This explanation – delineating the term microstructure both to large and small length scales – already shows some reason for the study of microstructures: the large scale properties of many materials, including technologically and commercially extremely relevant properties (e.g. hardness, fracture toughness, radiative properties, conduction, and many others) can be directly shaped by the microstructures of the materials, at least as much as by the phase-specific properties of the constituent phases. So, to understand how a material behaves in a given application or under given service conditions generally requires an understanding of its microstructure. To design a material that should show improved properties or performance, likewise requires an understanding of how these depend on the microstructures.

 

Nonetheless, this is by far not the whole reason for the study of microstructures. The spatial arrangement of the subunits of a material arises itself through a certain sequence of processes, and the physical shapes that these subunits have, are likewise the consequence of a history of processes that have acted on the material. This is easy to see for instance in the case of a slowly crystallizing multicomponent liquid: first formed crystals have opportunity to grow freely in solution and can develop their preferred morphological habitus, giving rise to idiomorphic crystals. Later or last formed crystals have to fill in remaining space between already present solids and often have irregular, not habitus oriented shapes. Thus, the microstructure – here, the shapes of various subunits (crystal species) of the material – reflects the sequence of events – the history – of the processes that formed the material. In this way, microstructures of materials contain development information that is contained in no other physical or chemical feature. In many cases of process analysis – forensic analysis of materials, or assessments of functioning or failure modes of industrial processes – this dynamic historical information embedded in a microstructure is of primary importance. Such information can be obtained by no other analysis or measurement, as it is encoded only in the microstructure itself. Yet, proper extraction of this information is not possible without adequate characterization and understanding of the microstructures. To show this in a particularly vivid example, let’s consider Figure 1.

Figure 1. SEM (scanning electron microscope) image of a material, taken in BSE (back-scattered electrons) imaging mode.

 

BSE imaging mode is very frequently employed in materials science. It shows the intensity of backscattered electrons under a given electron illumination as image brightness. This intensity is proportional to the mean atomic Z (atomic number) of a material, thus, its chemical composition. Brighter areas are composed of heavier atoms, darker areas – lighter atoms. Thus, it is not surprising to state that the image shows a segment of a metallic material (in this case, steel, with heavy iron atoms) embedded in a refractory material: the darker areas at the edges are elementally lighter, as is the case for refractories which are to a great part made from oxides in which the light element oxygen is about half of all atoms. The pitch black area at the left side is in fact porosity – a gap in the material, filled with an embedding medium from sample preparation. The entire material in view comes from a process step in liquid steelmaking, in which metal melts are prepared in a series of complex processes turning raw iron into liquid steel, and then cast. The proper understanding of these processes is a major challenge for operations.

 

This said, the image also shows a characteristic and very peculiar microstructure in the metal (the refractory parts are for now not relevant). Neutrally, it could be said that the bright (heavy) metal contains dark, elongate inclusions, which occur in this cross sectional as elongated thin but wavy or curved lines (thinking in three dimensions, they would be flakes, platelet or sheet like shapes). There are also numerous very tiny dark spots scattered. But the casual observer could be forgiven to think that this structure looks like the graphite-flake microstructure of a classical cast iron from hot metal, the near carbon saturated liquid iron that is the product of traditional carbon-based ironmaking and the input for many steelmaking operations. In fact, there is an entire industry that produces such carbon rich cast irons for industrial applications in which many bulk material properties of the cast workpiece depend to a large amount on the precise shape, size, and history of the carbon flakes or other forms of carbon produced in casting and tempering; and therefore there is an overwhelming scientific literature and typology of such graphite flakes and their genesis and engineering consequences. To show how good the first order similarity of the structure in Fig. 1 is to such graphite flakes in cast iron, we show here in Figure 2 a pretty much random case of a graphite flake microstructure taken from literature [Dinwiddle & Wang 1999, J. of Mat. Sci. 34:4775, DOI: 10.123/A:1004643322951]:

Figure 2: Image of a graphite flake microstructure in a cast iron [Dinwiddle & Wang 1999, J. of Mat. Sci. 34:4775, DOI: 10.123/A:1004643322951. The image is a reflected light image of a polished section surface, in which reflective iron appears bright and less reflective carbon (graphite) dark.

Comparison of the images suggests that there is a very good correspondence between the “apparent flake like looking” embeddings in the metal in Fig. 1 and the graphite flakes of a cast iron such as in Figure 2. So is the material in Fig. 1 a cast iron like material, which might be possible given that we said it comes from a steelmaking process? No, but instead of looking at small details at the given magnification, the matter becomes immediately clear if one zooms in a little bit on the apparent flake like bodies in Figure 1. We do this here as Figure 3.

