Interfacial Reactions and Inclusion Formations at an Early Stage of FeNb Alloy Additions to Molten Iron

Abstract

Nb is an important microalloying element in steelmaking. Its interaction with liquid Fe during an early stage of the alloying process has a considerable influence on the Nb recovery. In the present work, the inclusions in FeNb alloys were characterized using the electrolytic extraction method combined with SEM-EDS. The interfacial reactions between FeNb alloy and liquid Fe, as well as inclusion formations, were studied during an early stage of an alloy addition using a liquid-metal-suction method. The results revealed that a diffusion zone consisting of different regions of Fe–Nb phases was formed and that the thickness of the zone increased with time. Based on the experimental findings, the mechanism of the early dissolution process of FeNb alloys in liquid Fe was discussed. Moreover, the Nb rich regions formed after the alloy contacted with liquid Fe could modify the existing inclusions in the alloy, also their evolution mechanisms were studied. The addition of FeNb alloys can introduce inclusions, such as Al–O and Al–Ti–Nb–O inclusions to the liquid steel. Overall, this study has contributed to the understanding the behaviour of impurities from the FeNb source at the early dissolution process during the microalloying process of steels containing Nb.

1. Introduction

The discovery of niobium (Nb) during the 19th century and its ability to enhance the mechanical properties of carbon steels was explored by metallurgists.¹⁾ To date, it is widely used in weathering steels, rail steels, high strength low alloy (HSLA) steels.²⁾ The most compelling reasons for microalloying with Nb are the maximization of the yield strength through a combination of grain refinement and precipitation hardening.³⁾ Several methods are currently employed to add Nb to steels, such as niobium wires, pellets, and powders. The most common practice is to add ferroniobium (FeNb) during ladle treatment. Due to the high melting temperature of FeNb, it does not immediately melt but it rather sluggishly dissolves when added to liquid steel.²⁾ To maximise the beneficial effects of Nb, its content has to be precisely controlled. This requires thorough knowledge on the dissolution of FeNb alloys in the steel and on the interfacial reactions between the alloys and the melt. Furthermore, the inclusion behaviour at the early stage of alloying can help us to better understand the effect of impurities in FeNb alloys on the steel cleanliness.

In recent years, there has been a growing interest concerning the kinetics and mechanism of alloy melting and dissolution in liquid metals, e.g. FeSi,⁴⁾ FeMo,⁵⁾ Ti,⁶⁾ FeTi.⁷⁾ Moreover, interfacial reactions between alloying elements and molten steel have been investigated by several researchers. Van Ende et al.⁸⁾ studied the initial stage of Al deoxidation. They revealed that a reaction zone occurred and that it consisted of several layers of Al-rich intermetallic compounds. Microscopic investigations of the quenched samples demonstrated the formation of Al₂O₃ inclusions in the reaction zone. Pandelaers et al.⁹⁾ investigated the interfacial reactions between a solidified Fe shell and Ti as well as FeTi during the dissolution of Ti and FeTi in liquid Fe using the load cell measurement method. The results showed that a liquid reaction zone was formed between the Ti and the shell, which was governed by mass transport. Pande et al.¹⁰⁾ compared the dissolution behaviour of pure Ti, FeTi70 and FeTi35 in liquid Fe. They reported that the addition of FeTi70 and FeTi35 alloys can introduce Al₂O₃ and Al–Ti–O inclusions from these alloys to steel. Yan et al.¹¹⁾ studied the dissolution behaviour of FeMnSi alloy. Based on these studies, it can be found that most of them are focused on the dissolution behaviour of alloys which have a lower melting point compared to liquid steel. However, the publications on the FeNb–Fe interfacial reactions and inclusion behaviour after FeNb addition into the liquid melt are missing.

In this study, the cleanliness of FeNb alloy with respect to inclusions was first analysed. The interactions between FeNb and Fe shortly after the alloy addition were investigated based on quenched samples using the liquid-metal-suction method. The aim was to understand the early dissolution mechanism between FeNb and liquid Fe as well as the behaviour of inclusions from a FeNb alloy when the alloy had been added to liquid Fe.

2. Experimental

2.1. Experimental Procedures

The liquid-metal-suction method was used to investigate the diffusion process between the FeNb alloy and liquid iron (Fe) to represent the early stage of dissolution. The chemical composition of the commercial FeNb alloy is listed in Table 1.

Table 1. Main chemical composition of FeNb alloy.

