Inclusion Formation and Interfacial Reactions between FeTi ...

629 © 2013 ISIJ

ISIJ International, Vol. 53 (2013), No. 4, pp. 629–638

Inclusion Formation and Interfacial Reactions between FeTi Alloys and Liquid Steel at an Early Stage

Manish Marotrao PANDE, Muxing GUO* and Bart BLANPAIN

Dept. of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44 bus 2450, Leuven, BE-3001 Belgium.

(Received on December 13, 2012; accepted on January 8, 2013)

Titanium is usually added to the liquid steel in the form of ferroalloys with varying Ti concentrations.These titanium sources also contain Al, Ca and O as the main impurities. In the present work, three dif-ferent titanium sources, namely, pure Ti and two commercially produced Ti alloys i.e. FeTi70 and FeTi35are studied. The Ti or FeTi was brought in contact with the liquid iron using the suction method, for a pre-determined time and quenched. The reaction zone between the liquid Fe and the titanium source wassubjected to microstructural investigation. The high Ti concentration region obtained in a pure Ti–Fe reac-tion couple, the inclusion formation in liquid iron after coming in contact with FeTi70 and the evolution ofexisting inclusions in FeTi35 after coming in contact with liquid iron have been studied. The present studyhas helped in understanding the influence of impurities from the Ti source on the dissolution behavior andthe inclusion formation. On the basis of this study, it can be concluded that the Ti rich regions formedafter the introduction of pure Ti could modify the existing alumina inclusions in liquid steel, the impuritiesin FeTi70 contribute to the inclusion formation depending upon the availability of O while FeTi35 intro-duces inclusions to the liquid steel.

KEY WORDS: pure Ti; FeTi70; FeTi35; Al–Ti–O inclusions.

1. IntroductionTitanium is added to the ULC (ultra low carbon) steel to

bind the interstitial elements like carbon and nitrogen. Thetitanium requirement is met with the addition of the com-mercially available ferrotitanium grades like FeTi70 andFeTi35 in an Al deoxidized steel. These FeTi grades differnot only in their Ti contents but also in the nature of impu-rities present owing to their different processing routes.1)

However, it was shown with previous studies2–5) that theintroduction of titanium to the Al killed liquid steel resultsin the formation of Al–Ti–O inclusions. The formation ofAl–Ti–O inclusions takes place immediately after the titani-um source addition which with time reverts back to purealumina inclusions accompanied by a change in morpholo-gy. This behavior, in turn, affects the castability of steel.6) Itcan be ascertained that the formation of Al–Ti–O inclusionsdepends upon the local supersaturation of titanium andtherefore, the dissolution behavior of the titanium sourceaffects the inclusion formation.

The schematic representation of the dissolution/reactionafter the addition of ferroalloy/deoxidizer to the liquid steelis shown in Fig. 1. The dissolution process mainly consistsof (1) melting or dissolution depending on the melting tem-perature with the intermediate formation of a steel shell(stage I), (2) nucleation of inclusions in the vicinity of adeoxidizer depending upon the local supersaturation (stageII) (3) the growth and agglomeration of the inclusion parti-cles in liquid steel (stage III) (4) finally the removal of theseinclusions by various mechanisms (stage IV). As the forma-tion of inclusions like oxides is inevitable (stage II), most

of the research is directed towards the removal of inclusionsfrom the liquid steel (stage III and IV) to produce superiorquality steel. The formation of Al–Ti–O inclusions that canbe attributed to stage II, is due to the local supersaturationof Ti in liquid steel which reduces existing alumina inclu-sions. Pandelaers et al.7) studied the dissolution behavior ofpure Ti and FeTi70 alloy. The FeTi70 alloy used in hisexperiments was manufactured on a laboratory scale anddevoid of impurities like Al and Ca, which are typicallyfound in a commercial grade alloy. In his study, stage I wasmainly concerned (Fig. 1). Wang et al.3) compared the inclu-sion behavior after the addition of different titanium sourc-es, viz., pure Ti, FeTi70 and FeTi35 to the Al-killed liquidsteel on a laboratory scale, simulating the actual steelmakingprocess. Wang’s work can be attributed to the inclusionformation between stage II and stage III. On an industrialscale also, clusters of Al–Ti–O inclusions after the additionof FeTi to Al-killed steel at the end of RH treatment havebeen reported.8) The addition sequence of Al and Ti hasbeen studied in detail9) to analyze its influence on inclusionbehavior. It was observed that irrespective of the additiontime of Al or Ti, the final chemistry of the inclusions wasalumina. In our previous study,1) the investigation suggeststhat Al in FeTi70 is mostly present in its soluble formwhile that in FeTi35 is only partly soluble. The Al inFeTi35 was seen in the form of alumina and in the elementalform surrounding the unreduced titanium oxide. In the pre-vious studies,2–5,7,9) the formation of Al–Ti–O inclusionswas attributed to the local supersaturation of Ti, irrespectiveof its source. Very little attention, so far, has been paid tothe behavior of the impurities in FeTi upon its introductionto the liquid steel. Ferrosilicon containing impurities like Alis known to affect the inclusion formation in stainless steelssignificantly e.g. an increasing amount of Al in ferrosilicon

* Corresponding author: E-mail: [email protected]: http://dx.doi.org/10.2355/isijinternational.53.629

© 2013 ISIJ 630

ISIJ International, Vol. 53 (2013), No. 4

enhances the spinel formation while it reduces the MnOcontent of the inclusions.10,11)

In the present paper, the results of the reaction of liquidiron with commercially obtained FeTi70 and FeTi35 alongwith pure Ti as a reference source of titanium are shown toelucidate the phenomena occurring between stage I and IIi.e. the formation of the reaction zone between the Ti sourc-es and liquid iron including the nucleation of inclusions. Tothis purpose the liquid metal suction method was used. Theaim was to understand the effects of impurities on dissolu-tion as well as the mechanisms of the inclusion formationbetween the Ti/FeTi and liquid iron.

