Skip to main content

The structure of a hierarchically phase-separated metallic glass


Diffraction tomography images obtained at beamline ID11 show how the mesoscale structure is controlled by the cooling rate in a hierarchically phase-separated metallic glass with a small volume fraction of crystalline inclusions.

  • Share

The structure of matter at the nanoscale, in particular that of amorphous metallic alloys, is of vital importance for functionalisation. Local structures emerge in these materials when the different components separate from each other, much like oil and water. High-energy synchrotron radiation offers unique ways to see inside these glassy metals, especially when using new diffraction tomography methods such as X-ray diffraction nano computed tomography (XRD-nCT). The presence of small crystallites can generate artefacts in the reconstructions of smooth signals from different amorphous alloy components. Despite these challenges, an amorphous metallic alloy with meso-structure was imaged with sub-micron resolution. By reconstructing virtual X-ray diffraction patterns from different points in the sample, clean diffraction data was obtained, making it possible to acquire very localised structural information buried inside the material’s cross-section.

The sample studied was a hierarchically phase-separated Gd27.5Hf27.5Co25Al20 metallic glass ribbon, which was produced by ultra-fast cooling and contained a small volume fraction of (nano)crystals. The extremely high cooling rates are obtained by pouring a liquid metal onto the surface of a spinning Cu wheel, where the material solidifies as a ribbon with a thickness of about 40 microns. Across the thickness of the ribbon, there is a gradient in cooling rate from the front to the back of the ribbon. A cross-section of a small fragment of the ribbon was installed on the nanoscope station at beamline ID11, where XRD-nCT data were collected with a sub-micron focused X-ray beam.

Sharp Bragg reflections can introduce undesired artefacts for the tomographic reconstruction and so these were removed from the primary 2D diffraction images and analysed separately. This gave a cleaner dataset for the amorphous phases and also helped to identify the crystalline phases. Figure 1 shows a 2D XRD pattern as recorded by the FReLoN CCD detector. Wide rings characteristic of the amorphous nature of the Gd27.5Hf27.5Co25Al20 metallic glass ribbon are clearly visible. After azimuthal integration, a splitting of the first diffuse ring is clearly seen (1D XRD pattern in Figure 1), as well as tiny bright spots from the (nano)crystals. An automatic masking and filtering procedure was worked out to process the large volume of data.


Fig. 1: 2D XRD pattern and resulting 1D XRD patterns with and without masking and filtering. The figure also shows the subtracted XRD pattern (i.e., the Bragg peaks from the crystalline inclusions).

The cleaned, integrated 1D XRD data were arranged to obtain a sinogram in order to compute the tomographic reconstruction via an inverse Radon transform. This global sinogram is a 3D volume of data for the  NY  = 200 translations,  Nω = 180 rotations and 3350 bins in Q (0 – 108 nm-1) corresponding to the diffraction patterns. Reconstructions can be produced to give an image of the sample by integrating over any region of interest (ROI) in Q in order to extract a specific scattering contribution. For crystalline samples, the Bragg peaks are narrow and well separated from the background, and they can be separated easily for reconstruction. In an amorphous sample, the diffuse signals are continuous and generated by the disordered atomic arrangements, and are proportional to the probability of finding certain atoms at a certain position. This information is averaged over the entire Q space and it is not possible to make a unique separation for the contributions of each phase when their local compositions and structures are also unknown. Nevertheless, by careful selection of some Q values, it was possible to acquire images dominated by the different phases, as shown in Figure 2.


Fig. 2: Two regions of interest (ROIs), corresponding to a given scattering contribution and used to extract the relevant sinograms. These two regions correspond to a Gd-rich phase (lower Q value) and a Co,Hf-rich phase (higher Q value). The pictures show both the sinograms (global, around the Gd-rich phase, and around the Co,Hf-rich phase) and the corresponding reconstructed cross-sections.

The diffraction signals from voxels buried inside the sample can be recovered by reconstructing the sinogram for each of the Q values. Figure 3 shows two XRD patterns taken at the points marked by arrows A and B. The resulting XRD patterns illustrate that point A comprises the Gd-rich phase and point B the Co,Hf-rich phase. However, both patterns still reveal a small shoulder superimposed on the main broad peak at Q values characteristic of the complementary phase. This is to be expected, because in the reconstructed image there is a mixed contrast of both A and B zones.


Fig. 3: XRD patterns reconstructed for two points, A and B, as marked in the inset. The inset is a cross-section of the sample reconstructed by using all measured diffraction patterns (i.e., using the global sinogram). By deploying a suitable computer program, one can generate the XRD pattern in every point of the reconstructed cross-section by simply inserting the corresponding x and y coordinates.

The images obtained for this sample clearly demonstrate the viability of the method for imaging mesoscopic amorphous phase-separated structures. It is possible to see that the characteristic length scales for the spatial segregation are larger where the sample was cooled slower and that the crystalline inclusions were also found farther away from the more rapidly quenched surface. With the recent upgrade of the ESRF to the Extremely Brilliant Source, and the use of the new Eiger2 CdTe detector, it is expected that future experiments of this kind can be carried out far more quickly. This will enable measurements during relaxation, annealing, and mechanical deformation.


Principal publication and authors
X-Ray Diffraction Computed Nanotomography Applied to Solve the Structure of Hierarchically Phase-Separated Metallic Glass, M. Stoica (a), B. Sarac (b), F. Spieckermann (c), J. Wright (d), C. Gammer (b), J. Han (e), P.F. Gostin (f), J. Eckert (b,c), J.F. Löffler (a), ACS Nano (2021);
(a) Laboratory of Metal Physics and Technology, Department of Materials, ETH Zurich (Switzerland)
(b) Erich Schmid Institute of Materials Science, Austrian Academy of Sciences (ÖAW), Leoben (Austria)
(c) Chair of Materials Physics, Department of Materials Science, Montanuniversität Leoben, Leoben (Austria)
(d) ESRF
(e) Korea Institute for Rare Metals (KIRAM), Korea Institute of Industrial Technology (KITECH), Yeonsu-Gu, Incheon (South Korea)
(f) School of Metallurgy and Materials, University of Birmingham, Edgbaston (UK)