Element	Nb	Si	Mn	C	Al	Ca	Mg	Ti	O	S	P	N	Sn	Pb	Fe
[wt%]	66.3	1.04	0.2	0.103	0.1	0.03	<0.01	0.3	0.31	0.016	0.064	0.02	0.05	0.02	balance

At the beginning of the experiment, 400 grams of electrolytic Fe (99.9%, 67 ppm O) was melted in a magnesia crucible which was placed inside a graphite crucible. The experiments were performed in an induction furnace at 1600°C using an argon atmosphere. A schematic illustration of the experimental setup is shown in Fig. 1. Initially, a piece of FeNb alloy (~4 mm, ~0.6 g) was placed inside a quartz tube (5.8 mm inside diameter) before sampling (Fig. 1(b)). The bottom of the quartz tube (QT) was made narrower to maintain the alloy inside the tube. After holding the melt for 30 min at 1600°C to homogenize the temperature and composition, the quartz tube with a FeNb alloy piece was quickly introduced in the liquid Fe. At this time, a small volume of melt was sucked into the quartz tube and came into contact with the FeNb piece. After the alloy was held in the melt for the desired contact time (5, 10, 20 and 30 s), the quartz tube was rapidly withdrawn from the metal and quenched in cold water. It should be noted that two samples were taken for each contact time. The detailed conditions for the samplings are listed in Fig. 2.

Fig. 1.

Schematic illustration of the experimental setup.

Fig. 2.

Schematic illustration of the main operations.

2.2. Analysis and Characterization

The bottom part of the sample, which contained the alloy, was subjected to microstructure and inclusion investigations on polished cross-sections using a scanning electron microscope equipped with an energy dispersive spectrometer (SEM-EDS).

The electrolytic extraction (EE) method was applied for the extraction of inclusion from the FeNb alloy. The EE was carried out using a 10% AA (10 v/v% acetylacetone, 1 w/v% tetramethylammonium chloride-methanol) electrolyte. The following parameters were used during the EE process: electric currents between 45–60 mA, voltages between 3.3–4.2 V, and a charge of 500 coulombs. The total weight of the dissolved FeNb alloy was about 0.07 g. After extraction, the solution containing inclusions was filtrated through a polycarbonate (PC) membrane film filter with an open pore size of 0.4 μm. Thereafter, the characteristics of the extracted inclusions on the film filter were investigated using SEM-EDS.

The average size of non-spherical inclusions (d_V) was calculated according to Eq. (1):   

$$d_V=\frac{L_{max}+W_{max}}{2}$$(1)

where L_max and W_max are the maximum length and width of the investigated inclusion measured by ImageJ software, respectively.

The number of inclusions per unit volume (N_V) was calculated by using Eq. (2):   

$$N_V=n\cdot \frac{A_{filter}}{A_{observed}}\cdot \frac{{\rho }_{metal}}{W_{dissolved}}$$(2)

where n is the number of inclusions in the appropriate size interval, A_filter is the area of the film filter containing inclusions (1200 mm²), A_observed is the total observed area on the film filter, ρ_metal is the density of the metal matrix and W_dissolved is the dissolved weight of the metal during extraction.

3. Results and Discussion

3.1. Inclusions in FeNb Alloys

The characteristics of inclusions in FeNb alloys are shown in Table 2. It illustrates that four types of inclusions were observed based on the chemical composition determinations. These are Al–O, Ti–O, Al–Ti–O and Si–Al–Mg–O inclusions. The majority of the inclusions are pure Al–O inclusions (36%), which are divided into two groups according to their morphologies, namely single inclusions (type A1) and clusters (type A2). The second common inclusion is type B inclusions (30%), which are irregular Ti–O inclusions. The Ti content in the alloy is 0.3% (Table 1), which might be the reason for the presence of large-sized Ti–O inclusions (up to 69 μm). A third inclusion group is type C inclusions, consisting of Al–Ti–O clusters, where the Al–O inclusions are surrounded by Ti–O inclusions. The type D inclusions are irregular Si–Al–Mg–O inclusions, which might originate from the slag during the production of the alloy.

Table 2. Typical inclusions observed in FeNb alloys.

