10 times higher than the other two (DV56663 and DV56805).
Figure 9 Number of inclusions before and after vacuum degassing.
15
Chemical development of inclusions in a bearing steel Four heats of the bearing steel grade 803J were studied to establish the chemical development of inclusions over the process steps. The typical steel composition is presented in Table 2.
Table 2 Typical chemical composition of 803J bearing steel.
mass % C
mass % Si
mass % Mn
mass % Cr
mass % Ni
mass % Cu
mass % Al
mass % S
1.0 0.25 0.3 1.4 0.2 0.25 0.03 0.015
Inclusions found in the steel samples The inclusions found in the steel samples were categorized into 5 different types. The types are listed in the Table 3.
Table 3 Types of inclusions found at different stages of the process.
Composition Type B C D E F G
Al2O3 1 x x x x x
(CaO‐Al2O3) 2 x x x x x
Spinel + (CaO‐Al2O3) 3 x x
(CaO‐Al2O3) + CaS 4 x x x x
Spinel + (CaO‐Al2O3)+CaS 5 x x
Al2O3 (Type1) Inclusions of Al2O3 are detected in stage B and throughout the process to stage E. The inclusions are mostly configured like clusters in B stage and are decreasing in amount through the stages. After degassing (stage F), no clusters are found. Only some very small alumina inclusions are detected.
16
(CaOAl2O3) (Type2) This type of inclusions can be further classified into three different sub types,
namely the CaO‐Al2O3 liquid solution, the CaO⋅6Al2O3 compound and the
combination of the two. In the third case, the inclusion has the CaO⋅6Al2O3 compound in the centre surrounded by the liquid solution. The liquid nature of this phase at high temperature is suggested by its composition range and the round shape of the inclusions. Inclusions of Type‐2 are found in stages B to F.
Spinel + (CaOAl2O3) (Type3) Type‐3 inclusions are found in stages B and F. An inclusion of this type consists of two phases. The element mapping of a Type‐3 inclusion is presented in Figure 10.
Figure 10 Inclusion consisting of calcium aluminate with spinel core.
The outer phase is the same as liquid CaO‐Al2O3 phase in the type‐2 inclusions. The presence of a spinel core distinguishes it from the Type‐2 inclusions.
(CaOAl2O3) + CaS (Type4) This type of inclusion is basically the same as type‐2 except for the formation of a CaS shell around the inclusion. Inclusions of Type‐4 are found in the stages D to G.
Spinel + (CaOAl2O3) + CaS (Type5) Inclusions of Type‐5 differ only from Type‐3 by the formation of a CaS shell. In stages F to G, inclusions of Type‐5 are detected.
17
In supplement 3 the inclusion types are further explained.
Inclusion formation from attached slag layers It is commonly observed in all samples irrespective of the position and industry that the linings taken from the used ladles have four layers in general, viz. The original carbon bearing magnesium oxide, a decarburized layer, a slag infiltrated layer and the outer slag layer. The two outer layers are referred to as ladle glaze by previous studies[4]. For each steel plant, there is no substantial difference between the S‐samples. It is also true in the case of M‐samples.
The glaze layer found in the ladle of Ovako Steel has a thickness varying between 2 and 8 mm. The glaze is white‐grey in colour. It is noticed that the erosion of the refractory at the slag line is more profound than at the position a certain distance below this line.
Figure 11 Ovako S‐Sample taken from the slag line.
Figure 11 shows the cross section of a sample taken in the slag line of Ovako Steel. It can clearly be seen that the outer layer (to the right in the picture) consists of mostly ladle slag, while the slag infiltrated layer consists of both MgO matrix and slag. In fact, the slag region is composed of more than one phase. Figure 12 presents the SEM microphotograph with higher magnification. Two oxide phase, namely MgO.Al2O3 (spinel) and the 3CaO.Al2O3 calcium aluminate compound are found. It should be pointed out that in some cases it is difficult to clearly distinguish whether the calcium aluminate phase is 3CaO.Al2O3 or 12CaO.7Al2O3 by EDS analysis since
18
the two compounds are quite close to each other in chemical composition. In addition to these phases, there are regions consisting of a two phase mixture.
