Figure 1. Examples of complex Non-Metallic Inclusions observed in liquid steel during refining of a Ti-alloyed Al killed steel grade on a Ladle Furnace. SEM BSE images, of inclusions of 5-10 µm size.

 

 

An illustration or relics is depicted in Figure 1 above. In this case, the focus lies on the detailed reactions that take place during the processing of liquid steel on a ladle furnace before it is cast. These reactions involve the adjustment of the chemical composition of the steel in order to achieve the desired steel alloy grade during the refining process. The alloying of liquid steel with titanium is commonly achieved through the addition of required amounts of raw materials, such as ferrotitanium, to the steel melt at this stage. Such additions generally are cold and must first heat up in order to dissolve and effectively mix within the liquid steel bath. Also, these additions are not pure and often contain significant impurities. These impurities undergo reactions on their own, depending on the specific compositions present in their surrounding environment. The process of steel alloying is usually impossible to sample directly at its site of occurrence. Therefore, it is commonly simulated using chemical-physical modelling techniques, as outlined in the literature cited in Figure 2. Here, the formation of relics in the actual process allows the direct observation of process characteristics, even in a location that is otherwise inaccessible. In this case shown in Fig. 1, the alloy metal added, ferrotitanium, contains a significant amount of oxygen, mostly from minor oxide impurities. During heating, melting, and dissipation (mixing) of the alloy addition, the oxygen present reacts with all the metals available in the surrounding liquid. This reaction leads to the formation of oxide phases, which are determined by the transient local oxygen fugacity in the region of dissolving additions, which is high at first before the eventual homogenization.

 

Figure 2. Conceptual model of inclusion formation around raw alloy metals added to the liquid steel during refinement, from the referenced paper, ISIJ Intl. 53:629.

 

 

Such oxide particles in steel are generally termed ‘Non-Metallic Inclusions' (NMI) in steelmaking science and are largely unwanted. Many liquid steels, including the one in this case, have previously been alloyed with aluminium, in a process called 'killing'. As a result, the predominant equilibrium governing metal-oxide reactions in the liquid steel is the Al-Al2O3 equilibrium:

 

                                                                                                         4 Al + 3 O2 = 2 Al2O3

 

                                                                                       K = a^{^2}(Al2O3)^alumina / (a^{^4}(Al,^steel )a^³(O2,^steel))

 

Accordingly, prior to the addition of ferrotitanium, the vast majority of non-metallic inclusions (NMI) in steel are alumina grains, which remain as a residue from the initial deoxidation process. Upon the addition of ferrotitanium, a gradual process of heating, melting, and dissipation of the added alloy metal occurs, accompanied by the release of excess oxygen contained within the ferrotitanium. The dispersion of the addition results in a transient gradient of oxygen activities. Thus, the NMI formed along this transient gradient also exhibit a variable chemical composition. Initially, at the point of oxygen release, with initial enhanced oxygen concentration, there is a higher likelihood of the formation of oxidized phases, which may include phases with substantial quantities of titanium oxide (Ti2O3), due to the presence of an excess of titanium from the dissolving ferrotitanium. If the alloying reaction were to proceed to completion and maintain full equilibrium throughout the entire steel bath, the initially formed titania-containing inclusions would be completely eliminated, as the oxygen would undergo metal exchange according to:

 

                                                                                                        Ti2O3 + Al = Al2O3 + Ti

 

This reaction would proceed towards the right if the entire bulk steel composition were to reach equilibrium at the given very low overall oxygen concentrations. But to achieve this, it is necessary to retain comprehensive chemical communication between all components of the liquid and its suspended solids. Diffusion and microturbulent advection in the steel liquid bath ensure fairly consistent homogenization. However, the diffusion of oxygen in alumina (Al2O3) grains is very slow, even at high temperatures. The effective length scales for oxygen diffusion in alumina are found to be in the range of micrometres per hour, even at elevated temperatures of 1600°C. Accordingly, the formation of even a thin layer of alumina on the surface of an inclusion effectively blocks the transport of oxygen. Thus, the titania-rich NMI that are formed in the transient oxygen fugacity profile become chemically isolated when they form a thin pure Al2O3 shell as a result of equilibration with the larger steel melt. These inclusions can survive for an extended period of time with an unchanged interior, serving as the cores of complex, chemically zoned inclusions. Their existence can be used as evidence of the prevailing conditions during the transient oxygen/metal gradients that occurred while adding metal alloys in the steelmaking process.