Figure 3: Zoom in to Figure 1 (central part), again a SEM image in BSE mode. Phases distinguished by brightness contrast according to chemistry (mean Z). The flake-like elongated inclusions are shown to internally have multiple phases with different grey levels. The three most important of these are indicated. (Ceramic notation is used: C = CaO, A = Al2O3, F = Fe2O3 and so on).

 

Figure 3 is a simple magnification (zoom in) of a selected part of the flake-like enclosures in the (white) metal matrix. At this magnification, it is immediately apparent that the content of the flake like inclusions is not graphite at all – or any carbon: it is not nearly as dark as that light element phase would be in BSE mode, and moreover the content of the shapes is seen to consist of multiple, clearly separable phases. The phases found in the flake like shapes have intermediate grey levels typical for oxide phases (compare to Fig 1). Most of the tiny round inclusions also can now be made out as being similar to oxides. Care should be taken though, the grey level alone does not say which phase it is. However, once in the SEM, phase identification is easy by traditional point EDS (energy dispersive spectrometry) analysis. This was done in this case, and results in the following identifications - brightest is a magnesiowuestite (W,M; (Fe,Mg..)1-xO); mid grey a calcium ferrite (C2AF; nominally Ca2(Al,Fe)2O5); darkest a dicalcium silicate (C2S; nominally Ca2SiO4). (Note the ceramic notation – C for CaO, S for SiO2 etc, making for a very convenient shorthand for the phases).

 

People familiar with this particular metallurgical industry, steelmaking, will recognize these three phases as the main phases into which one particular slag from a specific processing step solidifies, namely “BOF slag” (“Basic Oxygen Furnace” slag, also called converter slag. A slag is an industry term for a high temperature molten oxide melt. Now, contrasting the shapes of the inclusions in the

Figure 4: Phase diagram of the iron-carbon system, showing the Fe-rich part as a T-X section from pure Fe up to the composition of cementite (6.7 % C). Temperatyures are giben in Fahrenheit at left edge and Celsius at right edge. The main compositional classifications are given at the lower edge. The phase assemblaged formed in different parts of the diagram are indicated as well.

 

observed metal, and their chemical composition, leads to a contradiction. Such a BOF slag is molten in the corresponding process at very high temperatures, in excess of 1300 °C and often up to 1650 °C or even hotter depending on composition. It is thus imaginable that the metal observed here is simply a sample of such a process. This process, in which oxygen ins blown onto liquid metal, is consequently highly oxidizing, which is expressed in the oxide phases crystallizing from the slag present in the process, all of which contain significant amounts of ferric (trivalent) iron. However, if this were a sample of such metal from processing, the shapes of the slag inclusions could not for any length of time have been what is observed in Fig. 1. When both metal and oxide melt are liquid, the shape of any emulsion of these liquids is dictated by free surface energy minimization. That means, they will form globules (spheres, droplets) of one melt in the other melt. Such highly non-spherical shapes as the oxide inclusions in the metal show in Fig. 1 could not exist for any length of time in a two liquid emulsion. How then did the oxides (formerly a homogeneous molten oxide melt) in this piece of metal obtain this form? A form moreover which is so strongly reminiscent of the well known graphite flake in cast iron microstructure? Hot metal (compositionally similar to cast iron) is what the BOF process starts out with, but such a carbon rich metal acquires the graphite flake structure only on cooling below 1150 °C. In the BOF process, the metal is strongly heated to high temperatures above 1500 °C. This contrast can be visualized by reference to the well-known iron carbon phase diagram, a version of which is reproduced here as Fig. 4. It shows the eutectic in the Fe-C system at circa 1150°C: this is where graphite flake microstructures form. But such carbon rich metal is extremely reduced (oxygen poor) and could not coexist with as highly oxidized phases as Fig. 3 shows. In contrast, as the carbon content of an iron melt decreases, its melting temperature (when below ~ 2 wt.% C) increases, through the peritectic point, until it reaches eventually the carbon free iron melting temperature at 1539°C. Such carbon free iron can coexist with the observed oxidic phases, but it could then obviously not form graphite flake microstructures.