Figure 12 Ovako S‐sample with higher magnification.
One of the phases is likely to be super cooled slag containing mostly CaO and Al2O3 with small amounts of SiO2 and MgO. The second phase in these two‐phase regions is a solid calcium aluminate. Due to that the size of the grains are small, it is very difficult to identify whether it is 3CaO.Al2O3 or 12CaO.7Al2O3. However, the EDS results seem to favour 12CaO.7Al2O3.
The refractory samples from the ladle of Uddeholm Tooling are a bit different in appearance compared to the samples from Ovako Steel. The slag layer is greenish in colour and glassier in appearance. Otherwise the attributes are quite the same between the two plants. As an example, Figure 13 shows the micrograph of the M‐sample. In addition to the precipitated MgO, two oxide phases and super cooled slag are identified. The two oxides are 2CaO.SiO2 and 3CaO.Al2O3 as marked in the figure. There are also regions consisting of two phases, namely a precipitated calcium aluminate compound (most likely 3CaO.Al2O3) along with super cooled slag.
19
Figure 13 Uddeholm M‐Sample with higher magnification.
The phases detected in different lining samples are listed in Table 4. It is seen in the table that the phases present in the glaze layers from the two steel plants differ somewhat. The calcium silicate phase is only detected in the samples of Uddeholm Tooling, while the MgO.Al2O3 phase is only found in the S‐sample of Ovako Steel. In the table the 3CaO.Al2O3 and 12CaO.7Al2O3 has been put in the same category for reasons mentioned earlier.
Table 4 Phases (besides MgO) in the samples detected by EDS X‐ray.
Ovako Steel Uddeholm Tooling
Phases S‐Sample M‐Sample S‐Sample M‐Sample
12CaO.7Al2O3 and/or 3CaO.Al2O3
X X X X
2CaO.SiO2 X X
CaO X X
MgO.Al2O3 X
Super cooled slag X X X X
20
Top phase entrainment in GaInSn model Samples, taken from 9 positions in the bulk phase to evaluate the existence of top phase entrapment in the open‐eye area, showed no visible entrapment. This might suggests that the entrainment of top liquid into the liquid metal in the area around the open‐eye does not play a significant role irrespective of the gas flow rate. More information regarding this can be obtained in Supplement 5.
Top slag entrainment in metal phase in an industrial ladle To verify the results from the laboratory study, three samples of slag‐metal interface taken around the open‐eye (see Figure 5) in the gas stirred ladle at Uddeholm Tooling were investigated to evaluate the presence of slag entrainment in the steel. Figure 14 present series of microphotographs taken from the three samples, respectively. While the first picture in each series shows the surface of the steel in contact with slag, the lowest picture shows the steel part about 20 mm from the slag‐metal interface. Note that almost all the dot spots are pores and no visible amount of slag is observed in any of the three samples.
Figure 14 Series of microphotographs of sample F, sample G and sample H.
21
22
DISCUSSION Evaluation of the AlO parameters in liquid iron at 1873 K A number of compositions were chosen in the Al content range of 0.01–10.0 mass %. After 1 hour equilibration, the collected samples were machined and prepared for composition analyses.
The equilibrium constant of the Al deoxidation of molten steel and interaction parameters at 1873 K were estimated on the basis of the present experimental results.
The reaction of Al deoxidation can ion (1): be written as Equat
2 3 (1)
Since the activity of Al2O3(s) can be regarded as unity because no other oxides are pres the , .ent in system the equilibrium constant is expressed as Eq (2).
· 2 % 3 % 2 3 (2)
In order to study the equilibrium between Al and O up to high Al content region, not only the first order parameters but also the second order was considered. With som p fe sim li ications (see Supplement 1) equation (3) is obtained as follows:
% % 0.086 % 0.528 % 0.0044 % %0.002 % 3 % 3.38 % 0.064 % % 3 %6.74 % % 3.57 % % 2 % (3)
Here, a multiple linear regression was attempted to simultaneously determine the first and second order parameters. With the experimentally known values put into
equation (3), each term of , , and were estimated by the multiple linear regression. With high relevance, the parameters could be determined as follows.