 

 

 

The NMI depicted in Figure 1 are illustrative examples of this. It is apparent that these inclusions exhibit complex microstructures within their cores, which include the existence of pores and sub-inclusions composed of various phases. Some of the sub-included phases, such as oxides and nitrides (e.g., TiN), would not be stable in the surrounding steel melt. The portion of the alumina grains that contains these sub-inclusions also shows a distinct brightening in grey-level when observed using backscattered electron imaging (BSE), in contrast to the alumina present at the outer edges of the grain. This slight enhancement in BSE brightness of the inclusion cores is due to the existence of a small amount of Ti2O3 (approximately 1-2 wt %) in solid solution in the alumina. It is shielded  from reduction by the presence of a protective layer of Ti-free pure Al2O3, which forms the thin outer margins of the NMI. Thus, the entire core of these complex-structured NMI is a relic of an earlier process (the melting and dissolution part of the alloying process) and not of the ‘current' steel bath equilibrium at the time of sampling. Conversely, the relic cores also contain sub-inclusions of metal, which appear bright in the BSE image. This metal, while also steel, consists of small fragments of the transient (and very different) steel compositions that surround the dissolving alloy. These fragments are unable to reach a state of equilibrium with the overall steel bath, as they are now encapsulated in the inclusions. The thin layers of pure alumina around the inclusion protects the cores as if it were a kind of armour protecting the insides, and thus, such relics are called ‘armoured relics' in much of microstructural literature. The surrounding liquid steel bath is strictly in equilibrium only with the surfaces of these complex NMI. This means that for overall calculations of steel bath equilibria, such as oxygen content, it is needed to disregard the interior of the inclusions. This is quantitatively insignificant for most metals, such as Al or Ti, but it holds quantitatively significant implications for light elements such as oxygen. The oxygen content of a bulk sample of steel can be dominated by the oxygen content present in encapsulated armoured relics, while the overall steel melt was really only in equilibrium witrh a small fraction of the oxygen. This can lead to an inaccurate depiction of the steel's chemical state in a thermodynamic process analysis if the inclusion microstructures are not correclty taken into account.

 

 

 

Microstructures can also serve as indicators of time progression in the process of material formation, allowing them to preserve a record of evolving conditions or circumstances. In certain scientific disciplines such as archaeology or geosciences, the application of this is enshrined in well-known principles like Steno's Law of Superposition: if something is deposited on top of something else, it is considered to be "younger" (e.g., rock strata). The deciphering of Earth's history was made possible through this understanding. The concept is also applicable to disciplines that do not typically emphasize the notion of deep time, such as materials science. A ‘stratigraphic' analysis of the sequence of formation of a material microstructure can often lead to information about process stages, or process conditions that are otherwise out of reach or are long gone. An example is shown in Figure 3.

Figure 3. SEM BSE images of materials from a converter duct, from BOF (Basic Oxygen Furnace, Converter) steelmaking. Top row: converter dust images, showing condensates from the converter off gas, next to some (much larger) pieces of entrained process materials. Bottom row: Process slag (and process steel/metal melt) droplets, splashed against the wall of the off-gas duct, which is water cooled, so that the splashed liquids are shock frozen (quenched). Note the decoration of splash surfaces by small metal bodies in the bottom right image, enlarged at left.

 

 

This figure (Fig. 3) shows materials obtained from the off-gas duct wall of a basic oxygen furnace, a widely utilized apparatus in the field of general steelmaking. In this installation, the process of converting carbon-rich hot metal into raw steel is achieved through the application of oxygen gas. The contact of oxygen to the iron melt leads to a reaction with dissolved carbon, resulting in the formation of carbon monoxide (CO). This carbon monoxide formation consumes the carbon content from the iron melt, turning it into raw steel. The location where oxygen is introduced to the raw iron melt experiences extremely high temperatures (exceeding 2000 C) and undergoes numerous chemical reactions. However, this specific location cannot be accessed using conventional sampling techniques. The study of these processes frequently depends on the utilization of indirect evidence and forward modelling techniques, given the difficulty in obtaining direct observations.