2. Experimental2.1. Procedure

The liquid metal suction method12,13) was used in order toobtain a reaction zone between the iron and the three differ-ent titanium sources (Tables 1 and 2) representative for theearly stage dissolution.

About 85 grams of electrolytic Fe (99.97% Fe) was filledin a magnesia crucible (30 mm inside diameter (ID), 35 mmoutside diameter (OD), 50 mm height (H)) which wasplaced inside a high temperature vertical tube furnace(GERO HTRV 100-250/18, MoSi2 heating elements) andmelted at 1 873 K under purified Ar atmosphere. The oxy-gen content in the off-gas was measured with a solid stateceramic oxygen sensor (Rapidox 2100, Cambridge SensotecLtd.). A typical value of the oxygen concentration in puri-fied Ar is about 10–20 ppm. The temperature profile in thefurnace ensures a hot zone, approximately 4 cm in length,in which the temperature variation is within 1°C. Three dif-ferent Ti sources were cleaned by pickling in 5 vol% NaOH

solution for 1 min, followed by washing in de-mineralizedwater, ultrasonic cleaning by dipping in acetone and drying.Each piece of Ti or FeTi was placed inside a silica tube (8mm ID, 10 mm OD) which was joined to another tube (6mm ID, 8 mm OD, 20 mm H) in order to fix these pieces.This is schematically shown in Fig. 2. After about 60 minof temperature homogenization of liquid iron at 1 873 K, thesilica tube containing Ti or FeTi piece at room temperaturewas immersed in the magnesia crucible containing liquidiron. A small volume of molten Fe was suctioned into thetube and brought into contact with Ti or FeTi for a prede-termined time of 10 s, 30 s and 60 s. The experimentalparameters for each test are given in Table 1. The tempera-ture of liquid iron just before the contact was 1 873 K. Thetotal oxygen measured in the bulk iron sample after theexperiment was 100 to 140 ppm.

2.2. Analysis and CharacterizationThe samples obtained after the experiments were subjected

to microstructural investigation by using a high resolutionscanning electron microscope (Philips SEM XL-30 FEG),equipped with an EDAX energy dispersive spectrometer(EDS) detector system. The total oxygen content was deter-mined in the bulk Fe samples after the experiments with aLeco combustion analyzer (TC-436DR). The thermodynam-ic calculations for FeTi35 experiments were carried outusing the ‘Equilib’ module of FactSage 6.2. The databasesused for these calculations were FACT, FSstel and FToxid.

3. Results and DiscussionThe early dissolution behavior of three different Ti sourc-

es was studied by bringing Ti or FeTi in contact with liquid

Fig. 1. Schematic representation of the typical stages during the alloying (and/or deoxidation) practice.

Table 2. Composition of the Ti sources used for the experiment (wt%).

Material Ti S Mn Al Si Ca V T.O. Fe

Ti 99.99 – – – – – – – –FeTi70 69.6 0.006 0.25 2.48 – 0.03 1.83 0.18 BalFeTi35 38.5 0.022 0.66 5.05 3.5 0.22 0.42 0.95 Bal

Electrolyticiron

C S Mn N Si Ca Cu T.O. Fe0.0005–0.0015 0.0001–0.0003 – 0.0005 0.0003 – 0.0001 0.0020–0.0050 Bal

Table 1. Experimental parameters.

Ti source(pure Ti or FeTi) Contact time (s) Sample identification

Ti (99.99%)10 Ti-0130 Ti-0260 Ti-03

FeTi7010 Ti70-0130 Ti70-0260 Ti70-03

FeTi3510 Ti35-0130 Ti35-0260 Ti35-03 Fig. 2. Schematic representation of the liquid metal suction method.


631 © 2013 ISIJ

iron at approximately 2 cm above the bottom of the crucible,inside the temperature zone (temperature variation ±1°C) ofthe vertical tube furnace. After allowing the Ti or FeTi incontact with liquid iron for the predetermined span, it wasfound that pure Ti and FeTi35 established a perfect contactwith liquid iron and macroscopically, a clear interface wasobtained (Fig. 3). However in case of FeTi70, instead of aclear interface, a macroscopically homogenous alloy wasobtained suggesting a complete melting of FeTi70 in the liq-uid Fe. The dissolution mechanisms and the inclusion for-mation in liquid steel for these systems (liquid Fe – pure Ti,FeTi70, and FeTi35) are discussed in the following sections.In order to understand the various phases formed during theliquid Fe and titanium interaction, the titanium concentra-tion of each phase is given in Table 3 obtained from the Fe–

Ti equilibrium phase diagram.14)

3.1. Fe-Pure Ti Reaction Couple3.1.3. Microstructure of the Fe–Ti Interface

Ti and Fe established a perfect contact as shown in Fig.3. The reaction couples Ti-01, Ti-02 and Ti-03 wereobtained after the predetermined times of 10, 30 and 60 srespectively. Titanium oxide inclusions were not observed inthe reaction zone of Ti-01, Ti-02 and Ti-03 samples. Theresult was unexpected. The probable reason is explained atthe later part of this section. Figures 4 and 5 indicate thatthe reaction zone widens with time. The dissolution of Tiinto liquid iron (Fig. 4) shows the phases analogous to theFe–Ti equilibrium phase diagram.14) The phases β -Ti, TiFeand TiFe2 are the products of metallurgical reactions i.e.eutectic reaction at 1 358 K and eutectoid reaction at 868 K,respectively. The reaction zone between the Fe and Ti wasclearly distinguishable for the different reaction times. Thecomposition in the reaction zone near pure Ti consisted ofthe eutectic mixture of TiFe and β -Ti (Fig. 4(a)). It meansthat the eutectic alloy (melting point 1 358 K) forms having

Fig. 3. Fe–Ti reaction couples obtained after introduction of Tisource to the liquid iron.