23
11.52 0.08
.0.231 0 018
0.026 0.006 228 61
As expected from the positive deviation of the present experimental results from others result in low Al content region, the equilibrium constant turned out to be
slightly smaller. On the other hand, the first order interaction parameter was as much as about one fifth of other values, i.e. ‐1.17 (Rohde et al.[18], JSPS[23], Itoh et
al.[24]) and ‐1.0 (Schenck et al.[17], Janke et al.[19]). The smaller value of could be justified by the present results from higher Al contents where the oxygen solubility is lower than that reported from other researchers. Also the Al‐O equilibrium in the present study does not show a drastic increase of O content at high Al content
owing to lower value of . In Supplement 1 also some other interaction parameters can be found.
It might be believed that the “gettering” effect, which is the tendency of some metals to grab gas when analyzing the O content, is responsible for the underestimation of O content. There have been a few reports that the O content determined by inert gas fusion infrared spectroscopy for a sample containing relatively high Al content might be inaccurate because of the “gettering” effect. Inoue et al. reported that O recovery depends on various extracting conditions such as the type of flux bath, temperature, and heating pattern in inert gas fusion infrared absorptiometry.[26] According to their results, it is necessary to use a suitable capsule and bath and have a sufficiently high extraction temperature in the determination of O content in an alloy of Al and 10–15 mass % Fe. Although the Al content considered in this work is lower than this, the O analysis might have been misled by the “gettering” effect.
Nevertheless, the determined parameters, which were determined by multiple linear regression, could successfully describe the Al–O equilibrium in the present experimental study.
24
Classification procedure with the help from INCA Feature Steel samples have been processed using an automatic feature detecting software, INCA Feature, connected to a SEM‐EDS x‐ray equipment. Thereafter the detected inclusions have been classified with a new classification approach. It should be mentioned that the classification classes, as they are now, have not been verified by manually counting the inclusion number and size in the light optical microscope and substantial errors might exist. However, the error would be consistent and thus a relative value could be obtained. Nevertheless, as mentioned in Supplement 2, the results are in good agreement with other researchers.
One assumption made in the classifying method was to derive the classes from the ternary phase diagram of Al2O3‐CaO‐MgO. It implies that the SiO2 would not influence the behaviour of the inclusions. This is not completely correct since the amount of SiO2 is quite large (0‐15 mass %). The SiO2 would somewhat decrease the liquidus temperature and thus increase the liquidus area at 1873 K. In Figure 15 the inclusions are categorized by the SiO2 content. As can be seen in the figure, it seems that the SiO2 content in the inclusions is increasing towards the liquid calcium aluminate region. This can be explained by the fact that spinel is solid and does not dissolve any SiO2 hardly. On the other hand, calcium aluminates at Uddeholm Tooling typically consist of about 10‐15 mass % SiO2.
Figure 15 Inclusions with various SiO2 content plotted in Al2O3‐CaO‐MgO diagram.
25
Figure 9 in the result part shows evidently an increasing trend of calcium aluminates. This is in good agreement with the observation from the size distribution in Figure 8. Calcium aluminates are known to be big. Hence, if the amount of calcium aluminates goes up, an increase in the bigger size classes should also be seen. The vanishing of the spinel inclusions is also in good agreement with previous work done by Tripathi et al. and Kristina et al.[5]
All heats except one show a change before and after degassing. Why heat DV56781 is not basically changing at all can be due to two reasons. (1), the heat is already close to equilibrium between top slag, steel and inclusions before degassing thus leaving a small change. (2), the samples have somehow been mixed up. During sampling several samples were taken from the same heat and stage to ensure good sampling quality, thus it is easy for one sample to by mistake end up in the wrong sample bag. To the author’s experience, the second reason is more likely. Nevertheless, the other heats show similar trends as each other in general and the heats DV56663 and DV56805 show it in particular.