Previously it has been reported by Bi¹²⁾ that Al–O inclusions were found in FeNb alloys, while no Ti–O containing inclusions were found. The possible reason for the presence of Al–O and Ti–O inclusions in this study was closely related to the Ti/Al ratio.^10,13) Also, it may also be explained by the different qualities of raw materials and the different technical methods for the production of the investigated ferroalloys. The sources of Al–O and Ti–O inclusions were most likely due to the deoxidation process which was controlled by aluminium and titanium during the FeNb alloy production. Besides, Al–O and Al–Ti–O clusters were formed due to the collision and aggregation of single Al–O and Ti–O inclusions. The particle size distributions of Al–O and Al–Ti–O inclusions are shown in Fig. 3. As can be seen, the peak in the particle size distributions of single Al–O inclusions of type A1 is about 11 μm, while that for the cluster type is about 24 μm. As a result, a waving curve obtained for Al–O inclusions has two main peaks. Moreover, the size range for Al–Ti–O inclusions is much wider (8–118 μm) compared to Al–O inclusions (8–46 μm). It should be noted that the number density increases a bit for large-sized Al–Ti–O inclusions, which can be explained by the random distribution of larger size inclusions on the film filter. It is well known that Al₂O₃ inclusions can significantly influence the steel properties and also cause nozzle clogging problems during casting.¹⁴⁾ Inclusions such as Ti–O and Al–Ti–O inclusions might also cause nozzle clogging and a decrease of the final product quality.¹⁵⁾ To conclude, the presence of these large-sized inclusions found in the FeNb alloy can reduce the quality of the steel product after the addition of this alloy into steel. Therefore, their transformations or modifications are important to understand and these are discussed in detail later in the paper.

Fig. 3.

Particle size distributions of Al–O and Al–Ti–O inclusions in FeNb alloys.

3.2. Overview of the Dissolution Phenomenon of FeNb Alloy

The early dissolution behaviour of an FeNb alloy in liquid Fe was studied by bringing FeNb in contact with liquid Fe at approximately 2 cm above the bottom of the crucible. Figure 4(a) shows an obtained typical QT sample after quenching (sample S2 as an example). It was found that the FeNb piece was located at the bottom of the sample. This is due to the larger density of the FeNb alloy (≈ 8.2 g cm⁻³)¹⁶⁾ compared to the liquid Fe (≈7.1 g cm⁻³).¹⁷⁾ The results showed that FeNb and Fe made good contact with each other, so that interactions could take place. Consequently, a diffusion zone was observed, which contained various phases. To identify these phases formed during the interactions, the range of Nb contents in each phase are given in the Fe–Nb equilibrium phase diagram in Fig. 4(b). It exhibits two eutectic reactions; one at 16 wt% Nb and 1375°C and the other at 75 wt% Nb and 1350°C. Moreover, two intermetallic phases Fe₂Nb and Fe₇Nb₆ are present and they have a limited solubility range.

Fig. 4.

(a) Typical photo of the QT sample and (b) Fe–Nb equilibrium phase diagram as calculated with FactSage. (Online version in color.)

During the short time of contact, an interdiffusion of Fe and Nb resulted in the formation of a diffusion zone. The microstructural evolution of this zone as a function of contact time is presented in Fig. 5. Five different regions are distinguished in the backscattered electron (BSE) image, as indicated by the dashed lines. Due to the fine structures in each region, the enlarged adjacent regions are clearly shown in Fig. 6. As shown in Fig. 6(a), there are three distinct phases in the commercial FeNb alloy before the contact with liquid Fe (region I), namely Fe₂Nb (45% Nb at average) and Fe₇Nb₆ (58% Nb at average) and a Nb-rich solid solution (96% Nb at average, not listed here). These phases are in accordance with those found in FeNb alloys according to previous studies.^18,19) The regions I and II after the contact with liquid Fe are shown in Fig. 6(b). It is found that the typical phases in region I after short contact with iron melt are the same with those observed before contact, which presents the unreacted part in the FeNb alloy. While region II consists of two phases, one containing 52–59% Nb and the other containing 11–15% Nb. It should be noted that region II only exists in some local areas, which is clearly shown in Fig. 6(c). In addition, region III consists of a 33–37% Nb phase and an 11–15% Nb phase. These phases might have been formed by the eutectic reaction of L⇋Fe₂Nb+δFe, depending on the local Nb contents and temperatures. Regions III and IV are illustrated in Fig. 6(d). Region IV also consists of two phases, which contain 11–15% Nb and 2–5% Nb, respectively. As the Nb content in the Fe-rich part decreases, invariant reactions δFe⇋Fe₂Nb+γFe and Fe₂Nb+γFe⇋αFe may take place. Figure 6(e) presents the regions IV and V, in which the 11–15% Nb phase in region IV has a dendritic morphology. This can be explained by a constitutional undercooling, due to the existence of temperature gradient between the alloy and liquid melt.