Table 3. Concentration of Ti in the various phases (wt%).14)

Phase α-Ti β-Ti TiFe TiFe2

Ti concentration 99.95 to 100 75.3 to 100 45.9 to 48.7 24.6 to 31.8

Fig. 4. The micrographs of the reaction zone between pure Ti and liquid Fe. The distances in μm are given from the pureTi towards pure Fe (a) Ti-01 (b) Ti-02 (c) Ti-03.

© 2013 ISIJ 632


a considerably lower melting point than pure Ti (meltingpoint 1 943 K) upon the contact. The reaction zone wasextended in the width and pure titanium crystals were seenin the β -Ti and TiFe matrix for the sample Ti-02 and Ti-03(Figs. 4(b) and 4(c)). This was due to the formation of Tirich liquid at the interface as the holding time increases,which resulted in the precipitation of pure Ti crystals duringcooling. The formation of Ti rich liquid can be attributed tothe melting of pure Ti due to the exothermicity of the Fe–Ti dissolution.15) The overall width of the reaction zone forTi-03 was comparable to Ti-02 (Fig. 5). However the extentto which different phases exist inside the reaction zone variesfor Ti-02 and Ti-03 samples. The measurement of distinctphases in the reaction zone for these samples is shown in Fig.5. It can be clearly seen from Fig. 5 that the width of theindividual phases varies showing the relatively high titaniumconcentration regions towards the Fe side with holding time.

The widths of the reaction zones are 1.3 mm for Ti-01,4.1 mm for Ti-02 and 4.3 mm for Ti-03 sample as measuredin the center of the samples. It can be speculated that the for-mation of Al–Ti–O inclusions after the addition of pure Tito the liquid steel can be due to the formation of these reac-tion zones in which the concentration of Ti is sufficientlyhigh to reduce the existing alumina inclusions in liquid steel.The concentration of dissolved Ti required to reduce theexisting alumina inclusions in the liquid steel containing aparticular dissolved aluminium content is discussed in theFe–FeTi35 section.

A large amount of data compiled by Cha et al.16) for Tideoxidation experiments from various researchers show thatas the Ti content increases above 1 wt%, the oxygen require-ment for the oxide formation also increases. In fact, exper-imental data of the oxygen saturation line corresponding tothe Ti concentrations above 10 wt% were not found. In thepresent experiments, the Ti concentration of the variousphases found in the reaction zone was much higher than 10wt% (Table 3 and Fig. 4). Therefore, the oxygen concentra-tion range (100–140 ppm) in the present experiments wasnot sufficient for the formation of titanium oxide.

3.2. Fe–FeTi70 Reaction CoupleThe FeTi70 contains a higher level of impurities as com-

pared to the pure Ti. The chemical composition of this alloyis given in Table 2 and the microstructure is shown in Fig.6. FeTi70 is manufactured by alloying titanium sponge andscrap (mainly Ti–6Al–4V) with iron. Therefore, elementalimpurities like Al and V are common in this alloy. Themicrostructure consists of three distinct phases i.e. β -Ti,TiFe and β -Ti+TiFe. These phases are in accordance withthe Fe–Ti equilibrium diagram for an alloy containing ~70wt% Ti. The phase TiFe can dissolve up to 4–8 wt% Al and2 wt% V.17)

Unlike pure Ti, the interface between Fe and FeTi70 alloyobtained after quenching was not macroscopically visible

(Fig. 3). The FeTi70 alloy (melting temperature 1 358 K),after coming in contact with liquid iron, melted and subse-quently, the mixing took place by convection and diffusion.

3.2.1. Microstructure of Fe–FeTi70 Reaction ZoneThe microstructure of the Fe–FeTi70 reaction zone, due

to complete melting of FeTi70, shows the boundary betweenthe Fe+FeTi70 mixture and Fe. All three samples consistedof dark regions distributed in a light grey matrix. The Ticoncentration of the dark region in Ti70-01 and Ti70-02samples (Figs. 7(a) and 7(b)), was intermediate between theβ -Ti and TiFe compositions while that of the light greymatrix was less than the Ti concentration of the TiFe2 phasei.e. it falls between the TiFe2 and pure Fe. The compositionsof each pure phase are listed in Table 3. In the Ti70-3 sam-ple (Fig. 7(c)), the dark areas were TiFe2 while the compo-sition of the light grey areas was less than 24 wt% Ti. Thisindicates a shift in the overall composition towards lower Ticoncentration from the original eutectic mixture of β -Ti andTiFe phases as observed in the as-received FeTi70. Thecomposition of the dark and light grey areas in theFeTi70+Fe mixture obtained in Ti70-1, Ti70-2 and Ti70-3samples are shown in Fig. 8. The phase indicated as Fe inTi70-1 and Ti70-2 samples has typically 2 to 5 wt% Ti andmarks the iron rich side.

The inclusions of Al–Ti–O type were seen in Ti70-2 andTi70-3 samples as shown in Figs. 7(b) and 7(c). In case ofTi70-3, in addition to the Al–Ti–O inclusions, Ca–Al–Ti–Oinclusions were also observed as shown in Fig. 7(c).

On comparing FeTi70 dissolution with pure Ti, it can beobserved that the FeTi70 started melting upon the contactdue to its low melting temperature (eutectic point) andresulted in the formation of inclusions due to the presenceof impurities like Al and Ca. While the dissolution of pureTi in liquid iron involves partial melting and the mixingmostly depends upon the diffusion of Ti species into the liq-uid iron. This suggests that the dissolution of FeTi70 ismuch faster than pure Ti.