A new classification algorithm has been tested. It successfully classifies 2642 inclusions of spinel, TP‐(CaO‐Al2O3) and (CaO‐Al2O3). The plotted values were positioning themselves nicely in the defined classes. It must be mentioned that it is extremely difficult to classify inclusions located at the boundary between two classes e.g. TP‐(CaO‐Al2O3) and (CaO‐Al2O3). The liquidus line is a natural boundary condition for the liquid calcium aluminate. However, it is hard to know how silica is affecting the liquidus line. It is also difficult to define classes when the liquidus lines are not following the iso mass % lines in the classification. In most cases the phases in a system is not divided according to iso mass % lines. Four classes should have been classified if looking at previous work. However, in this classification the two phase inclusions with an MgO or spinel core surrounded by calcium aluminate respectively has been incorporated into the same class (TP‐(CaO‐Al2O3)) because the distinction would be too difficult. Nevertheless, if the inclusion would be round, which can be measured by INCA Feature, the assumption could be made that the inclusion would be liquid; and if that round inclusion would be situated in a solid region if plotted (average composition) in a phase diagram the assumption could be done that the inclusion consists of multi phases. It would be valuable to continue the investigation along this line, so both multiphase inclusions could be identified by
26
the INCA feature analysis. Just to show the validity of the multiphase class the average composition of three clearly identified multiphase inclusions found by J. Björklund et al.[32] are plotted in Figure 8 (pink dots). It is seen that those inclusions are positioning themselves nicely among the other inclusion in the two phased area, thus suggesting that the classification is working good at least inside the classes.
The present technique would be a powerful tool to utilize the INCA analysis for the optimization of the ladle treatment, as the results of the huge number of inclusions could be examined consistently.
Origins and chemical development of inclusions It is important to find out the origins of the different inclusion types and their possible development during the process. In Supplement 3 the types found are furthered discussed, here is a summary.
Al2O3 (Type1) Alumina inclusions are generat ion reaction, ed by the deoxidat
2 3 (4)
The fact that big clusters and considerable amount of Al2O3 inclusions are found in stage B and the decrease of the cluster size as well as the amount of the Al2O3 further supports this argument. In the later stages of the process before degassing, small alumina inclusions are also found. They are likely to be the product of reoxidation.
(CaOAl2O3) (Type2) Calcium aluminate inclusions are generally found in ladle treatment. In fact, the calcium aluminate phase(s) are found in different types of inclusions throughout the ladle refining. The presence of the inclusions of Type‐2 just after tapping (step B) suggests that these inclusions could come from three possible sources, namely (1) from EAF, (2) ladle glaze, and (3) deoxidation. The low dissolved Ca content in the steel and the short time after deoxidation would probably rule out (3) thus suggesting that at least ladle glaze could be one source for generating calcium aluminates. In the case of EAF, more extensive investigations would be needed.
27
Spinel + (CaOAl2O3) (Type3) As in the case of CaO‐Al2O3 inclusions, inclusions of Type‐3 are unlikely formed by deoxidation because the shortage of supply of dissolved Ca. On the other hand, the coexistence of spinel phase and CaO‐Al2O3 phase in the ladle glaze (see Table 4) would make the formation of inclusions of Type‐3 from ladle glaze possible. Again, whether some of the inclusions of Type‐3 originate from EAF require further investigation.
(CaOAl2O3) + CaS (Type4) The inclusions of Type‐4 are found at later stages of the ladle treatment. The outer shell of CaS indicates that the inclusions are developed from CaO‐Al2O3 inclusions. The sulphur content is about 100‐150 ppm after degassing. By calculations (see Supplement 3 for details) it was shown that CaS would form at 1873 Kelvin at the aCaO=0.1 by the following n, reactio
(5)
Calculations based on the compositions of CaO‐Al2O3 inclusions using ThermoSlag[38] reveal that most of the inclusions have a CaO activities higher than 0.1. Reaction (5) indicates that lower oxygen activity favors the formation of CaS. The present finding is in good accordance with this aspect.