Fig. 5.

The interfacial microstructure with different contact time (a) 5 s, (b) 10 s, (c) 20 s, (d) 30 s.

Fig. 6.

The microstructure of different regions in the FeNb (a) and diffusion zone (b)–(e).

The different phases between the alloy and Fe are the result of diffusion by Nb and Fe. Figure 7(a) shows the line scan results from the alloy matrix to the Fe melt. It clearly illustrates that the Nb content has a decreasing tendency with some fluctuations, which are due to the existence of different FeNb phases. On the contrary, the opposite situation applies to the Fe content. It is assumed that a continuously shift in the overall composition towards a lower Nb concentration from the original phases in the FeNb alloy at the experimental temperature of 1600°C. Thereafter, element macro-segregation could occur which results in a phase separation into Nb-rich and Nb-less phases in the diffusion zones based on local Nb concentrations and temperatures during the solidification process. To conclude, the early dissolution of FeNb in liquid Fe mainly involves a partial mixing which mostly depends upon the diffusion of Nb atoms into the liquid Fe.

Fig. 7.

(a) Elemental line analysis of the diffusion zone, (b) the thickness and growth rate of different regions versus the contact time (c) predicted diffusion zone thickness versus time. (Online version in color.)

The total diffusion distance marked with the Nb content represents the thickness of the diffusion zone, which is found to grow with time. The measured thickness and the corresponding growth rate of different regions in the diffusion zone versus the contact time are plotted in Fig. 7(b). It can be seen that the thicknesses of regions II, III and IV increase with the contact time. Besides, the thickness of these regions increases in the following order at a given contact time: II, III and IV. As mentioned earlier, region II is located in some local areas inside region III. Therefore, their thicknesses are not included when calculating the total diffusion zone thickness. Overall, this results in the fact that the extended thickness of the diffusion zone increases from 360±70 μm to 1000±160 μm, as the contact time increases from 5 s to 30 s. The growth rate of the thickness is much higher at the beginning of the contact, especially before 10 s. Thereafter, the growth rate decreases with time due to the increasing diffusion zone thickness, which itself acts as a diffusion barrier.

The relationship between the diffusion zone thickness and the contact time looks parabolic. To predict the growth of the diffusion zone between added FeNb alloy and Fe melt, the diffusion process was estimated by fitting the measured diffusion zone thickness. The predicted diffusion thickness versus time and the obtained function equation are presented in Fig. 7(c), where t is the diffusion time and d is the diffusion zone thickness. It is found that the predicted diffusion zone thickness increases slightly with time and that it takes about 40 minutes to reach a thickness of 9 mm. Therefore, this is a rather slow diffusion process, which is controlled by the rate of mass transfer between the solid FeNb alloy and the liquid Fe. The costly FeNb alloys are generally added to the ladle during the last stage and their sizes are kept on the lower side, e.g. 10–40 mm. In actual steelmaking, it is quietly unlikely that the FeNb alloys would take such a long time to dissolve as the other much more dominating mechanisms like melting of FeNb, stirring and convection in the melt will enhance the mixing process. It should be noted that no stirring of the melt was applied in the present experiments. Thus, the predicted diffusion zone thickness is far from the real case of FeNb alloying. Moreover, the limited data and large variations in the thicknesses of the diffusion zones at different contact times, the practical diffusion process should be investigated more systematical in a future study.

Based on the considerations and observations with respect to the microstructures of the diffusion zones, the proposed development mechanism of the dissolution process is summarized as follows: (1) Due to the significant temperature difference between liquid Fe (1600°C) and solid FeNb alloy (25°C), a solid Fe shell was formed at the interface with the colder alloy. It should be noted that although the alloy was quickly introduced into the furnace, it was preheated to some extent before coming into contact with the liquid Fe. (2) The interdiffusion of Fe and Nb started between solid alloy and Fe shell. As time progressed, the thin solid Fe shell melted before the FeNb alloy due to the higher melting point of FeNb (1500–1550°C) compared to Fe. The alloy lump came into contact with the melt directly and started to dissolve faster. The interactions between solid FeNb and liquid Fe were intensified as the Fe and Nb transports were enhanced. (3) As a result of interdiffusion, a diffusion zone consisting of a continuously reduced Nb content from the alloy to the bulk Fe was formed. Then, several regions with different Fe–Nb phases were formed during cooling depending on the local Fe, Nb contents and temperatures. At the same time, the inclusion formations and transformations took place, which are discussed later.