The possibilities and the thermodynamic conditionsfavorable to formation of inclusions as observed in the Ti70-2 and Ti70-3 samples are discussed below.

Fig. 5. Schematic representation of the extent of the reaction zone and the various phases observed between the liquid Feand pure Ti (a) Ti-01 (b) Ti-02 (c) Ti-03.

Fig. 6. Distribution of phases in the FeTi70 alloy.


633 © 2013 ISIJ

3.2.2. Inclusion Formation in the Reaction ZoneThe interfacial reactions in the Fe–FeTi70 reaction zone

involve mainly liquid iron, dissolved oxygen, dissolved tita-nium and the impurities from the titanium source.

The inclusion characteristics i.e. size, morphology andchemical composition of the inclusions observed in Ti70-2and Ti70-3 samples (Fig. 9) are listed in Table 4.

The standard Gibbs free energies18–21) (J/mol) of forma-tion in liquid steel for these inclusions are given below. Thestandard state of the activities of dissolved elements (indi-cated by square brackets) was infinitely dilute solution inliquid iron while for the oxides, it was pure solid.

Al2O3(s) + [Ti] + 2[O] = Al2O3.TiO2(s)∆G° = –706 844 + 232.25T.................... (1)

Al2O3(s) + 3x[Ti] = 3TixO(s) + 2 [Al] ........... (2)

2[Al] + [Ti] + 5[O] = Al2TiO5(s)∆G° = –1 435 000 + 400.5T.................... (3)

[Ca] + [O] = CaO(s)∆G° = –629 998 + 144.75T.................... (4)

2[Al] + 3[O] = Al2O3(s)∆G° = –1 225 417 + 393.8T.................... (5)

CaO(s) + Al2O3(s) = CaO.Al2O3(s)∆G° = –19 246 – 18T......................... (6)

Fig. 7. Micrographs of the reaction zone between FeTi70 and liquid Fe for different reaction times (a) Ti70-1 (10 s) (b)Ti70-2 (30 s) (c) Ti70-3 (60 s).

Fig. 8. Composition of the dark and light grey regions observed inFe–FeTi70 reaction zone of Ti70-1, Ti70-2 and Ti70-3samples.

Fig. 9. SEM micrographs of the inclusions observed in samples:(a) and (b) Ti70-2; (c) and (d) Ti70-3.

© 2013 ISIJ 634


On the basis of the above reactions, the origin of inclusionformation after the introduction of the FeTi70 alloy is dis-cussed. As shown schematically in Fig. 1, when pure Ti orFeTi is added to the melt, a significant supersaturationoccurs at the periphery of the dissolving/melting Ti or FeTiand depending upon the oxygen concentration, the forma-tion of oxide inclusions takes place.

The formation of Al–Ti–O type of inclusions, as observedin the present experiments and the earlier work,2,5) is possibleby the Eqs. (1), (2) and (3). However, the possibility of theAl–Ti–O inclusion by Eq. (3) is more likely in the presentexperimental conditions because of the following reasons: (i)No alumina inclusion was present in the liquid Fe. Aluminainclusions were very rarely observed in the as-receivedFeTi70 and the few observed were large in size (> 20 μm)1)

while the Al–Ti–O inclusions obtained in the present sam-ples lie in the range of 2–3 μm (ii) all the reaction speciesi.e. dissolved Al, Ti and O are present at the reaction site intheir dissolved state in FeTi70 (iii) the appearance of theinclusions as observed in Figs. 9(a) and 9(b) is single phase.

The oxygen saturation lines, above which there is formationof Al–Ti–O inclusions, are calculated on the basis of Gibbsfree energy (Eq. (3)) and the interaction coefficients20,22,23)

data (Table 5) by varying Al and O concentration for a giv-en Ti concentration and the temperature.

Considering that the initial temperature of the liquid ironwas 1 873 K and the local temperature was likely to decreaseupon contact with FeTi70, the saturation lines are calculatedfor two temperatures. In the Ti70-2 and Ti70-3 samples, thetitanium concentration in the vicinity of Al–Ti–O inclusions(marked by circular regions in Figs. 7(b) and 7(c)) wasaround 3 to 5 wt% while the Al concentration was less than1 wt% (~2 000–7 000 ppm) as measured by the EDS. Thetotal oxygen concentration of the iron after the experimentswas 100 to 140 ppm determined by the Leco combustiontechnique. The dissolved O in the liquid iron was assumedto be the same as that of total oxygen. The melting point ofFe containing 3 to 5 wt% Ti is above 1 773 K. On combining

the compositional data with temperature, it further showsthat favorable thermodynamic conditions for the formationof Al–Ti–O inclusions exist for the oxygen concentrationbetween 100 to 140 ppm and Al concentration between2 000 to 7 000 ppm as marked by a rectangular region in Fig.10. This rectangular region lies above 2 wt% and below 4wt% Ti line for both 1 873 K and 1 773 K temperatures. Thisis in good agreement with the EDS measurement wherebyTi concentration was found to be 3 to 5 wt%.