 

 

 

Figure 3 illustrates the use of materials microstructures to gain knowledge of the inaccessible hot spot. The provided examples show SEM BSE images of dust samples obtained from an off gas washer of such an installation. The dust sample exhibits a diverse range of materials. Some of the splashes and fragments of process materials are relatively large, ranging from tens to hundreds of micrometres in size. However, they are still small enough to be entrained in and transported by the off-gas stream. These samples themselves serve as valuable process materials that warrant further study. However, the majority of the material comprises a fine to very fine assemblage of spherical grains, ranging in size from micrometre to sub-micrometre sizes. This material does not correspond to any of the primary process materials, such as steel and slag liquids or refractories. Instead, upon analysis, it is determined that they are small spherules composed of iron oxide. No parts of the primary process should contain liquid iron oxide (which dissociates at high temperatures). Instead, it is thought that these components are formed by the condensation of metal from the superheated vapor coming off the converter's hot spot. At the high temperatures within the hotspot, even iron exhibits a substantial vapor pressure, resulting in the vaporization of a certain quantity of iron. This vaporized iron then becomes entrained as a gas into the off-gas stream. However, the off gas rapidly cools, and the Fe vapor in the gas quickly condensates again. Not only iron does so, all other metal compounds in the bath (Mn, Si, Al, Ti, …) also have their own specific volatilities and vaporize as well in small amounts. These also precipitate from the gaseous phase as condensates when the temperature decreases. As the volatilities of the metals exhibit varying responses to temperature changes, the ratios of elements present in the off-gas metal vapor could potentially serve as a useful direct thermometer of hot spot conditions. However, the top row images in Figure 3 show samples obtained from a gas washer installation, where any recondensed metal undergoes re-oxidation as it is hot enough still to react with the water it encounters. As a result, the spherules visible in the images are FeOx, indicating that they are no longer the direct metal precipitate, but it reoxidation products. This induces unknown fractionation effects (as not every metal is eqally compatible into FeOx), making the chemical analysis of this dust by itself difficult to interpret. It would be advantageous to obtain direct samples of the recondensed metal, which is formed from the vapor phase, prior to its re-oxidation. The lower set of images depicted in Figure 3 demonstrates the fortunate occurrence of this situation. The gas duct, which leads the hot off gas away, is externally water cooled, resulting in a cold surface exposed to the gas stream. In the sections of the duct that are in close proximity to the reactor vessel, there are larger fragments of process slag (as well as some metal blobs) that are forcefully ejected from the reactor vessel. These fragments sporadically splash onto the cooled surface and get instantly frozen in (quenched). The image located in the lower right corner depicts a sample of such splashed-on material. Close inspection shows that individual slag splashes can be recognized, as they are sequentially built up on top of each other. At the same time, close inspection of the interface between these splashes reveals the presence of a delicate layer composed of small metal particles. This is enlarged in the lower left side of Figure 3. It can be seen that these are very tiny, dendrite like ‘trees' of metal, which by their brightness can be seen to still be metal, not oxide. These particles are formed by direct metal condensation from the off gas stream, which precipitates onto the cold sidewalls continually. In this particular context, the succession of slag splashes means that this metal condensate gets quickly covered under the next slag splash, which instantly freezes as well. Thereby the metal condenstate is ‘buried', and protected. Due to the water cooling of the duct sidewall, the slag splashes rapidly freeze, and become too cold to themselves oxidise the metal particles. Thus, recognizing the sequential buildup of the sidewall splash coating in this specific area, it becomes possible to identify material parts that serve as a direct representation of the process hot spot, which would otherwise be inaccessible. The metal precipitates observed in this study have extremely small sizes, ranging from 2 µm to 200 nm. However, advanced analytical techniques enable the accurate characterization of their composition, which yields process-relevant information for understanding hot spot processes. This example demonstrates how the time sequence of a stratigraphic buildup in a sample can be utilized to benefit the process engineer, given a sufficient comprehension of microstructures.