3.3. Inclusions in the Diffusion Zone

3.3.1. Inclusion Evolution with Time

The interfacial reactions in the diffusion zone involve mainly liquid iron, oxygen, niobium and the impurities from the FeNb source. During the initial stage of interactions, it can be observed that inclusions from the FeNb alloy changed with time. The typical inclusions observed in the diffusion zones are listed in Fig. 8. Overall, six types of inclusions were obtained. These were heterogeneous Nb–Ti–O inclusions with a Ti–O core covered by Nb–Ti–O outside layer (type I), homogeneous Nb–Ti–O (type II), Ti–Nb–Al–O inclusions containing Al–O center, Ti–O middle layer and Nb–Ti–O outside layer (type III), Ti–Nb–Al–O inclusions with an Al–O core and an Nb–Ti–O layer (type IV) as well as pure Al–O (type V) and Nb–O (type VI) inclusions. To better understand the inclusion transformations, the relationships between the frequencies, composition change of different types of inclusions and the contact time are shown in Fig. 9.

Fig. 8.

Typical inclusions found in the diffusion zones.

Fig. 9.

(a) The frequencies and (b), (c) composition changes of different types of inclusions versus the contact time.

No inclusions were found in the diffusion zone of sample S1 (5 s), which might be attributed to the short contact time and small diffusion zone. As the contact time increased to 10 s in sample S2, pure Ti–O inclusions from the FeNb alloy started to transform and resulted in the formation of Nb–Ti–O (type I) inclusions. The elemental mappings of type I inclusions are shown in Fig. 11(a). The Ti concentration is much higher in the center than that in the outer layer, Nb only exists in the outer layer and O distributes almost uniformly. Thus, it is concluded that an Nb–Ti–O outer layer forms. As shown in Fig. 9(a), the frequency of type I inclusions decreases significantly with the contact time. This is because some heterogeneous inclusions transform into homogeneous Nb–Ti–O (type II) inclusions. Specifically, the frequency of type II inclusions increases from 45% to 57% when the contact time increases from 10 s to 30 s. Moreover, it can be assumed that all Ti–O inclusions can transform into homogeneous Nb–Ti–O inclusions if the time is long enough.

Fig. 11.

Mapping results of different types of inclusions (a) type I, (b) type III, (c) type IV, and (d) type V. (Online version in color.)

Complex Al–Ti–O inclusions from the alloy experienced a similar transformation procedure, which resulted in the formation of Ti–Nb–Al–O inclusions (type III). It can be seen from the mapping results in Fig. 11(b) that the Al–O inclusions remain in their original form. Furthermore, a similar Nb–Ti–O layer occurs outside of the second Ti–O layer. As the contact time continued to increase, the depth of the Nb–Ti–O outer layer increased. In some cases, the layer of Ti–O disappeared and resulted in the Al–O core being surrounded by Nb–Ti–O inclusions (type IV). Their elemental mappings are shown in Fig. 11(c). In addition, their frequency increases from 8% for 10 s to 27% for 30 s due to these transformations.

In terms of composition changes, they obviously occur in the layer containing Nb and Ti, based on the observed results. As shown in Figs. 9(b) and 9(c), the average Nb content in inclusions evidently increases with an increased contact time except for type I inclusions. However, the average Ti content shows an obvious decrease with time. From another point of view, the increase of the Nb/Ti ratio is more pronounced for type II and IV inclusions and much less for type I inclusions. The detailed inclusion compositions are compared for type I and II inclusions, as depicted in Fig. 10. It is known that the metal matrix can influence the determination of the inclusion composition, especially for small-sized inclusions.²⁰⁾ It can be seen that the Fe content significantly decreases with an increased inclusion width. The Fe contents in inclusions are less than 5% when the width of inclusions larger than 4 μm. The effect of the metal matrix can be ignored in the determination of inclusion compositions, which has been reported before.²¹⁾ Furthermore, the atomic ratios of Nb/Ti in inclusions are higher in type II inclusions (1–1.5) compared to the out layer in type I inclusions (0.6–0.9). The possible reasons for these composition changes are discussed in detail later. Besides, the TiNb_(1–1.5)O_x solid solution can be considered as the equilibrium product of homogeneous Nb–Ti–O inclusions.

Fig. 10.