Also calcium aluminate inclusions containing up to 7–8wt% titanium (Ca–Al–Ti–O) were observed in the presentexperiments (Figs. 9(c) and 9(d) and Table 4). These com-plex oxides can be formed by the reduction of alumina orAl–Ti–O inclusions by Ca. The concentration of Ca inFeTi70 was 50–300 ppm.1) Approximately 0.8 g of FeTi70(starting weight) was melted after coming in contact with 4g of liquid Fe (the quantity suctioned into the silica tube)during the experiment. This results in a total calcium con-centration of 7 to 40 ppm in the resulting Fe–FeTi70 mix-ture. The thermodynamic calculations for the reactions (Eqs.(4) to (6)) shows that for Ca to be present in the elementalform, so that it can reduce alumina inclusion or Al–Ti–Oinclusion, the activity of oxygen in liquid iron has to beextremely small. The presence of high oxygen in liquid iron(100–140 ppm) and about 1 800 ppm of total oxygen inFeTi70, suggests that the existence of Ca in oxide is morefavorable than the elemental form. Therefore, it seems thatthe calcium aluminate inclusions may have already existedin the FeTi70.

Table 4. Inclusion characteristics observed in the Ti70-2 and Ti70-3 samples.

Sample Inclusion morphology Size (μm)Composition (wt%)

Al Ti Ca O Fe

Ti70-2Polygonal 2–3 46.23 11.25 42.52Polygonal 2–3 30.49 16.31 24.47 28.74Polygonal 1–2 27.64 14.56 20.90 36.89

Ti70-3

Dual phasespherical core(CaO.Al2O3)

~7 19.78 7.16 28.54 44.52

Polygonal periphery ~25 24.11 7.29 37.61 30.91

Dual phase(rounded)

Dark region(CaO.2Al2O3) ~10

31.30 7.23 23.92 37.55

Gray region 7.32 33.19 54.97 4.52Polygonal 2–3 19.82 21.04 29.59 29.54

Table 5. Interaction coefficients.

i j Reference

OAl –0.83 22)Ti –0.6 20)O –0.2 20)

AlAl 0.045 20)Ti 0.004 23)O –1.4 22)

TiAl 0.0037 23)Ti 0.013 20)O –1.8 20)

eij

Fig. 10. Oxygen saturation lines drawn for varying Al, O and Ticoncentration for 1 773 K and 1 873 K. The compositionalregion in which the inclusions were observed in the pres-ent experiments for Ti-02 and Ti-03 samples is indicatedby a rectangle. Al and Ti concentrations were measured byEDS while the O concentration was assumed to be 100–140 ppm as measured by the LECO combustion technique.


635 © 2013 ISIJ

The formation of Ca–Al–Ti–O inclusions by the reactionbetween dissolved Ti and already existing Ca–Al–O inclu-sions seems to be the most likely possibility based on theobservations and the thermodynamic calculations. However,the formation mechanism of Ca–Al–Ti–O inclusion requiresmore attention and needs to be investigated further. It is, sofar, clear that the impurities from FeTi70 contributestowards the inclusion formation.

3.3. Fe–FeTi35 Reaction CoupleFeTi35 was the most impure source of Ti introduced to

the liquid steel. The chemical composition of the alloy isgiven in Table 2. This alloy was quite fragile and porous innature. The variation in the chemical composition of thisgrade in the various samples taken within the same batchwas found to be very large.1) The presence of alumina inclu-sions and its size distribution in the as-received FeTi35(FeTi35-0) are shown in Fig. 11.

The inhomogeneous composition (Fig. 11(a)) can beattributed to the processing route of FeTi35 which is thereaction product of aluminothermic reduction of ilmenite.Therefore, the resultant product contains Ti, Al, alumina andunreduced Ti–Ox inclusions. The total oxygen content ofFeTi35 as measured with the LECO combustion techniquewas found to be very high (4 000 ppm to 1 wt%)1) and canbe largely attributed to the presence of alumina inclusions.The alumina inclusion size distribution is shown in Fig.11(b). The compositional inhomogeneity and the presenceof inclusions, makes FeTi35 quite distinct from the previoustwo Ti sources i.e. pure Ti and FeTi70. As per the Fe–Tiequilibrium diagram, the approximate melting temperatureof the FeTi35 (containing 38 wt% Ti) is found to be 1 623K which lies between the melting temperature of pure Ti(1 943 K) and FeTi70 (1 358 K). FeTi70 was found to becompletely molten while pure Ti was partially melted. Themicrographs of the FeTi35 samples on introduction to liquidiron, namely, Ti35-1, Ti35-2 and Ti35-3 are shown in Fig. 12.

The size and morphology of alumina inclusions were notchanged much in the Ti35-1 sample as compared to the as-received FeTi35-0 sample (Figs. 11(a) and 12(a)). In theTi35-2 sample, the microstructure of FeTi35 becomes quitehomogenous in appearance with a distribution of Ti–Ox andthe alumina inclusions (Fig. 12(b)). In Ti35-3 sample, Ti

rich regions (β -Ti) containing from 72 to almost 100% Tiwas seen in the coagulated microstructure. It shows thatthere was partial melting inside the FeTi35, as outlined inFig. 12(c). One shared phenomenon which was observed forall samples was the development of an oxide layer (mainlyAl and Ti oxide) between FeTi35 and the liquid iron whichwas grown in thickness with time. Due to the formation ofa complex oxide layer between the liquid Fe and FeTi35, onlya very low level of Ti (up to 2 wt%) was observed in theiron side within a distance of 200 μm from the oxide layer.

The development of the oxide layer between the FeTi35and liquid iron and the inclusion evolution in the FeTi35side are discussed in the following sections.

3.3.1. Oxide Layer Growth between FeTi35 and Liquid FeThe oxide layer consisting of Fe–Al–Ti–Si–O was grown

as a function of time from 10 to 60 s. The oxide layer thick-ness vs time is plotted and shown in Fig. 13. The oxide layermicrographs are shown in Fig. 14. The thin oxide layerbetween Fe and FeTi35 with a thickness of ~6 μm wasfound in Ti35-1 sample. This oxide layer consisted of loose-ly packed alumina and Al–Ti–O inclusions as can be seenin Fig. 14(a). In Ti35-2 sample, the oxide layer was uniformand about 35 μm thick. The composition of oxide layer inTi35-2 sample consisted of Al–Ti–O embedded in a contin-uous Ti–OX layer. The variation in thickness of the oxide

Fig. 11. FeTi35: (a) Distribution of phases and inclusions in theFeTi35 alloy (Matrix A – Al-4 to 10 wt%; Si-3 to 5 wt%;Ti-34 to 42 wt%; Fe-29 to 45 wt%; O-0 to 10 wt%) (b)Inclusion size distribution.