Spinel + (CaOAl2O3) + CaS (Type5) Since the CaS shell covers the CaO‐Al2O3 phase in the inclusion of this type, the discussion would be very similar as the previous one. The formation of the CaS is due to reaction (5).
Inclusion formation from attached slag layers As seen in the result part, there is no substantial difference between the phases found in the S‐sample and M‐sample in the case of both industries. The only difference is that spinel phase is found in the S‐sample but not in the M‐sample in the ladle of Ovako Steel. During teeming, ladle slag follows the descending steel. It adheres on the decarburized refractory surface (MgO) and reacts with it forming a slag layer. Hence, thermodynamically, the M‐samples would experience similar slag‐refractory reaction as the S‐sample. As revealed by Table 4 a number of oxide
28
phases precipitate during cooling. In order to gain a better understanding of the solidification process and the difference of the precipitations in the two steel plants, thermodynamic calculations were carried out using the commercial software THERMOCALC.
Since the assumption of global equilibrium would give substantial errors in the calculation, the Scheil‐Gulliver model is used to simulate the slag solidification. This model assumes infinitely rapid diffusion in the liquid phase, no diffusion in the solid phase and local equilibrium at the solid/liquid interface. After calculating the equilibrium at one temperature step, the software will disregard the precipitated phases and pass on the new liquid composition to the next temperature step. In order to calculate the amount of a specific solid phase at one temperature, the amount of the solid phase precipitated in the previous steps is added.
The results of the calculations for the slags of Ovako Steel and Uddeholm Tooling are presented in Figure 16 and Figure 17 respectively. In the case of the slag sample of Ovako Steel, 4 oxide phases (beside MgO) would precipitate according to the calculation. These phases are 3CaO.SiO2, 3CaO.Al2O3, MgO.Al2O3 and CaO.Al2O3. For the slag of Uddeholm Tooling, 3 oxide phases (beside MgO) would precipitate, namely, 3CaO.SiO2, CaO.Al2O3 and 3CaO.MgO.2SiO2.
Figure 16 THERMOCALC calculation of the typical slag composition of Ovako.
29
Figure 17 THERMOCALC calculation of the typical slag composition of Uddeholm.
The phases 3CaO.Al2O3 and MgO.Al2O3 predicted by the calculation for Ovako Steel (Figure 16) are found in the slag layer (Table 4) while calcium silicate phases and CaO.Al2O3 are not detected. It is possible the particles of these phases are too small to distinguish in the sample. However, it is more likely that in the case of calcium silicate, the phase has fallen off as fine powder during cooling of the ladle. The extremely low silica content in the slag layer in comparison with the ladle slag seems to violate the law of mass balance. As 3CaO.SiO2 is transformed into 2CaO.SiO2 + CaO at lower temperatures, it might lead to the formation of fine dust. The existence of tiny CaO particles in the slag layer of Ovako Steel would in some way support this reasoning. When it comes to the CaO.Al2O3 compound predicted by the calculation, this is more likely found as 12CaO.7Al2O3. The composition of this compound is situated between 3CaO.Al2O3 and CaO.Al2O3. However, this compound is not included in the thermodynamic database. The absence of this phase in the database would well explain the discrepancy between the calculation and observation. It is noted that spinel phase predicated by the calculation is only detected in the S‐samples. That the formation of MgO.Al2O3 requires longer reaction time might be a possible explanation. Moreover, in the S‐sample (see Figure 12), the 3CaO.Al2O3 is found as primary precipitation. In between, two‐phase regions are seen. EDS results indicate the two phase are possibly 12CaO.7Al2O3 (the dark grey phase) and super cooled slag.