Relationship between the Fe contents and atomic ratio of Nb/Ti in inclusions and the width of inclusions.

Pure Al–O inclusions (type V) remain unchanged and the percentages of them do not show a clear tendency with time. Also, Nb–O inclusions (type VI) were observed in all the four samples. Due to the high local concentrations of Nb, it reacted with O in liquid Fe to form Nb–O inclusions. The elemental mapping results of Nb–O inclusions are shown in Fig. 11(d). These inclusions were found not only in the diffusion zones but also in the bulk Fe. Another possible reason might be that they were formed during the melt solidification process.

Figure 12 shows the measured inclusion length for various contact times as a function of the distance from the interface between the diffusion zone and the alloy. As can be seen, heterogeneous Nb–Ti–O (type I) inclusions have significantly larger lengths compared to homogenous Nb–Ti–O (type II) inclusions. It takes a longer time for large-sized inclusions to reach a complete transformation. This, in turn, can explain the existence of small size homogeneous inclusions. Moreover, the majority of the inclusions were located inside region III in all the three samples, in which higher Nb contents (11–37%) were detected. It should be pointed out that there is no obvious difference for the inclusion characteristics in different regions even though there exist phases containing different Nb content change with the distance from the alloy matrix. One interesting point is that some homogenous Nb–Ti–O (type II) inclusions appear in some relatively concentrated areas. This might be related to the uneven distribution of Ti–O inclusions in the alloy. In addition, they are also likely to precipitate directly during solidification of the melt at the local areas having a larger concentration of Nb, Ti and O. In sample S4, the lengths of inclusions are significantly smaller than those in the other samples. The uneven distribution of inclusions and the random cutting surface might be the reason for their small sizes. As in these experiments, only a limited contact surface was available for observations, an area of alloy with relatively small-sized inclusions could have existed in sample S4.

Fig. 12.

Location of different types of inclusions in samples for different contact times. (Online version in color.)

3.3.2. Mechanism of the Inclusion Transformation

The possibilities and the thermodynamic conditions favourable to the formation of inclusions in the diffusion zones are discussed below. It is apparent that the Nb concentrations in the diffusion zone are significantly larger than those of Ti and Al. Therefore, during the initial stage the inclusions from the alloy piece transformed in this zone depending on the concentrations of Nb and temperatures. The formed Nb–Ti–O layer outside of the Ti–O inclusions was due to the reduction of Ti–O by Nb. However, pure Al₂O₃ inclusions remain unchanged, which indicates that they can not be reduced by Nb. Yoshihara et al.²²⁾ reported that when the Nb content exceeds the solubility limit in TiO₂, TiNb₂O₇ was formed. In addition, Nb was reported to substitute for Ti in TiO₂ as a cation with a valence of 5, while no Nb was present in Al₂O₃ in the oxidation behaviour of TiAl alloys.²³⁾ Besides, Al has been reported to easily reduce Nb from Nb₂O₅ to form niobium aluminides at a molar ratio of Nb₂O₅/Al=1.5 and 0.11 in the Nb₂O₅–Al system.²⁴⁾ Therefore, the present observations for the transformation of Ti–O to Nb–Ti–O inclusions can be explained from this point of view.

The reduction process is explained based on the thermodynamic calculations using the FactSage 7.1 thermodynamic software with databases of FactPS, FToxid and FTmisc. The simulation of a reduction of Ti–O and Al–O inclusions was carried out for a 100 g of iron (Fe) containing varying concentrations of Nb (0.1 to 10 wt%) and 0.5 wt% Ti₂O₃, 0.4 wt% Al₂O₃ at 1600°C. The initial amount of inclusions was selected based on the assumption that all Al and Ti contents (Table 1) are in the form of their oxides. The calculation results are shown in Fig. 13, it can be seen that the NbO_x and dissolved Ti concentrations increase and the TiO₂ decreases with an increased Nb content. However, Al₂O₃ keeps almost constant. Thus, these results can explain why Nb reduces Ti–O inclusions but not Al–O inclusions.

Fig. 13.

Dependence of the stability of TiO₂ and Al₂O₃ on the Nb content at 1600°C.