Fig. 12. Micrographs of the FeTi35 and liquid iron interface (a)Ti35-1 (10 s) (b) Ti35-2 (30 s) (c) Ti35-3 (60 s).

Fig. 13. Oxide layer thickness vs time of contact for FeTi35-1,Ti35-2 and Ti35-3 samples.

© 2013 ISIJ 636


layer formed in Ti35-3 sample was large as can be seen fromFig. 12 and the thickness was about 85 μm on average. Therelationship between the oxide layer thickness and the con-tact time looks parabolic with its axis of symmetry lying onthe y-axis suggesting accelerated growth of the oxide layerwith increasing contact time. This is in contrast to the gen-erally observed parabolic behavior with the axis of symme-try lying on the x-axis indicating that the oxidation ratedecreases with time due to the increasing oxide layer thick-ness which itself acts as a diffusion barrier. However, con-sidering (1) the limited data (only three points) and (2) alarge variation in the thickness of the oxide layer in Ti35-3sample (contact time = 60 s), a linear relationship betweenthe oxide layer thickness and contact time was assumedindicating that the oxide layer growth is reaction rate con-trolled rather than diffusion controlled.

Practically, the oxidation process is complex and in theseexperiments, an alloy (FeTi35) was involved. A mechanismof the oxide layer formation on the basis of observations andthe compositional data, is proposed:

(i) In Ti35-1 sample (contact time = 10 s), when liquidiron comes in contact with FeTi35, Al and Ti from FeTi35combines with the dissolved oxygen from the liquid iron toform small alumina and Al–Ti–O inclusions (≤2 μm) as canbe seen in the Fig. 14(a). The formation of a thin layer con-sisting of small inclusions between liquid Fe and FeTi35prevents significant diffusion of Ti in liquid iron. Addition-ally, the possibility of the attachment of a few inclusionsfrom FeTi35 to the interface also exist owing to (a) the pres-ence of a large amount of inclusions in FeTi35 and (b) care-ful observation of the interface after 10 s (Fig. 14(a)) showsloosely adherent inclusions with crystalline morphology.

Al and Ti (FeTi35) + O (liquid iron) = Al2O3 + Al–Ti–O

Al and Ti oxides (FeTi35) = TiOx + Al–Ti–O (at interface)

(ii) After 30 s (Ti35-2), further oxidation takes place on

the FeTi35 side forming a continuous Ti–OX layer with Al–Ti–O inclusions embedded in it at some locations (Fig.14(b)). Titanium, for which the concentration was signifi-cantly higher than the other elements in FeTi35, oxidizedforming a continuous Ti–OX layer. The oxide layer formedafter 10 s was very thin and mainly consisted of a looselyadherent particles. Therefore, there was a contribution ofdissolved oxygen from liquid iron. The possibility of adher-ing the already existing inclusions in FeTi35 to the oxidelayer also exist. The growth of the oxide layer can be attrib-uted to

Ti (FeTi35) + O (liquid iron) = TiOX

Al and Ti oxides (FeTi35) = TiOx + Al–Ti–O (at interface)

(iii) With longer holding times (60 s), the oxide layer wasdifficult to distinguish between different oxides within thereaction layer (Fig. 14(c)). The resultant oxide layer seemsto be the solid solution of mainly Al and Ti oxides. The ele-mental mapping of this (Ti35-3 sample) oxide layer isshown in Fig. 15. It can be seen that the strong oxide form-ing elements like Al and Ti with some Si in FeTi35 wereoxidized resulting in the formation of a complex oxide.

Al2O3 + TiOx + Al–Ti–O + Fe and Si (FeTi35)= Al–Ti–Si–Fe–O

On the basis of the the elemental mapping of the oxidelayer (Fig. 15), it can be deduced that the oxide layer asobserved in the micrographs (Fig. 14) has grown moretowards the FeTi35 side as the presence of the Fe in the bulkoxide layer was minimal. It means that the contribution ofoxygen and/or oxides from FeTi35 side was substantiallyhigher than that of the liquid Fe side. The possible sourcesfor oxygen and or oxides are dissolved oxygen (100–140ppm) in liquid iron, the air pockets in FeTi35 because of itsporous nature and the oxides/inclusions in FeTi35 (totaloxygen 4 000 to 10 000 ppm). In the beginning, the availabil-ity of oxygen can be attributed to the dissolved oxygen fromliquid iron side to a larger extent and; air infiltration due tothe porous nature of FeTi35 to a smaller extent up to 30 s.After 30 s, the growth of the oxide layer on the FeTi35 sidecan be attributed to the air infiltration to a larger extent andto a smaller extent, to the adherence of the existing inclu-sions in FeTi35. The inclusion evolution in the bulk of theFeTi35 is further discussed in detail in the following section.

3.3.2. Inclusion EvolutionThe micrographs of the inclusion evolution for Ti35-1,

Ti35-2 and Ti35-3 samples are shown in Fig. 16. The inclu-sion size distribution, average size and the area fraction ofthese inclusions as measured in FeTi35 side of these sam-

Fig. 14. Micrographs of the oxide layer (a) Ti35-1 (b) Ti35-2 (c)Ti35-3.

Fig. 15. Elemental mapping of the oxide layer in Ti35-3 sample.