 

The answer to this riddle is the recognition that the microstructure shown in the images is a dynamic microstructure (here – a replacement microstructure), that has obtained various parts of its microstructure at different times and has retained some of them while being chemically changed. For this, it helps to finally also refer to the surrounding of the main iron part of Fig. 1, namely, the entire metal piece is embedded in refractory materials, which in the industrial installations form the sidewalls of the metal making processes. These sidewalls are sometimes pierced by small injections of liquid metal from the main process, but they are also cooling places, in which the temperature drops off rapidly from the main process temperature of the reaction chamber. Thus, this piece of metal has started out as a standard hot metal (compositionally, similar to cast iron), of the sort the BOF process is charged with, which found its way into (not just onto) the refractory sidewall of the reactor. Here, it could solidify – not too rapidly, thus forming a standard grafite flake microstructure, and not a cementite (ledeburite) structure as would have happened if it had been rapid quenched. However, once solidified (below 1150 °C) in the wall, more happens to this metal. The sidewall, being pierced by this liquid metal, is also open to infiltration of gases from the reactor inside, in which the regular BOF process commences, which involves blowing of large amounts of oxygen gas. Such oxygen penetrating into the wall will then oxidize the metal that it encounters there in solidified form. That means, it will burn off the carbon from the hot metal. In the main reactor chamber this generates so much heat that everything melts down and any preexisting microstructures would be lost. Here however, the amount of oxygen reaching this place was just restricted enough, to gradually decarburize the metal, without ever remelting it. As the phase diagram in Fig. 4 shows, the material composed of graphite embedded in a C-rich iron phase can follow a temperature-carbon path that decarburizes it even while it is heating up; as long as it does not touch the solidus at any carbon content, it will not remelt. This however means that eventually, when all carbon is lost, the shapes of the former graphite flakes remain. Since the decarburization removes carbon in gaseous form (as CO), no residuals remain, the former flakes now represent physically open slots, a peculiar form of porosity, in the formerly carbon rich metal, that has now turned into solid steel. In this form, the location of this metal can then in the further progress of the reaction be reached by the melts of the reaction vessel, specifically, its molten-oxide (slag) melt, which is standard BOF slag. As the original hot metal liquid was able to penetrate into the wall to this location, so later also can do the slag. This slag meets the now-decarburized iron with its now-open slots of former graphite. These slots can be refilled with the slag melt, without losing their overall shape, as long as the temperature does not exceed 1540 °C (to melt the metal). At temperatures just below melting, steel becomes relatively ductile and prone to sintering, this can be observed in detail when comparing the actual shapes of the slots in Fig. 3 to the shapes of true graphite flakes in Fig. 2 – ubiquitous coarsening and rounding of the slots. Thus eventually, the process reaching its next cycle cooled down again, and at that stage, the largely fully oxidized BOF slag simply solidifies in its “container”, the formerly-graphite slots inside the now fully oxidized metal.

To summarize from the above set of images, the microstructures of the sample are a record of a multistage process.

 

• Stage I – C-Fe Hot Metal comes in as liquid, freezes in place.

 

• Stage II – While cold, the material oxidizes, removing all free C as gas (CO), leaving open slots instead former graphite, and decarburizing the hot metal, turning it into solid steel.

 

• Stage III – The place becomes hot again, hot enough for an FeO rich oxide melt (slag) to infiltrate, and fill up the open ex-graphite shapes, while remaining cold enough (< 1540 C) not to remelt the steel, preserving the shapes.

 

The stages of the microstructural development can be precisely correlated with the stages of this cyclic process. Charging with medium temperature hot metal, blowing of oxygen to remove the carbon, rising temperatures forming oxide rich slag at the end of the process that is mobile and can infiltrate the walls.

 

The point of this analysis is to illustrate the power of the information contained in microstructures. Nothing of the dynamic of the process could be deduced from a simple bulk analysis of the sampled materials. A chemical analysis would simply show the presence of BOF slag and steel. One could not deduce from it the former history as hot metal. The exact same bulk analysis results, whether by XRF, XRD, or chemistry, could also be obtained from a true slag-steel emulsion sample from somewhere in the center of the reactor vessel, which would have microstructures that would yield information about a completely different part of the industrial process under consideration. Dynamic historical information in materials is often preserved near exclusively in its specific microstructures, and neglecting to analyze and extract this information is equivalent to willful blindness on the part of the investigator. The sample does not have just “a composition”. It does not just have “a phase assemblage or phase makeup” that could be used as dataset into a thermodynamics package. It contains a wealth of dynamic information about precisely what happened in the sample to give it the state it now has, encoded in the microstructure of the material – shapes, locations, sizes, contact relationships etc. in correlation with composition and phase makeup. Only analyzing the microstructure of the sample can unlock this information. That is the purpose of microstructural microscopy.