30
The 3CaO.SiO2 phase predicted by THERMOCALC calculation for slag of Uddeholm Tooling is evidently found in the lining samples of the corresponding ladle as 2CaO.SiO2. The absence of spinel phase in the glaze layer (Table 4) of Uddeholm Tooling is also in good agreement with the model prediction. Instead of CaO.Al2O3 predicted by the model, the calcium aluminate phase found is more likely to be 3CaO.Al2O3; it is also a possibility that the phase might be 12CaO.7Al2O3 due to the earlier mentioned reasons in the result part. As also mentioned earlier, the absence of the 12CaO.7Al2O3 phase in the thermodynamic base could be a very good explanation for this discrepancy. The 3CaO.MgO.2SiO2 phase is not detected in the glaze layer. The amount of this phase predicted by THERMOCALC calculation is very small. It could be very difficult to detect by EDS. On the other hand, the amount of MgO in the remaining liquid assumed by the Scheil‐Gulliver model and the thermodynamic data could also be responsible for this discrepancy.
When deriving the Scheil equation one of the assumptions made is that the remaining melt always has a homogeneous composition. However, to justify this assumption the diffusion has to be either infinitely rapid, which is not possible, or there must be convection. This is supposed to only occur in volumes with a thickness larger than 1 mm[39]. The slag that penetrates the cracks and pores of the refractory would have very limited convection. Hence, the use of Scheil‐Gulliver model would be associated with uncertainties.
A recent study[10] shows that the infiltrated slag in the form of calcium aluminate can even stay in the MgO matrix for 2‐3 heats. It implies that the tiny calcium aluminate droplets in the MgO matrix would not be washed away completely by the steel during tapping. While it is likely that some of the CaO‐Al2O3 inclusions are brought over from the EAF to the ladle and some of them might be generated by deoxidation or reoxidation, the present results show that slag‐refractory reaction is at least a considerable source of non metallic inclusions.
Inclusion formation from slag entrainment at the openeye The Figure 14 demonstrates that no slag droplets are present in the steel bulk in the vicinity of the slag‐metal interface. Only a few non‐metallic inclusions are detected in these samples. Thus, the number of inclusions found in the region near the slag‐
31
32
metal interface is not bigger than the number of inclusions found in the steel samples obtained far away from the interface. The entrained slag droplets might have floated up to the interface before the metal solidified. However, the quenching of the samples was very fast (within a few seconds).[37,40] Droplets of small size at micro level would not have enough time to float up in such short time. The steel samples shown in Figure 14 clearly show the absence of the small slag droplets, indicating thereby the absence of entrainment of tiny slag droplets around the open‐eye in a gas stirred ladle. Bigger slag droplets (if there were in the liquid steel) might have time to float up to the surface. On the other hand, they would not pose any problem of generating inclusions, as they float back to the slag so fast.
The results of cold model experiments strongly support the findings in the industrial trials. No detectable amount of top liquid was found in any of the samples taken in the metal bulk. Generation of inclusions by entrainment during stirring has been the subject of many researchers.[12,13,14] However, the results of both the present laboratory experiments using Ga‐In‐Sn alloy and the industrial samples taken from slag‐metal interface do not suggest the entrainment of slag as the main source of the inclusions. Nevertheless, further well‐planned experiments are absolutely necessary to draw definite conclusions.