637 © 2013 ISIJ

ples is shown in Figs. 17(a) and 17(b).The inclusion size distribution for these samples indicates

that the number of large sized inclusions decreases as theholding time increases i.e. the number of large sized inclu-sions (60–90 μm) have decreased from Ti35-1 to Ti35-2sample while the inclusions less than 10 μm have increasedin number. In Ti35-3 sample the inclusions in the size range50 to 90 μm had disappeared (Fig. 17(a)). The decrease inthe large sized inclusions has affected the other inclusioncharacteristics e.g. it can be observed that the number, areafraction and the size of the inclusions continuously decreasefrom Ti35-1 to Ti35-3 sample (Figs. 17(a) and 17(b)).Another possibility for observing less inclusions in Ti35-3sample is that the distribution of the alumina inclusions wasinhomogenous throughout the microstructure in the as-received FeTi35. As in these experiments, only a limitedcontact surface was available as shown in Fig. 12, incident-ly, an area of Ti35-3 with a low number of inclusions couldhave been contacted.

The decrease in large sized alumina inclusions withincrease in holding time can be attributed to the reductionof alumina inclusions by the high Ti concentrations sur-rounding them. The unreduced Ti–Ox by Al and; aluminainclusions by Ti were reduced depending upon the localcomposition, specifically, the Ti/Al ratio (Ti and Al in wt%).

The reduction of oxides is explained on the basis of thethermodynamic calculations (FactSage). The local composi-tion i.e. Al and Ti concentration surrounding the aluminainclusions (Fig. 16) was measured by EDS. The aluminareduction reactions by Ti are shown below by Eqs. (7) and(8).

x/3 Al2O3 (s) + Ti = TiOx (s) + 2x/3 Al .......... (7)

Al2O3 (s) + m Ti = Al2–nTimO3(s) + nAl.......... (8)

It was also assumed that the elemental species like Al andTi were in the dissolved state because of the partial meltinginside the FeTi35 as can be seen from Fig. 12(c). The Ti/Alratio (concentration in wt%) as measured by EDS in theregion as indicated in Fig. 16 was found to lie in the rangeof 3 to 5 in Ti35-1 sample and 7 to 15 in Ti35-2 and Ti35-3 samples. This local compositional data was used for theFactSage calculations. The FactSage calculations were car-ried out for a 100 g of iron containing varying concentra-tions of Al (0.5 to 2 wt%) and Ti (0.2 to 25 wt%) with 0.05wt% alumina at two different temperatures i.e. 1 873 K and

Fig. 16. Microstructural evolution and the Al based inclusionsinside FeTi35 (compositional data was measured in thevicinity of inclusions as indicated by a circular ring).

Fig. 17. Evolution of Al based inclusions as a function of time inTi35-1, Ti35-2 and Ti35-3 samples (a) Inclusion size dis-tributions (b) Average inclusion size and area fraction.

Fig. 18. Dependence of the stability of pure Al2O3 on Ti/Al ratio at1 873 and 1 773 K.

© 2013 ISIJ 638


1 773 K. The calculation results show that with higher tita-nium concentration, especially when the Ti/Al (in wt%)ratio is greater than 10, the alumina starts diminishing andthe Ti2O3 and ilmenite starts forming as shown in Fig. 18.The pure Al2O3 was found to continuously decrease as theratio of Ti/Al approaches a value of 10 at 1 873 K and 12 at1 773 K while the Al content was increased. These calcula-tion results are found to be in very good agreement withMatsuura et al.2) and the present observations as indicatedin Figs. 16 and 17. The inclusions in the Ti35-3 were com-paratively small in size and a few in numbers. The Ti–Oxinclusions were found to be extremely small in size (lessthan 5 μm) while Al–Ti–O inclusions were found to be rel-atively large in size in the Fe–Ti matrix of Ti35-3 sample.

It is clear from the above discussion that the FeTi35 addi-tion introduces alumina and/or Al–Ti–O inclusions to theliquid steel.

4. Summary and ConclusionsGenerally, the formation of Al–Ti–O inclusions is attrib-

uted to the local supesaturation of Ti immediately after theaddition. However, the present study shows that the impuri-ties and the inclusions present in the Ti source alters their dis-solution behavior and subsequently, the inclusion formation.

The present work is summarized in the context of actualsteelmaking processing as follows:

(i) Pure Ti: The high titanium concentration regions afterthe reaction between pure titanium and liquid iron suggeststhat the titanium dissolution process is a relatively slow pro-cess owing to its high melting point. In the actual steelmak-ing process where the titanium addition is made in the Aldeoxidised liquid bath, soluble aluminium typically lies inthe range of the 300–600 ppm which ensures very low dis-solved oxygen content (~4–6 ppm). This in turn, makes theformation of pure titanium oxide inclusions difficult. How-ever, it can be speculated that if the dissolving Ti encountersan alumina inclusion, the alumina inclusion can be reducedto form Al–Ti–O because the Ti/Al (in wt%) ratio in liquidsteel surrounding the alumina inclusion can easily be above10. The micrographs of the Fe-pure Ti reaction zone revealsthe regions of very high Ti concentration in the vicinity ofdissolving pure Ti. The modification/reduction of aluminainclusions in liquid steel to Al–Ti–O inclusions as explainedby the previous researchers can be attributed to theseregions of high titanium concentration.

