Characterising microstructures

Materials World magazine
1 Oct 2017

Many materials undergo phase transformations during cooling. Understanding microstructures, therefore, is the route to predictability, as Neil Hollyhoke*, Dr Carl Slater and professors Claire Davis and Sridhar Seetharaman explain.

Characterisation of microstructures is important in metallic materials because of the impact on mechanical properties. A good understanding of microstructural development during processing will determine a product’s properties, but also optimise the processing route – and steel chemistry – to modify properties in a controlled and predictable manner. 

Many materials, particularly steels, undergo phase transformations during cooling, meaning that room temperature observations of microstructures are poor indicators of the microstructure of the material developed during casting and at the start of subsequent high-temperature deformation processing. Greater knowledge, via in situ observation, of the solidification structure will provide better understanding of the relationship between chemistry and processing, and help develop ideal processing routes with less need for trial and error.

The interaction of thermal and compositional gradients in solidifying mixtures leads to a variety of potential microstructures. For cast metallic alloys, the most common is a columnar dendritic front. As the mixture solidifies, different elements are incorporated into the crystal lattice, or are rejected and remain in the liquid, with a build-up of rejected alloying atoms surrounding the previously solidified material. This variation in composition changes the solidification range of the liquid, making a planar surface unstable. Any small perturbation grows more than the surroundings and becomes exaggerated. This positive feedback leads to numerous cellular structures. In metallic materials, these are typically narrow and long enough that there is space for the same mechanism to cause secondary branching from the central trunks, with the possibility of tertiary and higher order branches also developing. Once this branching begins, it is known as a dendritic structure. 

The chemical variation due to rejected solute can occur on two scales. Macrosegregation is large scale, giving rise to different compositions, and therefore properties, between the surface and centre of a cast. As this can occur over 10–100mm, it cannot be practically removed by heat treatment. Microsegregation refers to differences in composition between the central trunks of the dendritic structure and pockets of late solidifying liquid, which become trapped between the branches. 

Microstructure analysis – the old way

Microsegregation can be addressed by a homogenisation heat treatment – exposure to high temperatures increases atomic mobility enough for short-range diffusion to reduce the composition gradients within a practical timeframe. Rolling causes deformation that reduces the distance between solute depleted and enriched regions, further improving the effectiveness of homogenisation. However, homogenisation is expensive, in time and energy, while the increased atomic mobility can allow other potentially undesirable effects to occur, such as precipitation, oxidation and grain growth. Knowing the scale of the dendritic structure allows predictions to be made of suitable temperature and time for any homogenisation treatment and the rolling deformation on the final composition segregation. However, it is common, especially during the processing of steels, for grain growth, re-crystallisation and phase transitions to change the microstructure in such a way that the initial dendritic structure is obscured.

In some alloys, the measurement of secondary dendritic arm spacing (SDAS) can be trivial – such as in many aluminium alloys. Obtaining this information in other systems with less contrast between dendrite trunk and interdendritic compositions, or post solidification phase transformations, relies on more complex etching or elemental mapping. This hinders understanding of the solidification of many materials and causes their high temperature processing to be based on empirical relationships and assumptions. Because of the difficulty observing microstructures during solidification, the only in situ experiments that are commonly reported in metallography textbooks observe solidification in transparent organic materials as a point of comparison. While they have served well in obtaining a qualitative understanding of dendritic solidification, there are bound to be differences between solidification of molecular materials at low temperatures and solidification of atomic materials at high temperatures. 

Other work using X-ray tomography has observed metallic solidification in real time. This method has the advantage of providing details of the full 3D structure, but also requires equipment that may be hard to access and very specific alloy compositions to achieve sufficient X-ray contrast. The technique presented here allows for in situ observation of SDAS during solidification, even for opaque materials, at high temperatures. This confocal laser scanning microscope (CLSM) based technique can be used for many metallic alloys, although only surface observations can be made. The information obtained will give a better understanding of the solidification of metallic materials, as well as improving knowledge of the limitations of previous models, both physical models using alternative materials and computer modelling.