CONCLUSIONS The present study on the Al‐O relation in liquid iron at 1873 K has shown that the dissolved oxygen content would be lower at higher Al contents than the previously reported values. On the other hand, the O content has been found slightly higher when it came to lower Al contents. The classification of inclusions was developed using the commercial software INCA Feature. Three classes were made, spinel, TP‐(CaO‐Al2O3) and (CaO‐Al2O3). The results from Uddeholm tooling samples showed that the spinel phase disappeared after degassing along with a reduction in numbers for the two phase inclusion (TP‐(CaO‐Al2O3)). Pure calcium aluminates however showed an increasing trend for all heats. Steel and slag samples were also taken at OVAKO Steel to study the origin and development of inclusions. The Inclusions were classified into 5 different types according their chemical compositions and morphologies. Alumina inclusions were generated by deoxidation reaction. While ladle glaze was at least a potential source generating CaO‐Al2O3 inclusions (Type‐2). In the inclusions of Spinel + (CaO‐Al2O3) (Type‐3), spinel phase was always found in the centre surrounded by CaO‐Al2O3 phase. To meet thermodynamic constraints, a layer of CaO‐Al2O3 phase was formed. Type‐4 inclusions were identified from Type‐2 by the formation of a CaS shell. Similarly, Type‐5 inclusions were different from Type‐3 by the CaS shell. MgO refractory samples with attached ladle slag were also taken from old ladles at the two plants studied. The precipitated phases in the slag layers, and the cracks and pores of the refractory were examined. The phases found were 3CaO.Al2O3 (and/or 12CaO.7Al2O3), MgO.Al2O3 and CaO in Ovako Steel case and 3CaO.Al2O3 (and/or 12CaO.7Al2O3) and 2CaO.SiO2 in the case of Uddeholm Tooling. THERMOCALC calculations, on one hand showed similar trends of the phase precipitations, and on the other hand showed discrepancy. The absence of 12CaO.7Al2O3 in the database could be the explanation of this discrepancy. Studies were made on the slag entrainment in a gas stirred ladle. Ga‐In‐ Sn alloy was used to simulate the liquid steel, while MgCl2‐Glycerol(87%) solution was used as ladle slag. No amount of top liquid was observed in any of the samples. This finding was in good agreement with the results from industrial trials where no entrapped slag droplet was found either. Thus it might be concluded that the entrainment of slag into the steel bulk around the open‐eye would not be the major source of inclusions.
33
34
FUTURE WORK In the present work, formation and chemical development of non‐metallic inclusions has been investigated by taking samples at various stages during the production and looking at the reaction between slag and refractory material. Also the possible slag entrainment from the open‐eye has been investigated. Many topics have been touched and the author suggests that the following areas need to be investigated further:
• Laboratory study about the slag refractory reactions to in more detail establish the precipitated phases during solidification and also the effect of changing the slag composition.
• Verifying the classification algorithm by going back and look at the classified inclusion to validate the procedure. Also develop the classification approach to include other process steps.
• The chemical development of inclusions made in this study was on bearing steel with high sulphur content. It would be interesting to extend the study to also include the bearing steel with low sulphur content.
• Some more, well lanned, industrial experiments are suggested to really establish the valid of the slag entrainment results in this study. Samples could be taken at different distances from the open‐eye rim.
pity
35
36
REFERENCES 1. S. Riaz. K. C. Mills and K. Bain: Ironmaking and Steelmaking: 2(2002), 107. 2. G. J. Hassall, K. G. Bain, N. Jones, and M. O. Warman: Ironmaking and
Steelmaking, 5(2002), 383. 3. K. Beskow, J. Jia, C. H. P. Lupis and Du Sichen: Ironmaking and Steelmaking,
29(2002), 427. 4. K. Beskow and Du Sichen: Ironmaking and Steelmaking, 5(2004), 393. 5. K. Beskow, N. Sano and Du Sichen: AIST Iron & Steel Technology, 2(2006), 103. 6. N. Tripathi, M. Nzotta, A. Sandberg, and Du Sichen: Ironmaking and
Steelmaking, 31(2004), 235. 7. K. Beskow, N. N. Tripathi, M. Nzotta, A. Sandberg and Du Sichen: Ironmaking
and Steelmaking, 31(2004), 514. 8. J.‐H. Son, I.‐H. Jung, S.‐M. Jung, H. Gaye and H.‐G. Lee: ISIJ International,
11(2009), 1542. 9. N. Tripathi, M. Nzotta, A. Sandberg and Du Sichen: Steel Grips, 2(2004), 40. 10. M. Song, M. Nzotta and Du Sichen: Unpublished work, “Study of the
formation of non‐metallic inclusions by ladle glaze and the effect of slag on inclusion composition using tracer experiments”, 2009.
11. K. Beskow, and Du Sichen: Scandinavian Journal of Metallurgy, 6(2003), 320. 12. S. Cho, S. Hong, J. Han and B. You: Effect of Slag Layer on Flow Patterns in a
Gas Stirred Ladle, Proceedings of the 7th international Symposium on eco‐materials processing and design (ISEPD‐7), Chengdu, China, 8‐11 Jan. 2006.