In addition, a similar reduction process was also found in the complex Al–Ti–O inclusions. Figure 14 shows a schematic diagram for the modification of an Al–Ti–O inclusion into an Al–Ti–Nb–O inclusion; it can be divided into the following steps: (1) When the Nb starts to diffuse into the complex inclusion/Fe interface, Ti–O in the inclusions is reduced by Nb to form an Nb–Ti–O outside layer. (2) With the diffusion of dissolved Nb, the NbO_x content in the Nb–Ti–O outside layer increase and this layer becomes thicker. Meanwhile, content gradients of NbO_x and TiO_y are formed in this layer. Then, the TiO_y content reduces and the NbO_x content increase along the radial direction within the layer. This can explain the increased NbO_x content in the outside layer with increasing time, as shown in Fig. 9(b). (3) With a further reduction, the Ti–O layer starts to transform into a Ti–Nb–O layer, where the TiO_y concentration is higher than that of the outside Nb–Ti–O layer. Finally, the Ti–O layer fully transforms into Ti–Nb–O and only an Al–O core remains.

Fig. 14.

Schematic of the formation mechanism of Al–Ti–O inclusions.

It should be noted that the determined core composition contained some Nb and Ti contents. One possible reason is that they are affected by the outside Nb–Ti–O layer due to the small core size (<4 μm). Moreover, the shapes and sizes of the Al–O cores seem changed compared to those in type III inclusions. Another possible reason is that there might be some diffusions or reactions happened between the TiNb_(1–1.5)O_x solid solution and the Al–O core. Therefore, a detailed future study is needed to verify this possibility.

According to the experimental results and thermodynamic calculations, the evolution mechanisms of the inclusions in the diffusion zones are schematically shown in Fig. 15. As for pure Al–O inclusions, they do not change during this short contact time. While for complex Al–Ti–O inclusions, the Ti–O layer is firstly reduced by Nb to form an Nb–Ti–O layer. As the reduction continues, the pure Ti–O layer disappears and fully transforms into Ti–Nb–O. This results in inclusions with an Al–O core surrounded by an Nb–Ti–O layer. In terms of pure Ti–O inclusions, a reduction layer of Nb–Ti–O first appears and their thickness increases with time. Then, the Ti–O layer transforms into Ti–Nb–O finally changes to homogeneous Nb–Ti–O inclusions. In addition, small size homogeneous Ti–Nb–O inclusions can precipitate during the solidification due to the dissolved Nb, Ti and O in the diffusion zone.

Fig. 15.

Schematic illustration of the evolution mechanism of different types of inclusions.

To summarize, it is important to understand if these inclusions are removed or not when FeNb alloys are added during the final stage in the ladle refining process just before continuous casting. This study is more focused on the possible evolutions of these inclusions during the early stage of FeNb dissolution process, and their behaviours in the melt after the additions of FeNb alloys will be discussed in a separate paper.

4. Conclusions

The characteristics of inclusions in commercial FeNb alloys were investigated. The interfacial reactions between solid FeNb and liquid Fe and the mechanism of the early dissolution process of FeNb were investigated. The inclusion formation and transformation in the diffusion zone were also studied. The following specific conclusions were obtained:

(1) Four types of inclusions were observed in FeNb alloys using the EE method, namely Al–O (8–46 μm), Ti–O (5–69 μm), Al–Ti–O (12–118 μm) and Si–Al–Mg–O (9–22 μm) inclusions. The proportion of Al–O, Ti–O and Al–Ti–O inclusions accounted for 87% of the total number of inclusions.

(2) The early dissolution mechanism of the FeNb alloy in liquid Fe was proposed, which was controlled by diffusion. Three main regions were identified in the diffusion zone based on their microstructure and composition, which was attributed to the interdiffusion of liquid Fe and solid FeNb alloy. Each of them contained two different Fe–Nb phases, i.e. Fe-55% Nb and Fe-13% Nb (at average) in region II, Fe-35% Nb and Fe-13% Nb in region III and Fe-13% Nb and Fe-3% Nb in region IV. Moreover, a shift in the overall composition towarded lower Nb concentration from the FeNb alloy to the Fe matrix was observed. The thickness of the diffusion zone increased with the contact time and the growth rate of the thickness was much higher at the beginning of the contact.

(3) The Ti–O inclusions transformed to heterogeneous inclusions with a Ti–O core covered by an Nb–Ti–O outside layer. Thereafter, they changed to homogeneous Ti–Nb–O inclusions, due to a reduction which was caused by a high Nb concentration surrounding the inclusions. The Ti–O in the Ti–Al–O inclusions experienced the same transformation way and finally formed the inclusions with the Al–O core surrounded by an Nb–Ti–O outside layer. However, pure Al–O inclusions remained their original form without changes. The addition of FeNb alloys in steel certainly introduces Al–O and Al–Ti–Nb–O inclusions.