(ii) FeTi70: Apparently, FeTi70 meets two requirementsto reduce the local supersaturation in order to prevent theAl–Ti–O formation i.e. its low melting point makes it a fast-er melting alloy and as compared to pure Ti, it is a moredilute source of titanium. However, FeTi70 contains ele-mental impurities like Al and Ca. Until now, the behaviorof these impurities was unclear. The formation of Al–Ti–Oinclusions after FeTi70 addition gives two possibilities (a)the reduction of alumina inclusions by dissolved Ti and or(b) the formation of Al–Ti–O inclusions by combining thedissolved Ti, Al and O as observed in the present experi-ments. FeTi70, being a low melting eutectic alloy, wasfound to be dissolved much faster than pure Ti. Therefore,after the addition of FeTi70, the type of inclusion formationis controlled by the availability of dissolved oxygen. Thesource for oxygen can be either reoxidation or the FeTi70alloy itself. Nonetheless, in the present experiments, it is clearthat the impurities from FeTi70 contribute towards the inclu-sion formation like Al–Ti–O and Ca–Al–Ti–O inclusions.

(iii) FeTi35: The oxide layer at the interface betweenFeTi35 and iron grows in thickness with increase in the con-tact time. The growth of oxide layer can be attributed to the

local oxidation of the elements like Al and Ti as well as theattachment of inclusions from FeTi35 to the interface. It hasprevented the significant mixing of the FeTi35 alloy in liq-uid iron. However, the evolution of the alumina inclusionsinside the FeTi35 microstructure was observed. The size ofpure alumina inclusions were found to have been decreasedwith time. At 1 873 K, if the Ti/Al (in wt%) ratio exceeds avalue of 10, the alumina inclusions can be reduced by Ti asshown with the thermodynamic calculations. The high Ticoncentration at the edges surrounding the alumina inclu-sions was found to reduce them to Al–Ti–O inclusions. Inactual steelmaking, it is quite unlikely that the FeTi35 wouldtake such a long time to dissolve as the other much moredominating mechanisms like eddy diffusion and convectionwill enhance the mixing process. However, the short timefor dissolution means less time for the modification of exist-ing alumina inclusions. Therefore, it is certain that theFeTi35 addition to the liquid steel introduces alumina andAl–Ti–O inclusions. This has also been observed and report-ed in our earlier study at the industrial scale.24)

If FeTi35 is produced in the same manner as FeTi70 i.e.by alloying, the advantage of lower local supersaturationcan be taken due to Ti dilution and faster melting as com-pared to the FeTi35 produced by aluminothermic reduction.It will also prevent the introduction of alumina/Al–Ti–Oinclusions maximizing the overall Ti yield/recovery.

AcknowledgementsThis work was performed with the financial support of

ArcelorMittal Industry Gent (Sidmar) and the IWT (projectno. 070277).

REFERENCES

1) M. M. Pande, M. Guo, X. Guo, D. Geysen, S. Devisscher, B. Blanpainand P. Wollants: Ironmaking steelmaking, 37 (2010), 502.

2) H. Matsuura, C. Wang, G. Wen and S. Sridhar: ISIJ Int., 47 (2007),1265.

3) C. Wang, N. T. Nuhfer and S. Sridhar: Metall. Mater. Trans. B, 40(2009), 1022.

4) C. Wang, N. Verma, Y. Kwon, W. Tiekink, N. Kikuchi and S.Seetharaman: ISIJ Int., 51 (2011), 375.

5) M. Van Ende, M. Guo, R. Dekkers, M. Burty, J. Van Dyck, P. T.Jones, B. Blanpain and P. Wollants: ISIJ Int., 49 (2009), 1133.

6) S. Basu, S. K. Choudhary and N. U. Girase: ISIJ Int., 44 (2004), 1653.7) L. Pandelaers: PhD thesis, KU Leuven, Belgium, (2011).8) M. Wang, Y-P. Bao, H. Cui, H-J. Wu and W-S. Wu: ISIJ Int., 50

(2010), 1606.9) M-K. Sun, I-H. Jung and H-G. Lee: Metal. Mater. Int., 14 (2008),

791.10) J. H. Park and Y. B. Kang: Metall. Trans. B, 37B (2006), 791.11) J. H. Park and H. Todoroki: ISIJ Int., 50 (2010), 1333.12) M-A. Van Ende, M. Guo, J. Proost, B. Blanpain and P. Wollants: ISIJ

Int., 50 (2010), 1552.13) M-A. Van Ende, M. Guo, J. Proost, B. Blanpain and P. Wollants: ISIJ

Int., 51 (2011), 27.14) ASM Metals Handbook, Vol. 3, Alloy Phase Diagram, ASM, Ohio,

(1992).15) S. A. Argyropoulos and R. I. L. Guthrie: Metall. Trans. B, 15 (1984),

47.16) W. Y. Cha, T. Nagasaka, T. Miki, Y. Sasaki and M. Hino: ISIJ Int.,

46 (2006), 996.17) O. Heinen, D. Holland-Moritz and D. M. Herlach: Mater. Sci. Eng.

A, 449–451 (2007), 662.18) W. Y. Cha, T. Miki, Y. Sasaki and M. Hino: CAMP-ISIJ, 19 (2006),

760.19) H. Ito, M. Hino and S. Ban-ya: Tetsu-to-Hagané, 83 (1997), 773.20) G. K. Sigworth and J. F. Elliontt: Met. Sci., 8 (1974), 298.21) R. H. Rein and J. Chipman: Trans. Metall. Soc. AIME, 233 (1965),

415.22) H. Ohta and H. Suito: The Final Report of Ultra Clean Steel Research

Group, Thermodynamics of Production of Ultra Clean Steel, ISIJ,Tokyo, (1999), 66.

23) G. Yuanchang and W. Changzhen: Metall. Trans. B, 21B (1990), 543.24) M. M. Pande, M. Guo, S. Devisscher and B. Blanpain: Ironmaking

Steelmaking, 39 (2012), 519.

Inclusion Formation and Interfacial Reactions between FeTi ...

Documents

Transcript of Inclusion Formation and Interfacial Reactions between FeTi ...