13. M. F. Caregnato, F. P. Coutinho, G. P. Dias, D. Andreis, J. M. Pereiral, M. G. Francois and A. C. F Vilela: Physical modelling on behaviour of slag phase in the bath surface in gas stirred ladle, 60DG Congresso Annual da Associacao Brasiliera de Metalurgia e Materials, Belo Horizonte‐MG, Brasil, 25‐28 July 2005.
14. K. Krishnakumar and G. A. Irons: A Model Study of Slag Eye Formation during Gas Purging at the Surface of a Metal Bath, AISTech 2005 Volumes I & II and ICS 2005 Conference Proceedings, Volume I, Vol. I & II, 833.
15. D. C. Hilty and W. Crafts: J. Met., 2(1950), 414. 16. R. J. Fruehan: Metall. Trans., 1(1970), 3403. 17. H. Schenck, E. Steinmetz and K. K. Mehta: Arch. Eisenhüttenwes., 41(1970),
131. 18. L. E. Rohde, A. Choudhury and M. Wahlster: Arch. Eisenhüttenwes., 42(1971),
165.
37
38
19. D. Janke and W. A. Fischer: Arch. Eisenhüttenwes., 41(1976), 195. 20. A. McLean H. B. Bell: J. Iron Steel Inst., 203(1965), 123. 21. J. C. d’Entremont, D. L. Guernsey and J. Chipman: Trans. Metall. Soc. AIME,
227(1963), 14. 22. J. H. Swisher: Trans. Metall. Soc. AIME, 239(1967), 123. 23. Recommended Values of Equilibrium Constants for the Reactions in
Steelmaking, Japan Society for the Promotion of Science, 19th Committee, 1984.
24. H. Itoh, M. Hino and S. Ban‐ya: Tetsu‐to‐Hagane, 83(1997), 773. 25. N. A. Gokcen and J. Chipman: J. Met., 5(1953), 173. 26. R. Inoue and H. Suito: Mater. Trans. JIM, 32(1991), 1164. 27. J. F. Elliott, M. Gleiser and V. Ramakrishna: Thermochemistry for Steelmaking,
Addison‐Wesley, 1960. 28. E. T. Turkdogan: Physical chemistry of high temperature technology,
Academic Press, New York, 1980. 29. C. H. P. Lupis and J. F. Elliott: Acta Metall., 14(1966), 529. 30. Y. Kang, M.Nzotta and Du Sichen: Steel Grips, 5(2007), 18. 31. N. Tripathi and Du Sichen: Ironmaking and Steelmaking, 33(2006), 213. 32. J. Björklund, M. Andersson, M. Nzotta and P. Jönsson: Steel Research Int.,
79(2008) No. 4, 261. 33. K. Steneholm, M. Andersson, M. Nzotta and P. Jönsson: Steel Research Int.,
78(2007), 522. 34. K. Steneholm, M. Andersson and P. Jönsson: Steel Research Int., 77(2006),
392. 35. Celox® information folders, Heraeus Electro‐Nite, Södra Kungsvägen 66, 181
32 Lidingö, Sweden, www.electro‐nite.com, 2006. 36. J. Björklund, M. Andersson and P. Jönsson: Ironmaking and Steelmaking,
34(2007), 312. 37. K. Beskow. D. Pranesh. J. Björkvall. M. Nzotta and Du Sichen: Ironmaking
Steelmaking, 33(2006), 74. 38. Du Sichen, J. Zhang, and S. Seetharaman: ASIA STEEL INTERNATIONAL
CONFERENCE, Jamshedpur, India, April 9‐12, 2003. 39. H. Fredriksson and U. Akerlind: Materials Processing During Casting, John
Wiley and Sons Ltd., Chichester, 2006, 188. 40. K. Beskow, D Pranesh, J. Björkvall and Du Sichen: Ironmaking Steelmaking,
33(2006), p. 454