Research Project

3D Microstructural Characterization of Materials

At the very heart of materials science and engineering is the idea that the properties of a material are governed by its structure, and that we can control that structure (and therefore the properties) by how we process the material. So it should be no surprise that new ways to characterize and understand the structure of materials are of great interest.

At the very heart of materials science and engineering is the idea that the properties of a material are governed by its structure, and that we can control that structure (and therefore the properties) by how we process the material. So it should be no surprise that new ways to characterize and understand the structure of materials are of great interest.

 

Traditional techniques for characterizing microstructure fall into two categories. In microscopy, we typically examine a planar section through the material, which is exposed by cutting the sample and polishing the surface. Different techniques, such as optical microscopy, scanning electron microscopy, and electron backscatter diffraction (EBSD) provide different information, but they all are limited to essentially two-dimensional views of the structure. On the other hand, we have techniques like x-ray diffraction (XRD) that can reveal structural information from a volume of material, but average over that volume – they don’t provide any spatial resolution.

 

But many aspects of structure can’t be gleaned from simple 2D sections or volume-average measurements – even such basic things as the true size and shape of grains, or the connectivity of phases in multiphase materials. Furthermore, some important aspects of material behavior – such as fatigue crack initiation – may depend on rare, “contingent” microstructural features that are unlikely to be revealed in a 2D slice taken at random through the material. To understand these aspects of structure we need a true, three-dimensional representation of the material.

 

In our group, we are tackling this problem in two ways. First, we have built an instrument for automated serial sectioning and EBSD analysis of materials. In serial sectioning we image a section through a material, then remove a layer of controlled thickness to reveal what was under the original surface and image that. We can work our way down through the material, taking images of each section, and then reconstruct the images into a 3D representation of the structure, as illustrated here:

 

adapted from S. Ghosh et al., Computer-Aided Design 40, 293 (2008).

 

There are several ways to do serial sectioning, but in our case, we have integrated a pulse femtosecond laser with a scanning electron microscope:

 

 

To image the material in 3D, we raster the laser beam over the sample surface. The extremely short laser pulses vaporize the surface without significantly heating the material underneath. We then tilt the surface towards the EBSD detector, and by scanning the electron beam over the surface we can determine the crystallographic orientation of the material at every point. We then repeat the laser ablation/EBSD process many times to generate a stack of EBSD images, which we then turn into a 3D crystallographic map of the structure. The example at right above shows the grain morphology in a sample of rolled aluminum. Finally, before moving on we should note that we didn’t invent this technique – it was originally developed by Tresa Pollock and McLean Echlin, first at the University of Michigan and later at the University of California – Santa Barbara.

 

The other approach to 3D microstructural characterization we are pursuing is high-energy diffraction microscopy (HEDM). This is a spatially-resolved approach to x-ray diffraction that uses high-energy x-rays (60-120 keV) to penetrate bulk (>1 mm) specimens. There are two basic approaches to HEDM, as illustrated here:

 

Credit for HEDM data from J.C. Schuren et al., Curr Opin Solid State Mater. Sci. 19, 235 (2015) and T.J. Turner et al., Integr. Mater. Manuf. Innov. 5, 235 (2016)

 

In near-field HEDM (nf-HEDM) the detector is placed very close to the sample so that the position of a diffraction spot depends not only on the scattering angle and the crystallographic orientation of the diffracting grain (as it does for all XRD experiments) but also on the location of the grain in the sample. By scanning a narrow x-ray beam across the sample and rotating the sample, we collect many diffraction patterns. From the positions of the diffraction spots on the patterns we (or, to be honest, a clever computer algorithm) can determine the crystallographic orientation of every volume element (voxel) in the sample, from which we can deduce the location, size, shape, and orientation of the grains.

 

The other approach is far-field HEDM (ff-HEDM), in which a high-resolution detector is placed farther from the sample (figure (b) above). This reduces the spatial resolution but improves the angular resolution, so that we can determine the complete elastic strain tensor, averaged over each grain.

 

Together, nf and ff-HEDM provide unparalleled insight into the 3D microstructure of polycrystalline metals, and they have the advantage over the serial sectioning techniques of being non-destructive. This means that after characterizing the structure via HEDM we can then conduct mechanical tests on the same sample or, more ambitiously, perform HEDM on samples while they are being deformed. More generally, the 3D data we collect (via serial sectioning or HEDM) can be used as the basis for computer simulations of material behavior, such as crystal plasticity modeling of mechanical deformation.

 

This work is conducted under the auspices of the Hopkins Extreme Materials Institute [http://hemi.jhu.edu], which is supported (in part) by the Army Research Laboratory through the Center for Materials in Extreme Dynamic Environments. If you’re a prospective graduate student or post-doc and think it would be fun to do these kinds of experiments, by all means, drop an email to Prof. Hufnagel.

 

/**/
Back to top