Acknowledgements

Yong Wang acknowledges the financial support from the China Scholarship Council (CSC). Joo Hyun Park, acknowledges the LG Yonam Foundation, Korea, for his staying and collaborative research in KTH Royal Institute of Technology, Sweden. Also, the authors would like to thank Jun Yin, Hongying Du and Wangzhong Mu for their kind help during the experiment.

References

• 1)   B. W. A.  Den: Ph.D. thesis, McMaster University, (2013), http://hdl.handle.net/11375/13268, (accessed 2013-10).
• 2)   S. A.  Argyropoulos and  P. G.  Sismanis: Metall. Trans. B, 22 (1991), 417.
• 3)   S.  Liu,  V. S. A.  Challa,  V. V.  Natarajan,  R. D. K.  Misra,  D. M.  Sidorenko,  M. D.  Mulholland,  M.  Manohar and  J. E.  Hartmann: Mater. Sci. Eng. A, 683 (2017), 70.
• 4)   Y. E.  Lee,  H.  Berg and  B.  Jensen: Ironmaking Steelmaking, 22 (1995), 486.
• 5)   J.  Leitner,  C.  Hauler,  F.  Oeters,  L.  Zhang and  S.  Eriksson: Stahl Eisen, 120 (2000), 69 (in German).
• 6)   L.  Pandelaers,  F.  Verhaeghe,  B.  Blanpain,  P.  Wollants and  P.  Gardin: Metall. Mater. Trans. B, 40 (2009), 676.
• 7)   H. E.  Kellner,  A. V.  Karasev,  A.  Memarpour and  P. G.  Jönsson: Steel Res. Int., 87 (2016), 1461.
• 8)   M.-A.  Van Ende,  M.  Guo,  J.  Proost,  B.  Blanpain and  P.  Wollants: ISIJ Int., 50 (2010), 1552.
• 9)   L.  Pandelaers,  D.  Barrier,  P.  Gardin,  P.  Wollants and  B.  Blanpain: Metall. Mater. Trans. B, 44 (2013), 561.
• 10)   M. M.  Pande,  M.  Guo and  B.  Blanpain: ISIJ Int., 53 (2013), 629.
• 11)   P. C.  Yan,  L.  Pandelaers,  L. C.  Zheng,  B.  Blanpain and  M. X.  Guo: ISIJ Int., 55 (2015), 1661.
• 12)   Y.  Bi,  A.  Karasev and  P. G.  Jönsson: Steel Res. Int., 85 (2014), 659.
• 13)   M.-A.  Van Ende,  M.  Guo,  R.  Dekkers,  M.  Burty,  J.  Van Dyck,  P. T.  Jones,  B.  Blanpain and  P.  Wollants: ISIJ Int., 49 (2009), 1133.
• 14)   L.  Zhang and  B. G.  Thomas: ISIJ Int., 43 (2003), 271.
• 15)   D. C.  Park,  I. H.  Jung,  P. C. H.  Rhee and  H. G.  Lee: ISIJ Int., 44 (2004), 1669.
• 16)   B.  Bhardwaj: The Complete Book on Ferroalloys, NIIR Project Consultancy Services, Delhi, (2014), 387.
• 17)   I.  Jimbo and  A. W.  Cramb: Metall. Trans. B, 24 (1993), 5.
• 18)   S. J.  Shah,  H.  Henein and  D. G.  Ivey: Mater. Charact., 78 (2013), 96.
• 19)   S.  Voß,  M.  Palm,  F.  Stein and  D.  Raabe: J. Phase Equilib. Diffus., 32 (2011), 97.
• 20)   Y.  Wang,  A.  Karasev and  P. G.  Jönsson: Steel Res. Int., 91 (2020), 1900669.
• 21)   H.  Doostmohammadi,  A.  Karasev and  P. G.  Jönsson: Steel Res. Int., 81 (2010), 398.
• 22)   M.  Yoshihara and  K.  Miura: Intermetallics, 3 (1995), 357.
• 23)   H.  Jiang,  M.  Hirohasi,  Y.  Lu and  H.  Imanari: Scr. Mater., 46 (2002), 639.
• 24)   A.  Mansurova,  V.  Chumarev,  L.  Leont’ev,  R.  Gulyaeva and  N.  Sel’menskikh: Russ. Metall. (Met.), 2012 (2012), 929.