Introducing the laser scanning microscope

The confocal laser scanning microscope is a flexible tool, first pioneered by Emi and co-workers at Tohoku University, Japan. Its use in steel solidification has been growing over the last decade and has since been used in a variety of applications, ranging from imaging direct-reduced iron melting to oxidation at Carnegie Mellon University, USA, by Seetharaman and co-workers, and at Wollongong University, Australia, by Professor Rinn and co-workers, as well as inclusion agglomeration, peritectic transformation and solid/liquid interface velocities. However, its use for more universal problems is becoming evident, in particular the ability to measure the primary and secondary dendrite arm spacing in materials, for which data is otherwise difficult to obtain.

The basic premise is that a laser is raster scanned across the sample and, after being reflected, passes through a pinhole at the focal point before being detected, obscuring out-of-focus planes. Scanning at multiple depths, and combining the images can achieve a wide depth of field. This can also provide detailed topographic data of a static surface. The use of a short wavelength (≈ 405nm) laser, allows observation of high temperature surfaces – molten steel in this case – without black body radiation obscuring the image. This second application is used for in situ observation of solidification phenomena, providing real-time data that can be used to analyse growth rates and for observation of the solidification structure before it is obscured by subsequent phase changes.

For in situ observation of high Al and 304 stainless steel solidification, 5mm cubes of several distinct steel grades were placed in alumina crucibles, which were in turn placed in a vacuum chamber below the CLSM. The chamber was then evacuated and filled with high-purity argon (N5 scrubbed to < 2ppm). The samples were melted by the concentrated radiative heat from a halogen bulb, and the surface recorded during solidification. Samples were then sectioned, polished and etched to provide a baseline result using conventional techniques in order to compare surface and bulk observations.

Since the equipment currently does not allow for a high degree of quenching, samples of the size used can be cooled at up to 40K/s. In the work presented here, cooling rates of 0.1-25K/s were used. This fairly broad range of cooling rates effectively covers the rates in the steel industry, from thick slab or ingot casting to thin slab and belt casting. 

Nucleation occurs most readily on the crucible surfaces, and thermal modelling determined that the thermal gradient that develops in the sample is parallel to the free surface. This means that dendrites will usually grow from points in contact with the crucible, parallel to the surface, either with the main trunk visible or – more often – the secondary dendritic arms visible as they grow up from a sub surface primary dendrite trunk.

By using steel grades that maintain a single phase through solidification and down to room temperature, it has been possible to compare in situ measurements of SDAS on the surface during solidification to bulk measurements made on the same samples using the more conventional sectioning, polishing and etching approach. While the small observable area in the current setup makes achieving high statistical significance more time consuming than conventional techniques, it allows researchers to obtain results in situations where conventional techniques fail, whether due to insufficient contrast when etching or phase changes obscuring the cast structure. 

Opening doors 

This work has used fully ferritic low-density steel and fully austenitic 304 stainless steel. Both have shown positive results in SDAS between in situ surface measurements and results obtained by conventional techniques. Therefore, it seems likely that results for a broad selection of steel grades will be equally reliable – an image of an X70 pipeline steel is also included to demonstrate the potential to obtain data for grades that more mainstream methods struggle with. This can validate results obtained through conventional techniques, or confirm theoretical behaviour when such techniques fail to produce results.

It was noted that because this method looks at the last solidifying material, it is possible that macrosegregation will occur and cause a difference between the bulk and surface measurements, or that surface tension will also affect results. However, neither seems to have any significant impact, and the surface results fall within experimental error of the bulk results. 

The equipment used is limited to cooling rates of < ≈40K/s, although further adjustments could increase this range and there is no reason why the CLSM must be paired with this particular heating mechanism. Therefore, other setups can be considered, potentially for cooling rates seen in newer techniques, such as twin roll casting, or an increase in the observable area.

Overall, the CLSM allows for in situ observations of solidification structures and quantitative measurements of parameters that are essential for the development of modelling capability. Many other applications are also possible, including observations of slag-molten metal interactions, transformations, and interface mobility. 

The project has shown that surface measurements of dendritic structures correlate well with bulk values. This opens doors for future analyses of steel solidification across a broad range of grades and, more broadly, expands understanding of the solidification phenomena.

*Neil Hollyhoke is a final year PhD student in the Warwick Manufacturing Group at the University of Warwick, UK. He works closely with Dr Carl Slater, and both him and Slater work with professors Claire Davis and Sridhar Seetharaman.